The biological clock enhancer nobiletin ameliorates steatosis in genetically obese mice by restoring aberrant hepatic circadian rhythm

Abstract

Nonalcoholic fatty liver disease (NAFLD) is associated with disruption of homeostatic lipid metabolism, but underlying processes are poorly understood. One possible mechanism is impairment in hepatic circadian rhythm, which regulates key lipogenic mediators in the liver and whose circadian oscillation is diminished in obesity. Nobiletin enhances biological rhythms by activating RAR-related orphan receptor nuclear receptor, protecting against metabolic syndrome in a clock-dependent manner. The effect of nobiletin in NAFLD is unclear. In this study, we investigate the clock-enhancing effects of nobiletin in genetically obese (db/db) PER2::LUCIFERASE reporter mice with fatty liver. We report microarray expression data suggesting hepatic circadian signaling is impaired in db/db mice with profound hepatic steatosis. Circadian PER2 activity, as assessed by mRNA and luciferase assay, was significantly diminished in liver of db/db PER2::LUCIFERASE reporter mice. Continuous animal monitoring systems and constant dark studies suggest the primary circadian defect in db/db mice lies within peripheral hepatic oscillators and not behavioral rhythms or the master clock. In vitro, nobiletin restored PER2 amplitude in lipid-laden PER2::LUCIFERASE reporter macrophages. In vivo, nobiletin dramatically upregulated core clock gene expression, hepatic PER2 activity, and ameliorated steatosis in db/db PER2::LUCIFERASE reporter mice. Mechanistically, nobiletin reduced serum insulin levels, decreased hepatic Srebp1c, Acaca1, Tnfα, and Fgf21 expression, but did not improve Plin2, Plin5, or Cpt1, suggesting nobiletin attenuates steatosis in db/db mice via downregulation of hepatic lipid accumulation. These data suggest restoring endogenous rhythm with nobiletin resolves steatosis in obesity, proposing that hypothesis that targeting the biological clock may be an attractive therapeutic strategy for NAFLD.

NEW & NOTEWORTHY NAFLD is the most common chronic liver disease, but underlying mechanisms are unclear. We show here that genetically obese (db/db) mice with fatty liver have impaired hepatic circadian rhythm. Hepatic Per2 expression and PER2 reporter activity are diminished in db/db PER2::LUCIFERASE mice. The biological clock-enhancer nobiletin restores hepatic PER2 in db/db PER2::LUCIFERASE mice, resolving steatosis via downregulation of Srebp1c. These studies suggest targeting the circadian clock may be beneficial strategy in NAFLD.

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) is the most common chronic liver disease estimated to affect more than 80 million individuals in the United States (1, 2). Despite its enormous prevalence, no Food and Drug Administration-approved therapy exists specifically for NAFLD, highlighting an urgent need for further investigation into potential disease-causing mechanisms. NAFLD comprises a continuum of disease states from steatosis in its earliest forms to inflammation, fibrosis, and ultimately cirrhosis in end-stage patients (3). As evidence of its diverse pathology, NAFLD is associated with pathological remodeling of a large number of hepatic signaling networks including metabolic, inflammatory, and fibrogenic processes (4–6) and identifying unifying mechanisms among these complex systems has proven elusive.

The circadian clock is a highly conserved mechanism intimately involved in regulation of the liver transcriptome. The mammalian clock is a molecular network of genes and regulatory proteins that function via transcriptional-translational feedback loops to control the rhythmic activity of a large number of physiological processes (7, 8). The core clock consists of activating and repressive components initiated by the formation of a brain and muscle ARNT-like-1 (BMAL1): Circadian locomotor output cycles kaput (CLOCK) heterodimer that binds to E-box DNA motifs to augment expression of clock-responsive genes including notably Period-2 (Per2) and Cryptochrome-1 (Cry1). The protein products of Per2 and Cry1 accumulate in the cell over time forming heterodimer complexes that translocate back to the nucleus and suppress BMAL1:CLOCK activity to function as the primary feedback inhibition loop of the molecular clock. Other regulatory feedback loops exist such as RAR-related orphan receptor (ROR) nuclear receptor RORα, which upregulates Bmal1 and Clock via competitive binding with REV-ERBα/β repressor complexes at ROR response elements. Clock output genes such as D-box binding PAR BZIP transcription factor (Dbp) modulate the effects of the biological clock via characteristic high-amplitude circadian oscillations and can be used to assess intrinsic clock function.

Peripheral oscillators in mammalian cells are synchronized to the central master clock in the suprachiasmatic nucleus (SCN) of the hypothalamus by humoral and neuronal inputs. However, in the liver, core machinery is further linked with key metabolic regulators such as AMP-activated protein kinase (AMPK) (9), peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α) (10), and Sirtuin (SIRT1/6) (11, 12), suggesting an important nutrient-sensing purpose of the hepatic clock. The hierarchal structure of the central and peripheral clocks allows behavioral zeitgebers (ZT) such as daily feeding and light-dark cycles, in conjunction with local nutrient sources, to entrain cell-autonomous clocks in hepatocytes to fine-tune energy utilization in the liver. The molecular clock plays a substantial role in liver physiology with RNA-sequencing studies reporting more than 10% of the hepatic transcriptome cycles in a circadian manner (13, 14).

It is reasonable then to hypothesize that dysfunction in endogenous hepatic circadian rhythm is detrimental to nutrient homeostasis, and indeed, disruption of the core clock is associated with metabolic liver disease. Clock mutant mice demonstrate profound impairment in circadian rhythmicity and develop the full characteristics of metabolic syndrome with accompanying hepatic steatosis (15–17). Specific disruption of hepatic circadian rhythm using liver-specific Bmal1 deletion exacerbates hepatic steatosis, increases hepatic insulin resistance, and impairs hepatic gluconeogenesis in obese mice (18, 19). Conversely, entraining endogenous circadian rhythm using time-restricted feeding prevents steatosis in isocaloric mice fed a high-fat diet (20), suggesting a causal relationship between hepatic circadian rhythm dysfunction and NAFLD.

An area of recent intense study in circadian biology has been pharmacological modulation of core clock machinery. Elegant studies by He et al. (21) using high-throughput chemical screening recently identified nobiletin, a naturally occurring citrus flavone, as a potent enhancer of endogenous circadian rhythm. Nobiletin was shown to increase circadian PER2 oscillation by directly augmenting ROR in obese mice with dampened PER2. Interestingly, nobiletin also decreased liver size and attenuated hepatic steatosis in a Clock-dependent manner (21). However, underlying mechanisms conferring the beneficial effects of nobiletin on circadian rhythm in NAFLD are unclear.

In this study, we report nobiletin ameliorates hepatic steatosis in genetically obese PER2::Luciferase reporter mice with fatty liver via downregulation of lipogenic and proinflammatory signaling mechanisms, proposing the intriguing hypothesis that pharmaceutical augmentation of hepatic circadian rhythm may be a novel potential therapeutic strategy for NAFLD.

MATERIALS AND METHODS

Animal Studies

All animal investigations were conducted in an American Association for the Accreditation of Laboratory Animal Care-accredited facility with studies approved by the Institutional Animal Care and Use Committee (IACUC). Control and obese PER2::LUCIFERASE reporter mice were generated by crossing transgenic male PER2::LUC reporter mice (gift in kind from Dr. Daniel Rudic) with female C57BL6 mice heterozygous for leptin receptor mutation (db/+, Jackson Laboratory, Strain No. 000697) and interbred for at least seven generations. Male mice were used in all experiments. Pair-fed male littermates heterozygous for leptin receptor (db/+^-) were used as lean controls. Mice were housed 2–4 per cage in a temperature-controlled environment (20 ± 2°C) with standard 12:12-h light/dark cycle unless otherwise discussed, with times of subjective day corresponding with 7 AM–7 PM period (ZT 0–12). Mice were maintained on regular chow diet with tap water available ad libitum. Genotyping was performed by PCR on purified toe clippings in 3-wk-old mice during weaning using Taq DNA polymerase and a Bio-Rad C1000 Touch Thermal Cycler with ethidium bromide visualization on 2% agarose gel. For terminal procedures, mice were anesthetized in an induction chamber using 5% isoflurane (1 L/min O₂) and then immediately decapitated via guillotine. Excised tissue was snap frozen in liquid nitrogen and stored at −80°C.

Metabolic Phenotyping

Serum analytes in fasted animals including hemoglobin A₁c (PD Diagnostics), serum glucose (Zoetis, Inc., Kalamazoo, MI), and insulin (ALPCO, Salem, NH) were obtained from trunk blood collected immediately after euthanasia (ZT 2) using manufacturer’s instructions. Serum aliquots were centrifuged at 10,000 RPM for 30 s, and supernatant was stored at −80°C for further analysis. Hepatic triglyceride content was assessed using the Folch-Lees method (22) with the following modifications. Briefly, lipids were extracted from 100-mg samples of pulverized liver and separated by thin layer chromatography using silica gel 60 plates. Gas chromatography was performed with an Agilent 7890 A system using capillary columns. Hepatic triglyceride content was determined by comparing lipid retention times to known standards. Serum aspartate aminotransferase (AST) and alanine aminotransferase (ALT) concentrations were measured using colorimetric assay kits (Sigma-Aldrich) according to the manufacturer’s instructions.

Continuous Food Intake Measurements

Mouse food intake was monitored in plexiglass respiratory chambers using open-circuit Oxymax Comprehensive Lab Animal Monitoring Systems (CLAMS; Columbus Instruments). Singly housed 12-wk-old male mice (n = 8 or 9 per group) were fed regular chow and water ad libitum and continuously monitored in 15-min intervals over a 5-day period after an initial 24-h acclimation period for noninvasive assessment of daily food intake.

Body Composition

Whole body and tissue biopsy composition was performed using a Bruker Minispec Body Composition Analyzer LF90II TD-NMR (Bruker, Billerica, MA). Percent fat composition by weight was assessed in freshly extracted 100-mg right hepatic lobe biopsy samples obtained on day of euthanasia.

Liver Histology

Mouse liver histology was assessed using the slow-frozen method where representative 0.5-cm² right hepatic lobe liver biopsies were embedded in Tissue-Tek OCT cryostat molds and suspended in a metal beaker containing isopentane cooled by dry ice. Cryosections (5 µM) were prepared by institutional core services and stained with hematoxylin and eosin (H&E), oil red-O (ORO), or trichrome using standard methods in blinded fashion. Cryosections were visualized on a Zeiss Axioplan-2 microscope using Axiocam imaging technology with staining quantified in ≥8 randomly selected fields per animal using ImageJ software (NIH).

Assessment of Oxidative Stress

Hepatic oxidative stress was assessed using 8-hydroxydeoxyguanosine (8-OHdG) and nitrotyrosine immunofluorescence staining. Cryosections were cut from right hepatic lobe biopsies slow frozen in OCT and postfixed in acetic alcohol formalin fixative for 1 min. Sections were incubated with either rabbit anti-nitrotyrosine or goat anti-8-OHdG antibody for 12 h at 40°C in a covered humidity chamber followed by goat anti-rabbit or rabbit anti-goat Alexa Fluor 594-conjugated secondary antibody (Thermo Fisher) for 30 min at 23°C. Background Sniper (Biocare Medical) was used as blocking agent to reduce nonspecific staining with PBS used as primary antibody in negative controls. DAPI was used to visualize nuclei mounted using Fluorescence VectaShield media. Slides were stained simultaneously to reduce variability in immunofluorescence intensity. Slides were visualized using a Zeiss LSM780 confocal microscope captured using Zen software with staining quantified in ≥5 randomly selected fields per animal using ImageJ software (NIH).

Gene Expression

Total RNA from pulverized snap-frozen liver tissue was extracted using Qiagen RNeasy Mini Kit treated with DNase. RNA yield was assessed by absorbance spectrophotometry using Nanodrop 2000 (Thermo Fisher). Complementary DNA was synthesized from 1 to 2 μg of purified RNA using SuperScript III (Thermo Fisher). Primers were designed using BLAST (NIH). Oligonucleotide sequences are listed in Supplemental Table S1 (all Supplemental Tables are available at https://doi.org/10.6084/m9.figshare.20294445.v1). Quantitative real-time PCR was performed in triplicate in CFX Connect Real-Time PCR Detection System (Bio-Rad) using SYBR green (Bio-Rad). Gene products were amplified at 95°C for 3 min followed by repetitive condensing cycles at 95°C for 15 s, 58.5°C for 15 s, and 72°C for 15 s for total of 40 cycles. Gene expression was calculated using the 2^−ΔΔCT method normalized to 18 s as the internal control.

Nobiletin Studies In Vitro

Nobiletin (Sigma-Aldrich) was externally validated in PER2::LUC peritoneal reporter macrophages as described by He et al. (21) with minor modifications. Briefly, 12-wk-old lean control PER2::LUC reporter mice were injected with 1 mL of 3% Brewer thioglycollate medium intraperitoneally and harvested 96 h later. Peritoneal macrophages (10⁵) were seeded on 96-well plates in DMEM for 2 h at 37°C. Nonadherent cells were removed by gently washing three times with warm PBS. Cells were synchronized with dexamethasone (0.1 µM) and treated by increasing doses of nobiletin (5 µM, 15 µM, and 50 µM) or DMSO vehicle. For palmitate studies, 10⁵ cells were plated in DMEM on 96-well plate on day 1, treated with palmitate (250 µM; Sigma-Aldrich) on day 2, and nobiletin (15 µM) on day 3 with media changed to luminescence buffer (0.1 mM luciferin-containing medium; Promega, Madison, WI). Bioluminescence was recorded every 2 h for a 24-h period in a Lumistar Galaxy luminometer (BMG) maintained at 37°C.

Nobiletin Studies In Vivo

Twelve-week-old PER2::LUC reporter mice bred on a db/db background were treated with either nobiletin or polysorbate-80 vehicle via oral gavage at 1800 h every day for a 15-day period. Nobiletin was dosed using a step-up method where nobiletin was initially administered at a 100 mg/kg dose for the first 10 days and subsequently increased to a 200 mg/kg dose for the final 5 days, resulting in a total nobiletin load of 2,000 mg/kg per mouse over the 15-day treatment period. A step-up strategy was used to minimize the risk of developing tolerance to nobiletin and avoid potential toxicity at higher doses. Mice were euthanized at midnight (2400 h), with cages maintained in darkness until immediately before euthanasia. Endogenous PER2::LUCIFERASE reporter activity was evaluated using Renilla luciferase assay in pulverized 50-mg samples (Promega, Madison, WI).

Constant Dark Studies

A cohort of 12-wk-old heterozygous control (db/+) and obese (db/db) PER2::LUC mice were maintained in either standard 12:12-h light:dark (LD) or constant dark (DD) conditions for a 12-wk period. Light:dark cycles corresponded with 7 AM–7 PM period (ZT 0–12). Mice were housed 3–4 animals per cage in a temperature-controlled environment (20 ± 2°C) with standard chow available ad libitum using heterozygous (db/+) pair-fed littermates as controls. All mice were euthanized at 1200 h (ZT 5) with tissue collected as described previously.

Microarray Studies

Large-scale gene array studies were performed in heterozygous control (db/+) and obese (db/db) mice euthanized at 1200 h and 2400 h. Total hepatic mRNA were extracted using TRIzol reagent (Invitrogen) and analyzed using Affymetrix technology (Santa Clara, CA). Circadian variation was defined as [fold change] > 2.0 between the 1200 h and 2400 h time points. Gene-set enrichment analysis was performed using Ingenuity Pathway Analysis (Qiagen) or Reactome (http://reactome.org) technology for genes with [fold change] >2.0 or >2.25, respectively.

Statistical Analysis

Continuous variables are represented as mean ± standard error and compared using Student’s t test or one-factor ANOVA with SNK test for multiple comparisons, as appropriate. Differences in mRNA expression over time were assessed using two-factor ANOVA and SNK post hoc test. Circadian rhythmicity in PER2::LUCIFERASE reporter macrophages was evaluated by cosinor analysis using Cosinor Periodogram v2.3, as previously described (23). Cosinor analysis is a statistical technique that calculates a least-squares regression for a cosine function to provide circadian parameters such as period length and cycle amplitude. Statistical tests were performed using SigmaPlot v12.0 or GraphPad Prism v9.2.0. A P < 0.05 was considered statistically significant.

RESULTS

Obese db/db Mice Recapitulate the NAFLD Phenotype

Baseline anatomic profiling is listed in Table 1, showing obese (db/db) mice have significantly increased body weight (+65%), whole body adiposity (+619%), and decreased muscle mass (−58%) versus heterozygous control mice. Fasted db/db mice were profoundly hyperglycemic (increased fasting blood glucose, and hemoglobin A₁c), hyperlipidemic (increased serum cholesterol and triglycerides), and hyperinsulinemic compared with control mice (Table 1), recapitulating the obese dysmetabolic phenotype characteristic of NAFLD.

Table 1.

Hyperphagic db/db mice are diabetic, dyslipidemic, and obese

Parameter	Control	db/db
Anatomic profile
Body weight, gm	32.2 ± 2.4	53.1 ± 3.9*
Whole body adiposity, %	8.2 ± 3.4	59.0 ± 1.7*
Whole body lean mass, %	83.3 ± 3.1	34.7 ± 1.7*
Serum metabolic indices
Glucose, mg/dL	213.8 ± 50	363.3 ± 130*
Hemoglobin A1c, %	4.6 ± 0.3	8.0 ± 1.2*
Cholesterol, mg/dL	55.8 ± 17	133 ± 46*
Triglycerides, mg/dL	31.9 ± 8.4	52.8 ± 26*
Nonesterified fatty acids, mEq/mL	0.3 ± 0.1	1.2 ± 0.5*
Insulin, ng/mL	0.06 ± 0.03	2.6 ± 3.4*
Leptin, pg/mL	3,528 ± 2,015	11,840 ± 3,093*

Data shown as mean ± SE; n = 6–8/group. Baseline anatomic phenotyping and fasting serum indices for control and db/db mice. *Significantly different from control (P < 0.05) using Student’s t test.

Markers of liver injury are presented in Fig. 1, showing db/db livers are significantly larger, grossly steatotic, with increased hepatic ORO staining versus control mice (Fig. 1, A–D). NMR spectroscopy revealed significantly greater intrahepatic fat (+101%) and triglyceride content (+240%; Fig. 1, E and F) in db/db mice. Hepatic oxidative stress, as assessed by 8-OHdG and nitrotyrosine immunofluorescence staining, was significantly increased in db/db mice (Fig. 1, G and H). Serum ALT and AST concentrations were significantly increased in db/db mice (Fig. 1I), consistent with the chronic liver injury phenotype of human fatty liver.

Figure 1.

Obese db/db mice have steatosis, oxidative stress, and hepatic injury. Markers of liver injury in control and db/db mice. A: representative liver morphology; scale bar: 1 cm. B: representative hepatic H&E and ORO staining; scale bar: 50 µm. C: liver size indices; n = 5. D: ORO quantification in >6 fields per animal using ImageJ; n = 4. E: liver biopsy fat content (%) as assessed by NMR; n = 5. F: hepatic triglyceride content as assessed by GC/MS; n = 5. G: representative images of hepatic oxidant stress as assessed by 8-OHdG and nitrotyrosine immunofluorescence staining. H: staining quantification in >5 fields per animal using ImageJ; n = 3. I: serum ALT and AST concentrations; n = 5. Data shown as means ± SE. *Significantly different from control (P < 0.05) using Student’s t test. ALT, alanine aminotransferase; AST, aspartate aminotransferase; H&E; hematoxylin and eosin; ORO, oil red-O.

Circadian PER2 Activity is Diminished in Obese PER2::LUC Mice

Luciferase assay was used to evaluate endogenous PER2 activity in PER2::LUC reporter mice bred on db/db background. As shown in Fig. 2A, hepatic PER2 activity in control mice gradually increased over the course of the day to peak at 2400 h (midnight local time). In contrast, no such PER2 circadian variation was observed in db/db mice, demonstrating instead a disordered pattern of PER2 oscillation significantly diminished from controls at 2400 h. Luciferase data were validated using PCR, which similarly revealed a gradual rise in hepatic Per2 expression that peaked at 2400 h in control mice (Fig. 2B). However, hepatic Per2 expression was significantly decreased at 2400 h in db/db mice, indicating obese PER2::LUC mice have diminished hepatic PER2 activity.

Figure 2.

Obese db/db mice have depressed hepatic Per2. A: hepatic PER2 reporter activity in control and db/db mice at time points indicated. Reporter activity was assessed using Renilla luciferase assay in 50-mg samples using luminometer. Hepatic Per2 (B) and Dbp (C) mRNA expression in control and db/db mice at time points indicated; 18 s was used as internal control. D: renal PER2 reporter activity in control and db/db mice at time points indicated. Reporter activity was assessed using Renilla luciferase assay in 50-mg samples using luminometer. Shaded areas represent time of subjective night. Data shown as means ± SE; n = 3 or 4 per group for each time point. *Significantly different (P < 0.05) using two-factor ANOVA.

As PER2 is a negative regulator of the circadian cycle, we tested the hypothesis that diminished PER2 activity was associated with concomitant increase in hepatic Dbp, a major clock output gene. Hepatic Dbp was maximally expressed at 1800 h in control mice, coinciding with nadir PER2 activity (Fig. 2C). However, db/db mice demonstrated significantly more robust Dbp expression (+50%) during same 1800 h timepoint, suggesting that diminished PER2 rhythm is associated with exuberant and possibly deranged cycle output. To investigate if diminished PER2 is a phenomenon unique to the liver, circadian PER2 reporter activity was evaluated in kidneys of PER2::LUC reporter mice (Fig. 2D), which also demonstrated a gradual increase in PER2 activity peaking at 2400 h (midnight) in control mice. In contrast, renal PER2 activity was significantly blunted at 2400 h in db/db mice, indicating diminished endogenous PER2 activity in multiple peripheral oscillators of obese PER2::LUC mice.

We evaluated if diminished peripheral PER2 was due to impairment in the central clock by assessing diurnal mouse feeding behavior, whose rhythm is primarily controlled by the SCN and can be used to evaluate master clock function. As expected, PER2::LUC db/db mice had significantly greater total daily food intake as well as total nighttime food intake than control mice (Fig. 3A). However, the usual hours of food intake for control and db/db mice were both preserved to times of subjective night (ZT 12–24; Fig. 3B), suggesting that the central clock behavioral rhythm dictating daily feeding patterns is not disrupted in db/db mice and thus the primary circadian defect in obesity must lie within peripheral oscillators. To further exclude confounding behavioral rhythms, we measured intrahepatic fat content (%) in control and db/db mice subjected to light-dark (LD) or constant dark (DD), finding control mice with normal hepatic rhythm had no significant difference in fat content in LD versus DD conditions (Fig. 3, C–F). However, db/db mice in constant dark (DD) developed greater steatosis and intrahepatic lipid content than db/db mice in LD, suggesting peripheral oscillators in the liver are impaired during obesity to promote intrahepatic lipid accumulation.

Figure 3.

Diminished hepatic Per2 in db/db mice is not due to alteration in behavioral feeding patterns in the central clock. Behavioral rhythms were evaluated using food intake monitoring in CLAMS following initial animal acclimation period. A: total average food intake and total average daytime/night time food intake in control and db/db mice; n = 8 or 9. B: hourly food intake in control and db/db mice; n = 8 or 9. Shaded areas represent time of subjective night. ZT: Zeitgeber time with ZT0 corresponding to beginning of daylight. Although db/db mice had significantly greater food intake, time of food ingestion was similar between groups, suggesting central clock function is preserved between groups. Liver mass (C) and liver biopsy fat content (%; D) as assessed by NMR in control and db/db mice subjected to 12-h light:dark (LD) or 24-h constant dark (DD) for 12-wk period; n = 5. Representative hepatic H&E (E) and ORO (F) staining; scale bar: 75 µm. Data shown as means ± SE. ns, not significant. *Significantly different (P < 0.05) using two-factor ANOVA. H&E; hematoxylin and eosin; ORO, oil red-O.

To investigate if derangement in circadian rhythm is associated with global alterations in hepatic transcription, microarray analysis was used to compare gene expression at 1200 h and 2400 h in control and db/db mice. Using stringent criteria, this analysis revealed 328 hepatic transcripts cycled in a circadian manner in db/db mice compared with only 158 transcripts in control mice (P < 0.001). Notably, these circadian transcripts included 55 genes in db/db mice classified by IPA as directly involved in proinflammatory signaling networks versus only 14 genes in control mice (P < 0.001). Gene-set enrichment analysis identified “circadian rhythm signaling” as one of the most enriched pathways in control but not db/db mice (Table 2). In contrast, the most enriched pathways in db/db mice focused on immune signaling activation. A separate analysis comparing expression profiles in control and db/db mice at 1200 h again centered on proinflammatory signaling mechanisms in db/db mice with “LPS/IL-1-mediated inhibition of RXR function” the top canonical pathway and “Inflammatory Response” the top disease process. Similar results were obtained using Reactome gene-set enrichment analysis (Supplemental Table S2). Collectively, these data suggest circadian control of the hepatic transcriptome is impaired in db/db mice, which is associated with pathological upregulation of inflammatory signaling mechanisms.

Table 2.

IPA gene-set enrichment analysis in liver of control and db/db mice

Top IPA Canonical pathways in control mice (2400 h vs. 1200 h)	Overlap	P Value
FXR/RXR activation	7/138	5.33E-06
LPS/IL-1 mediated inhibition of RXR function	8/224	1.48E-05
Circadian rhythm signaling	4/35	2.54E-05
STAT3 pathway	5/74	3.17E-05
Acute phase response signaling	6/171	1.99E-04

Top diseases and biological functions	No. of Molecules	P Value
Metabolic disease	38	1.33E-03–5.12E-12
Neurological disease	28	1.07E-03–1.25E-10
Endocrine system disorders	42	1.33E-03–8.08E-10
Cancer	95	1.46E-03–1.61E-09
Organismal injury and abnormalities	95	1.46E-03–1.61E-09

Top canonical pathways in db/db mice (2400 h vs. 1200 h)	Overlap	P Value
Altered T cell and B cell signaling in rheumatoid arthritis	11/88	5.46E-09
Agranulocyte adhesion and diapedesis	14/190	4.57E-08
Type 1 diabetes mellitus signaling	11/111	6.40E-08
T helper cell differentiation	9/72	1.38E-07
Role of NFAT in regulation of immune response	13/177	1.45E-07

Top diseases and biological functions	No. of Molecules	P Value
Endocrine system disorders	77	9.25E-08–3.80E-45
Gastrointestinal disease	88	2.99E-05–3.80E-45
Immunological disease	98	4.60E-05–3.80E-45

Top canonical pathways in db/db vs control mice (at 1200 h)	No. of Molecules	P Value
LPS/IL-1 mediated inhibition of RXR function	77	9.25E-08–3.80E-45
Nicotine degradation II	88	2.99E-05–3.80E-45
Xenobiotic metabolism signaling	98	4.60E-05–3.80E-45
LXR/RXR activation	84	9.25E-08–3.80E-45
FXR/RXR activation	84	2.71E-05–4.06E-22

Top diseases and biological functions	No. of Molecules	P Value
Inflammatory response	155	1.16E-04–5.83E-16
Metabolic disease	111	8.95E-05–9.09E-15
Dermatological diseases and conditions	246	5.99E-05–1.09E-14
Endocrine system disorders	98	8.95E-05–3.28E-13
Gastrointestinal disease	371	1.14E-04–3.28E-13

Top IPA canonical pathways and biological functions in control, db/db, and db/db vs. control mice (at 1200 h) for genes with [fold change] >2.0 at indicated time points.

Nobiletin Enhances Endogenous Circadian Rhythm In Vitro

We then investigated the effect of nobiletin, a recently described circadian clock enhancer, on endogenous PER2 oscillation in vitro (21). Nobiletin activity was validated in PER2::LUC peritoneal macrophages, where cosinor analysis revealed nobiletin significantly increased PER2 amplitude in a dose-dependent manner with no effect on circadian period (Fig. 4, A–C). Investigating the effect of nobiletin on cell culture models of fatty acid overload, cosinor analysis revealed lipid-loading cells with palmitate [PA; (250 μM)] profoundly degraded PER2 amplitude (Fig. 4, D and E). However, nobiletin (15 μM) significantly restored PER2 amplitude in PA-treated cells with no effect on PER2 period (Fig. 4F), supporting the clock-enhancing properties of nobiletin in vitro.

Figure 4.

Nobiletin improves palmitic acid-induced dampening of circadian rhythm in PER2::LUC reporter macrophages. Peritoneal macrophages were harvested from 12-wk-old PER2::LUC reporter mice as described in materials and methods. A: PER2 activity in dexamethasone (0.1 µM)-synchronized PER2::LUC reporter macrophages treated with nobiletin (5 µM, 15 µM, and 50 µM) or vehicle for 24-h period. Nobiletin had dose-dependent increase in circadian PER2 amplitude (B) but not period (C). D: PER2 activity in PER2::LUC reporter macrophages treated with palmitate (250 µM), nobiletin (15 µM), or vehicle. Palmitate significantly decreased PER2 amplitude (E) in PER2 reporter cells, which was partially restored with Nobiletin (15 µM). F: nobiletin had no additional effect on circadian period in palmitate-treated cells. Circadian parameters calculated using cosinor analysis. Data represent two independent experiments; data shown as means ± SE; n = 5 wells per group. *P < 0.05 vs. vehicle using ANOVA; †P < 0.05 vs. palmitate-only group using ANOVA.

Nobiletin Enhances Endogenous Circadian Rhythm to Resolve Hepatic Steatosis In Vivo

We then tested the hypothesis that nobiletin ameliorates hepatic steatosis in genetically obese (db/db) PER2::LUC reporter mice with fatty liver. Consistent with its amplitude-enhancing properties in vitro, nobiletin significantly increased PER2 reporter activity in both liver and kidney of PER2::LUC db/db mice with diminished peripheral PER2 (Fig. 5, A and B). PCR confirmed nobiletin dramatically upregulated hepatic circadian signaling including significantly increased Clock and Dbp expression in PER2::LUC db/db mice (Fig. 5C), supporting the clock-enhancing effects of nobiletin in vivo.

Figure 5.

Nobiletin restores diminished endogenous PER2 in PER2::LUC db/db mice. Nobiletin restores diminished PER2 activity in liver (A) and kidney (B) of PER2::LUC db/db reporter mice. C: nobiletin upregulates hepatic clock gene expression in PER2::LUC db/db reporter mice. Data shown as means ± SE; n = 3–5. *Significantly different (P < 0.05) using ANOVA.

In the liver, nobiletin dramatically ameliorated hepatic steatosis, significantly reduced ORO staining, and revealed a trend toward attenuated liver size and intrahepatic lipid content in PER2::LUC db/db mice (Fig. 6). Nobiletin significantly decreased liver mass normalized to the change in body weight of db/db mice (−445 mg vs. 1,128 mg; P < 0.05). Mechanistically, nobiletin significantly decreased hepatic sterol regulatory element-binding transcription factor 1c (Srebp1c), acetyl-CoA carboxylase-1 (Acaca1), and tumor necrosis factor-α (Tnfα) expression in db/db mice, suggesting nobiletin resolves NAFLD by downregulating lipogenic and proinflammatory signaling cascades (Fig. 7). Additional mediators of hepatic lipid accumulation such as carnitine palmitoyltransferase-I (Cpt1), perilipin-2 (Plin2), perilipin-5 (Plin5), and fibroblast growth factor 21 (Fgf21) were not upregulated with nobiletin. Moreover, metabolic phenotyping revealed nobiletin had no significant effect on body weight, adiposity, or lean body mass in db/db mice, indicating improvements in steatosis with nobiletin were independent of these factors (Supplemental Table S3). Interestingly, nobiletin decreased serum insulin concentration (−37%) in db/db mice with no significant effect on fasting triglyceride, cholesterol, or glucose levels. Collectively, these data suggest restoring endogenous hepatic PER2 with nobiletin resolves steatosis in db/db mice, proposing that targeting the biological clock may be an attractive therapeutic strategy for alleviation of NAFLD.

Figure 6.

Nobiletin ameliorates hepatic steatosis in PER2::LUC db/db mice. A: representative hepatic H&E, ORO, and trichrome straining in control and PER2::LUC db/db mice; scale bar: 50 μM. B: ORO and trichrome quantification in >6 fields per animal using ImageJ. C: liver biopsy fat content (%) as assessed by NMR. D: liver size indices. Data shown as means ± SE; n = 3–5. *Significantly different (P < 0.05) from control using ANOVA. H&E; hematoxylin and eosin; ORO, oil red-O.

Figure 7.

Nobiletin effect on hepatic gene expression in PER2::LUC db/db mice. Data shown as means ± SE; n = 3 or 4. *Significantly different (P < 0.05) using ANOVA.

DISCUSSION

NAFLD is the most common chronic liver disease in the Western hemisphere, yet efforts to develop effective medical therapies have been thwarted. In this study, we show that endogenous hepatic PER2 is diminished in genetically obese PER2::LUCIFERASE mice with fatty liver. Nobiletin, a recently described biological clock enhancer, restores endogenous PER2 oscillation in vivo and in vitro. Intriguingly, nobiletin resolves steatosis in db/db mice by downregulating key lipogenic and proinflammatory hepatic signaling mechanisms, identifying the hepatic clock as an important determinant of steatosis whose pharmacological manipulation may be an exciting emerging therapeutic strategy for NAFLD.

Nobiletin is a highly permeable polymethoxylated flavone under preinvestigation for an array of conditions such as neurodegenerative disease (24), ischemic stroke (25), neointimal hyperplasia (26), atherosclerosis (27), diabetic cardiomyopathy (28), gastric malignancy (29), and experimental colitis (30). However, while generally considered beneficial in obesity, molecular mechanisms underpinning the positive effects of nobiletin are debated. Nobiletin was reported to increase Srebp1c expression in white adipose tissue of high-fat diet (HFD)-fed mice (31), whereas another closely related flavonoid, naringenin, increases SREBP1 nuclear trafficking in HepG2 cells (32). In contrast, others report nobiletin reduces HFD-induced hepatic Srebp1 expression in low-density lipoprotein receptor (LDLR) knockout mice (25, 33) and attenuates hyperglycemia-induced Srebp1 expression in HepG2 cells (34). Similar controversies exist regarding mechanisms of hepatic lipid extrusion. For instance, CPT1 catalyzes the formation of acyl-carnitines to function as the rate-limiting step of mitochondrial fatty acid β-oxidation. Nobiletin was shown to increase hepatic Cpt1 expression and fatty acid oxidation as determined by [³H]palmitate conversion to [³H]H₂O in LDLR knockout mice with diet-induced obesity (14, 33). These reports are incongruent with microarray studies revealing no significant improvement in Cpt1 expression in HFD-fed mice treated with nobiletin (21), potentially deriving from differences in animal models, nobiletin dosing strategies, treatment durations, and timing of experiments.

Hyperphagic (db/db) mice, which lack functional leptin receptor, manifest components of the metabolic syndrome including increased visceral adiposity, dyslipidemia, and hyperglycemia with insulin resistance (35). Moreover, the db/db mouse is also a recognized model for the experimental study of hepatic steatosis in the context of obesity and insulin resistance, the most common human phenotype of NAFLD. In regard to circadian rhythm, db/db mice demonstrate blunted circadian rhythm along with altered behavior and feeding. For instance, db/db mice are generally arrhythmic with respect to locomotor activity, and as a consequence of leptin deficiency, exhibit distinctly elevated total and daytime food intake. db/db mice also display disrupted sleep architecture as evidenced by greater total sleep time, sleep fragmentation, and bouts of nonrapid eye movement (NREM) (36). Photic induction of the master clock, as determined by c-Fos, and phosphorylated extracellular signal-regulated kinase1/2 (P-ERK1/2) induction in the SCN in late night, are also increased in db/db mice to suggest greater responsiveness to light (37). However, circadian oscillation of core clock proteins such as PER2 and BMAL1 are not altered in the SCN of db/db mice versus lean control (db/+) mice, suggesting altered behavioral rhythms in db/db mice are not associated with circadian alteration of core clock oscillators in the SCN (38). In agreement with this finding, we show here that attenuating the influence of behavioral rhythm using constant dark (DD) conditions exacerbates hepatic steatosis in db/db mice (Fig. 3), whereas restoring oscillation of hepatic PER2 with nobiletin ameliorates hepatic steatosis in db/db mice (Fig. 5). Thus, we propose that the primary circadian deficit responsible for altered endogenous hepatic rhythm in db/db mice with fatty liver lies not within the master clock of the SCN but rather peripheral oscillators intrinsic to the liver. The peripheral oscillators of the liver are entrained not only by neuronal and glucocorticoid inputs from the central clock but also strongly by local nutritional and hormonal zeitgebers that function to synchronize environmental cues such as daily feeding with metabolism. For example, postprandial insulin potently induces PER2 protein abundance, augments circadian amplitude, and shifts the phase of endogenous hepatic rhythm to communicate time-of-feeding to peripheral oscillators in the liver (39). Although further studies are needed, we speculate nobiletin may function to similarly augment hepatic PER2 to synchronize environmental cues such as feeding with nutrient metabolism in the liver. Of course, implicit in this function is the importance of optimizing nobiletin administration with timing of food intake.

An important limitation of this study is that food intake was not directly measured, precluding the ability to evaluate the effects of nobiletin on total food consumption. However, a previous study reported 8 wk of nobiletin (200 mg/kg body wt administered via oral gavage in every other day dosing) did not significantly alter food intake in mice with diet-induced obesity (21). In addition, we specifically chose a shorter nobiletin treatment duration (15 days) to minimize potential changes in mouse weight that may occur over the study period. Thus, although we cannot exclude food intake as a potential confounding factor contributing to the improvement in hepatic steatosis in db/db mice treated with nobiletin, we believe this effect is likely to be small. It should also be acknowledged that measuring gene expression at a single time point, such as in the data presented in Fig. 5, cannot unequivocally determine if nobiletin enhances oscillator amplitude or instead leads to phase advances. However, given that hepatic expression of Dbp, a clock output gene, was significantly increased with nobiletin (Fig. 5C) suggests nobiletin most likely augments oscillator amplitude rather than altering circadian phase.

We show here that nobiletin reduces serum insulin concentration and hepatic expression of Srebp1c and target-gene Acaca1, suggesting nobiletin resolves steatosis in db/db mice by downregulating Srebp1c programming. Srebp1c is a key mediator of hepatic lipid metabolism whose activity oscillates in a circadian manner (40). Moreover, Srebp1 is known to be upregulated in humans with NAFLD, and pharmacological or genetic inhibition of Srebp1 attenuates hepatic steatosis in mouse models of obesity (41). Other mechanisms of hepatic triglyceride catabolism such as lipid droplet trafficking and fatty acid β-oxidation are unlikely to be implicated, as hepatic Cpt1, Plin2, Plin5, and Fgf21 expression are not upregulated with nobiletin. Interestingly, Fgf21 expression, a hepatokine involved in energy expenditure whose transcription is regulated in circadian manner (42–43), was downregulated with nobiletin, speculating that nobiletin may attenuate obesity-induced FGF21 resistance (44). These data show nobiletin regulation of intrahepatic lipid metabolism is indeed complex, and more mechanistic studies are needed to precisely define the beneficial effects of nobiletin in NAFLD.

Epidemiological studies have shown an association between circadian-related sleep disorders such as insomnia and obstructive sleep apnea and increased risk of NAFLD (36–38). Possible mechanistic explanations include worsening of metabolic risk factors for NAFLD such as weight gain and insulin resistance with chronic sleep disturbances. Sleep disruption has been associated with increased production of both reactive oxygen species and proinflammatory serum cytokines such as TNFα (45–47). TNFα signaling plays a central role in NAFLD pathogenesis by promoting hepatic insulin resistance and proinflammatory gene expression (48–52). Serum TNFα levels predict patients at risk for developing incident NAFLD, whereas inhibition of TNFα activation with pentoxyfylline attenuates histological markers of liver inflammation in humans with nonalcoholic steatohepatitis (NASH) (53–60). We show here that elevated hepatic TNFα expression in obese db/db mice is prevented by nobiletin, suggesting a novel potential link between impaired hepatic circadian rhythm and proinflammatory cytokine production. Although db/db mice are an excellent model for studying hepatic steatosis in the context of obesity and insulin resistance, nobiletin effects on inflammation should be investigated in more faithful NASH conditions such as high-fat or Western diet. TNFα has also been shown to accelerate development of fatty liver by upregulation of SREBP1c-mediated lipogenesis (61), representing another possible mechanism whereby nobiletin attenuates SREBP1c activity via downregulation of TNFα.

We propose a model whereby the widespread gene dysregulation intrinsic to NAFLD may derive or exacerbate from impaired hepatic circadian rhythm. In healthy individuals, endogenous hepatic circadian rhythm functions to constrain lipogenic and proinflammatory gene transcription. However, during obesity, the hepatic clock is diminished, resulting in uninhibited and chaotic activation of pro-NAFLD mechanisms such as SREBP1c-mediated lipogenesis. Restoring circadian oscillation of key clock regulators with nobiletin may prevent disinhibition of pathological hepatic signaling. Our data implicate impaired hepatic circadian rhythm as a causative mechanism for fatty liver in obesity, supporting pharmacological manipulation of the hepatic clock as an emerging therapeutic strategy in NAFLD.

SUPPLEMENTAL DATA

Supplemental Tables S1–S3: https://doi.org/10.6084/m9.figshare.20294445.v1.

GRANTS

This research was supported by National Institutes of Health (NIH) Grant F31HL154646 (to C.A.P.), American Heart Association Grant 17POST33410322 (to J.T.B.), and NIH Grants R01HL124733 and NIH R01HL147159 (to D.J.F. and D.W.S.). Presentation of this data was supported by a Caroline tum Suden/Frances Hellebrandt Professional Opportunity Award (to S.L.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.L., C.A.P., D.J.F., and D.W.S. conceived and designed research; S.L., C.A.P., J.T.B., J.D.M., and D.J.F. performed experiments; S.L., C.A.P., J.T.B., J.D.M., D.J.F., and D.W.S. analyzed data; S.L., C.A.P., J.T.B., and D.W.S. interpreted results of experiments; S.L., C.A.P., and J.T.B. prepared figures; S.L. and C.A.P. drafted manuscript; S.L., C.A.P., J.T.B., J.D.M., D.J.F., and D.W.S. edited and revised manuscript; S.L., C.A.P., J.T.B., J.D.M., D.J.F., and D.W.S. approved final version of manuscript.