Down-regulation of hepatic CLOCK by PPARα is involved in inhibition of NAFLD

This work aimed to investigate the role of nuclear factor peroxisome proliferator-activated receptor α (PPARα) in modification of circadian clock and their relevance to development of nonalcoholic fatty liver disease (NAFLD). Both male wild-type (WT) and Pparα-null (KO) mice treated with high-fat diet (HFD) were used to explore the effect of PPARα and lipid diet on the circadian rhythm. WT, KO, and PPARα-humanized (hPPARα) mice were treated with PPARα agonist fenofibrate to reveal the hPPARα dependence of circadian locomotor output cycles kaput (CLOCK) down-regulation. The mouse model and hepatocyte experiments were designed to verify the action of PPARα in down-regulating CLOCK and lipid accumulation in vivo and in vitro. Strongest NAFLD developed in mice fed 45%HFD, and it was inhibited in WT mice. The activity rhythm of WT mice was found to be different from that of the KO mice on normal diet and HFD. The core circadian factor CLOCK was down-regulated by HFD in both WT and KO mice in the liver, not in the hypothalamus. More interestingly, hepatic CLOCK was down-regulated by basal PPARα and activated PPARα in dose dependence of fenofibrate. Accordingly, CLOCK down-regulation dependent of PPARα activity was involved in inhibition of lipid metabolism in hepatocytes. Down-regulation of hepatic CLOCK by basal PPARα contributes to tolerance against development of NAFLD. Inhibition of CLOCK by activated PPARα is involved in inhibition of NAFLD by PPARα agonists. • PPARα inhibited NAFLD development induced by HFD. • PPARα mediated modifications of circadian rhythm and the hepatic circadian factor CLOCK in NAFLD models. • Down-regulation of hepatic CLOCK by basal PPARα contributed to tolerance against development of NAFLD. • Inhibition of CLOCK by activated PPARα was involved in therapeutic actions against fatty liver diseases by PPARα agonists.


Introduction
Nonalcoholic fatty liver disease (NAFLD) is a major challenge due to its complex pathogenesis, difficult diagnosis, and high prevalence without approval treatment [1,2]. NAFLD is closely associated with obesity and usually occurs with type 2 diabetes that is partly due to increased insulin resistance [3,4]. It leads to cirrhosis and liver cancer, which were predicted to be a major public health problem worldwide [5].
These metabolic diseases are known to be most related with the over-intake of high-calorie food. High-fat diet (HFD) is widely used to induce metabolic diseases since energy intake is the most important etiological factor. The phenotype induced by HFD is similar to the human metabolic disease, characterized by obesity, glucose intolerance, and fatty liver [6]. However, the lipid percentage of fat component in HFD varied in published studies. Male Sprague Dawley rats fed HFD containing 30% fat for 17 weeks successfully developed NAFLD [7]. Fatty liver was induced by feeding a diet containing 30% fat for 2 months in male C57BL/6 mice [8]. Mice fed 60%HFD for 12 months developed glucose tolerance [9]. Besides, Jia Luo and Zheng Yan contributed equally to this work.
* Aiming Liu liuaiming@nbu.edu.cn 1 School of Medicine, Ningbo University, Ningbo, China 2 Ningbo College of Health Sciences, Ningbo, China obesity was induced in mice fed a 53%HFD for 6 weeks [10]. But the optimal lipid content for inducing the strongest NAFLD is still unknown. The 2017 Nobel Prize in Medicine declared the important role of circadian clock proteins in physiology and diseases [11]. Circadian rhythms were cycles of gene expression, metabolism, and behavior, produced by internal clock factors [12]. Circadian oscillations were generated by transcriptional regulatory feedback loops, which includes factors such as circadian locomotor output cycles kaput (CLOCK) and brain and muscle ARNT-like 1 (BMAL1) [13]. The heterodimer CLOCK/BMAL1, as the central elements in both central clock and peripheral clock, is involved in the regulation of downstream genes as a transcription factor [14,15]. Their overall action in coordinating metabolism is closely related to many diseases including NAFLD [11]. When Clock or Bmal1 was knocked out, the mice developed fatty liver, impaired glucose tolerance, hyperlipidemia, and other metabolic diseases [16,17]. Thus, artificial disruption of the core circadian gene Clock or Bmal1 promote development of multiple metabolic abnormalities. However, little is known how circadian clock was modified in development NAFLD, as well as the underlying molecular mechanism.
Peroxisome proliferator-activated receptorα (PPARα) regulates inflammation, energy homeostasis, and metabolism of lipid and glucose [18]. Compared with other tissues, hepatic PPARα was the most abundant and played a key role in maintaining lipid metabolism homeostasis [19]. Both CLOCK and BMAL1 were reported to directly regulate the diurnal expression of PPARα gene by an E-box-dependent mechanism. In CLOCK-mutant mice and BMAL1 knock-out mice, the mRNA level of PPARα was down-regulated in hepatic tissues, and its circadian oscillation disappeared [20,21]. Conversely, PPARα was found to regulate BMAL1 via PPRE site in its promoter region [20]. It is not clear whether regulation of circadian CLOCK by PPARα is involved in NAFLD or how it occurs.
In this study, male wild-type (WT), Pparα-null (KO), and PPARα humanized (hPPARα) mice were used to investigate the differential tolerance of mice on HFD. The modification of CLOCK and BMAL1 in the hepatic tissues and the hypothalamus by PPARα was explored. Hepatocytes were used to validate the action of PPARα in CLOCK down-regulation and the downstream effect in lipid accumulation.

Reagents and materials
All the critical reagents and materials were listed in supplementary Table 1.

Mice and treatments
Male 129/Sv WT, KO, and hPPARα mice with the same genetic background were described in earlier studies [22,23]. The mice were housed in standard conditions at 24 ± 1 °C, with humidity of 50 ~ 70%. Twelve-hour light/12-h dark cycle (8:30-20:30 was defined as daytime) was set in the house. Pure water and commercial diet were accessed ad libitum. All the procedures were performed in compliance with the ARRIVE guidelines [24].
Both WT and KO mice were divided into four groups randomly (Con, 10%HFD, 45%HFD, and 60%HFD, n = 5). The mice were fed normal diet (ND) and diet containing 10%, 45%, and 60% fat for 12 weeks, and their body weight was recorded every week. The oral glucose tolerance test (OGTT) was performed before the end of the experiment.
Since 45%HFD was the strongest in inducing the NAFLD, it was used to explore the relationship between HFD and circadian rhythm. Eight-week-old WT and KO mice were assigned to Con and HFD groups (n = 5). Mice in both groups were fed ND for the first week, and those in the HFD groups were fed diet containing 45% fat for another 2 weeks. The activity and metabolic rhythm of mice were recorded after feeding HFD for 2 weeks.
In the third experiment, WT, KO, and hPPARα mice were divided into 4 groups, respectively (n = 4). They were treated with 5, 25, and 125 mg/kg fenofibrate (F5, F25, F125) twice a day by oral gavage for 5 days, to explore the direct action of PPARα on CLOCK and the effect on steatosis development.
To explore functional involvement of CLOCK in PPARαmediated steatosis regulation, male 129/Sv WT mice were arranged into Con, CDAHFD, and H + FENO group (n = 7). CDAHFD mice were fed a choline-deficient, L-amino aciddefined, high-fat diet (CDAHFD) for 3 weeks. H + FENO mice were fed CDAHFD and simultaneously gavaged 25 mg/kg fenofibrate twice a day for another 3 weeks.
All the above mice were sacrificed at 8:30 in the morning. The serum of fresh blood samples was collected after centrifugation. Liver and hypothalamus tissues of mice were separated and collected. The biggest liver lobe was cut off and fixed in 10% neutral formalin buffer. All the fresh tissues left were frozen in dry ice and kept under − 80 °C pending analysis.

Rhythmic data acquisition
Mice were singly housed in OxyletPro Physiocages (Panlab Oxylet, America) and allowed a 48-h acclimation. Measurements of the volume of oxygen (VO2) consumed, the volume of carbon dioxide (CO2) produced, the energy expenditure (EE), and activity rhythm were performed over 48 h. Values of VO2, CO2, and EE were normalized to the body weight raised to the power 0.75.

Biochemical and histopathological analysis
The content of total cholesterol (TC) and triglyceride (TG) in serum and liver were determined using end-point method according to the instructions in the kits. Formalin-fixed liver was dehydrated in ethanol gradients. After transparency in xylene, the tissues were embedded by paraffin. Four-micrometer sections were prepared for hematoxylin and eosin staining. The pathological changes were imaged using Olympus BX43 light microscope.

Isolation of mouse primary hepatocytes and hepatocyte culture
Primary hepatocytes were isolated from mouse livers using Hank's balanced salt solution with 0.4 mg/ml type IV collagenase perfusion. After the cellular viability (> 85%) was confirmed by using the Trypan blue exclusion method, hepatocytes were cultured in Williams-E supplemented with 10% fetal bovine serum (FBS). HepG2 cells were cultured in RPMI 1640 medium or DMEM with high glucose supplemented with 10% (v/v) FBS and 100 U/ml penicillin-streptomycin.
HepG2 cells were firstly stimulated with WY-14643 (0, 10, 50, 200 uM) for 30 h, to validate the action of PPARα on CLOCK down-regulation. Primary hepatocytes and HepG2 cells were challenged with oleic acid (OA, 0.1 mM) and WY-14643 (50uM) for 72 h. Cell lipid accumulation was visualized by oil red O staining and quantified by Image J 18.0.

Western blot assay
The protein separated in 10% SDS-polyacrylamide gel was transferred to PVDF membrane for incubation of antibodies. Method of detail was described in earlier studies [25]. The antibody information is listed in supplementary Table 2.

Quantitative PCR analysis
Total RNA extracted from the liver was detected using protocols used in our previous work [25]. One microgram of total RNA was taken and added to the reverse transcription mixture reagent. Primer sequences used in this work are shown in supplementary Table 3.

Statistical analysis
All experimental data were presented as mean ± SD, and SPSS 17.0 was used for data analysis. Student's t test was used for difference examination between two groups. Means of more than two groups were compared using analysis of variance (ANOVA, one-way or two-way). And Bonferroni test was used for post hoc analysis of differences in each group. The difference was considered statistically significant when p values were below 0.05.

PPARα inhibited NAFLD development under HFD treatment
The body weights of HFD groups were higher than those of Con group in both WT and KO mice. In WT mice, the body weight of 45%HFD group and 60%HFD group increased significantly from the third week. Interestingly, the former generated more body weight gain than the latter during the last 3 weeks (p < 0.05) (Fig. 1A). The body weight gains of 45%HFD group and 60%HFD group from the 10th week were higher than that of Con group in KO mice (p < 0.05) (Fig. 1B). These data indicated that the high-energy diet induced body weight gain in WT and KO mice, with strongest effect by 45%HFD.
In the OGTT, the glucose concentration in 45%HFD and 60%HFD groups at 30 min and 60 min was higher than those in Con group for WT mice (p < 0.05). A similar result was found with KO mice fed 45% and 60%HFD, supporting that 45% and 60%HFD induced impaired glucose tolerance with equal potency in WT and KO mice (p < 0.05) (Fig. 1C, D). For serum TC, the level in 45%HFD and 60%HFD groups was significantly elevated compared with that of Con group two mouse lines (p < 0.05) (Fig. 1E). No difference of TG level in the serum between WT and KO mice was observed (Fig. 1F). These data show that strongest metabolic disorders were induced by 45%HFD.
Liver TC and TG were increased under 45%HFD and 60%HFD treatment in the WT and KO (p < 0.05) ( Fig. 2A,  B). In 45%HFD groups, liver TC and TG of KO mice were higher, indicating that PPARα deficiency increased liver lipid deposition. Hepatic pathological analysis showed normal hepatic structures in both WT and KO groups. Hepatic tissue structure of the 10%HFD group was not modified. Both 45%HFD and 60%HFD groups showed steatosis, and the former was more obvious (Fig. 2C). Between the two mice lines, fatty liver was the more serious in KO mice. These data indicated that 45%HFD group was the strongest in inducing fatty liver and PPARα was tolerance against HFD-induced NAFLD in WT mice. Fig. 1 Strongest metabolic disorders were induced by 45%HFD. Male WT and KO mice were randomly divided into four groups (Con, 10%HFD, 45%HFD, and 60%HFD, n = 5). They were fed normal diet and diet containing 10%, 45%, and 60% fat for 3 months. A, B Body weight change of WT and KO mice. C, D Change of oral glucose tolerance test of WT and KO mice. E, F Change of serum TC and TG in WT and KO mice. Data are expressed as mean ± SD. Single asterisk indicated p < 0.05, in comparison with WT-Con/ KO-Con. # indicated p < 0.05, in comparison between 45%HFD group and 60%HFD group. Difference at individual time points was determined by two-way ANOVA and Bonferroni post hoc test Data are expressed as mean ± SD. Single asterisk indicated p < 0.05, in comparison with WT-Con/KO-Con, # indicated p < 0.05, in comparison with WT. Difference in each group was determined by twoway ANOVA and Bonferroni post hoc test

HFD modified circadian rhythm in mice
In the diet intake profiles of the second experiment, the nighttime and daytime diet consumption were significantly decreased for both the WT and KO mice after being treated with 45%HFD for 1 and 2 weeks (Fig. S1A, B).
Whole body energy metabolism was tested to explore the effect of HFD on circadian rhythm. We checked the VO2, CO2, EE, and activity within 48 h for WT and KO mice. The activity rhythm of WT mice was different from that of KO mice on both the ND and HFD treatment (Fig. 3A, B). The variation of VO2, VCO2, and EE in HFD group was decreased between dark and light phases in the two mouse lines (Fig. 3C, D and S2A-D). Statistically, activity was increased in the WT-HFD and KO-HFD groups, and the latter was higher than the former during the night phase (p < 0.05) (Fig. 3E). And HFD induced a decrease in VO2 and EE in WT and KO mice during both dark and light phases (p < 0.05) (Figs. 3F and S2E). Compared to WT-Con group, the production of CO2 was reduced in WT-HFD and KO-Con group during night (p < 0.05) (Fig. S2F). These data suggested that HFD decreased diet intake and circadian rhythm in mice.

CLOCK was down-regulated by basal PPARα
In the 2nd experiment, compared with their control groups, the core circadian factor CLOCK in the liver tissues were significantly decreased in WT-HFD and KO-HFD groups. This sharp decrease was supposed to be caused by HFD. In contrast, no significant change in the other core circadian factor BMAL1 was observed in challenged groups. Different from the data in the liver tissues, the CLOCK was not down-regulated by HFD in the hypothalamus, in either the WT mice or the KO mice. Additionally, the BMAL1 level kept unchanged in both WT-HFD and KO-HFD groups compared with the control group, similarly as those in liver tissues (Fig. 4A). Therefore, the changes of core clock protein CLOCK caused by HFD mainly occurred in liver rather than hypothalamus.
After the protein level analysis, the transcriptional regulatory circadian factors Clock and Bmal1 were analyzed. Fig. 3 HFD modified circadian rhythm in mice. Eight-week-old WT and KO mice were assigned to Con and HFD groups (n = 5). Mice in both groups were fed ND for the first week, and those in the HFD groups were fed diet containing 45%HFD for another 2 weeks. A, B Measurements of activity monitored over a 48-h period in ND-fed and HFD-fed group for WT and KO mice. C, D Measurements of EE (normalized to body weight) monitored over a 48-h period in different group for WT and KO mice. E, F Average measurements of activity and EE (normalized to body weight) monitored between dark and light phases in two mouse lines. Data are expressed as mean ± SD. Single asterisk indicated p < 0.05, in comparison with WT-Con/ WT-Con. # indicated p < 0.05, in comparison with WT-HFD. Difference in each group was determined by one-way ANOVA and Bonferroni post hoc test Different from the tendency of protein level, the mRNA levels of Clock and Bmal1 in the hepatic tissue of HFDtreated group were not modified in comparison with the control group in either the WT mice or the KO mice (Fig. 4B,  C). To avoid false negative results, another two primers in different exons were designed to explore Clock gene transcription. The results also showed no modification of Clock transcription (Fig. S3A). Thus, the modification of CLOCK by HFD occurred at protein degradation rather than gene transcription. In the determination of PPARα activity, the mRNA levels of PPARα target gene Cyp4a10 and Ehhadh were not modified (Fig. 4D, E). More importantly, CLOCK in WT mice was reduced compared with KO mice under both Con and HFD treatments (Fig. 4A). This indicated that CLOCK was down-regulated by basal activity of PPARα at protein level.

PPARα activation down-regulated CLOCK and inhibited lipid accumulation
Considering the action of basal PPARα on CLOCK level, three dose levels of fenofibrate were administered to WT, KO, and hPPARα mice to activate PPARα. With the increase of fenofibrate dose level, the hepatic CLOCK was downregulated in WT and hPPARα mice, but it was not modified in KO mice. Similarly, there was no significant change of BMAL1 in any group (Figs. 5A and S4A). And the mRNA levels of Clock and Bmal1 were not modified at different concentrations of fenofibrate in three mouse lines (Figs. 5B, C and S4B, C). These results were consistent with the Clock mRNA transcription on different exons (Fig. S3B). The mRNA levels of PPARα target gene Cyp4a10 and Ehhadh were significantly elevated in dose-independent manner in hPPARα mice, not in KO mice (p < 0.05) (Fig. 5D, E).
In mice were treated with CDAHFD and fenofibrate, fenofibrate significantly inhibited steatosis induced by CDAHFD (Fig. 6A). The mRNA levels of PPARα target gene Cyp4a10 and Ehhadh were significantly elevated in H + FENO group (p < 0.05) (Fig. 6B, C). Both CDAHFD and H + FENO mice down-regulated CLOCK, and the latter was more obvious in which inhibition of steatosis was observed (Fig. 6D).
HepG2 cells were firstly stimulated with PPARα agonist WY-14643 (0, 10, 50, 200 uM) for 30 h. Consistent with the results of in vivo experiments, CLOCK protein was down-regulated in a concentration dependence of WY14643. Besides, CLOCK level was significantly inhibited under both the stimulation of OA and WY14643 (50 uM) in both HepG2 cells and primary hepatocytes (Fig. 7A). Lipid accumulation occurred in both HepG2 cells and primary hepatocytes stimulated by OA. WY14643 significantly reduced the area of lipid droplets (p < 0.05) (Fig. 7B-D). These data

Discussion
NAFLD is a disease characterized by accumulation of TG in hepatocytes without excess alcohol intake [14]. The  prevalence of NAFLD is increasing in tandem with the incidence of obesity due to high-calorie diets and sedentary lifestyles [26]. Importantly, metabolic disorders are one of the strongest independent risk factors for developing NAFLD [27]. But few experiments compared NAFLD induced by diets with different fat content. In this study, both 45%HFD and 60%HFD successfully induced body weight gain, impaired glucose tolerance, and higher liver TC and TG. In terms of fatty liver, 60%HFD induced less ballooning and steatosis, compared with 45%HFD groups. This occurred probably because the diet taste decreased too much and the diet consumption decreased, when the fat percent increased.
HFD had a deleterious effect on the circadian system of rodents by blunting the feeding/fasting cycle, where the peripheral clock was sensitive to the composition and timing of food [28]. The expression of Per2 and Bmal1 genes increased in liver and brown adipose tissue after restricted feeding on HFD, and the oscillation amplitude of circadian rhythm increased [29]. Switch from low-fat diet to HFD was reported to affect both the central and peripheral clocks in human beings. And the dietary fat altered the functions of the central and peripheral biological clocks [30]. In this study, the core circadian factor CLOCK in the liver not hypothalamus was significantly down-regulated, under the HFD treatment. This suggested that modification of the peripheral clock by HFD was uncoupled from the central clock. Accordingly, the hepatic clock is more important in development of NAFLD.
The CLOCK/BMAL1 heterodimer was involved in the regulation of liver metabolism as a transcription factor [31]. It was once reported that the HFD only minimally modified the circadian gene expression in mouse visceral adipose tissue and liver tissues [32]. However, the opposite results were also reported. The Clock amplitude was significantly reduced in adipose tissue, but little in the liver. Additionally, the rhythmic amplitude of Bmal1 was decreased in adipose tissue and the liver after consumption of HFD [33]. In this study, however, the core circadian factor CLOCK, not BMAL1 of liver tissues, was significantly reduced in both WT-HFD and KO-HFD groups. Similar as the report mentioned above, HFD affected quite little in the transcription of clock transcription. Therefore, it was supposed that the core circadian factor CLOCK was modified directly by HFD, and it occurred in protein level rather than the gene transcription.
PPARα regulated lipid metabolism and inhibited the occurrence of fatty liver. PPARα agonist, fenofibrate and and primary hepatocytes (100 ×). C, D Relative oil red positive area in HepG2 and primary hepatocytes. Data were expressed as mean ± SD. Single asterisk indicated p < 0.05, in comparison with Con group. # indicated p < 0.05, in comparison with CDAHFD group. Difference in each group was determined by one-way ANOVA and Bonferroni post hoc test WY14643, promoted fatty acid oxidation, reversed adipocyte differentiation, and reduced fat storage [34,35]. Interestingly, protective effects of PPARα against fatty liver were observed, since the steatosis was attenuated in WT mice, agreeing with previously reported [8]. However, the underlying mechanism remained to be investigated. Current studies on the mechanism of PPARα inhibiting of NAFLD focused on anti-inflammatory, antioxidant stress, and other aspects. In this study, basal PPARα mediated CLOCK down-regulation contributed to tolerance against NAFLD. Accordingly, basal PPAR was associated with tolerance to NAFLD development regardless of ND or HFD treatment (Fig. 8).
PPARα and circadian factors were reported to regulate each other. The rhythm of PPARα expression disappeared when Clock was knocked out [36]. Similarly, oscillation of PER3 and BMAL1 was changed in PPARα-null mice [37]. CLOCK/ BMALl could bind the E-box of the second intron of Pparα to activate its expression and regulate downstream lipid metabolism [20]. When PPARα was activated, CLOCK was downregulated dose-dependently, and the lipid accumulation was inhibited. As expected, the transcription of CLOCK was not modified by activated hPPARα either. This modification also occurred in protein level, like the modification by HFD mentioned above. However, the molecular mechanisms under this modification independent of transcription remain to be investigated. Since the in vivo response was triggered by PPARα activator fenofibrate in humanized model, it was supposed to be of good relevance in the clinic.