Bitter Orange (Citrus aurantium Linné) Improves Obesity by Regulating Adipogenesis and Thermogenesis through AMPK Activation

Obesity is a global health threat. Herein, we evaluated the underlying mechanism of anti-obese features of bitter orange (Citrus aurantium Linné, CA). Eight-week-administration of CA in high fat diet-induced obese C57BL/6 mice resulted in a significant decrease of body weight, adipose tissue weight and serum cholesterol. In further in vitro studies, we observed decreased lipid droplets in CA-treated 3T3-L1 adipocytes. Suppressed peroxisome proliferator-activated receptor gamma (PPARγ) and CCAAT/enhancer binding protein alpha indicated CA-inhibited adipogenesis. Moreover, CA-treated primary cultured brown adipocytes displayed increased differentiation associated with elevation of thermogenic factors including uncoupling protein 1 and PPARγ coactivator 1 alpha as well. The effects of CA in both adipocytes were abolished in AMP-activated protein kinase alpha (AMPKα)-suppressed environments, suggesting the anti-adipogenic and pro-thermogenic actions of CA were dependent on AMPKα pathway. In conclusion, our results suggest CA as a potential anti-obese agent which regulates adipogenesis and thermogenesis via AMPKα.


Introduction
Obesity (body mass index, BMI > 30 kg/m 2 ) has risen into a serious health threat worldwide, especially in developed countries [1]. According to an analytic study of 19.2 million participants from 1975 to 2014, age-standardized BMI increased from 21.7 to 24.2 kg/m 2 in men and 22.1 to 24.4 kg/m 2 in women globally. Speculations based on this trend expects obesity prevalence of 2025 as 18% in men and 21% in women worldwide [2]. This global trend in concerning, as obesity is associated with high risks of comorbidities such as type 2 diabetes mellitus [3], cardiovascular diseases [4] and even certain cancers [5].
Obesity is a complex status in which multiple factors participate. Above all, when energy consumption exceeds expenditure, the surplus energy is stored as lipid, specifically triglyceride, in white adipose tissues (WATs) [6]. While WAT works as a storage unit to save excessive energy, brown adipose tissue (BAT), on the other hand, functions in an opposite way. BAT is an organ which can dissipate energy as heat in response to cold exposure or other stimuli such as pharmacological

Oil Red O Staining
Intracellular lipid accumulation was measured by an Oil Red O staining assay as described previously [29]. The absorbance was measured at 500 nm in a VERSAmax microplate reader (Molecular Devices, Sunnyvale, CA, USA).

RNA Extraction and Real-Time Reverse Transcription Polymerase Chain Reaction (RT-PCR)
RNA was extracted using a GeneAllR RiboEx total RNA extraction kit (GeneAll Biotechnology, Seoul, Republic of Korea), cDNA reverse-transcription was performed using a Power cDNA synthesis kit (iNtRON Biotechnology, Seongnam, Republic of Korea), and Real-Time RT-PCR was performed in a Step One Real-Time PCR System (Applied Biosystems, Waltham, MA, USA). as previously reported [31]. The primers used in this study are shown in Supplementary Table S2.

Protein Extraction and Western Blot Analysis
Cells were harvested and then lysed in ice-cold radioimmunoprecipitation assay (RIPA) buffer. The protein concentration was determined using a protein assay reagent (Bio-Rad Laboratories Inc., Hercules, CA, USA). Western blot analysis was performed as previously described [31]. Briefly, equal amounts (20 µg) of total protein were resolved by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (10-12% based on the size of detecting protein) and transferred to a polyvinylidene difluoride (PVDF) membrane, which were incubated with primary antibodies (overnight, 4 • C), and then incubated with the proper horseradish peroxidase-conjugated secondary antibody diluted at 1:10000 dilution (1 h, room temperature (RT)) (Jackson Immuno Research, West Grove, PA, USA).

Mitochondrial Microscopic Analysis and Immunofluorescence Staining
The mitochondrial analysis was performed using a Mito-Tracker Red probes CM-XRos (Invitrogen, Carlsbad, CA, USA) as previously described [30]. Fluorescence signal images were obtained using an IX71 confocal microscope (Zeiss, Oberkochen, Germany).
The immunofluorescence staining was performed as previously described [33]. Briefly, after treatment with or without CA, brown adipocytes were fixed with 4% formaldehyde and permeabilized with PBS containing 0.25% Triton X-100. After blocking non-specific binding with 5% BSA/PBS, the cells were incubated with UCP1 and PGC1α antibodies in 5% BSA/PBS (overnight, 4 • C), and incubated with fluorescent secondary antibody Alexa Fluor 488 and Alexa Fluor 546 (1 h, RT). Nuclei staining was performed by incubation with 4 ,6-diamidino-2-phenylindole (DAPI). The immunofluorescence-stained images were acquired using a fluorescence microscope (Logos Biosystems, Anyang, Republic of Korea).

Statistical Analysis
All data were expressed as mean ± S.E.M., and processed statistically using SPSS 20 for Windows (SPSS Inc., Chicago, IL, USA). Values with p < 0.05 calculated by a Kruskal-Wallis H test followed by a post hoc test of Bonferroni's method was considered statistically significant.

Chromatographic Characterization of CA
The immature dried fruits of Citrus aurantium Linné were extracted by 80% ethanol. The result from HPLC chromatogram showed that CA contains two abundant compounds, naringin and neohesperidin ( Figure 1). The naringin and neohesperidin was 20.132% (0.916 mg/mL) and 14.440% (0.657 mg/mL) of the 80% ethanol extract of CA 4.55 mg/mL, respectively. Figure 1. HPLC chromatogram of CA. Each peak displays CA (above) and naringin and neohesperidin (below). CA was extracted with 80% ethanol. CA, bitter orange (Citrus aurantium Linné).

CA Suppressed Body Weight Gain in HFD-induced Obese C57BL/6 Mice
To investigate whether CA can regulate weight of HFD-fed mice, after inducing obesity with HFD for 4 weeks, CA of 100 mg/kg/day with HFD was orally administered to C57BL/6 mice for additional 8 weeks. As shown in Figure 2A, HFD group and HFD plus CA group showed significant differences in the gain of body weight at eleventh week (24.89 ± 0.21 g vs. 15.08 ± 0.97 g). At end of the experiment, the body weights of the two groups were showed significant difference (50.31 ± 0.84 g vs. 40.91 ± 0.91 g, respectively), adipose tissue weights decreased by 35.64% in CA-treated obese mice (HFD, 0.56 ± 0.14 g vs. CA, 0.36 ± 0.08 g, p = 0.0143) ( Figure 2B), and the serum total cholesterol level was significantly attenuated by CA in HFD-fed mice ( Figure 2C) (p < 0.05).

CA Suppressed Body Weight Gain in HFD-induced Obese C57BL/6 Mice
To investigate whether CA can regulate weight of HFD-fed mice, after inducing obesity with HFD for 4 weeks, CA of 100 mg/kg/day with HFD was orally administered to C57BL/6 mice for additional 8 weeks. As shown in Figure 2A, HFD group and HFD plus CA group showed significant differences in the gain of body weight at eleventh week (24.89 ± 0.21 g vs. 15.08 ± 0.97 g). At end of the experiment, the body weights of the two groups were showed significant difference (50.31 ± 0.84 g vs. 40.91 ± 0.91 g, respectively), adipose tissue weights decreased by 35.64% in CA-treated obese mice (HFD, 0.56 ± 0.14 g vs. CA, 0.36 ± 0.08 g, p = 0.0143) ( Figure 2B), and the serum total cholesterol level was significantly attenuated by CA in HFD-fed mice ( Figure 2C) (p < 0.05).

CA Decreased Lipid Accumulation by Inhibiting Adipogenesis of 3T3-L1 White Adipocytes
First, to determine the concentration dependent effect of CA (10-1000 µg/mL) on cell viability, 3T3-L1 preadipocytes were applied to MTS reagent. Treatment with 10-1000 µg/mL CA for 48 h did not display any cytotoxic effect in 3T3-L1 cells (p > 0.05, Figure 3A), thus we chose concentrations of 100, 100, and 1000 µg/mL for further investigations. Next, to determine whether CA can block white adipogenesis, 3T3-L1 preadipocytes were differentiated with MDI, and the effects of CA were analyzed. As shown in Figure 3B, CA (1000 µg/mL) suppressed lipid accumulation, and this was confirmed by the reduced absorbance of Oil Red O (p < 0.05). At the same time, the mRNA expression involved in white adipogenesis (Pparg, Cebpa, Fabp4, Adipoq and Retn), which were markedly increased during MDI-induced adipogenesis, were attenuated by CA (1000 µg/mL of CA concentration; p = 0.032 for Pparg, p = 0.041 for Cebpa, p = 0.004 for Fabp4, p = 0.002 for Adipoq, and p = 0.001 for Retn, respectively) ( Figure 3C). In addition, CA treatment reduced levels of PPARγ and C/EBPα, which are both well-known key regulators of adipocyte differentiation ( Figure 3D). diet-fed control group; HFD, high fat diet-fed obese group; CA, high fat diet supplemented with 100 mg/kg/day bitter orange (Citrus aurantium Linné) -fed obese group.

CA Decreased Lipid Accumulation by Inhibiting Adipogenesis of 3T3-L1 White Adipocytes
First, to determine the concentration dependent effect of CA (10-1000 μg/ml) on cell viability, 3T3-L1 preadipocytes were applied to MTS reagent. Treatment with 10-1000 μg/ml CA for 48 h did not display any cytotoxic effect in 3T3-L1 cells (p > 0.05, Figure 3A), thus we chose concentrations of 100, 100, and 1000 μg/ml for further investigations. Next, to determine whether CA can block white adipogenesis, 3T3-L1 preadipocytes were differentiated with MDI, and the effects of CA were analyzed. As shown in Figure 3B, CA (1000 µ g/ml) suppressed lipid accumulation, and this was confirmed by the reduced absorbance of Oil Red O (p < 0.05). At the same time, the mRNA expression involved in white adipogenesis (Pparg, Cebpa, Fabp4, Adipoq and Retn), which were markedly increased during MDI-induced adipogenesis, were attenuated by CA (1000 µ g/ml of CA concentration; p = 0.032 for Pparg, p = 0.041 for Cebpa, p = 0.004 for Fabp4, p = 0.002 for Adipoq, and p = 0.001 for Retn, respectively) ( Figure 3C). In addition, CA treatment reduced levels of PPARγ and C/EBPα, which are both well-known key regulators of adipocyte differentiation ( Figure 3D). PPARγ and C/EBPα protein levels were analyzed by a western blot analysis. Quantification of the protein bands was measured using Image J. Results were expressed relative to GAPDH. All data are presented as the mean ± SEM. # p < 0.05 versus the MDI-uninduced preadipocytes, * p < 0.05 versus the MDI-induced adipocytes, ** p < 0.01 vs. the MDIinduced adipocytes. MDI, differentiation medium; CA, bitter orange (Citrus aurantium Linné).

CA Increased Differentiation of Primary Cultured Brown Adipocytes
Brown fat specializes in energy expenditure to reduce weight gain through thermogenesis, thus, we evaluated the effect of CA on thermogenesis to explore their involvement with respect to weight loss in HFD-induced obese mice. First, we assessed the cytotoxicity of CA (10-1000 μg/ml) in primary PPARγ and C/EBPα protein levels were analyzed by a western blot analysis. Quantification of the protein bands was measured using Image J. Results were expressed relative to GAPDH. All data are presented as the mean ± SEM. # p < 0.05 versus the MDI-uninduced preadipocytes, * p < 0.05 versus the MDI-induced adipocytes, ** p < 0.01 vs. the MDI-induced adipocytes. MDI, differentiation medium; CA, bitter orange (Citrus aurantium Linné).

CA Increased Differentiation of Primary Cultured Brown Adipocytes
Brown fat specializes in energy expenditure to reduce weight gain through thermogenesis, thus, we evaluated the effect of CA on thermogenesis to explore their involvement with respect to weight loss in HFD-induced obese mice. First, we assessed the cytotoxicity of CA (10-1000 µg/mL) in primary cultured brown adipocytes obtained from interscapular BAT. As in Figure 4A, no specific decrease in brown adipocyte viability was observed in treatment with 10-1000 µg/mL CA for 48 h. We next investigated whether CA treatment affected the development of primary cultured brown adipocytes Nutrients 2019, 11, 1988 8 of 16 by treating them with CA (10 or 1000 µg/mL) during brown adipocyte differentiation. According to our data, the lipid accumulation at a concentration of 1000 µg/mL of CA were obviously increased (p < 0.05) than that of the non-treated control during brown adipocyte differentiation ( Figure 4B). investigated whether CA treatment affected the development of primary cultured brown adipocytes by treating them with CA (10 or 1000 µ g/ml) during brown adipocyte differentiation. According to our data, the lipid accumulation at a concentration of 1000 µ g/ml of CA were obviously increased (p < 0.05) than that of the non-treated control during brown adipocyte differentiation ( Figure 4B).
To explore whether the brown adipocyte function agreed to the effect of CA on differentiation, we further focused on the mitochondrial number or mass. We stained brown adipocytes with Mito-Tracker Red, which is a red fluorescent dye that specifically stains mitochondria [34], to observe significantly strong signals in the cytoplasm of CA-treated cells (p = 0.001), which indicates increased mitochondrial copy number ( Figure 4C).

CA Induced Thermogenic Factors in Primary Cultured Brown Adipocytes
Multiple factors participate in the differentiation and thermogenic activation of BAT [17]. Our next goal was to elucidate the effect of CA on the related factors in such process. As expected, CA enhanced the mRNA levels of both thermogenic and mitochondrial specific genes, including Pgc1a, Ucp1, Prdm16, Sirt3, Cidea, and CytC (1000 µ g/ml of CA concentration; p = 0.043 for Pgc1a, p = 0.050 for Ucp1, p = 0.01 for Prdm16, p = 0.050 for Sirt3, p = 0.021 for Cidea, and p = 0.012 for CytC) ( Figure 5A). We also analyzed protein levels of UCP1 and PGC1α, which were significantly increased by CA at 1000 µ g/ml compared to non-treated control during differentiation ( Figure 5B), demonstrating the thermogenic effects of CA. After observing an upregulation of UCP1 and PGC1α, we further examined the expression levels of those by using immunofluorescence assay. We observed that the fluorescent protein levels of UCP1 and PGC1α were markedly induced by CA treatment (1000 μg/ml) during brown adipocytes differentiation (p = 0.001 for PGC1α and p = 0.043 for UCP1) ( Figure 5C). To explore whether the brown adipocyte function agreed to the effect of CA on differentiation, we further focused on the mitochondrial number or mass. We stained brown adipocytes with Mito-Tracker Red, which is a red fluorescent dye that specifically stains mitochondria [34], to observe significantly strong signals in the cytoplasm of CA-treated cells (p = 0.001), which indicates increased mitochondrial copy number ( Figure 4C).

CA Induced Thermogenic Factors in Primary Cultured Brown Adipocytes
Multiple factors participate in the differentiation and thermogenic activation of BAT [17]. Our next goal was to elucidate the effect of CA on the related factors in such process. As expected, CA enhanced the mRNA levels of both thermogenic and mitochondrial specific genes, including Pgc1a, Ucp1, Prdm16, Sirt3, Cidea, and CytC (1000 µg/mL of CA concentration; p = 0.043 for Pgc1a, p = 0.050 for Ucp1, p = 0.01 for Prdm16, p = 0.050 for Sirt3, p = 0.021 for Cidea, and p = 0.012 for CytC) ( Figure 5A). We also analyzed protein levels of UCP1 and PGC1α, which were significantly increased by CA at 1000 µg/mL compared to non-treated control during differentiation ( Figure 5B), demonstrating the thermogenic effects of CA. After observing an upregulation of UCP1 and PGC1α, we further examined the expression levels of those by using immunofluorescence assay. We observed that the fluorescent protein levels of UCP1 and PGC1α were markedly induced by CA treatment (1000 µg/mL) during brown adipocytes differentiation (p = 0.001 for PGC1α and p = 0.043 for UCP1) ( Figure 5C). and PGC1α protein levels were analyzed by a western blot analysis. Quantification of the protein bands was measured using Image J. Results were expressed relative to GAPDH. (C) Immunofluorescence staining of UCP1 (red) and PGC1α (green) were performed (magnification 200×, scale bar 100 μm). Quantification of the immunofluorescence signaling was measured using Image J from five separate slides. All values are means ± SEM of data from three separate experiments. # p < 0.05 versus DM-uninduced preadipocytes; and * p < 0.05 and ** p < 0.01 versus DM-induced adipocytes. DM, differentiation medium; CA, bitter orange (Citrus aurantium Linné).
The recruitment of functional brown adipocytes, also called beige or brite adipocytes, is another rising target for obesity treatment [35]. However, in contrast to results from brown adipocytes, CA failed to induce the thermogenic potential in mature white adipocytes. There was no significant upregulation in both UCP1 and PGC1α by CA treatment in 3T3-L1 adipocytes (Supplementary Figure  S1).

CA Regulated Adipogenesis and Thermogenesis via AMPK Activation
AMPK, the energy sensor and also the master regulator of metabolic homeostasis, acts to increase glucose uptake, fatty acid oxidation, and mitochondrial biogenesis [36]. In addition, AMPK is involved in both thermogenic activation of BAT and browning of WAT as well as conversion of preadipocytes to mature fat cells [37]. Based on these reports, to elucidate whether the effect of CA on promoting adipogenesis and thermogenesis was dependent on the activation of AMPK, 3T3-L1 adipocytes and primary adipocytes were treated with CA at indicated concentration during differentiation. As a result, CA (1000 µ g/ml) in both 3T3-L1 cells and primary brown adipocytes enhanced the levels of pLKB1, the upstream signal of AMPK, pAMPKα, and pACC, which is phosphorylated by AMPK [38]. without significant changes in total AMPK (p < 0.001 for pAMPKα Quantification of the immunofluorescence signaling was measured using Image J from five separate slides. All values are means ± SEM of data from three separate experiments. # p < 0.05 versus DM-uninduced preadipocytes; and * p < 0.05 and ** p < 0.01 versus DM-induced adipocytes. DM, differentiation medium; CA, bitter orange (Citrus aurantium Linné).
The recruitment of functional brown adipocytes, also called beige or brite adipocytes, is another rising target for obesity treatment [35]. However, in contrast to results from brown adipocytes, CA failed to induce the thermogenic potential in mature white adipocytes. There was no significant up-regulation in both UCP1 and PGC1α by CA treatment in 3T3-L1 adipocytes (Supplementary Figure S1).

CA Regulated Adipogenesis and Thermogenesis via AMPK Activation
AMPK, the energy sensor and also the master regulator of metabolic homeostasis, acts to increase glucose uptake, fatty acid oxidation, and mitochondrial biogenesis [36]. In addition, AMPK is involved in both thermogenic activation of BAT and browning of WAT as well as conversion of preadipocytes to mature fat cells [37]. Based on these reports, to elucidate whether the effect of CA on promoting adipogenesis and thermogenesis was dependent on the activation of AMPK, 3T3-L1 adipocytes and primary adipocytes were treated with CA at indicated concentration during differentiation. As a result, CA (1000 µg/mL) in both 3T3-L1 cells and primary brown adipocytes enhanced the levels of pLKB1, the upstream signal of AMPK, pAMPKα, and pACC, which is phosphorylated by AMPK [38]. without significant changes in total AMPK (p < 0.001 for pAMPKα and p = 0.001 for pACC) ( Figure 6A). Phosphorylated AMPKα by CA was successfully blocked after an AMPK inhibitor CC treatment in the both cells ( Figure 6B,C). In addition, perturbed induction of PPARγ and C/EBPα by CA during the differentiation was reversed by CC treatment in 3T3-L1 cells. Concurrently, the CA-induced PGC1α and UCP1 expression were attenuated by CC treatment in primary brown adipocytes ( Figure 6D,E). These results denote that adipocytes with AMPK deficiency is specifically affected in the expression of adipogenesis-and thermogenesis-related genes and proteins in both 3T3-L1 adipocytes and primary cultured brown adipocytes.
Nutrients 2019, 11, x FOR PEER REVIEW 10 of 16 and p = 0.001 for pACC) ( Figure 6A). Phosphorylated AMPKα by CA was successfully blocked after an AMPK inhibitor CC treatment in the both cells ( Figure 6B and C). In addition, perturbed induction of PPARγ and C/EBPα by CA during the differentiation was reversed by CC treatment in 3T3-L1 cells. Concurrently, the CA-induced PGC1α and UCP1 expression were attenuated by CC treatment in primary brown adipocytes ( Figure 6D and E). These results denote that adipocytes with AMPK deficiency is specifically affected in the expression of adipogenesis-and thermogenesis-related genes and proteins in both 3T3-L1 adipocytes and primary cultured brown adipocytes.

Discussion
Since obesity is a common chronic disease that results from dysregulated adipogenesis caused by imbalance of food intake and EE, therapeutic strategies targeting regulation of adipogenic differentiation or increase EE are both attractive approaches for combating obesity [39,40]. Generally, classic brown adipocytes increase EE by producing heat, which protects against the development of excess lipid accumulation, and eventually prevents overweight and obesity. Approaches to elevate the thermogenic activity of brown and beige adipocytes are potentially promising new strategies for preventing and treating obesity because the activity of these thermogenic adipocytes is inversely correlated with fat mass and positively related to EE [41].
Medications for obesity treatment vary, but only five are currently approved by the Food and Drug Administration (FDA) of USA [42]. However, the search for alternative treatments is still an ongoing challenge due to the various side effects of current medications. In context, some natural compounds such as polyphenol and flavonoids offer an alternative therapeutic strategy, as they can reduce the accumulation of lipid by inhibiting lipogenesis or inducing lipolysis, and subsequently decrease adiposity [43], or induce EE by increasing UCP1 in BAT [30,33]. However, there has been little evidence that CA can directly affect adipogenesis of white adipocytes and thermogenesis of brown adipocytes which are both attractive targets of anti-obese strategy. The current study evaluated the effect of CA on adipocyte differentiation in WAT and thermogenic capacity of BAT in both 3T3-L1 white adipocytes and primary cultured brown adipocytes. Our findings revealed that administration of CA significantly reduced body weight in obese mice, suggesting that CA may be potential option as an intervention to fight obesity.
Adipogenesis is a complex process of which several factors involve. Among these factors, PPARγ and C/EBPα are considered as the most crucial regulators of adipogenesis [11]. They mutually induce expressions, cooperate as transcription factors, and eventually leads to synergistic activation of other adipocyte factors [12]. In our study, yet the effect of CA on lipid accumulation may not seemed to be impressive (17.6% decrease at 1000 µg/mL concentration), but the adipogenesis regulators, PPARγ and C/EBPα, were suppressed by 21.6% and 34.6% respectively by CA treatment (1000 µg/mL), indicating a potentially strong effect on the adipogenesis cascade. The results indicated that the CA-reduced lipid accumulation is attributed to a decrease in these crucial factors associated with white adipogenesis. In addition, the genes those encode adipokines such as adipocyte protein 2 (aP2), adiponectin and resistin, were significantly suppressed by CA treatment, indicating reduced adipocyte development.
Considering the benefits and safety of BAT-mediated EE, this unique tissue can be the core of next generation for anti-obesity agents. Several factors are related to the process of non-shivering thermogenesis [44]. UCP1 in the inner membrane of mitochondria is preferably abundant in brown adipocytes. By uncoupling oxidative phosphorylation from ATP production, UCP1 converts triglyceride to heat [45]. The PR-domain containing 16 (PRDM16), transcribed from the gene Prdm16, is the inducer of the thermogenic phenotype, and activates the brown fat program in adipocytes via Pgc1a, Cidea and CytC [45,46]. On the other hand, sirtuin 3 is a member of the sirtuin family of protein deacetylases which preferentially localize in mitochondria [47,48], and is recently recognized as an important regulator of non-shivering thermogenesis program [49]. We found that CA promotes activation of BAT function and differentiation by upregulation of genes related to BAT thermogenesis, mitochondrial biogenesis and mitochondrial activation such as Pgc1a, Ucp1, Prdm16, Sirt3, Cidea and CytC, thereby suppress obesity by promoting EE. However, in the concept of 'browning', a rising concept of approach in obesity care of which strategy is aimed to recruit functional brown adipocytes within WAT [50], CA failed to alter any thermogenic factors including UCP1 and PGC1α. This allowed us to conclude CA may benefit obesity by suppressing adipogenesis of white adipocytes but not inducing 'browning' of it. The thermogenic action of CA may only happen in BAT, yet it is known that UCP1 activation in BAT can systemically affect lipolysis of WAT to produce fatty acid for thermogenic fuel supply [51].
According to previous reports, CA displays potentially beneficial effects on metabolic diseases such as obesity and type 2 diabetes [25][26][27][28]. However, the underlying mechanism of the anti-obese effect of CA remained obscure so far. In this study, based on the crucial role of AMPK in metabolic diseases, we attempted to reveal the effect of CA on adipogenesis and thermogenesis and assess the role of AMPK in this process. AMPK is a serine/threonine kinase of a heterodimer complex comprising a catalytic α subunit (α1, α2) and two regulatory subunits β (β1, β2) and γ (γ1, γ2, γ3). When AMPKα is phosphorylated, ATP consumption (anabolism) is inhibited and ATP production (catabolism) are activated. The overall effect of AMPKα activation is therefore to synthetize ATP and restore AMP:ATP and ADP:ATP ratios for the maintenance of cellular energy homeostasis [52][53][54]. AMPK acts as a regulator of white adipogenesis, brown adipogenesis, and brown adipocyte activation. The different role of AMPK in these three processes has been established previously. In white adipocyte differentiation (or adipogenesis), AMPK negatively regulates this process. Treatment of AICAR, the AMPK activator, in 3T3-L1 or F442A preadipocytes led to inhibition of differentiation accompanied by decreased PPARγ and C/EBPα [37,55]. On the other hand, in brown adipocytes, AMPK is essential for their development. Deletion of AMPK β subunits or AMPKα1 in mice resulted in defected brown adipocyte development [56,57]. Moreover, AMPK plays a crucial role in the activation of non-shivering thermogenesis in brown fat as well [58,59]. Thus, we attempted to identify whether AMPK pathway is involved in the action of CA. In consistent with previous reports on AMPK, we could observe that CA elevated phosphorylation levels of AMPKα and ACC both in 3T3-L1 adipocytes and primary cultured brown adipocytes. Furthermore, inhibition of AMPKα pathway by pre-treatment of CC lead to abolishment of the anti-adipogenic and pro-thermogenic effect of CA. Thus, we could assume the anti-obese effect of CA was dependent on the activation of AMPKα pathway.
Hit identification is considered as the basic level of small molecule discovery. Thus, it is necessary to identify effective constituents which can represent the effects of natural products. CA contains many compounds, including p-Octopamine and synephrine alkaloids. [26] The flavonoids, also isolated from CA, of which hesperidin and naringenin are major active compounds have been used to treat cardiovascular diseases [60]. Through and HPLC analysis, we identified two constituents which may possibly be responsible for the effect of CA. Naringin is a component derived from most citrus fruits including grapefruit and CA. Recent studies by Pu et al. [61] and Sui et al. [62] has shown that naringin can activate AMPK and thereby reduced body weight in HFD-fed obese C57BL/6J mice. Neohesperidin, another well-known compound of the Citrus genus, also has an AMPK-dependent lipid-regulating effect in vivo and vitro [63]. Further studies reported the anti-adipogenic or anti-obese effect of naringin [64][65][66][67] and neohesperidin [68,69], nevertheless, none of these have established the effect of either compound on UCP1-mediated thermogenesis. Our LC-MS data suggests that these two constituents may be responsible for the AMPKα-dependent anti-adipogenic and thermogenic effect of CA, however, to clarify, further investigation should be carried out with different fractions and/or identified constituents.

Conclusions
In conclusion, our study revealed the underlying mechanism of the anti-adipogenic effect of CA, of which AMPKα acts as a crucial factor. Moreover, a new therapeutic possibility of CA; induction of non-shivering thermogenesis, was suggested by our study regarding primary cultured brown adipocytes. The activation of brown adipocytes was also dependent on AMPKα as well. Overall, we suggest CA as a new potential anti-obese agent which can inhibit white adipogenesis and induce brown adipocyte thermogenesis via activation of AMPKα.