Introduction

Obesity is associated with disorders such as hypertension, dyslipidemia, endothelial dysfunction, and diabetes [1, 2]. It is mainly characterized by excessive deposition of adipose tissues, and induction of lipolysis is considered as a targeted therapy for the prevention and improvement of obesity [3, 4]. White adipose tissue (WAT) stores excess energy as triacylglycerol (TG) during fasting, and plays a key role in energy metabolism by providing free fatty acids (FFA) and glycerol through hydrolysis of TG [5]. In contrast, brown adipose tissue (BAT) is specialized for non-shivering thermogenesis through fat burning and high expression of mitochondrial uncoupling protein 1 (UCP1), which uncouples respiration and dissipates chemical energy as heat [6, 7].

Intriguingly, rediscovery of BAT in adult humans has led to a renaissance in the study of this tissue as a target for the development of therapeutic treatments for obesity [8]. Besides classical brown adipocytes, beige or “brite” (brown in white) adipocytes are developmentally distinct and are formed from transdifferentiation of mature white adipocytes or differentiation of brown adipogenic progenitors [9, 10]. Transcriptional regulators important for controlling expression of the BAT phenotype include PRDM16, PGC-1α, and PPARγ [9]. Among them, PGC-1α controls mitochondrial biogenesis and heat production by inducing expression of UCP1 and other respiratory factors. Most studies on induced browning of white adipocyte have been conducted under cold exposure and pharmacological activation of β3-adrenergic receptors in WAT via promotion of WAT remodeling by adipocyte progenitors [11, 12].

Meanwhile, there has been considerable investigation of pharmacological nutritional/dietary agents as well as signaling pathways that contribute to browning of WAT [13]. Several dietary compounds such as capsaicin, berberine, orexin, and irisin have been reported as browning promoters by energy expenditure and lipolysis [14–16]. Here, we focused on curcumin, a natural compound extracted from rhizomes of Curcuma species, since it is a naturally occurring curcuminoid of turmeric, which is a member of the ginger family (Zingiberaceae) [17]. Moreover, dietary curcumin administration was reported to prevent weight gain and play an anti-inflammatory role in obesity [18]. Our previous study indicated that curcumin along with a browning cocktail could synergistically induce the brite phenotype and drive BAT thermogenesis through significant elevation of brown fat-specific genes [19].

For comprehensive understanding of the role of adipose tissue in obesity, several proteomic studies using rodent adipose tissues have been carried out, and proteins in BAT and WAT from rodents subjected to high fat diet have been compared [20, 21]. Furthermore, comparative proteomic analysis was also carried out during brown preadipocyte differentiation or adipose tissue lipolysis [22, 23]. These studies demonstrated several mechanistic pathways that control expression of thermogenesis proteins in BAT. Previously, we reported that the expression of several thermogenic and lipogenic proteins varies in adipose tissues of obese prone and obese-resistant rats [20]. Another research group compared expression of proteins in WAT and BAT between calorie-restricted and control rats [24]. The results showed that ATP-citrate synthase (ACLY) was up-regulated by caloric restriction, which altered activities of proteins associated with energy metabolism.

In this study, we observed that curcumin has a browning effect on white adipocytes based on its ability to alter expression of brown adipocytes marker proteins in white adipocytes, thereby inducing acquisition of BAT features. To obtain more detail, we performed comparative proteomic analysis of primary white adipocytes between control and curcumin-treated adipocytes, as curcumin and a browning cocktail synergistically induced the brite phenotype. In addition, network analyses suggested that enhanced expression of HSL and other browning-specific markers by curcumin treatment reduces body fat by increasing lipolysis and suppressing lipogenesis by trans-differentiation from white adipocytes into beige adipocytes.

Materials and methods

Culture and differentiation of primary white adipocytes

Ten-day-old sprague-dawley (SD) rats were purchased from Daehan Experiment Animals (Seoul, Korea). Rats were used for isolation of primary stromal vascular fractions (SVFs) from rat inguinal fat pads, cultured, and differentiated into mature primary adipocytes as described previously [25]. In brief, inguinal fat pads were minced with scissors and digested with type I collagenase (1 mg/ml; Sigma-Aldrich) at 37 °C in a water bath shaker. After 90 min, contents were immediately filtered in mesh and centrifuged at 1000×g for 10 min. The layer of floating cells was then washed three times with PBS. Adipocytes were incubated in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10 % FBS and 100 units/ml of penicillin–streptomycin (Sigma-Aldrich) under a 5 % CO2 atmosphere at 37 °C. After reaching confluence (day 1), cells were initiated for adipocyte differentiation by incubation with differentiation-inducing medium consisting of 10 µg/ml of insulin (Sigma, St. Louis, MO, USA), 0.25 µM dexamethasone (Sigma), and 0.5 mM 3-isobutyl-1-methylxanthine (IBMX, Sigma) in DMEM, followed by maturation medium containing 10 % FBS and 10 µg/ml of insulin. During treatments, unless otherwise stated, cells were maintained in complete medium containing 20 µM curcumin for 6–8 days before further analysis, and maturation medium was changed every 2 days. The range of curcumin doses (1–20 µM) used in this study was determined based on previous reports [19, 26]. For induction of adipocyte browning, differentiation-inducing medium was supplemented with 50 nM triidothyronine, whereas maturation medium was supplemented with 50 nM triidothyronine and 1 μM rosiglitazone (browning cocktail).

Preparation of protein samples for two-dimensional electrophoresis (2-DE)

Cells were lysed in 200 µL of rehydration buffer solution containing 7 M urea, 2 M thiourea, 4 % CHAPS, 20 mM DTT, 1 mM PMSF, 2 % IPG buffer (Ampholyte 3/10, Bio-Rad), and a trace of bromophenol blue. Lysed tissues were then homogenized by a homogenizer (PT 1200E, Kinematica, Luzern, Switzerland) on ice, after which extracts from homogenized WAT were centrifuged at 13,000×g for 20 min. The supernatant was then stored at −80 °C until analysis. Protein content of cells was determined using RC DC™ protein assay (Bio-Rad).

2-DE

2-DE was performed in duplicate using protein samples from rat primary white adipocytes, which consisted of control and curcumin-treated cells. 2-DE experiments were conducted using previous methods outlined by our laboratory [20, 27, 28]. Briefly, immobilized pH gradient (IPG)-isoelectric focusing (IEF) of WAT samples was performed on pH 3–10, 18 cm IPG DryStrips (GE Healthcare, Buckinghamshire, UK) in a PROTEIN IEF cell (Bio-Rad) using the protocol recommended by the manufacturer. IPG strips were rehydrated passively for 12 h in strip holders with 350 µL of rehydration solution, which included 7 M urea (Bio Basic, Ontario, Canada), 2 M thiourea (Sigma, St. Louis, MO, USA), 4 % CHAPS (Bio Basic), 1 mM PMSF (Sigma), 20 mM DTT (GE Healthcare), 2 % IPG buffer (Bio-Rad), and 150 µg of primary white adipocyte proteins. IEF was executed as previously described [28]. Gel strips were then placed onto a 20 × 20 cm 12 % polyacrylamide gel for resolution in the second dimension. Fractionation was performed using the Laemmli SDS discontinuous system at a constant voltage of 15 mA per gel for 14 h. For image analysis and peptide mass fingerprinting (PMF), a total of six gels, including two gels per group with separated proteins, were visualized by silver staining [28].

Image acquisition and data analysis

Gels were imaged on a UMAX PowerLook 1120 (Maxium Technologies, Akron, OH, USA), and modified ImageMaster 2-D software V4.95 (GE Healthcare) was used for comparison of images. A reference gel was selected from gels of the control group, and detected spots from other gels were matched with those in the reference gel. Relative optical density and relative volume were calculated in order to correct differences in in-gel staining. Each spot intensity volume was processed by background subtraction and total spot volume normalization, and the resulting spot volume percentage was used for comparison.

Protein identification

For protein identification by PMF, protein spots were excised, digested with trypsin (Promega, Madison, WI, USA), mixed with CHCA in 50 % ACN/0.1 % TFA, and subjected to MALDI-TOF analysis (Microflex LRF 20, Bruker Daltonics). Spectra were collected from 300 shots per spectrum over an m/z range of 600–3000 and calibrated by two-point internal calibration using trypsin auto-digestion peaks (m/z 842.5099, 2211.1046). Peak list was generated using Flex Analysis (ver 3.0). The threshold used for selecting peaks was executed as previously described [28]. Protein score is −10× Log (p), where p is the probability that the observed match is a random event, and greater than 61 is significant (p < 0.05).

Immunoblot analysis

Cell lysates were prepared using RIPA buffer (Sigma) by homogenization and centrifugation at 14,000×g for 20 min. Immunoblot analyses were performed as previously described [28]. Briefly, the membrane was rinsed three times consecutively with TBS-T buffer, followed by incubation for 1 h with 1:500–1000 dilutions of primary monoclonal antibodies against ANXA2, ATP5B, HSP90, CA3, MDH, β-actin, PPARγ, p-AMPK, and HSL (1:200–500, Santa Cruz Biotechnology) in TBS-T buffer containing 1 % skim milk. After three washes, the membrane was incubated for 1 h with horseradish peroxidase-conjugated anti-goat IgG or anti-rabbit IgG secondary antibody (1:1000, Santa Cruz Biotechnology) in TBS-T buffer containing 1 % skim milk. Development was carried out using enhanced chemiluminescence (West Zol, iNtRON Biotechnology). Quantification of band intensities was performed by using ImageJ software (NIH).

Network analysis

For gene/protein classification and network analysis, identified proteins obtained from proteomic analysis together with major fatty acid metabolic proteins were analyzed using the Protein ANnotation THrough Evolutionary Relationship (PANTHER) and Database for Annotation, Visualization, and Integrated Discovery (DAVID) classification systems [29, 30]. These public databases allow the systematic division of large gene lists in an attempt to assemble a summary of related biological functions, pathways, and protein classes. DAVID provides a comprehensive set of functional annotation tools to understand the biological relevance of such a large list of genes/proteins. Identification of gene enrichment categories was carried out according to SP-PIR keywords combining the annotation from both Swiss-Prot (SP) and Protein Information Resource (PIR). Associations of differentially expressed genes/proteins with broadly defined molecular networks were combined and visualized using the STRING database.

Real-time RT-PCR

Transcription levels of genes were quantitatively determined by real-time RT-PCR (Stratagene Mx 3000p QPCR System, Santa Clara, CA, USA). Total RNA was isolated from WAT tissues using an easy-spin TM (DNA-free) Total RNA Extraction kit, and reverse-transcription was carried out using a Maxime RT Premix kit (iNtRON Biotechnology). FastStart Universal SYBR Green Master (Roche Diagnostics, Indianapolis, IN, USA) was employed for real-time RT-PCR. Transcription levels of each gene (Acca2, Atp5b, Atp5h, Cebpb, Cs, Fabp4, Gpd, Hadha, Hspd1, Hspe1, Lipe, Mdh2, Ppard, Pparg, Ppargc1a, Prdm16, Ucp1, Ucp2, Ucp3, and Vdac1) were normalized to Gapdh transcription levels, and relative changes in gene expression were quantified using the ∆∆CT method [31]. Oligonucleotide sequences of the primers used are shown in Table 1.

Table 1 List of primers for RT-PCR used in this study
Full size table

Co-immunoprecipitation (Co-IP)

To identify UCP1/PGC-1α-interacting partner proteins in primary white adipocytes in response to curcumin treatment, Co-IP was performed with slight modifications as outlined in our laboratory [32]. Primary white adipocytes containing 500 µg of protein were incubated with 15 μl of anti-UCP1/PGC-1α antibody (Santa Cruz Biotechnology) at 4 °C for 3 h on a rotary shaker, after which 20 μl of A/G PLUS agarose conjugate suspension (Santa Cruz Biotechnology) was added and allowed to mix at 4 °C for 12 h on a rotary shaker. After centrifugation at 4000×g for 5 min at 4 °C, the supernatant was collected and the beads washed three times with RIPA buffer (Sigma). Then, 2 × Laemmli buffer was added and boiled for 5 min. For the negative control sample, control rabbit IgG (Santa Cruz Biotechnology) was used. After Co-IP, all samples were separated by 8–12 % SDS-PAGE, after which samples were transferred to a PVDF for immunoblot analysis.

Statistical analysis

All experimental results were compared by One-way Analysis of Variance (ANOVA) using the Statistical Package of Social Science (SPSS, version 17) program; data are expressed as the mean ± SEM. A protected least-significant difference (LSD) test, which is a method for multiple comparison consisting of single-step procedures in One-way ANOVA, was used to demonstrate significant differences between means (p < 0.05).

Results

Proteomic analysis and classification of rat primary white adipocyte proteome

For identification of differentially expressed proteins between control (CN) and curcumin-treated adipocytes, all protein samples were separated by 2-DE (Fig. 1a) and identified by MALDI-TOF–MS (Table 2), and database searches were performed with high confidence based on high score and sequence coverage using MASCOT server. A total of 325 individual matched spots ranging in mass from 6 to 240 kDa between pH 3-10 were detected. Fifty-eight spots were identified with statistically different intensities between the CN and CR groups. These differentially altered spots showed dramatic alterations in response to curcumin treatment, suggesting that curcumin possibly plays a pivotal role in obesity prevention and lipolysis in primary white adipocytes. Next, differentially expressed proteins identified from the two groups were subjected to functional analysis using the PANTHER database. Functional analysis classified the identified proteins as related to biological processes, molecular function, and protein class by PANTHER classification (Fig. 1b). These proteins were involved in metabolic processes (50 %), cellular processes (18.1 %), and localization (12.5 %) based on GO Biological Process. These proteins were also involved in catalytic and transporter activities based on GO Molecular Function. In PANTHER Protein Class, identified proteins were classified as oxidoreductase (24.5 %), hydrolase (14.3 %), cytoskeletal protein (8.2 %), transferase (8.2 %), and chaperone (8.2 %). Among the differentially expressed spots, 25 spots were significantly down-regulated, whereas 33 spots were up-regulated in curcumin-treated adipocytes (Fig. 2).

Fig. 1
figure 1

Representative silver-stained 2-DE gel images of rat primary white adipocyte proteome. a Differentially regulated proteins and proteins of interest are marked with circles and arrows. The numbers in gels are listed in Table 1. b Fifty-eight identified proteins were classified into biological process, molecular function, and protein class using the PANTHER classification system

Full size image
Table 2 List of identified proteins showing differential expression in rat primary white adipocytes
Full size table
Fig. 2
figure 2

Differentially regulated proteins in rat primary white adipocytes from curcumin-treated (CR) group compared with control (CN) group. a Down-regulated and b Up-regulated proteins. Each bar shows an average volume density (%) in 2-DE analysis. Statistical significances between control and curcumin-treated groups were determined by One-way ANOVA test, where p < 0.05

Full size image

Differential expression of proteins involved in lipid metabolism and mitochondrial biogenesis

Prior to comparison of differentially altered proteins between the CN and CR groups, we classified proteins based on DAVID bioinformatics resources. To identify enriched functional gene groups in DAVID, identified proteins were converted from a UniProt protein ID into a DAVID gene ID. Candidate proteins were classified according to gene ontology entries in UniProt. After submitting the converted list into DAVID, top affected gene functions in curcumin-treated adipocytes were related to a central role in mitochondrial metabolism (Fig. 3a). Fourteen (ACCA2, ACO2, ALDH2, ATP5B, ATP5H, CS, DLD, GLUD1, HADHA, HSPD1, HSPE1, MDH2, PRDX5, and VDAC1) proteins were mainly localized to mitochondria and were shown to exert a variety of mitochondrial functions; proteins required for transport or transfer of fatty acids or proteins involved in lipid metabolism were observed frequently. DAVID analysis showed that three (ALDH2, ATP5B, PRDX5) out of 14 proteins were down-regulated in adipocytes by curcumin treatment, whereas ATP5B expression was elevated by Western blot analysis. In functional pathway analysis by DAVID, two proteins (ACCA2 and HADHA) were shown to be involved in fatty acid degradation in mitochondria from the KEGG database (Fig. 3b). This process includes three major steps as follows: (1) lipolysis of and release from adipose tissue; (2) activation and transport into mitochondria; and (3) β-oxidation. Fatty acid degradation is the process in which fatty acids are broken down into their metabolites as the main energy supply of animals. Fatty acids must be activated before they can be transported to mitochondria for fatty acid oxidation.

Fig. 3
figure 3

Discovery enriched functional-related gene groups from differentially regulated proteins between control (CN) and curcumin-treated (CR) groups. a Functional classification by DAVID and b protein pathway in KEGG pathway

Full size image

To elucidate interactions between the identified proteins and brown fat-specific markers in primary white adipocytes, six proteins associated with fatty acid metabolism and mitochondrial function were verified by Western blot analysis (Fig. 4a). Two mitochondrial proteins (ATP5B (Atp5b) and MDH (Mdh2)) were significantly up-regulated in response to curcumin treatment, whereas expression of HSP90 was reduced. Two fatty acid metabolic proteins (HSL (Lipe) and CA3) were significantly up-regulated in primary white adipocytes with curcumin treatment, whereas ANXA2 was down-regulated. Based on these results, enhanced lipolysis mediated by increased HSL induction upon curcumin treatment affects lipid metabolism through alterations of adipose differentiation and adipose-derived hormone levels.

Fig. 4
figure 4

Validation of fatty acid metabolism and mitochondrial proteins by Western blotting. a mRNA expression levels of mitochondrial protein-coding genes b and thermogenesis genes by real-time PCR c in control (CN) and curcumin-treated (CR) groups. Real-time PCR and immunoblot analysis results are expressed as relative intensity normalized to Gapdh/β-actin and represent the mean ± S.E.M of triplicate independent experiments. Statistical significance was determined by one-way ANOVA test, where *p < 0.05 and **p < 0.01. Relative changes in gene expression were quantified using the ∆∆CT method. Fold changes refer to expression ratios of CR to CN

Full size image

Validation network analysis using real-time RT-PCR

To evaluate the results of the DAVID classification, mitochondrial function-associated genes were further confirmed using real-time RT-PCR (Fig. 4b). Among the mitochondrial protein-coding genes in adipocytes, ten protein-coding genes (Acaa2, Atp5b, Atp5 h, Cs, Gpd, Hadha, Hspd1, Hspe1, Mdh2, and Vdac1) were selected from the list of identified candidates by DAVID. Their mRNA levels were significantly altered, and expression patterns of all genes were in line with the proteomic results. Moreover, curcumin treatment led to significant elevation of mitochondrial Acaa2 and Hadha levels, suggesting enhanced fatty acid degradation in mitochondria.

To confirm the activities of key browning and thermogenic markers together with other mitochondrial proteins, we measured expression levels of Ucp1, Ucp2, Ucp3, and Ppard in response to curcumin treatment (Fig. 4c). As expected, higher mRNA expression of Ucp1 was observed in curcumin-treated adipocytes compared with control cells, whereas Ppard expression was not significantly changed by curcumin treatment. This result suggests that curcumin dissipates energy as heat by uncoupling the mitochondrial respiratory chain from ATP formation through expression of Ucp1.

Interactions of browning markers in primary white adipocytes

To understand interactions between the identified proteins and brown fat-specific markers in primary white adipocytes, six important browning markers (Cebpb, Prkaa2, Ppargc1a, Pparg, Prdm16, and Ucp1) were subjected to functional analysis using STRING network analysis. The network analysis results revealed possible interactions between the identified proteins and brown fat-specific markers (Fig. 5). Combined results from the proteomic and network analysis identified eight genes as protein-coding genes, and two candidates (HSL/Lipe and FABP4/Fabp4) showed possible interactions with six browning markers (zoom in image in Fig. 5). These results suggest that two candidates (HSL and FABP4) associated with UCP1 play significant roles in controlling the browning phenotype in white adipocytes.

Fig. 5
figure 5

Discovery of novel fat-browning markers by network analysis from differentially regulated proteins in response to curcumin treatment. Interaction of identified proteins with main fat-browning markers using the web-based software tool STRING as well as two proteins (FABP4, HSL (LIPE)) highly connected with six fat-browning markers in STRING. Zoom in image (red question markers) shows that HSL is tightly connected with two proteins (PGC-1α and UCP1). (Color figure online)

Full size image

Validation of browning marker interactions in rat primary adipocytes

Interestingly, STRING network analysis revealed interactions between HSL (Lipe) and main browning players. To validate these results, six protein-coding genes (Cebpb, Fabp4, Lipe, Pparg, Ppargc1a, and Prdm16) were selected for further real-time PCR analysis. Expression levels of three protein-coding genes (Fabp4, Lipe, and Pparg) were significantly up-regulated (16, 8, and 7-fold changes, respectively) by curcumin treatment in primary white adipocytes (Fig. 6a). Furthermore, to verify interaction targets, HSL (Lipe) was immunoprecipitated with primary adipocyte lysates from each group (Fig. 6b) and validated further by Western blotting. The results show that UCP1 and PGC-1α were co-expressed with AMPK, C/EBPβ, HSL, and PPARγ. Interestingly, four interacting proteins (AMPK, C/EBPβ, HSL, and PPARγ) were up-regulated along with UCP1 expression by curcumin treatment, whereas these four proteins were down-regulated with PGC-1α expression. These data suggest that HSL is concomitantly expressed along with UCP1 during fuel metabolism and trans-differentiation from white to brown adipocytes during adipocyte differentiation. Interaction of UCP1 and HSL was further validated in primary white adipocytes by dual-fluorescent staining (Fig. 6c).

Fig. 6
figure 6

Validation of proteomic data and predicted targets generated by network analysis. a Real-time RT-PCR, b interaction of PGC-1α, UCP1, and HSL validated by co-immunoprecipitation-based Western blot analysis, and c co-localization of UCP1 and HSL in rat primary white adipocytes. Real-time PCR and immunoblot analysis results are expressed as relative intensity normalized to Gapdh/β-actin and represent the mean ± S.E.M of triplicate independent experiments. Statistical significance was determined by one-way ANOVA test, where *p < 0.05 and **p < 0.01

Full size image

Discussion

In this proteome study, we observed differential expression of multiple proteins involved in mitochondrial function in primary white adipocytes upon curcumin treatment, suggesting enhanced fatty acid oxidation and mitochondrial biogenesis. To further investigate the roles of the identified proteins in primary white adipocytes, we performed analyses using the PANTHER, DAVID, and STRING databases. Interestingly, our proteomic results revealed differential expression of multiple proteins involved in lipid metabolism and mitochondrial function.

A recent study by Wang et al. [33] demonstrated that curcumin can potentially prevent obesity by inducing browning of inguinal WAT via the norepinephrine-β3AR pathway. Consistent with these data, we found that curcumin significantly elevated expression of brown fat-specific markers (Tmem26 and Tbx1) as well as proteins such as UCP1, PGC-1α, and PRDM16 [19]. Taken together, these findings suggest that curcumin plays a dual modulatory role in inhibition of adipogenesis as well as induction of the brown fat-like phenotype. However, these recent studies have not been able to identify correlations between brite-specific markers and other browning-associated proteins in WAT via promotion of WAT remodeling in response to curcumin treatment.

Our results show that curcumin obviously elevated expression of proteins/genes involved in both fatty acid oxidation and the TCA cycle, which may contribute to reduced adiposity. Two (CS and HADHA) out of 14 mitochondrial proteins in primary white adipocytes showed higher expression upon curcumin treatment. CS is a Krebs TCA cycle enzyme that catalyzes synthesis of citrate from oxaloacetate and acetyl coenzyme A. HADHA catalyzes the last three steps of mitochondrial β-oxidation of long-chain fatty acids and plays important roles in control and regulation of β-oxidation [34]. In addition, partial inhibition of fatty acid oxidation has been reported as a therapy for non-insulin dependent diabetes mellitus (NIDDM) [35]. A recent study found that these two proteins show reduced mitochondrial expression in human omental fat from obese subjects and are directly involved in energy metabolism of cells [36, 37]. ACAA2 is also involved in the fatty acid metabolism pathway and catalyzes the last step of the mitochondrial fatty acid β-oxidation pathway [38]. Our data support an increased number of mitochondria and dramatically increased expression of ACAA2/Acaa2, which also activates mitochondrial fatty acid β-oxidation.

ATP synthase is localized to the plasma membrane in many different cell types, including endothelial cells and adipocytes, and they may contribute to regulation of lipid metabolism and cellular proliferation [39]. Although the function of the ATP5B gene product is not clear, reduced expression of the ATP5B gene was detected in various types of human tumors [40]. Catalytic sites of ATP synthesis are present in the β subunit, which plays a crucial role in energy metabolism and directly participates in energy production. According to our findings, ATP5B presumably contributes to mitochondrial oxidative phosphorylation and mitochondrial biogenesis.

It is well known that two essential malate dehydrogenases, MDH1 (cytosolic malate dehydrogenase) and MDH2 (mitochondrial malate dehydrogenase), play important roles in the Krebs cycle for energy metabolism. The native dimeric form of this enzyme is specifically activated by citrate in the NAD + leads to NADH direction and inhibited by citrate in the NADH leads to NAD + direction [41]. Furthermore, Zou et al. [42] reported that breed-, gender-, and tissue-specific expression patterns of Mdh2 expression are positively correlated with adipocyte volume across six adipose tissue types. We observed consistent up-regulation of MDH/Mdh2 in primary white adipocytes by curcumin treatment. Interestingly, a recent report highlighted the role of lactate in energy metabolism and revealed an effect on WAT browning [43]. Induction of Ucp1 by an increased redox state (NADH, H+-to-NAD+) favors its dissipation through increased uncoupled respiration. Browning of white adipocytes by intermediate metabolites constitutes an adaptive mechanism to alleviate redox pressure. Further study is needed to fully understand the mechanism and to further determine whether or not this browning mechanism plays a significant role in energy expenditure.

Fatty acids must be esterified to Coenzyme A before they can undergo oxidative degradation, and two major lipases (ATGL and HSL) are involved in lipid degradation. Activation of HSL in adipocytes has been proposed to play an essential role in the regulation of body weight and fat mass since HSL is responsible for release of FFA from stored TG in adipose tissues [44, 45]. In contrast to our expectations, it was suggested that mice lacking HSL expend more energy during thermogenesis than control mice [44]. Furthermore, it was also reported that the fasting-induced loss of body weight is correlated with UCP2 expression but not UCP1 expression in BAT [44]. However, we observed that enhanced lipolysis mediated by increased HSL induction upon curcumin treatment affected lipid metabolism through alterations of adipose differentiation and adipose-derived hormone levels. In addition, curcumin treatment led to increased expression of FABP4, which might promote binding to newly released fatty acids [46]. Our results suggest that HSL is probably involved in the determination of white versus brown adipocytes during adipocyte differentiation, which was in line with earlier result reported by other investigators [47]. Although the exact mechanisms underlying this novel role of HSL remain to be elucidated, it seems clear that HSL is, at least in part, participates in the control of browning of white adipocytes.

The multifunctional protein ANXA2 is down-regulated after exercise or diet intervention in human adipose tissue, and it has been suggested to enhance GLUT4 translocation to membranes [48]. Our recent study found that thiodigalactocide (TDG) binding to galectin 1 (GAL1) may restrict formation of the GAL1-ANXA2-VIM (vimentin) interactome, as these proteins were all down-regulated, leading to lower lipid droplet formation [49]. Down-regulation of ANXA2 and VIM together with GAL1 led to reduced lipogenesis or TG synthesis in TDG-treated rats. These findings suggest that curcumin plays a modulatory role in inhibition of adipogenesis as well as transformation of lipid droplets via alteration of lipid droplet-associated proteins in primary white adipocytes. In summary, the molecular mechanisms underlying curcumin-induced increased differentiation of primary white adipocytes seem to involve induction of increased fatty acid metabolism by ACAA2. Additionally, translocation of HSL to lipid droplets is the critical event regulating lipolysis induced by curcumin treatment. These FA are used by mitochondria, as demonstrated by increased expression of enzymes involved in β-oxidation and mitochondrial biogenesis (ATP5B, ACAA2, and HADHA). Enhanced FA utilization induces production of acetyl-CoA as well as a proton gradient between the two mitochondrial membranes. This gradient is dissipated by UCP1, thereby increasing heat production in primary white adipocytes upon curcumin treatment. ANXA2 is down-regulated after curcumin treatment and has been shown to enhance GLUT4 translocation to membranes. These results suggest that curcumin plays a modulatory role in inhibition of adipogenesis and lipogenesis by increasing protein levels associated with lipolysis, fatty acid β-oxidation, and energy expenditure (Fig. 7).

Fig. 7
figure 7

Proposed schematic diagram of differentially regulated proteins/genes determining curcumin-induced browning in rat primary white adipocytes. For abbreviated protein names, see Abbreviations and Table 2. (Arrow) stimulatory, (dashed line) translocation, (up arrow) up-regulation

Full size image

In conclusion, curcumin promotes fatty acid oxidation, mitochondrial activity, and Ucp1 expression possibly by activating a regulator of mitochondrial function. Finally, we systematically identified candidates involved in lipolysis, fatty acid β-oxidation, and energy expenditure.