Expression Pattern and Regulatory Role of microRNA-23a in Conjugated Linoleic Acids-Induced Apoptosis of Adipocytes

Natural conjugated linoleic acids (CLAs) are a mixture of positional and geometric isomers (cis or trans) of linoleic acid with conjugated double bonds. Accumulating evidence suggests that CLAs have a diverse range of biological functions including the enhancement of immunity [1], prevention of tumor formation [2,3], and the reduction of body fat deposition [4,5,6].

Currently, obesity and related metabolic disorders caused by excessive fat deposition are becoming a growing public health crisis. CLAs show strong anti-obesity actions in humans and animals. They have been shown to inhibit the differentiation process of preadipocytes and reduce the triglyceride content of mature or newly differentiated adipocytes as well as induce adipocytes apoptosis [7,8,9,10]. Cis-9, trans-11 (c9, t11)-CLA and t10, c12-CLA are two primary isomers of CLA that have different biological properties. T10, c12-CLA generally has a much stronger anti-adipogenesis effect than c9, t11-CLA.

As apoptotic inducers, CLAs have strong anti-cancer effects on various tumor cells. C9, t11-CLA and t10, c12-CLA either directly activate the caspase cascade of apoptosis or promote apoptotic signal transmission [11,12]. However, little is known about the effects of CLAs on apoptosis in adipocytes. Several pieces of evidence suggest that CLAs directly induce apoptosis in adipocytes such as the 3T3-L1 cell line or mouse fat cells [8,13,14]. Furthermore, we have also found that dietary supplementation with CLAs decreases back fat accumulation in piglets partially through apoptosis induction in adipocytes [15]. However, the underlying details of this mechanism are unclear and require further studies.

MicroRNAs (miRNAs) are small non-coding RNAs that are 18-25 nt in length. miRNAs are widely expressed in a diverse range of animal tissues and are involved in the post-transcriptional expression regulation of numerous target genes [16]. Depending on their regulatory target, many miRNAs play important role in the apoptosis of various cell types via the control of pro- or anti-apoptotic molecules [17,18,19]. It has also been recently shown that nutritional and dietary factors such as CLA alter the expression levels of miRNAs [20,21]. However, it remains unclear which miRNAs and regulatory pathways are affected by CLA.

This study aimed to explore and compare the differential effects of the two CLAs on adipocyte apoptosis and which miRNAs were implicated in the process. We found that c9, t11-CLA had a stronger apoptosis induction effect than t10, c12-CLA. This difference likely occurred because t10, c12-CLA, but not c9, t11-CLA, activated nuclear factor (NF)-κB, which is a suppressor of apoptosis. Moreover, miR-23a expression was significantly down-regulated by the CLAs, whereas miR-23a over-expression attenuated the apoptosis induction effect of CLA by targeting the APAF1 gene. An in vivo study in mice also showed that dietary supplementation with CLA increased the number of apoptotic adipocytes and decreased fat deposition at least in part through the altered expression of miR-23a/APAF1. Together, these findings indicate that miR-23a plays an important role in the adipocyte apoptosis induced by CLA and has the potential to be a new target for obesity therapy.

3T3-L1 cells were seeded in Dulbecco's modified Eagle's medium (DMEM)/F12 (Gibco, NY, USA) supplemented with 10% fetal bovine serum (FBS) and cultured in 6-well plates at 37°C in a 5% CO₂- humidified atmosphere. For adipogenesis differentiation, the cells were stimulated with an adipogenic inducer cocktail (0.5 mM 3-isobutyl-1 methylxanthine, 1 µM dexamethasone, and 5 µg/mL insulin, Sigma-Aldrich, MO, USA) for 2 days followed with 5 µg/mL insulin for another 2 days. After the inducers were removed from the medium, the cells were treated with either c9, t11-CLA or t10, C12-CLA (50 µM, Sigma-Aldrich, USA) for subsequent experiments.

293T cells were cultured for the dual-luciferase activity assay. The cells were grown in DMEM supplemented with 10% FBS, and the medium was changed every 2 days.

Apoptosis in adipocytes was observed and detected by various methods, including caspase 3/7 activation analysis, TUNEL assay, and flow cytometry. Caspase 3/7 activity was quantitated by using a Caspase-Glo® 3/7 Assay Kit (Promega, CA, USA) according to the manufacturer's instructions. A DeadEnd™ Fluorometric TUNEL System (Promega) was used to observe and measure the adipocyte apoptosis. The number of apoptotic adipocytes was quantitated by flow cytometry assay (Gallios, Beckman Coulter, USA) with the Annexin V/PI double-staining method according to the manufacturer's instructions. (The Annexin V/PI apoptosis assay kit was purchased from Life Technologies, USA).

To compare the differences in miRNA expression in CLA-treated cells and normal cells, a commercial mouse miRNA microarray which contained probes for 1900 mouse miRNAs from the Sanger miRBase database v.21.0, was used in this study. Microarray hybridizations were performed by the LC Sciences Biotech Company (Hangzhou, China). The tagged miRNAs were purified and hybridized with the LC Sciences miRNA Microarray-Single following the manufacturer's instructions. After hybridization, the arrays were subjected to a stringent wash, and fluorescence data were collected using an Axon laser scanner model 4000B (Axon Instruments). The arrays were scanned at a pixel size of 10 µM with the Cy3 gain at 460 and the Cy5 gain at 470. Data extraction and image processing were performed using ArrayPro™ software equipped with a morphological filter (Media Cybernetics).

To explore the effect of miR-23a on apoptosis, a miR-23a-specific mimic, an inhibitor of miR-23a and negative control oligos were custom-designed by the RIBOBIO Biotech Company (Guangzhou, China) and transfected into cells for over-expression (miR-23a mimic) or knock-down (miR-23a inhibitor) of miR-23a.

For the in vivo study, 60 healthy Kunming white mice (initial body weight, 15±1 g) were randomly divided into 3 groups and fed either a high-fat diet alone (control group and antagomir group, 45% kcal from fat) or high-fat + CLA diet (CLA group, CLA dose in diet, 15 g/kg; c9, t11-CLA:t10, c12-CLA=1:1 in the mixtures). To explore the effect of miR-23a on fat deposition in mice, a stable specific antagonist was used to silence the endogenous miR-23a. After high-fat feeding for 30 days, all mice in the antagomir group were treated with a miR-23a antagomir (3 nM per mouse, RIBOBIO, Guangzhou, China) by tail intravenous injection. The mice were injected 3 times, with one injection every 5 days. During the experimental period, the body weight and food intake of the mice were measured every 15 days. At the 45th day, all mice were bled and weighed, and then 8 mice from each group were sacrificed. The fat tissues were peeled off, weighed and then quick-frozen for subsequent experiments. All animal experiments were performed according to Chinese laws and were approved by the Institutional Animal Care of the Chongqing Academy of Animal Science.

Total RNA was extracted from cultured cells or fat tissues of mice with RNAiso Reagent (TaKaRa, Dalian, China), and 1.0 µg of each sample was reverse-transcribed to cDNA by using a PrimeScript™ RT reagent Kit (TaKaRa). PCR was performed using a Step One system (ABI, USA) with SYBR Premix Ex Taq™ II (TaKaRa). The 18S RNA and GAPDH genes were used as reference genes with similar results.

Small RNA was extracted from cultured cells or fat tissues of mice with a special Small RNA Reagent (TaKaRa) and initially reverse-transcribed and amplified with gene-specific primers using a SYBR® PrimeScript™ miRNA RT-PCR kit (Clontech, CA, USA). The kits included a general reverse transcription primer. U6 and 5S RNA served as endogenous reference RNAs for normalizing the cellular content of other miRNAs.

Cultured cells or fat tissues of mice were lysed in RIPA buffer (Beyotime Biotech, Beijing, China), and the protein expression levels were analyzed using a standard western blot procedure. Tubulin was used as a loading control. The PPARγ, caspase 3, cleaved caspase 3, cytochrome C and cleaved PARP antibodies were obtained from Abgent (USA), and the P-IKKα, IκBα, P-IκBα, and APAF1 antibodies and horseradish peroxidase-conjugated secondary antibody were obtained from CST (MA, USA).

The nuclear translocation of activated NF-κB was detected by immunofluorescence labeling according to the manufacturer's instructions by using a Cellular NF-κB Translocation Kit (Beyotime Biotech, China). Briefly, the cultured cells were washed with PBS three times, fixed in 4% neutral formalin for 15 min, and then incubated with a blocking buffer for 1 h. Next, cells were incubated with the primary p65 antibody for 1 h, followed by incubation with a Cy3-conjugated secondary antibody for 1 h and a final incubation with DAPI for 5 min before observation. The positive p65 protein fluoresced red and nuclei fluoresced blue, and both could be viewed simultaneously with a fluorescence microscope at an excitation wavelength (350 nm for DAPI and 540 nm for Cy3). To create a two-color image, the red and blue images were overlaid, and the purple fluorescence represented the co-localization areas of NF-κB and nuclei.

NF-κB activity was determined by using an NF-kappaB p65 PhosphoTracer Immunoassay Kit (Abeam, MA, USA) following the manufacturer's protocol. Briefly, cells were seeded in 96-well plates and treated with different CLAs for 24 h at 37 °C. After the CLAs were removed and the cells were washed with PBS, the cells were incubated with primary antibody overnight at 4 °C followed by the appropriate secondary antibody. Cells were then exposed to substrate solution for 30 min, and the signal was measured using a fluorescent plate reader (BioTek, VT, USA).

The 3'-UTR of APAF1 containing an intact miR-23a recognition sequence was cloned into the pGL3-control-mcs2 reporter vector (constructed by our laboratory). Oligonucleotides (200 bp) harboring wild-type or mutant miR-23a binding sites from the mouse APAF1 3'-UTR were annealed and ligated into the EcoR I and Pst I sites of the pGL3-control mcs2 reporter vector. For the luciferase assays, 293T cells were transfected with the appropriate plasmids in 24-well plates and harvested for luciferase activity assays using the dual-luciferase reporter assay system (Promega) at 48 h after transfection. The relative luciferase activity was normalized to that of firefly luciferase.

Each experiment was performed three times independently. The data are presented as the means ± S.D. The SPSS 19.0 statistics software package was used for one-way ANOVA and Student's two-tailed unpaired t-test. Differences were considered significant at P < 0.05.

Treatment with the two CLA isomers significantly inhibited differentiation in the adipocytes. Both isomers decreased the intracellular triglyceride content, down-regulated fatty acid synthetase (FAS) and up-regulated adipose triglyceride lipase (ATGL) mRNA expression levels, and decreased PPARγ protein level in 3T3-L1 cells during the differentiation process (Fig. 1A-E). T10, c12-CLA showed a stronger inhibition effect on differentiation and lipid accumulation in the cells than c9, t11-CLA. Moreover, t10, c12-CLA significantly decreased the viability of 3T3-L1 cells under proliferation conditions, while c9, t11-CLA only showed a very weak influence (Fig. 1F). These findings were consistent with previous reports [6,7,8].

Importantly, there were obvious apoptotic characteristics, such as cell shrinkage, fluorescence enhancement, DNA fragmentation, and cell separation, in the differentiated 3T3-L1 cells following CLA treatment for 6 h (Fig. 2A and B). The CLAs increased the expression of the pro-apoptotic gene BAX and induced the release of cytochrome C into the cytoplasm, which activated caspase 3/7, a key player in apoptosis (Fig. 2C and D). Additionally, tumor necrosis factor α (TNFα) was highly expressed in CLA-treated cells. The expression of the TNFα receptor (TNFR) was also increased at the same time (Fig. 2E). TNFα directly induces apoptosis in many cell types by activating the death receptor signaling cascade [22,23]. Therefore, changes in the expression of these regulators suggested that apoptotic signaling was activated by CLA in adipocytes and was transmitted through the mitochondrial pathway and death receptor pathway simultaneously. Moreover, both a TUNEL assay and flow cytometric analysis indicated that c9, t11-CLA had a stronger apoptotic induction effect on the differentiated adipocytes than t10, c12-CLA (Fig. 2A, B and C), though t10, c12-CLA showed more powerful inhibition effects on fat production and adipogenesis differentiation.

To further explore the differential apoptotic induction effects between the two CLA isomers, we analyzed the effect of the CLAs on NF-κB, which suppresses apoptosis in many cell types [24,25,26]. We found that t10, c12-CLA, but not c9, t11-CLA, induced the activation and translocation of NF-κB in adipocytes (Fig. 3). Following 12 h of t10, C12-CLA treatment, IκB/IκB kinase (IKK) α and Iκβ were activated (Fig. 3B and C), causing significant increases in the nuclear translocation and expression of phosphorylated p65 (Fig. 3A and D). This finding suggested that the apoptotic induction effect of t10, c12-CLA may be partly attenuated by NF-κB activation.

We next examined the effect of CLAs on miRNA expression profiles. A microarray assay showed that 10 miRNAs had a notably different expression (Fig. 4A and B, P < 0.05). C9, t11-CLA significantly down-regulated the expression levels of five miRNAs (miR-215, Let-7a, miR-23a, miR-28c, and miR-1898) and up-regulated the expression levels of the other five (miR-760, miR-6963, miR-6904, miR-7063 and miR-8114). T10, c12-CLA also down-regulated miR-23a, miR-28c, and Let-7a expression levels, whereas it had the opposite effect on the other seven miRNAs. The qRT-PCR data corroborated these changes in miRNA expression as shown in Fig. 4C. Furthermore, Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis showed that all 10 miRNAs were involved in the advance and onset of apoptosis.

miR-23a expression was decreased the most among the 10 identified miRNAs. Additionally, previous studies have reported that miR-23a participates in apoptosis in other cell types [27,28,29]. Therefore, we further examined the role of miR-23a in adipocyte apoptosis. A specific mimic and inhibitor were used to over-express and knock-down miR-23a, respectively. In the mimic group, miR-23a expression was up-regulated 14.5-fold in 3T3-L1 adipocytes, while it was down-regulated more than 90% following inhibitor transfection (Fig. 5A). Moreover, the mimic rescued the expression of miR-23a in adipocytes treated with CLA, and the compensatory expression greatly impeded the apoptosis induced by the CLA isomers (Fig. 5B and C). However, the miR-23a inhibitor did not further enhance the apoptosis induced by CLA.

We found that the miR-23a inhibitor also directly increased the expression of cleaved caspase 3/7, cytochrome C, and cleaved poly (ADP-ribose) polymerase (PARP) protein (Fig. 5D); enhanced the activity of cellular caspase 3/7 (∼1.6-fold of control levels, Fig. 5B); and increased the number of apoptotic cells (Fig. 5E). These results suggested thatmiR-23a plays a direct role in adipocyte apoptosis. Therefore, CLA-induced apoptosis in adipose tissue may occur partly through altered miR-23a expression.

Our target gene predictive assay using the TargetScan online tools (http://www.targetscan.org/mmu_71/) indicated that the apoptosis-related genes Bcl-2 and APAF1 were potential targets of miR-23a. However, qPCR and western blotting results showed that the miR-23a mimic significantly decreased APAF1 mRNA and protein expression but did not obviously affect BCL2 expression (Fig. 6A and B). Furthermore, our dual-luciferase reporter assay showed that co-transfection of Luc-APAF1 with the miR-23a mimic decreased luciferase activity by 58% compared to control. Transfection with the APAF1 mutant did not cause any significant changes in luciferase activity, which suggested that miR-23a directly regulated APAF1 expression by binding its 3'-untranslated region (Fig. 6C and D). Therefore, we believe APAF1 is a direct target of miR-23a during the apoptotic process.

To further evaluate the effects of CLA and miR-23a on obesity and fat deposition, an in vivo study was conducted in mice (Fig. 7A). As expected, supplementing a CLA mixture into high-fat chow (15 g/kg) for 45 days significantly decreased fat deposition (∼65%, P < 0.01) compared with control mice fed high-fat chow only (Fig. 7B, C and D). Additionally, as expected, fat tissues of CLA-fed mice had lower levels of miR-23a, higher caspase 3/7 activity, and more apoptotic adipocytes than control mice (Fig. 7E-G).

Next, we silenced endogenous miR-23a in mice via intravenous injection of the specific antagomir. The miR-23a expression level in the fat tissues was significantly decreased by approximately 80% 12 h after the antagomir injection, and the effect of one injection lasted 3-5 days. After three injections (every five days), mice showed a lower body weight and significant lower fat deposition (∼45%, P <0.05) compared with control mice, even though they were fed the same high-fat chow. Similar to CLA treatment, miR-23a silencing also increased caspase 3/7 activity (1.7-fold) and the number of fat cells undergoing apoptosis compared with control mice. As expected, we also observed increases in APAF1 mRNA and protein expression following miR-23a silencing (Fig. 8). These results suggested that the in vivo anti-obesity effects of CLA occur, at least in part, through apoptosis induced in fat cells and alterations in the expression of apoptosis-related miRNAs and their targets, such as miR-23a/APAF1.

Research into the apoptosis of adipocyte will help us combat obesity and related metabolic syndromes, but little is known about how CLA induces adipocyte apoptosis or which target genes are involved. Here, we found that t10, c12-CLA inhibited cellular differentiation and lipid production in pre-adipocytes and adipocytes to a greater extent than c9, t11-CLA. However, contrary to earlier work [13,14], we showed that c9, t11-CLA exhibited stronger induction effects than t10, c12-CLA on mature adipocytes with the same dose and at the same time. These inter-study differences may reflect the extent of adipocyte differentiation or differences in detection methods. Indeed, c9, t11-CLA also exhibited a strong apoptosis induction action in other cells [30,31].

Activated NF-κB plays a crucial role in the initiation and development of apoptosis in many cell types. Tamai et al (2000) found suppression of NF-κB rendered preadipocytes susceptible to TNF-α-induced apoptosis [32]. Activated NF-B could confer adipocyte resistance to TNF-α-induced apoptosis. Similarly, NF-κB showed negative regulated function in cellular apoptosis in other instances [24,25,26,33]. Our results indicated that only t10, c12-CLA effectively phosphorylated IKKα and IκB in adipocytes, leading to the activation and nuclear translocation of NF-κB. It is conceivable that activation of NF-κB feedback weaken the apoptosis induced by t10, c12-CLA. Other studies also reported that only t10, c12-CLA could activate the NF-κB signaling pathway in muscle and peripheral blood mononuclear cells [34,35]. These reflect the differences in biological properties and related molecular mechanisms of the two CLAs and partly answer the reason for differential apoptosis induction effects between different CLA isomers.

An increasing number of miRNAs have been confirmed in adipocytes and them shown to regulate the cell proliferation, differentiation, and apoptosis [16,36,37]. In the present study, a total of 10 miRNAs showed significant changes following CLA treatment. Our GO and KEGG pathway assays indicated that all 10 miRNAs and their targets were involved in the regulation of apoptosis. The observed differential expression profiles of seven of these miRNAs between the two CLA-treated cells may explain why the two CLA isomers distinct differences on function.

MiR-23a has previously been proposed as a key apoptosis-related molecule, and we observed it to be significantly down-regulated by both CLAs in the present study. Some studies have suggested it was a molecular marker for a variety of cancers because of its high expression in tumor cells [38,39]. Moreover, it is also known to control the initiation and development of apoptosis in various types of tumor cells [40,41,42,43]. Our results clearly indicated that it also played a pivotal role in adipocyte apoptosis through control of APAF1 expression.

In conclusion, our findings indicate that c9, t11-CLA has a stronger apoptosis induction on mature adipocytes, though t10, c12-CLA exhibits more forceful inhibition effects on differentiation and fat production in pre-adipocytes. MiR-23a is down-regulated by CLAs in adipocytes, which increases the expression of APAF1, leading to apoptosis.

This work was supported by the National Basic Research Program of China (Project Number 2012CB124702), the National Nature Science Foundation of China (Project Number 31302055 and Project Number 31470117) and the Chongqing Fundamental Research Project (Project Number 16418 and Project Number 14403).

The authors declare that they have no competing interests.

R. Qi and Q. Wang contributed equally to this work.