|Home | About | Journals | Submit | Contact Us | Français|
Conjugated linoleic acid (CLA) is a term that refers to positional and geometrical isomers of linoleic acid (LA) . Previous mouse studies have shown that the trans-10, cis-12 isomer of CLA has mostly delipidative effects [2–5] while the cis-9, trans-11 isomer has mostly anti-carcinogenic effects . The trans-10, cis-12-CLA isomer elicits delipidative effects not only in mice, but also in rats, chicks, pigs, and humans [7–9]. Effects of CLA on obesity vary between and within species; in mice, CLA reduces fat-body mass by 50 to 70% . These changes in adipose and body mass may be caused by reduced size and number of adipocytes [10–13], differences in energy expenditure [14–17] and/or feed intake . In addition to changes in adipose tissue, increases in size, cytoplasmic vacuolization, and fatty acid oxidation in liver of CLA fed mice have been reported [4, 7, 8, 19]. Much of the increased liver mass may be explained by accumulation of lipids [19, 20] derived from mobilization of delipidated adipose tissue stores [21, 22].
Attempts have been made to elucidate the molecular mechanisms of action of CLA. To date, four microarray studies using gene expression profiling to examine the effects of CLA have been reported [10, 19, 23, 24]. Regulatory pathways involved in development, signal transduction, and fatty acid metabolism affected by CLA have been identified. In a previous study , we reported microarray data from adipose tissue of a polygenic obese line of mice fed 1% trans-10, cis-12-CLA. It was concluded that the delipidative effects of CLA occur through a pleiotropic reduction in fatty acid and triglyceride translocation and storage, decreased glucose availability, and increased fatty acid oxidation. In that experiment adipose mass was 30% less in the epididymal depot of CLA fed mice, 27% less in the mesenteric depot and 58% less in brown adipose tissue after 14 days of CLA treatment . In that same study we found that livers of the CLA-fed mice accumulated more fat and mass than the livers from LA-fed mice. Here we report gene expression and fatty acid composition data from livers of these male mice that were fed trans-10, cis-12-CLA for 14 days . Based on these data it was concluded that CLA is associated with differential regulation of many genes in the murine liver, particularily those regulated by the hepatocyte nuclear factor 4a transcription factor (HNF-4a), which is thought to be important in controlling metabolic status of the liver .
M16 mice were maintained according to an approved North Carolina State University’s Institutional Animal Care and Use Committee protocol. The M16 line is a moderately obese line that was selected for 3–6 week post-weaning gain .
Animals were fed purified AIN93G pellets (Harlan Teklad, Madison, WI) formulated with either 1% trans-10, cis-12-CLA or 1% LA as a treatment control. The trans-10, cis-12-CLA was kindly donated by BASF (Ludwigshafen, Germany), and LA was purchased from Nu-Chek-Prep (Elysian, MN). Dietary trans-10, cis-12-CLA and LA content were analyzed, confirming 92% and 99% purity, respectively. All animals were fed the LA diet for 7 days and then given ad libitum access to their assigned diet (either CLA or LA) for 14 days. A total of 36 nine-week-old male mice were used for this experiment, with 18 mice receiving each diet. Mice in the fed state were euthanized by CO2 asphyxiation and liver samples collected, weighed, and stored at −80°C.
Lipids were extracted from liver samples of eight mice per treatment group and fatty acids were quantified by gas-liquid chromatography as previously described . Dietary treatment effects (LA vs CLA) were analyzed according to a completely randomized design using the general linear models procedure of SAS (SAS Institute, Cary, NC). Differences were considered significant when P < 0.05.
Total RNA was isolated from liver using TriReagent (Sigma, St. Louis, MO) and further purified with Qiagen RNeasy Mini kit (Qiagen, Valencia, CA) following the manufacturer’s protocol with modifications previously described .
Agilent (Palo Alto, CA) G4121A mouse oligo microarray slides containing more than 20,000 probes were used for this experiment. Pooled total RNA was labeled with cyanine 3 (Cy3)- and cyanine 5 (Cy5)-labeled dCTP using the Agilent Fluorescent Direct Label kit (Palo Alto, CA) following the manufacturer’s protocol. Pooled samples were used to reduce costs of the experiment. For pooling, equal amounts of total RNA from 4 or 5 different mice, were combined for labeling. Because there were 36 mice from each treatment group, 4 biological replicates were represented on the microarrays. Four microarrays were hybridized using RNA pooled from the 36 animals (18 CLA-fed, 18 LA-fed). Microarrays were hybridized for 17 hours and washed as previously described . Slides were scanned with an Agilent G2565BA microarray scanner (Palo Alto, CA).
Microarray data were collected and analyzed as previously described . Briefly, microarray data extraction was performed using Agilent G2567AA Feature Extraction software, following Agilent’s direct labeling protocol. Rosetta Resolver version 3.2, build 220.127.116.11.33 (Rosetta Biosoftware, Kirkland, WA) was used to analyze data as previously described . Intensity plots were produced and genes were identified as candidate genes if P < 0.01 using the Rosetta Resolver error model . Raw data were deposited in the National Center for Biotechnology Information Gene Expression Omnibus in accordance with “minimum information about microarray experiments” guidelines and given accession numbers GSM272840, GSM272907, GSM272908, GSM272909 and GSE10865. P-values from Rosetta Resolver were converted to false discovery rate (FDR) P-values .
Real-time RT-PCR primers were designed to assay 12 genes and the ribosomal protein S18 housekeeping gene using Beacon Designer 7 (PREMIER Biosoft International, Palo Alto, CA) or selected from primers previously designed  (Table 1). PCR products were sequenced to confirm their identity. Genes targeted for quantitative real-time RT-PCR verification were selected based on their involvement in signal transduction pathways (socs6 and stat5b) or because they had greater than 2-fold differential expression based on microarray results (bcl6, pdcd8, tspyl4). Others were selected because they had previously been studied in the liver of female mice fed CLA (acox1, egfr, fasn, igfbp2, me)  or are important components of metabolic pathways (agpat2 and dgat2).
A total of 500 ng RNA was reverse transcribed using Applied Biosystems High Capacity cDNA Reverse Transcription Kit (Foster City, CA) for each mouse. Real-time RT-PCR reactions for 12 target genes and the S18 housekeeping gene were performed using Applied Biosystems Power SYBR® Green PCR Master Mix (Foster City, CA) in a BioRad iCycler (Hercules, CA) with minor modifications. Fluorescein was added at a final concentration of 10 nM as the reference dye. Cycling conditions were as follows: 95°C for 7 min, 60 cycles of (95°C for 30 sec, appropriate annealing temperature (Table 1) for 30 sec, 72.0°C for 30 sec), 72°C for 5 min, 95°C for 1 min, 55.0°C for 1 min, followed by a melt curve analysis of 80 cycles of 10 seconds at 55°C with a 0.5°C increase every cycle. SYBR Green gene expression data were collected from two replicates of each sample and three replicates of each standard curve point. Replicate samples were averaged and analyzed using the Pfaffl  method to quantify expression ratios between the CLA and LA cycle threshold values. Determination of statistical significance was done by ANOVA (45). The Relative Expression Software Tool  calculated fold changes adjusted for differences in PCR amplification efficiencies.
GeneGo’s MetaCore software suite (St. Joseph, MI) was used to further interrogate the differentially expressed genes from the experiment. Gene abbreviations for the 775 genes exceeding the FDR threshold of P < 0.05 and their fold changes were uploaded into the MetaCore workflow to identify potential pathways or networks associated with CLA treatment. The most relevant signaling or metabolic pathways and networks were identified using the “enrichment analysis—pathway maps” option. In addition, networks were built using network building tools provided in the software. The “analyze network (transcription factors)” tool built networks connecting differentially expressed gene targets of transcription factors.
The average amount of fat in the CLA and LA livers was 8.54g and 5.34g, respectively, per 100 g of wet tissue, indicating accumulation of 62.5% more fat in the CLA-fed livers (P < 0.002; Table 2). As we reported previously , there was a 33% increase in liver weight from mice fed the CLA diet for 14 days. Accumulation of fat accounted for approximately 20% of this increase. Because feeding CLA increased hepatic lipid levels, we examined the fatty acid profile. The trans-10, cis-12-CLA isomer (18:2) was only present in livers of mice fed the CLA supplemented diet (Table 2). The C18:0/C18:1 ratio was significantly different (P < 0.05) between the livers of mice fed CLA and LA.
Microarrays were used to identify genes differentially expressed due to CLA treatment. Analysis of gene expression data identified 1393 differentially expressed genes at a nominal P-value of 0.01. After FDR correction, 775 genes were differentially expressed (P < 0.05; a list of these genes can be found with the GEO accession GSE10865). Quantitative real-time RT-PCR assays designed for 12 genes and a housekeeping gene (Table 1) were used to validate our results or to compare with results reported in a similar study  in which CLA was fed and liver gene expression examined. Of the 12 target genes evaluated (Table 3), results from 1 gene disagreed between techniques indicating 92% of the genes tested had results congruent between microarray and RT-PCR assays. The result for stat5b was likely a false-positive from the microarray, in which the more sensitive real-time PCR method indicated no statistical differential expression for this gene.
Our list of 775 differentially expressed genes was imported into GeneGo software to identify pathways affected when CLA is added to the diet. Five pathways were identified that may be affected by CLA treatment (P < 0.03) (Table 4). These pathways involve cell signaling cascades, while more traditional metabolic pathways were unaffected (Table 5).
In this study, we examined differential gene expression using microarrays and real-time PCR in the livers of mice fed either CLA or LA. We found that livers of CLA fed mice weighed more and contained more lipid. Conversion of saturated fatty acids to their corresponding monounsaturated fatty acids is catalyzed by stearoyl CoA desaturase 1 (SCD1); hence, comparing ratios of unsaturated to monounsaturated fatty acids may provide an indication of enzymatic activity . Some groups have shown SCD1 levels to be reduced [34, 35], while others have found levels to be unaffected [23, 36] or increased  in murine liver when mice were fed CLA. In our evaluation of liver samples, the 16:0/16:1 ratio was unchanged (P = 0.372) while the ratio of 18:0/18:1 was decreased (P = 0.045). Therefore we cannot determine if SCD activity was affected by CLA treatment based solely on fatty acid ratios. In liver the 18:0/18:1 ratio was decreased and the 16:0/16:1 ratio was unchanged; however, in adipose we found the opposite, with 16:0/16:1 increased and 18:0/18:1 unchanged in these same mice . It is possible that hepatic steatosis was caused by accretion of fatty acids originating from delipidated adipose tissue during the 14 days of CLA treatment [21, 22]. Putative translocation of fatty acids from adipose to liver tissues could have masked changes in the ratios of saturated to unsaturated fatty acids result ing from de novo hepatic synthesis. Although we did not measure circulting non-esterified fatty acid concentration or composition, Oikawa and coworkers  observed an attenuation of CLA-induced hepatic lipidosis via arachidonic acid and this effect was not associated with altered NEFA concentration.
Using GeneGo we were able to identify biological pathways that appear to be affected when CLA is added to the diet. The five identified pathways involve cell signaling cascades, while more traditional metabolic pathways appear to be unaffected (Table 5). These results support use of microarrays, because initially we hypothesized metabolic pathways such as fatty acid and glycerol lipid synthesis and/or degradation, would be affected. However, these pathways were not significantly affected, with most of the genes associated with these pathways being unchanged. While carbon-flux through these hepatic pathways may well be altered by dietary CLA feeding, our data would suggest that changes in flux, by and large, are not mediated by changes in gene expression of enzymes in the pathways.
CLA treatment appears to affect the protein kinase A (PKA) signal transduction pathway containing a total of 23 objects, such as proteins and genes, as defined by GeneGo (Table 4). Eight of these objects were differentially expressed in our experiment, with four genes down-regulated in the livers of the CLA-fed mice and four genes up-regulated. Many of these same genes are found in the cyclic adenosine monophosphate (cAMP) signaling pathway, which regulates PKA and was another pathway significantly affected by CLA treatment (Table 4). Down-regulation of these four genes may indicate that less cAMP is available in the liver, which may lead to less degradation of stored glycogen . Additional evidence supporting CLA’s affect on the cAMP/PKA signaling pathway was shown in rat cardiac gene expression, in which Tappia and co-authors suggest that abnormalities in this signaling pathway are modulated by dietary CLA .
The G-protein coupled receptor protein signaling pathway associated with smooth muscle contraction was also affected by CLA treatment. This pathway contains a total of 54 objects, with ten of those being differentially expressed in our experiment (Table 4). Genes in this pathway include many of the same genes in the cAMP and PKA signaling pathways discussed above. This includes adenylate cyclase, whose transcript was down-regulated in the CLA-treated livers, and whose gene product normally increases levels of cAMP, which in turn, activates PKA . CLA was previously shown to affect vascular smooth muscle cells through inhibition of nuclear factor κB (NF-κB) activation via a peroxisome proliferator-activated receptor γ-dependent mechanism, preventing an inflammatory response and development of atherosclerosis .
An immune response pathway was affected in our CLA experiment (Table 4). This pathway, involving antigen presentation by major histocompatibility complex class I proteins, contains 6 out of 26 objects differentially expressed when CLA was fed. CLA isomers have been shown to have anti-inflammatory properties  and affected genes in this pathway were generally down-regulated, adding further evidence of CLA’s anti-inflammatory effects.
The angiopoietin-Tie2 signaling pathway was also affected by CLA treatment. This pathway contains 34 objects, 7 of which were differentially expressed in our experiment (Table 4). Angiopoietin down-regulates NF-κB by binding to endothelial cell receptor tyrosine kinase (Tie2), which eventually results in failure of NF-κB activation. Wunderlich and co-workers  reported repression of NF-κB in the liver of mice fed a high fat diet was associated with accumulation of lipid in hepatocytes, similar to what was observed in the liver of mice fed CLA. CLA modulates NF-κB activation in macrophages , and may have a similar effect in hepatocytes.
A study similar to our study was conducted feeding 0.5% trans 10, cis 12-CLA to female mice . Rasooly and colleagues  targeted genes involved in fatty acid oxidation and we used several of the primers they reported  to directly compare real- time PCR results (Table 3). They found a tendency toward differential expression of acetyl CoA oxidase 1 (acox1), while we found significant down-regulation of this gene. Rasooly and coworkers  found down-regulation of epidermal growth factor receptor (egfr) in their CLA-fed mice and our results support that finding. Insulin-like growth factor binding protein 2 (igfbp2) was differentially expressed in both studies  while fatty acid synthase (fasn) was not differentially expressed in our real-time PCR data, but was in their study. We found no differential expression of malic enzyme (me), while their microarray result was unconfirmed using real-time PCR. Therefore, it is most likely that me is not differentially expressed in liver when trans-10, cis-12-CLA is fed. The few discrepancies between these studies may be due to gender differences because we examined males and they  examined females. Clodfelter and co-workers  reported extensive gender gene expression differences in murine liver.
Networks built around transcription factors using the differentially expressed gene list identified a common transcription factor, hepatocyte nuclear factor 4a (HNF-4a), which regulates gene expression for 98 of the 775 genes (Figure 1). This transcription factor regulates a vast number of genes in hepatocytes and pancreatic islets and mutations within HNF-4a cause the type 1 form of maturity-onset diabetes of the young . HNF-4a is an extensive controller of liver function and previous studies showed that conditional knock- out of this gene in the mouse liver  produces offspring that accumulate lipid in the liver, similar to what is seen when CLA is fed. In addition, pathway analysis indicated that the PKA pathway was affected in this experiment and Viollet and co-workers  reported that PKA phosphorylates HNF-4a, modulating the DNA-binding activity of this important hepatic transcription factor.
Holloway and co-workers  examined murine liver-specific HNF4a-deficiency and found lack of this transcription factor has a greater impact on gene expression in males. They suggest HNF-4a regulates gender-specific gene expression in the liver using mechanisms active in males and primarily inactive in females . We only examined males which may explain why we identified genes regulated by HNF-4a as being differentially expressed.
Co-enzyme A esters of long-chain fatty acids modulate the transcriptional activity of HNF-4a based on chain length and degree of saturation , but no data exist on specific effects of different CLA isomers. Khan and Heuvel  hypothesized that different CLA isomers act as peroxisome proliferator-activated receptor ligands, suggesting specific nuclear receptors might recognize distinct fatty acid structures leading to different phenotypes, perhaps explaining the unique effects of CLA feeding.
Analysis of the amount of fat in our liver samples indicated that only 20% of the increase in liver mass was due to increase in fat. Some of the weight accretion may be due to cell proliferation through inhibition of HNF-4a. Lazarevich and co-workers  showed loss of HNF-4a led to cell proliferation and dedifferentiation in hepatocellular carcinomas. Support for HNF-4a inhibition is found in examination of transcript levels for the genes affected by CLA treatment. Examination of putative functions of the 98 genes regulated by HNF-4a found four to be involved in inflammation and ten to be involved in fatty acid and bile acid metabolism (Table 6). The sodium bile acid cotransporter, slc10a1, was down-regulated in the CLA-fed mice, and this gene is directly activated by HNF-4a . HNF-4a also stimulates two cytochrome P450 genes (cyp27a1 and cyp1a1)  , which were down-regulated in the CLA-fed livers. It appears that bile acid metabolism and transport may be affected in these samples, which was previously found in murine liver accumulating fat after mitochondrial β-oxidation was inhibited . Hnf4a was not differentially expressed (P value = 0.11) in our study, but the HNF-4a coactivator, peroxisome proliferative activated receptor, gamma, coactivator 1 alpha (ppargc1, also known as pgc-1a), was down-regulated in the CLA livers and sterol regulatory element binding factor 2 (srebp2), a competitive inhibitor of PPARGC1 recruitment to HNF-4a , was up-regulated. Therefore, lack of HNF-4a activation, and not differential gene expression of this gene, may be responsible for some of the effects on gene expression we observed.
Previously  we used mic roarrays to identify differentially expressed genes in adipose tissue from these mice after 5 and 14 days of CLA feeding and herein have reported differential expression in the livers after 14 days of CLA treatment. Use of microarrays is one approach to studying the effects of dietary supplementation on gene expression and may provide insight into the major responses that occur. In this study we have begun to elucidate some of the molecular mechanisms of CLA’s effects on the male murine liver, realizing we did not capture the entire picture because we did not measure gene product activity. A comparison of affected GeneGo pathways across the adipose and liver tissues finds that different pathways and genes are differentially expressed. Therefore, the effect of feeding CLA to a polygenic obese line of mice does not appear to be a simple case where one metabolic pathway is up-regulated in one tissue and down-regulated in the other tissue and continued research is needed to clarify the effects of CLA treatment.
This work was funded in part by the North Carolina Agricultural Research Service and the NIH—NIEHS Intramural Research Program.
The authors thank Drs. Ben Corl and Lin Xi for conducting the fatty acid analysis and Peggy Overton for coordinating the phenotypic data collection. BASF kindly donated the trans-10, cis-12-CLA used in this experiment. The authors gratefully acknowledge Shelly Nolin and Audrey O’Nan for technical assistance. During redaction of this article another hepatic microarray paper was published by Guillen et al., identifying other potentially important genes altered during various models of hepatic steatosis.
The authors declare no conflicts of interest.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.