Trans-10, cis 12-Conjugated Linoleic Acid-Induced Milk Fat Depression Is Associated with Inhibition of PPARγ Signaling and Inflammation in Murine Mammary Tissue

Exogenous trans-10, cis-12-CLA (CLA) reduces lipid synthesis in murine adipose and mammary (MG) tissues. However, genomewide alterations in MG and liver (LIV) associated with dietary CLA during lactation remain unknown. We fed mice (n = 5/diet) control or control + trans-10, cis-12-CLA (37 mg/day) between d 6 and d 10 postpartum. The 35,302 annotated murine exonic evidence-based oligo (MEEBO) microarray and quantitative RT-PCR were used for transcript profiling. Milk fat concentration was 44% lower on d 10 versus d 6 due to CLA. The CLA diet resulted in differential expression of 1,496 genes. Bioinformatics analyses underscored that a major effect of CLA on MG encompassed alterations in cellular signaling pathways and phospholipid species biosynthesis. Dietary CLA induced genes related to ER stress (Xbp1), apoptosis (Bcl2), and inflammation (Orm1, Saa2, and Cp). It also induced marked inhibition of PPARγ signaling, including downregulation of Pparg and Srebf1 and several lipogenic target genes (Scd, Fasn, and Gpam). In LIV, CLA induced hepatic steatosis probably through perturbations in the mitochondrial functions and induction of ER stress. Overall, results from this study underscored the role of PPARγ signaling on mammary lipogenic target regulation. The proinflammatory effect due to CLA could be related to inhibition of PPARγ signaling.


Introduction
Dietary nutrients influence the quantity and composition of milk during lactation. Specifically, dietary lipids regulate milk lipid synthesis and milk fatty acid composition in different species of animals. Recently we showed that trans fatty acids (FA) and, in particular, trans-containing conjugated linoleic acid (CLA) isomers regulate murine mammary lipid metabolism to different extents [1]. CLA isomers are the positional and geometric isomers of linoleic acid, an 18-carbon FA with two double bonds. The conjugated double bonds in the CLA are responsible for their biological and biochemical activities. Of the different CLA isomers, the role of trans-10, cis-12-CLA in decreasing milk fat synthesis is well established.
The effects of dietary trans-10, cis-12-CLA on lipid metabolism in adipose and liver have been examined previously [2,3]. Gene expression profiling studies in rodent adipose [2,4], liver [5][6][7], and macrophages [8] have been conducted to help elucidate the molecular mechanisms elicited by trans-10, cis-12-CLA. In adipose and liver, trans-10, cis-12-CLA reduces adipogenesis, increases hepatic steatosis, and leads to insulin resistance, hyperinsulinemia, and inflammation [3,9]. Studies on the effects of trans-10, cis-12-CLA in the mammary tissue of rodents and cows have largely focused on the changes pertaining to lipid metabolism. However, in vitro cell culture and in vivo studies have reported induction of mammary epithelial cell apoptosis at supraphysiological doses of CLA [10,11]. We hypothesized that the effects of trans-10, cis-12-CLA in mammary tissue are not limited to lipid metabolism pathways and may involve other functional networks as has been observed in adipose tissue [4]. The specific objective of this study was to use microarrays and bioinformatics to characterize mammary and liver gene networks that are sensitive to supplemental trans-10, cis-12-CLA during lactation in mice.

Materials and Methods
The experimental procedures were approved by the Institutional Animal Care and Use Committee of the University of Maryland. The mammary and liver samples harvested from a previous experiment [1] were used in the present study. The details about the animals, diets, fatty acid composition of the diets, experimental design, and sample collection are described earlier [1]. Briefly, C57BL/6J mice fed the control diet from day 2 postpartum were randomly allotted to either control or a CLA-supplemented diet on day 6 postpartum. The control diet consisted of sucrose 590 g/kg, vitamin-free casein 200 g/kg, alphacel 50 g/kg, fat 100 g/kg (cocoa butter 14.3 g/kg, corn oil 30.9 g/kg, olive oil 34.8 g/kg, oleic acid 20 g/kg), AIN 76 mineral mixture 40 g/kg, AIN 76 vitamin mixture 15 g/kg, DL-methionine 3 g/kg, and choline bitartrate 2 g/kg. Oleic acid was replaced at 30% (wt/wt basis) by trans-10, cis-12-CLA in the CLA treatment. Oleic acid is incorporated into milk FA, but it was chosen because it has little effect on mammary lipid metabolism [12]. Litter size was adjusted to 6 or 7 pups to maintain uniform milk synthesis among mice. Lactating mice were fed the experimental diets from day 6 to day 10 postpartum. Milk samples were collected on day 6 and day 10 postpartum. On day 10 postpartum, the mice were sacrificed using isoflurane, and individual liver and mammary tissues were collected from dams, fast-frozen in liquid nitrogen, and stored at −80 ∘ C until RNA or lipid extraction.  Set, Illumina Inc.) platform containing 38,467 single-spotted oligonucleotides targeting 35,302 genes was used for transcript profiling. Methods for aminoallyl labeling of cDNA, microarray hybridizations, and scanning were as described earlier [14].

Statistical
Analysis. Data from 40 microarrays (two dyes, two tissue samples/mouse, for a total of 10 animals) used for analysis were initially normalized for dye effects using the median of control elements on the microarray. Subsequently, the log2 normalized ratios of mammary versus reference (i.e., RNA mixture of different tissues including mammary) signal intensities were adjusted for global dye and microarray effects and normalized by Lowess. The data were analyzed using MIXED procedure (SAS Institute Inc., Cary, NC, USA). The fixed effects of the model included tissue and dye while the random effects were mouse and microarray. Raw values for the tissue effect were adjusted using Benjamini and Hochberg's False Discovery Rate (FDR). Differences in relative expression between CLA and control were considered significant at an FDR-adjusted ≤ 0.20. For a more stringent characterization of differentially expressed genes (DEGs), a ≥1.5-fold difference in mRNA expression was set as threshold. We used a more liberal cutoff for the liver data set as very few genes met the criterion used for the mammary tissue. The liver data were assessed using a < 0.05 without an FDR adjustment but with a threshold of ≥1.4-fold difference to determine affected genes. Visualization of within-group variance for DEG in mammary gland tissue and for the entire set of transcripts in liver tissue was performed via hierarchical clustering and box-plot analysis using Gene-Spring GX (Agilent Technologies; Figures 1-3). Data from qPCR after normalization with internal control genes (using the geometric mean of Stx8, Plod3, and Ngb) were analyzed using MIXED procedure (SAS Institute Inc., Cary, NC, USA). The complete dataset files have been deposited at the MIAME database (http://www.ebi.ac.uk/arrayexpress/; username: reviewer E-MEXP-3622, password: veitxmnm).

Bioinformatics Data
Mining. Bioinformatics was conducted using the Dynamic Impact Approach (DIA) according to Bionaz et al. [15]. Briefly, this analysis relied on the publicly available resources Kyoto Encyclopedia of Genes and Genomes (KEGG) database and the Database for Annotation, Visualization and Integrated Discovery (DAVID v6.7,20). Entrez gene IDs were used to identify individual sequences.  GO:0045721 ∼ negative regulation of gluconeogenesis GO:0042026 ∼ protein refolding GO:0045583 ∼ regulation of cytotoxic T cell differentiation GO:0045585 ∼ positive regulation of cytotoxic T cell differentiation GO:0032799 ∼ low-density lipoprotein receptor metabolic process GO:0032802 ∼ low-density lipoprotein receptor catabolic process GO:0032803 ∼ regulation of low-density lipoprotein receptor catabolic process GO:0001887 ∼ selenium metabolic process GO:0016081 ∼ synaptic vesicle docking during exocytosis GO:0009449 ∼ gamma-aminobutyric acid biosynthetic process GO   The KEGG resource has pathway information associated with each gene ID in Mus musculus. There are six major pathway categories, namely, metabolism, genetic information processing, environmental information processing, cellular processes, organismal systems, and human diseases. In addition, we used Ingenuity Pathways Analysis (IPA) relying on the entire microarray data set with associated statistical values.
The DAVID resource provides typical batch annotation and gene-GO term enrichment analysis to highlight the most relevant GO terms associated with a given gene list. The latest version provides extended annotation content coverage including GO terms, protein-protein interactions, protein functional domains, disease associations, biological pathways, sequence general features, homologies, gene functional summaries, and tissue gene expression. The ToppGene suite (http://toppgene.cchmc.org/) was used for functional enrichment analysis of data from liver tissue at an uncorrected < 0.05 and a threshold of ≥1.4-fold in CLA versus control. ToppGene suite is a web-based portal that provides 17 categories of annotations, including GO-molecular function, GO-biological process, GO-cellular component, human phenotype, mouse phenotype, protein domains, pathways, pubMed cocitations, protein-protein interactions, cytoband, transcription factor binding sites, gene family coexpression, computational expression correlations, micro-RNA targets, drug, and disease. The database details for each category of annotations are presented in the ToppGene web portal (http://toppgene.cchmc.org/).

Results
We have previously reported the effect of trans-10, cis-12-CLA on milk production, milk fat content, milk fatty acid composition, liver weight, and liver FA composition [16]. Briefly, compared with the control, CLA decreased milk fat percentage by 44% and reduced the proportion of de novo synthesized FA (FA ≤ C 16:0) while increasing 20:4n-6 ( Table 2). The trans-10, cis-12-CLA was preferentially taken up by the mammary tissue (1.42 g/100 FAME in milk) compared with liver (0.20 g/100 FAME). However, the liver weight increased by 32% in mice fed CLA as a consequence of increased lipid content (0.47 g FAME/g liver, dry) compared with control (0.24 g FAME/g liver, dry). Furthermore, the proportions of FA 18:0, 18:3, and 20:4n-6 decreased ( < 0.01) in the liver of mice fed CLA diet ( Table 2).

Bioinformatics.
A total of 217 pathways in the KEGG database (murine genome) contained 30% or more genes represented on the microarray platform; these were deemed appropriately to be covered for further analysis. Among the most-impacted pathways (average impact 5.78 ± 2 SD (16.4)) that were affected due to feeding CLA versus control were biosynthesis of unsaturated fatty acids, PPAR signaling pathway, adipocytokine signaling pathway, and nitrogen metabolism ( Figure 2). Overall, those pathways were inhibited due to dietary CLA. Pathways are considered moderately affected by feeding CLA (5.78 ± 1 SD (11.1)) and with clear activation included glycosphingolipid biosynthesis and vasopressin-regulated water reabsorption. Other pathways are considered modestly affected by feeding CLA (5.78 ± 0.5 SD (8.4)) including riboflavin metabolism, fatty acid elongation in mitochondria, ErbB signaling, and pentose and glucuronate interconversions, which were activated ( Figure 2).
The DAVID categories biological process (GOTERM BP ALL), molecular function (GOTERM MF ALL), and cellular component (GOTERM CC ALL) are shown in Figure 3. Within biological processes are "biological goals" such as regulation of gluconeogenesis which is accomplished by ordered assemblies of molecular functions. Molecular Function describes the tasks performed by individual gene products, for example, calcium channel inhibitory activity. The cellular component classification type involves subcellular structures, localization, and macromolecular complexes, for example, Golgi cisterna.
The top 10 GO terms from our analysis are shown in Figure 3, and the approach for calculating impact and flux within each GO category has been described previously [14,15]. Briefly, the length of the impact bar denotes the relative biological significance of the gene identifiers in each category; that is, a bigger bar reflects a more important biological term based on DEG. Similarly, the relative direction of the flux (activation or inhibition) is denoted by the intensity of color as reported in Bionaz et al. [15]. Results using DIA indicated that, within biological processes, only 2 of the top 10 mostimpacted categories were inhibited, that is, negative regulation of gluconeogenesis and selenium metabolic process, whereas protein refolding and regulation of cytotoxic T cell differentiation were highly impacted and activated ( Figure 3). Interestingly, the GO terms within cellular component were the least impacted among the three GO categories and mostly seemed to be inhibited in response to feeding CLA, for example, golgi cis cisterna and beta-catenin destruction complex ( Figure 3). Within molecular functions, feeding CLA led to an activation of calcium channel inhibitor activity, myosin head/neck binding, and ATP dependent protein binding, whereas it inhibited retinol transporter activity and nicotinamide phosphoribosyltransferase activity.

Microarray Verification.
The microarray results were verified by quantitative real time RT-PCR assays for genes regulating lipid metabolism, apoptosis, inflammation, acute-phase response signaling, and transcription regulation ( Table 5). Of the 21 genes tested, 17 genes were similar to those of microarray and four were false-positives (Table 5).

3.2.
Liver. The effect of trans-10, cis-12-CLA on the liver transcriptome was modest at the same criterion used for selecting DEG of mammary tissue. Hierarchical clustering (not shown) across individual animals and the box-plot analysis (not shown) revealed a high-degree of animal variation in response to CLA or control, which is not unexpected based on previous studies with lactating mice [17]. We were unable to cluster by gene due to the large size of the dataset (38,185 elements); that is, there were no statistically significant genes in liver at the stringent cutoff used with mammary data. There were 170 DEGs when the selection criteria were relaxed (uncorrected < 0.009, 1.5-fold difference in CLA versus control) with 72 upregulated and 98 down-regulated. The CLA diet did not affect genes related to FA synthesis (e.g., Fasn, Acc, and Scd1; data not shown) even though there was increase lipid accumulation in the liver. Some of the genes related to FA oxidation (Acox1, Cpt1c, and Fmo3) were downregulated (data not shown). Among the genes related to FA transport, CD36, Fabp1, and Fabp2 were upregulated while Fabp4 was down-regulated (data not shown).  Feeding CLA increased liver mass and altered hepatic FA composition ( Table 2). The greater liver mass was due in part to lipid accumulation and was associated with several alterations in gene expression. For instance, the "mouse phenotype" feature of the ToppGene suite used for the gene enrichment analysis identified hepatic steatosis, abnormal Kupffer cell morphology, abnormal liver sinusoid morphology, abnormal mononuclear phagocyte, abnormal circulating creatinine level, and abnormal circulating amino acid level as the top terms affected by CLA feeding ( Table 6). The most impacted pathways by CLA were electron transport chain, glucose regulation of insulin, glyoxylate cycle, biosynthesis of unsaturated FA, alanine and aspartate metabolic pathway, erythrocytes differentiation pathway, and tyrosine degradation ( Table 7). The finding of an overall upregulation of the unsaturated FA biosynthesis pathway was due to the fact that the pathway includes several upregulated genes (e.g., Acot2, Fads1, Fads2, Agpat2, and Elovl5; data not shown) as identified by ToppGene.
The GO categories GO: molecular function, GO: biological process, and GO: cellular component are presented in Table 8. Feeding CLA activated molecular functions related to oxidoreductase activity, chaperone binding, unfolded protein binding, pyrophosphatase activity, but inhibited cofactor binding, pyridoxal phosphate binding, and vitamin B6 binding. CLA feeding upregulated the biological processes related to oxoacid-, carboxylic acid-, organic acid-, and ketonemetabolic processes in addition to cellular catabolic process. Mitochondrion was identified as the most-impacted cellular component in the liver of mice fed CLA.

Discussion
We observed contrasting effects of trans-10, cis-12-CLA on liver and mammary tissue, two predominantly lipidsynthesizing tissues in lactating mice. In the mammary tissue, the reduction in milk lipid content was characterized by lower concentrations of de novo synthesized fatty acids (FA ≤ C 16:0) as a consequence of inhibition of several genes related to lipid synthesis (data not shown). Lipid metabolism was the main cellular function affected by trans-10, cis-12-CLA, and the large impact and inhibition of biosynthesis of unsaturated fatty acids KEGG pathway confirmed such effect. The bioinformatics analyses underscored that a major effect of trans-10, cis-12-CLA on mammary tissue encompassed alterations in signaling pathways and phospholipid species biosynthesis.
Substantial regulation of milk lipid synthesis in lactating mice occurs at the mRNA level, and there is evidence suggesting that part of the pathway is regulated by SREBP1 [18,19]. The lower expression of Srebp1 and its downstream genes in previous studies provided evidence of its role in murine milk lipid synthesis. However, the effect of trans-10, cis-12-CLA on SREBP1 is indirect as FA cannot bind to this transcription factor [20]. In contrast, FA are natural ligands of the nuclear receptor PPAR which is known to regulate lipogenesis in some cell types [21][22][23]. The overall inhibition of PPAR signaling KEGG pathway, the reduction in Ppar expression and its downstream genes regulating FA uptake (Fabp4), FA synthesis (Fasn), FA desaturation (Scd), and lipid droplet formation (Cav1) due to trans-10, cis-12-CLA, all support a role for PPAR in murine mammary lipid metabolism.
We have previously shown that activation of PPAR could upregulate mammary lipogenic gene networks in bovine mammary epithelial cells [13]. CLA has also been shown to act as a PPAR activator in some cell types [24]. However, an important consideration in the context of PPAR regulation of murine mammary lipid synthesis is the fact that the mammary gland of this species contains a substantial amount of adipocytes which may in fact be an important source of PPAR and also adipocytokine mRNA expression [25].
Glycosphingolipid biosynthesisganglio series was the most activated KEGG pathway in addition to being one of the most-impacted pathways by trans-10, cis-12-CLA. Glycosphingolipids are key components of the apical plasma membrane and subcellular components including vesicles [26]. Glycosphingolipids are involved in molecular functions such as differentiation, apoptosis, and cell-cell interaction [26]. In murine mammary tissue, ganglioside GD1 is specifically increased during lactation and is highly concentrated in the milk fat globule membranes and helps prevent aggregation of milk fat globules [27]. We have previously observed a shift in the lipid globule distribution towards an increase in the number of smaller lipid droplets in the secreted milk as well as intracellular mammary epithelium in mice fed trans-10, cis-12-CLA [17], and this could be due to altered glycosphingolipid biosynthesis in addition to reduced lipogenesis.
Based on microarray data from murine mammary tissue Rudolph et al. [19] suggested that gluconeogenesis per se is not an important function in murine mammary epithelium. However, the enzyme PCK1 which is a key component of this pathway in liver also is central for the generation of glycerolphosphate for esterification during the process of glyceroneogenesis [28]. Thus, it is not surprising that the top mostimpacted GO biological process due to dietary CLA was inhibition of negative regulation of gluconeogenesis; that is, supply of glucose to mammary gland might have been decreased with a consequent increase in use of other carbon sources (e.g., lactate) to generate glycerol-phosphate.
In addition to its effects on lipid metabolism, in the IPA analysis trans-10, cis-12-CLA affected cellular processes related to cell cycle progression, cell-cell interactions, and cell assembly and organization (data not shown). These effects of CLA were probably through signaling pathways regulating ErbB receptors and the actin cytoskeleton ( Figure 2). ErbB signaling was identified as one of the top 10 impacted KEGG pathways in our study. The ErbB signaling pathway in murine mammary gland is involved in the control of cell survival, cell-cell interaction and cellular differentiation, cell cycle progression, and morphogenesis [29,30]. Activation of the ErbB pathway is considered beneficial for cellular proliferation in damaged tissue [30]; thus, it could be possible that the activation of this pathway in our study was a compensatory response due to CLA. Proliferation of mammary epithelium and ductal hyperplasia has also been observed in CLA-fed Table 6: Top terms associated with the "mouse phenotype" and "disease" functions in the liver of mice fed trans-10, cis-12-CLA using the ToppGene suite application.   mice independent of alterations of ErbB expression [31]. Ovariectomized mice fed CLA experienced an estrogenindependent allosteric mammary growth characterized by ductal elongation due to enlarged proliferative terminal end buds [32]. In addition, CLA feeding was associated with increased incidence of tumor development and progression [31,33]. Regulation of actin-based motility by rho was one of the top 5 canonical pathways identified by IPA analysis. Dietary CLA upregulated expression of actin B (Actb) and actin G (Actg), which play an important role in formation of stress fibers connecting extracellular matrix to the intracellular medium. Profilin (Pfn) is involved in actin filament polymerization [34]. Rho proteins regulate Actb, Actg, and Pfn [35] and can affect actin polymerization which can actively drive vesicle movement in cells [36]. The ras homolog D (RhoD), which is involved in endosomal dynamics and rearrangement of actin cytoskeleton, was also upregulated by CLA in this study. RhoD is involved in the regulation of both membrane traffic and cytoskeleton in the cell [37]. Overexpression of RhoD induced remodeling of actin cytoskeleton accompanied by increased endosomal fission and scattering of vesicles in the cells [37]. Thus, our results suggested that CLA either directly or indirectly caused a substantial remodeling of cellular structures. It could be possible that such effects were partly due to changes in unsaturated FA availability, for example, for cellular membrane formation.

ID
Trans-10, cis-12-CLA causes marked dilation of endoplasmic reticulum (ER) in the mammary epithelial cells and upregulates x-box binding protein 1 (Xbp1) expression. In addition, CLA is also known to induce the ER stress response by increasing splicing of XBP1 mRNA and activation of c-Jun N-terminal kinase (JNK) signaling [38,39]. In our study, a greater expression of Xbp1 and Jnk1, genes coding key proteins in the ER stress response, due to CLA suggested the induction of an unfolded protein response (UPR) in the ER. That idea is supported by the downregulation of calnexin (Canx), a molecular chaperone which mediates the proper folding of nascent proteins in ER, coupled with upregulation of BiP and HSP90AA1 which bind to misfolded proteins. Thus, the activation of UPR with CLA might have been associated with greater ER stress and underscored a marked derangement in posttranslational modification of proteins. A key feature of the UPR is cellular inflammation [40], and the activation of cytotoxic T cell differentiation along with several acute-phase proteins in our study seems to confirm that CLA exerted a proinflammatory state in mammary cells (Figure 4). In this context, the inhibition of golgi cis cisterna, beta-catenin destruction complex, and nuclear exosome could have been a response to ER stress and inflammation.
Trans-10, cis-12-CLA inhibits cellular proliferation and induces apoptosis in primary mammary epithelial cells [10]. Different mechanisms have been proposed for the induction of apoptosis by CLA: (a) p53-dependent [41,42]; (b) mitochondrial pathway targeting Bcl2 [43]; and (c) ER response involving induction of Xbp1, phosphorylation of eIF2 , induction of CHOP, and the cleavage of caspase 12 [38]. In this study, in addition to the induction of ER stress response, a gene related with the mitochondrial pathway, for example, Bcl2, was also affected. The Bcl protein decreases the activation of caspase 3 involved in apoptosis [44], and a mouse Bcl2-mutant strain has greater rates of apoptosis in cells [45].
The murine mammary tissue is characterized by an upregulation of acute-phase proteins during the initial phase of involution, but the exact role of such response has not been determined [52][53][54]. In our study, a proinflammatory effect induced by CLA on mammary tissue could have led to decreased milk synthesis capacity of the epithelial cells and a subsequent effect on the growth of pups nursing mice fed trans-10, cis-12-CLA [1]. The marked accumulation of 20:4n-6 in milk of CLA-fed mice might have played a role in the proinflammatory response observed, for example, by enhancing the production of oxidized lipid products [55]. The accumulation of 20:4n-6, however, contrasts with the marked decrease in its concentration in milk of lactating dairy cows receiving an exogenous infusion of CLA that reduced milk fat synthesis [56].
The acute-phase response proteins are closely regulated at the transcriptional level by STAT3 and NFkB [57]. ER stress can induce upregulation of acute-phase proteins through NFkB activation [58]. None of these transcriptional regulators was upregulated with CLA. However, there was upregulation of the proinflammatory cytokines IL6 and TNF. IL6 is known to exert its action through increasing STAT3 [52]. During the acute-phase response, TNF decreases the expression of nuclear receptors PPAR , PPAR , RXR , and LXR [43] which in turn could affect the lipid-synthesizing transcriptome. In addition to its effects on lipid metabolism, PPAR dampens inflammation via transrepression of proinflammatory molecules [21]. For instance, PPAR inhibits inflammation by direct interaction with AP1 and NFkB, thus, preventing their binding to response elements on inflammatory genes [21]. Activation of PPAR by synthetic PPAR agonists reduces the expression of inflammatory cytokines and chemokines in bovine mammary epithelial cells [59]. While FA are considered as natural ligands of PPAR [21][22][23], the lower PPAR expression and greater acute-phase protein expression, which confirmed the existence of a marked degree of inflammation, suggested that trans-10, cis-12-CLA antagonizes PPAR .
Further work is needed to determine if the ability of PPAR to repress proinflammatory molecules is diminished due to its interaction (e.g., competitive inhibition) with trans-10, cis-12-CLA. Overt inflammation might reduce transcription of anti-inflammatory nuclear receptors (e.g., feedback inhibition) leading to negative effects on cellular mechanisms including lipid synthesis and secretion. In lactating mouse mammary tissue targeted deletion of Ppar resulted in greater concentration of proinflammatory lipids in milk leading to reduced pup growth and induction of inflammation. Those results highlighted the importance of PPAR signalling during lactation [55]. Some of the effects of trans-10, cis-12-CLA in our study might have been regulated through its effects on PPAR .
The effect of trans-10, cis-12-CLA on liver was dramatic as CLA caused accumulation of lipid in the span of four days. The associated increase in liver mass due to TAG accumulation suggested induction of hepatic steatosis in CLA-fed animals. This was confirmed with the gene functional analysis as "hepatic steatosis" was one of the top terms associated with the "mouse phenotype" category. Primary biliary cirrhosis and liver cirrhosis were the top terms in the "disease" category suggesting that trans-10, cis-12-CLA feeding predisposed the mice to these conditions. The onset of hepatic steatosis could predispose liver to inflammation and oxidative stress leading to steatohepatitis and eventually to liver cirrhosis [60,61].
The development of hepatic steatosis is specific for trans-10, cis-12-CLA feeding but not to other CLA isomers [62]. Hepatic steatosis has been previously reported in growing mice after 14 days of trans-10, cis-12-CLA feeding [63].
Potential mechanisms involved in the onset of hepatic steatosis due to trans-10, cis-12-CLA have been reviewed recently [3], but it is important to recognize that all those previous studies have been conducted with nonlactating animals. The reduction in FA oxidation coupled with increased uptake of FA may be one of the reasons for hepatic lipid accumulation [3]. Decreased FA oxidation could lead to accumulation of intracellular diacylglycerol concentration which is implicated in hepatic insulin resistance through activation of protein kinase C [26]. In a recent study, CLA-induced hepatic steatosis was associated with an increase in diacylglycerol content and membrane associated protein kinase C [64] suggesting that CLA feeding could potentially lead to hepatic insulin resistance.
In this study, mitochondria, the primary organelle involved in FA oxidation, were the most-impacted cellular component followed by ER and peroxisomes. Mitochondrial dysfunction causes hepatic steatosis and has been implicated in the progression from nonalcoholic fatty liver disease (NAFLD) to non-alcoholic steatohepatitis (NASH) [65]. The top affected pathways (e.g., electron transport chain and glucose activation of insulin secretion), molecular functions (e.g., oxidoreductase activity), biological processes (e.g., oxoacid metabolic process, and cellular catabolic process, cellular ketone metabolic process), and diseases (e.g., Leigh disease, Leigh syndrome) are related to mitochondria further confirming that trans-10, cis-12-CLA feeding affected hepatic mitochondrial functions.
The ER was the second most affected cellular organelle in the liver of trans-10, cis-12-CLA-fed mice. The activation of molecular functions such as chaperone binding and unfolded protein binding suggests induction of ER stress response in hepatic tissue. Induction of ER stress could further deteriorate the steatotic condition by augmenting hepatic lipogenesis via activating proteolytic cleavage of SREBP1 [66], decreasing hepatic secretion of TAG by reducing apoB secretion [67] and upregulation of Xbp1 [68]. The transcriptional regulator Xbp1, a known regulator of the unfolded protein response, regulates the expression of lipogenic genes Scd1, acetyl-CoA carboxylase beta (Acc2), and diacylglycerol acetyltransferase 2 (Dgat2), independent of SREBP1 and ChREBP [68]. Also, a shift in hepatic FA composition towards saturated FA disrupts ER homeostasis and promotes ER stress [69,70]. An increased ratio of saturated to unsaturated FA is considered as a secondary hit in the progression of steatosis to steatohepatitis [69,71]. Trans-10, cis-12-CLA-fed mice in our study had greater hepatic palmitate concentration (67% over control) thereby increasing the ratio of saturated FA. Thus, those data confirm previous reports that the induction of ER stress could be caused in part by altered hepatic FA profile.
As indicated above, the presence of adipocytes in the mammary tissue used for transcriptomics is an important consideration when interpreting our data, particularly as it relates to PPAR and adipocytokine signaling. Adipose tissue constitutes ca. 20% of the lactating murine mammary tissue in early lactation [69] and in rodents is indispensable for the normal function of the mammary epithelium [16,72]. It has been suggested that the gene expression changes in a complex tissue like rodent mammary could be a function of changes in the gene expression within a particular cell type (i.e., mammary or adipose) or the proportion of the given cell type (i.e., mammary or adipose) [25]. This could be an important factor during different stages of mammary function and development when the abundance of the two cell types varies substantially [25]. Our study was conducted during peak lactation when the adipose content is at its lowest, but further studies are needed to delineate more closely the impact of trans-10, cis-12-CLA on the relative abundance of mammary and adipose cell types in the rodent mammary gland during peak lactation.

Conclusions
During lactation the effect of CLA on the liver was similar to those observed previously in growing mice. The hepatic gene enrichment functional analysis revealed induction of hepatic steatosis, perturbations in the mitochondrial functions, and induction of ER stress. The major effect of CLA on mammary tissue encompassed alterations in cellular signaling pathways, phospholipid species biosynthesis, and a marked degree of inflammation and ER stress. Such responses might be associated with enhanced cellular apoptosis and precocious involution of the mammary gland. Inhibition of PPAR signaling was associated with decreased milk fat synthesis and uncontrolled inflammation.