A functional nutrigenomics approach
Promotor Prof. dr. Michael Müller
Hoogleraar Nutrition, Metabolism and Genomics Humane Voeding, Wageningen Universiteit
Co-promotor Dr. Guido J.E.J. Hooiveld
Universitair docent
Humane Voeding, Wageningen Universiteit Promotiecommissie Prof. dr. Jaap Keijer
Wageningen University Prof. dr. Ulrich Beuers University of Amsterdam Prof. dr. Hannelore Daniel Technical University of Munich Prof. dr. Ivonne Rietjens Wageningen University
A functional nutrigenomics approach
Meike Bünger
Proefschrift
ter verkrijging van de graad van doctor op gezag van de rector magnificus
van Wageningen Universiteit, Prof. dr. M.J. Kropff, in het openbaar te verdedigen op vrijdag 12 september 2008 des namiddags te vier uur in de Aula.
Meike Bünger. Probing the role of PPARα in the small intestine: A functional nutrigenomics approach.
PhD Thesis. Wageningen University and Research Centre, The Netherlands, 2008. With summaries in English and Dutch.
Background The peroxisome proliferator-activated receptor alpha (PPARα) is a ligand-activated transcription factor known for its control of metabolism in response to diet. Although functionally best characterized in liver, PPARα is also abundantly expressed in small intestine, the organ by which nutrients, including lipids, enter the body. Dietary fatty acids, formed during the digestion of triacylglycerols, are able to profoundly influence gene expression by activating PPARα. Since the average Western diet contains a high amount of PPARα ligands, knowledge on the regulatory and physiological role of PPARα in the small intestine is of particular interest.
Aim In this thesis the function of PPARα in the small intestine was studied using a combination of functional genomics experiments, advanced bioinformatics tools, and dietary intervention studies.
Results Detailed analyses on the expression of PPARα in small intestine showed that PPARα is most prominently expressed in villus cells of the jejunum, coinciding with the main anatomical location where fatty acids are digested and absorbed. Genome-wide transcriptome analysis in combination with feeding studies using the synthetic agonist WY14643 and several nutritional PPARα agonists revealed that PPARα controls processes ranging from fatty acid oxidation and cholesterol-, glucose- and bile acid metabolism to apoptosis and cell cycle. In addition, we connected PPARα with the intestinal immune system. In a more focussed study we showed that PPARα controls the barrier function of the intestine. By comparing the intestinal and hepatic PPARα transcriptome we found that PPARα controls in these two organs the expression of two distinct, but overlapping sets of genes. Finally, by performing a range of functional studies deduced from the transcriptome analysis, we demonstrated that PPARα controls intestinal lipid absorption.
Conclusion By maximally utilizing the unique possibilities offered in the post-genome era, the studies described in this thesis reported on the function of PPARα in small intestine. We conclude that intestinal PPARα plays an important role, is relevant for nutrition, and its effects are distinguishable from the hepatic PPARα response. Our results provide a better understanding of normal intestinal physiology, and may be of particular importance for the development of fortified foods, and prevention and therapies for treating obesity and inflammatory bowel diseases.
The small intestine plays a critical role in nutrition, since it is the primary site of food digestion and nutrient absorption. Another function is to prevent the translocation of bacteria and foreign antigens to extra-intestinal sites by forming a selective barrier. It has been demonstrated that the transcription factor PPARα can be activated by natural fatty acids and their activated derivatives (acyl-CoA esters) [1-4]. PPARα is expressed in a variety of tissues including the small intestine [5, 6]; however, its function has been almost exclusively studied in liver. Little is known about PPARα and PPARα target genes in non-hepatic tissues. Knowledge on the regulatory and physiological function of PPARα in the small intestine is of particular interest, since the average Western diet contains a high amount of triacylglycerols [7] that are hydrolyzed to monoacylglycerol and free fatty acids before entering the enterocyte [8]. Consequently the small intestine is frequently exposed to high levels of PPARα agonists, and therefore an important functional role of this transcription factor may be envisioned. The aim of the research described in this thesis was therefore to characterize the function of PPARα in the small intestine, with special emphasis on nutritional relevance, utilizing a nutrigenomics approach.
In chapter 1 an overview of the literature is given concerning the gastrointestinal tract, the concept of nutrigenomics research, the peroxisome-proliferator-activated receptors (PPARs), and the technology used in this thesis. The results of published microarray studies on PPARα-dependent gene regulation performed in different organs are also summarized in this first chapter. In chapter 2 the genome-wide effect of PPARα activation in the small intestine is reported, with focus on intestinal epithelial cells. The commonalities and differences of PPARα dependent gene regulation in small intestine and liver was the subject of the studies described in chapter 3. In chapter 4 and 5 we focus on two main functions of the small intestine, the nutrient absorption (chapter 5) and the selective barrier function (chapter 4). In chapter 4 the effect of PPARα-activation by three different fatty acids and the synthetic compound WY14643 is examined, with emphasis on effects on intestinal transporters and phase I/II metabolic enzymes. In chapter 5 the effect of PPARα activation on intestinal dietary lipid absorption is investigated. Finally, in chapter 6 the general discussion, conclusions and recommendations are presented.
Abstract
Aim and outline of this thesis 1 General introduction
2 Genome-wide analysis of PPARα activation in murine small intestine 3 Organ-specific function of PPARα as revealed by gene expression profiling 4 PPARα-mediated effects of dietary lipids on intestinal barrier gene expression 5 PPARα regulates intestinal lipid absorption
6 General discussion Appendix
References
Summary of this thesis Samenvatting Dankwoord Curriculum Vitae List of publications Educational programme 7 10 32 58 82 102 126 132 134 144 148 154 160 162 5 164
Apated from:
Meike Bünger, Guido Hooiveld, Sander Kersten and Michael Müller, “Exploration of PPAR functions by microarray technology – a paradigm for nutrigenomics”,Biochimica et Biophysica Acta - Molecular and Cell Biology of Lipids, Volume 1771, Issue 8, pages 1046-1064.
12
The gastrointestinal tract: function and anatomy
The gastrointestinal tract is the system of organs which serves two main functions — assimilation of nutrients and elimination of waste. The gut anatomy is organized to serve these functions. A schematic overview of the gastrointestinal tract and accessory organs is given in Figure 1. The gastrointestinal tract starts with the mouth, followed by pharynx, esophagus, stomach, small intestine, colon and ends with the rectum (Figure 1). Hence, anatomically the small intestine is localized in-between stomach and colon and is directly connected with the accessory digestive organs, pancreas and liver via small ducts attached to the upper part of the small intestine.
The physical and chemical digestion of foods starts in the mouth by chewing and the salivary enzyme amylase. This mixture is conducted to the stomach via the esophagus, where it is mixed with the acidic gastric juice for further degradation. The acid environment in the stomach also kills most bacteria that might have entered together with the food. Proteins are hydrolyzed into polypeptides by the protease pepsin in the stomach.
Most of the enzymatic hydrolysis of lipids and other food components occurs in the small in-testine. However, pancreas and liver both contribute to this process. The enzymes secreted by the exocrine pancreatic tissue help break down carbohydrates, fats, and proteins in the small intestine. The liver has a wide variety of functions, among others to produce bile to assist in the digestion of food. Both pancreas and liver release their contents via the greater duodenal papilla (comprised of the ampulla of Vater and the sphincter of Oddi) into the upper small intestine, the duodenum.
Next to digestion and nutrient absorption a healthy gut incorporates another vital function. By forming a selective barrier the translocation of bacteria and other foreign antigens to extra-intes-tinal sites is prevented.
Figure 1. The diges-tive tract. (Adapted from
Campbell NA, 1987, Biol-ogy, the Benjamin/Cum-mings Publishing compa-ny, Inc, Redwood City.)
13
This barrier function is part of the innate immune system. Together with the host response, which is triggered through the activation of specific transcription factors controlling chemokine and cy-tokine expression, both innate and adaptive defense mechanisms protect the body against patho-gens. An overview of small intestinal anatomy is shown in Figure 2. The mucosa is the inner lining of the small intestine and consists of three layers: the epithelium, the lamina propria, and the muscularis mucosae. To enlarge the absorptive area of the small intestine, the mucosa and submucosa is arranged into high folds, called plicae circulars (valves of Kerckring).
Numerous villi, which are finger like structures, further increase this area (Figure 2a). A villus consists of a single line of epithelial cells. This epithelium undergoes perpetual renewal, fueled by a population of multipotential stem cells located at the base of the crypts of Lieberkühn. There are four main cell lineages the multipotential stem cells differentiate into: (1) enterocytes, (2) goblet cells, (3) enteroendocrine and (4) paneth cells. Enterocytes are the dominant cell popula-tion of small intestinal villi. They are polarized cells with a basal nucleus and an apical brush border called microvilli. These cells are responsible for the (selective) absorptive properties of the small intestine, thus are essential for nutrient and drug uptake (Figure 2c). Goblet cells are mucin-secreting cells, protecting and lubricating the mucosa.
The enteroendocrine cells form a deceptively large population and are suggested to “communi-cate” with the intestinal lumen. They contain numerous basally-sited, dense core neurosecretory granules, which contain secreted peptide hormones. Enteroendocrine cells are scattered through-out the whole intestinal epithelium, unlike the Paneth cells, which are found exclusively at the crypt base. Paneth cells contain large secretory granules, and express a number of proteins, in-cluding lysozymes and cryptins with antibacterial properties.
Figure 2. The anatomy of the small intestine.
(Adapted from Marieb EN 2004, the Digestive Sys-tem in Human Anatomy & Physiology. Pearson Benjamin Cummings, San Francisco, USA.)
14
Cellular differentiation in the small intestine is found along two distinct axes, the proximal-distal (also termed longitudinal or cephalocaudal axis) and the crypt-villus axis. Along the proximal-distal axis, the small intestine can be anatomically and functionally divided into three parts, the duodenum, jejunum, and the ileum. Cytological changes along this proximal-distal axis occur, in that (a) the goblet cell number increases (b) villi become more finger-like and decrease in length, (c) lymphoid tissue increases and (d) plicae circulares diminish.
The crypt-villus axis is a vertical axis with epithelial cells extending bidirectional upwards and downwards from the stem cells anchored adjacent to the crypt base. Enterocytes, mucous produc-ing goblet cells, and enteroendocrine cells migrate out of the crypt towards the villus, while Pa-neth cells migrate towards the crypt-base. Structural differentiation and functional specialization of each of the four cell lineages occurs along this crypt-villus axis.
The cytological and functional changes translate into differential gene expression along the prox-imal-distal and crypt-villus axis. Differential gene expression along these axes and between the four different cell types has been examined in some studies [1-3].
15
Nutrigenomics: Molecular nutrition + genomics
Traditionally, nutritional science was mainly concentrated on nutrient deficiencies and their effects on health and disease. However, over the past few decades, research emphasis has gradually shifted to the link between (over)-nutrition and chronic diseases, including cancer, obesity, cardiovascular disease, and diabetes [4-8]. In parallel, there has been increasing interest in the molecular mechanisms underlying the beneficial or adverse effects of foods and food components. More recently, driven by the continuing and accelerating discoveries in omics technology, unique possibilities have emerged to investigate the genome-wide effects of nutrients at the molecular level. This research field of gene-nutrient interactions is called nutritional genomics or nutrigenomics, and encompasses the fields of biotechnology, genomics, molecular medicine and human nutrition [5, 7, 8]. It is widely recognized that nutrigenomics has the potential to increase our understanding of how nutrition influences metabolic pathways and homeostasis, how this regulation is disturbed in a diet-related disease, and to what extent individual genotypes contribute to such diseases [4-8].
However, as opposed to pharmacological research, nutritional research needs to take into account specific problems inherent to nutritional interventions. The rather small deviations from homeostasis that characterizes the early phase of a metabolic disease [9, 10], and the complexity as well as the variability of consumed foods in general are just two examples. The body has to handle a variety of nutrients at the same time, each of which can have numerous targets with different affinities and specificities. This contrasts starkly with pharmacology, where single agents are used at low concentrations and act with relatively high affinity and selectivity for a limited number of biological targets. The challenge of nutrigenomics research is to dissect the important but complex research problems into small and feasible projects. This requires simplification of our model systems in order to obtain correct and clear answers to the research question addressed. A fruitful strategy is to borrow methods that are well established in medical or pharmacological research but are rather new in the field of nutritional research. For example, in analogy to pharmacology, nutrients can be considered as signaling molecules that are recognized by specific cellular sensing mechanisms [5]. Since the property that allows nutrients to activate specific signaling pathways is carried in their molecular structure, small changes in structure can have a profound influence on which sensor pathways are activated.
16
PPARs and nutrition
Nutrients impact gene expression mainly by activating or suppressing specific transcription factors [5, 11]. The most important group of transcription factors involved in mediating the effect of nutrients and their metabolites on gene transcription is the superfamily of nuclear receptors, which consists of 48 members in the human genome [12]. This superfamily is subdivided into six families [13], of which the NR1 family is most relevant to nutrition. Nuclear receptors govern gene expression via several distinct mechanisms that involve both activation and repression of DNA transcription. After site-specific DNA binding, their final transcriptional activity depends on physical interactions with a set of associated proteins, the so-called coactivators and corepressors. These coregulators are not exclusive to nuclear receptors and are recruited in a similar manner by numerous other DNA-binding transcription factors [13-15].
One important group of receptors that mediates the effects of dietary fatty acids on gene expression are the Peroxisome Proliferator Activated Receptors (PPARs, NR1C) [13, 16, 17]. Three PPAR isotypes, α (NR1C1), δ (also called β) (NR1C2), and γ (NR1C3) can be distinguished and characterized by different biological roles. Transcriptional regulation by PPARs requires heterodimerization with the retinoid X receptor (RXR; NR2B) [18], which is also part of the nuclear receptor superfamily [13, 19]. When activated by an agonist, the PPAR/RXR heterodimer stimulates transcription via binding to DNA response elements (PPRE) present in and around the promoter of target genes. Besides upregulating gene expression, PPARs are also able to repress transcription by directly interacting with other transcription factors and interfere with their signaling pathways, a mechanism commonly referred to as transrepression [20].
Although much is already known about PPARs, gaps in our knowledge remain. In so far as the biological role of a particular PPAR is directly coupled to the function of its target genes, probing PPAR-regulated genes via the application of genomics tools can greatly improve our understanding of PPAR function. By combining transgenic animal models with elaborate microarray analyses, a comprehensive understanding of the in vivo role of PPARs can be obtained. See e.g. Michalik et al [18] for a current overview of PPAR agonists and models. However, it should be realized that transcriptome profiling does not unequivocally demonstrate that differentially expressed genes are direct PPAR targets. Direct regulation has to be shown through additional methods, such as chromatin immunoprecipitation, yeast one-hybrid and transactivation assays.
17
Fundamentals of microarray technology
One of the most powerful technologies to emerge from the age of genome sequencing are DNA microarray slides, which enable the comparatively scanning of genome-wide patterns of gene expression for any organism with a sequenced genome. Nowadays applications of microarrays fall into three main categories: studies of genomic structure, gene expression profiling, and profiling of protein-DNA interactions. The application of arrays for genomic studies primarily involves the search for single-nucleotide polymorphisms and DNA copy-number variation, which may have considerable importance regardless of whether they cause an overt disease [21-23]. Gene expression profiling, or transcriptomics, is extensively used to study how cells respond to certain stimuli in order to identify altered molecular pathways or to diagnose and predict clinical outcomes [24-26]. Protein-DNA interaction-profiling, or ChIP-on-chip [chromatin immunoprecipitation-on-chip] or location analysis (LA), is an emerging technique that enables the systematic investigation of the precise location of genomic protein-binding sites [27, 28]. Here, we will focus on gene expression profiling since the other applications of array technology have yet to be applied in PPAR research.
Microarray technology, analyses and interpretation of data
Microarray experiments are, in principle and practice, extensions of hybridization-based methods that have been used for decades to identify and quantify nucleic acids in biological samples [29-31]. Microarray technology utilizes gene-specific probes that represent individual genes which are arrayed on an inert substrate. Several types of microarray platforms are available [32, 33], but currently the most commonly used arrays are manufactured by companies such as Agilent [34] or Affymetrix [35]. The experimental procedure involves extraction of RNA from biological samples, followed by labeling with a detectable marker (typically a fluorescent dye). After hybridization to the array and subsequent washing, an image of the array is acquired by determining the extent of hybridization to each gene-specific probe [26, 36]. The data are then normalized to facilitate the comparison between the experimental samples. Next, a set of objective criteria is applied, for example the elimination of genes with minimal variance between the samples.
The most common aim of transcriptome analysis is to find genes that are differentially expressed between the various experimental samples. Although early microarray papers used a simple ‘fold change’ approach to generate lists of differentially expressed genes, most analyses now rely on more sound statistical tests to identify differences in expression between groups [37]. Unfor-tunately, there are no standards at the level of data filtering, which is often done according to personal preference and experience. This leads to discrepancies and prevents a high degree of reproducibility. It should also be emphasized that statistical significance is not the same as bio-logical significance. Moreover, the classical approach of treating genes as independent entities is increasingly being criticized. The main reasons are first the arbitrariness of the chosen cut-off (e.g. false discovery rate, fold change) and second the disregards of the broader context in which gene products function. Although by design this approach will enable the identification of genes that show large changes in gene expression, it might not reveal small yet coordinated changes in
18
gene expression in a set of related genes, which is often the case in nutrigenomics research. In response, testing for gene classes is becoming increasingly popular. Gene classes are tradition-ally based on Gene Ontology categories [38], but recently also relationships based on e.g. meta-bolic pathways or signal transduction routes are taken into account [39]. Gene class testing thus improves the identification of affected biological processes in microarray data sets, promoting greater understanding of the underlying mechanisms driving the observed differences between samples.
Reproducibility, standardization and accessibility of microarray studies
Doubt has often been cast on the reliability, reproducibility and cross-platform concordance of DNA microarrays [40-43]. This has led to the launch of the microarray quality control (MAQC) project, a large effort involving over 125 participants representing academia, industry and gov-ernment, which carefully scrutinized the microarray technology and its use. The first set of results from this project has recently been published, addressing inter- and intra-platform agreement and reproducibility [44]. The results revealed that the distinct platforms and test sites perform comparably, generating similar lists of differentially expressed genes, undeniably establishing the robustness of the technology, provided the proper statistical analyses are applied [44]. Given the plethora of platforms and protocols for class-comparison, prediction or discovery, it is often hard to reconstruct the experimental methods in a study from the published paper, often mak-ing it impossible to enable other researchers to verify conclusions based on array data [45]. The Microarray Gene Expression Data (MGED) society therefore developed the concept of reporting minimum information about a microarray experiment (MIAME) to reduce the widespread confu-sion [46, 47]. The adoption of these guidelines by many journals has prompted, among others, the submission of detailed protocols and microarray data sets the major public repositories Gene Expression Omnibus [48] and ArrayExpress [49]. However, as we also experienced during the preparation of this manuscript, a huge point of concern is that not all data submissions provide adequate detail for reanalysis, or even worse, have not been submitted to public databases [50].
19
PPARα biology and microarray analysis
PPARα was first described as a receptor that is activated by peroxisome proliferators [51]. Shortly after this discovery, it was found that PPARα could be activated by natural fatty acids [52]. Tissue expression patterns indicated that PPARα was highly expressed in organs that carry out signifi-cant catabolism of fatty acids such as the liver, brown adipose tissue, heart, intestine and kidney [52-54]. It is therefore not surprising that the identification of PPARα target genes has concentrat-ed mainly on cellular lipid metabolism in the context of the hepatocyte. Indeconcentrat-ed, the first PPARα target gene identified was acyl-coenzyme A oxidase [55] which is involved in peroxisomal fatty acid β-oxidation. This discovery was soon followed by the identification of many more PPARα target genes involved in many key functions of lipid metabolism, such as transport and cellular uptake of fatty acids, intracellular fatty acid binding and activation, microsomal ω-oxidation, per-oxisomal β-oxidation, mitochondrial β-oxidation and ketogenesis, synthesis of lipoproteins, and glycerol metabolism. In addition, it was demonstrated that PPARα activation attenuated inflam-matory responses [52-54]. More recently, the availability of microarrays, specific agonists and transgenic animal models has opened the possibility to comprehensively study the functional role of PPARα in other, less obvious processes in liver and other tissues.
Liver
Several studies have been performed aiming to identify genes regulated by PPARα other than those connected with fatty acid catabolism. In one of the first studies that applied array technol-ogy to discover novel PPARα-regulated pathways, Kersten et al. [56] used Affymetrix GeneChips to show that PPARα influenced the expression of several genes involved in trans- and deamination of amino acids, and urea synthesis. Activation of PPARα using the synthetic ligand WY14643 decreased mRNA levels of these genes in wild type but not in PPARα-null mice, suggesting that PPARα is directly implicated in the regulation of their expression. Consistent with these data, plasma urea concentrations were modulated by PPARα in vivo. Thus, in addition to oxidation of fatty acids, PPARα also regulates metabolism of amino acids.
In a similar study, Patsouris et al. [57] compared mRNA of livers of fed and fasted PPARα-null and wild-type mice on Affymetrix GeneChips. As was expected, a fasting-induced increase in expression of fatty acid oxidative and ketogenic genes was observed that was PPARα-dependent. A similar type of regulation was observed for cytosolic and mitochondrial glycerol 3-phosphate dehydrogenase, which are involved in the conversion of glycerol to glucose. A combination of molecular biological and physiological follow-up studies was applied to functionally confirm these data, demonstrating that PPARα directly stimulates hepatic glycerol metabolism and, via this and other mechanisms, importantly influences hepatic glucose production during fasting. This effect of PPARα may account for the pronounced hypoglycemia observed in fasted PPARα-null mice. Combined with another study [58], the results from both papers underscore the key function of PPARα in controlling hepatic intermediary metabolism during fasting.
While PPARs are primarily studied because of their pharmacological relevance, it should be real-ized that PPARs likely evolved as dietary lipid sensors. Accordingly, it can be hypothesreal-ized that
20 Research question Model 1 Species 2 Strain 3 Platform 4 Design Conclusion GEO id 5 Year Reference PP AR α-Liver Identification of genes that are PP AR α-dependently
regulated during fasting.
IV M WT(SV129), PPAR α-null AF , Mu6500 Mice were fasted for 0h or 24h PP AR α represses amino acid metabolism and is a key controller of intermediary metabolism during fasting. 2001 [56] Study the molecular mechanisms responsible for the pleiotropic ef fects of a potent peroxisome proliferator (toxicogenomics study). IV M WT(SV129), PPAR α-null
Incyte, UniGene mouse cDNA microarrays
Mice were treated with 0.125% (w/ w) WY14643 for 2 weeks PP AR α regulates many genes not associated with peroxisomes. Delayed onset of enhanced expression of some genes may be the result of metabolic events occurring secondary to PP AR α
activation and alterations in lipid metabolism.
2001 [60] Study the mechanisms through which peroxisome proliferators regulate hepatocyte proliferation and apoptosis (toxicogenomics study). IV M WT(SV129) P PP AR α-null TB
Mice were treated with 1
150mg/kg/day
diethylhexylphthalate (DEHP) for 2 days
12 genes dif ferentially expressed between WT vehicle and DEHP treated mice. Iron binding proteins, e.g. lactoferrin, may play a role in DEHP-mediated liver growth, but not in peroxisome proliferation. 2002 [74] Examine whether gene expression is altered in dif ferent ways by WY14643 or fenofibrate; to find genes whose expression has not been previously reported to be af fected by PP AR α agonists. IV M WT(C57BL/ 6N) and CD-1 AF , U74v2
Mice were treated with 100 mg/kg fenofibrate, or 30 mg/kg or 100 mg/kg WY14643 for 2 or 3 days
Changes in gene expression by WY14643 and fenofibrate were almost identical. Serum amyloid
A2 was repressed; metallothionein-1
and -2 were
induced by WY14643, but not fenofi
brate. 2002 [75] To evaluate whether arrays can be used to distinguish between the ef fects of dif ferent peroxisome proliferators (toxicogenomics study). IV R
Sprague- Dawley VAF + albino cDNA NIEHS, Rat Chip Animals were dosed orally via gavage with 250 mg/kg/day of clofibrate, 250 mg/ kg/day of WY14643, 100 mg/kg/day of gemfibrozil, or 120 mg/kg/day of pheno
-barbital for 24h or 2 weeks
Changes in gene expression were similar , but distinguishable for the fibrates, whereas the non-related compound phenobarbital altered
expression of a distinct set of genes.
2002
[61]
Table 1: Papers that used micr
oarray technology to investigate (parts) of PP
21 Identification of ciprofibrate related genes that are important for liver growth; to examine ciprofibrate ef fects on the diverse metabolic systems of liver in rats (toxicogenomics study). IV R F HM Rats were treated with 50mg/kg/day ciprofibrate for 8 weeks Several known PP AR α-regulated genes and some potentially new targets, as well as indirectly af fected non-target genes, were identifi ed. Af fected cellular processes were lipid and sugar metabolism, cell growth and proliferation, stress-, immune and inflammatory response, transcription. GSE335 2003 [76] Examine carcinogenic po -tential of PP AR α agonists using immortalized hepa -tocyte cell lines generated from WT and PP AR α-null mice (toxicogenomics study). CL M
Immort. hepatocytes from WT (MuSHa +/+) & PPAR α-null mice (MuSHa -/-) O
Immortalized cells were incubated with WY14643 for 6h
Kupf fer cells are not required for ef fects of WY14643. PP AR α-dependently regulated genes are involved in carcinogenesis and cell cycle control (ubiquitin COOH-terminal hydrolase 37, cyclin T1) 2003 [77] Compare ef fect of clofibrate on mRNA expression profiles in primary hepatocytes of rat, mouse, and human (toxicogenomics study). PC M, H, R AF Primary rat and mouse hepatocytes (n=3), 6 human liver donors (3 female, 3 male) were incubated with 250 µM clofibrate for 72h Genes involved in peroxisome proliferation and apoptosis were up- resp. downregulated in rodent, but not human hepatocytes. Induction of hepatocyte nuclear factor 1α was specific for human hepatocytes 2003 [65] To identify novel pathways regulated by PP AR α. IV M WT(SV129), PPAR α-null AF , MOE430A Mice were fasted for 0 or 24 h The fasting-induced increase in expression of fatty acid oxidative and ketogenic genes was PP AR α dependent. In addition, a similar type of regulation was observed for cytosolic and mitochondrial glycerol 3-phosphate dehydrogenase, which are involved in the conversion of glycerol to glucose. Combined with additional (functional) data, it was shown that PP AR α directly governs glycerol metabolism in liver 2004 [57] Elucidate the link between the gene regulation by PP AR α agonists and the concomitant changes on plasma lipid parameters; to identify smaller gene sets that are predictive of the function of these li-gands. IV R SD HM Rats received 8 dif ferent PP AR α
agonists for 4 days
A set of 19 genes, regulated in at least three of the seven compounds, was found to be capable in predicting the triglyceride-lowering capacity of
each of these drugs.
2004
22 Identification of co-regulators required for PP AR α function. CL H
cells over- expressing CITED2
O CITED2+ vs. con -trol transfected cells were treated with PP AR α agonists CITED2 was identifi ed as coregulator for PP AR α. Microarray analysis of CITED2-overexpressing cells exposed to PP AR α agonists identifi ed key genes that af fected activity of PP AR α, namely angiopoietin-like protein 4, forkhead C2, hypoxia-inducible factor-1a and MAPK phosphatase 1. Both proteins may participate in signaling cascades of
hypoxic response and angiogenesis.
see table legend 6
2004 [78] Study the ef fect of ciprofibrate treatment in primates. IV P Cynomolgus monkey AF , HG-U95A v2 Monkeys received
ciprofibrate, either for 4
days (400mg/kg/ day) or 15 days (3, 30, 150 or 400mg/kg/ day). Genes related to ribosome and proteasome biosynthesis were upregulated; downregulated genes were in the complement and coagulation cascades; no transcriptional signal for DNA damage or oxidative stress was observed; transcriptional signals were consistent with an anti-proliferative and pro-apoptotic ef fects. Key regulatory genes (JUN, MYC, NF κB families) were downregulated, which is in contrast to rodents. Finally , the magnitude of induction in δ-oxidation pathways was substantially greater in
rodents than primates.
GSE2853 2005 [69] Developing in vitro -based gene expression assays for two prototypical toxi -cological classes: aryl hy -drocarbon receptor (AhR) and PP AR α agonists (toxi -cogenomics). IC R AF , RAE230A, TLDA Primary hepatocytes were treated with 15 AhR or PP AR α agonists. Transcriptome analyses on Af fymetrix arrays identifi ed core gene signatures consisting of 8 resp. 11 genes for AhR and PP AR α agonism. This small set of genes can be used to quantitatively assess the degree to which a compound falls into
a certain mechanistic toxicological class.
2006 [79] Determine if delivery of PP AR α-siRNA using hydrodynamic tail vein injection would result in functional inhibition of target gene expression in mouse liver . IV M WT(SV129), PPAR α-null HM, ~23,000 Mice were injected with PP AR α-siRNA, control siRNA or buf fer; responses were compared with WT and PP AR α-null mice fed fenofibrate (200mg/kg/day) for 1 or 7 days. siRNA-mediated knockdown of PP AR α resulted in a transcript profile and metabolic phenotype comparable to those of PP AR α-null mice. Combining the profiles from mice treated with fenofibrate, the specificity of the RNAi response was confirmed and new candidate genes proximal to PP AR α
regulation were identifi
ed. 2006 [80] Examine gene expression profiles in wild type and PP AR α -/- mice fed various plant seeds and grains, because PP AR α may be important for inducing detoxification systems. IV M WT(SV129), T(C57BL6), PPAR α-null
AG, whole mouse genomic oligo microarray
Mice were fed a control diet, or the same diet enriched with various plant seeds and grains for one week. Only sesame seeds killed PP AR α-null mice. PP AR α plays a vital role in inducing various xenobiotic metabolizing enzymes. Since a PP AR α agonist alone could not induce most of these enzymes, it is suggested there is an essential crosstalk among PP AR α and other xenobiotic nuclear receptors (CAR, PXR) to induce a
detoxification system for plant compounds.
2006 [81] 1: Model: CL=Cell line, IV= in vivo, PC= primary isolated cells, 2: Species: H=Human, M=Mouse, R=Rat, P=Primate, 3: W=W istar , SD=Sprague-Dawley , =Fish -er , CDIGS= Charles River Laboratory CD (SD) IGS BR rats, 4: AF=Af fymetrix, AG=Agilent, SA=Sigma-Aldrich, O=Operon, I=Incyte, AB=Amersham Biosciences, CT=Clontech, CG=Compugen, HM=home made, RG=Research Genetics, TB=T oxBlot, DSB= Display Systems Biotech, SupA=SuperArray , TIGR=The Institute for
Genomic Research 5: GEO=Gene Expression Omnibus, http://www
.ncbi.nlm.nih.gov/geo/, 6: http://www
.jbc.or
23
PPARα is activated by changes in dietary fat load, for example by high fat feeding. However, because the effects of dietary fat on (PPARα-dependent) hepatic gene expression are minor, it was shown that changes in gene expression only appear significant after comprehensive analysis of gene expression using microarrays. Indeed, comparative microarray analysis of PPARα-depen-dent gene regulation by a synthetic PPARα agonist, by prolonged fasting, and by high fat feeding indicated that although all treatments caused activation of PPARα, pronounced differences in the magnitude of PPARα-target gene induction between these pharmacological, physiological and nutritional stimuli was observed [59].
Ideally, genome-wide analyses should include gene knockouts to allow unambiguous interpre-tation of the role of PPARα in regulating differentially expressed genes. However, this has not always been the case. Using Incyte cDNA arrays, Cherkaoui-Malki et al. [60] profiled changes in gene expression in livers of mice exposed to WY14643 for two weeks. PPARα activation in-creased expression of genes involved in lipid and glucose metabolism and genes associated with peroxisome biogenesis, cell surface recognition function, transcription, cell cycle, and apoptosis. The last two processes may be mechanisms underlying PPARα ligand-induced toxicity. The au-thors did not study nor mention the genes that were repressed by WY14643.
Hamadeh et al. [61] compared the gene expression profiles elicited by three PPARα agonists (clofibrate, gemfibrozil, WY14643) and one unrelated compound (phenobarbital) in livers of rats after exposure to the respective compounds for 24 hours or 2 weeks. Not surprisingly, they found a greater similarity in gene expression profile among the PPARα agonists than between the PPARα agonists and phenobarbital. Approximately 25% of the probes present on their rat cDNA array were significantly altered by at least one of the PPARα agonists for at least one time point. Unfortunately, no data on differentially expressed genes were provided for each agonist sepa-rately. Overall, genes that were significantly altered by PPARα agonist treatment were involved in fatty acid transport, fatty acid synthesis and catabolism, cell cycle, cell proliferation, and the acute phase response. Based on the analysis of early (24hrs), late (2wks) and time-independently regulated genes, a relationship between expression profiles and gene function was proposed. Most of early regulated transcripts corresponded to signaling related genes, whereas transcripts regulated at the latter time point predominantly reported on adaptation events, or were related to the observed hypertrophy. Remarkably, no genes were identified that were commonly regulated by all three agonists.
Another comparative analysis between different PPARα agonists was performed using prima-ry mouse hepatocytes [62]. Hepatocytes were exposed to multiple concentrations (10, 20 and 100µM) of bezafibrate, fenofibrate or WY14643 for 24 hours, followed by gene expression pro-filing on Agilent whole mouse genome arrays. Treatments with the highest concentration resulted in 151, 149, and 145 differentially expressed genes for bezafibrate, fenofibrate or WY14643, respectively. Hierarchical clustering analysis showed that expression profiles clustered according to dosage rather than specific drugs, indicating that a common effect exists across this class of compounds. Gene function analysis of 121 genes regulated by at least two out of three fibrates
24
showed that the majority of these genes were involved in lipid catabolism. The authors speculate that this may lead to elevated generation of hydrogen peroxide and the production of reactive oxygen species, which may be linked to the development of liver cancer.
Clinically, fibrate drugs are used mainly for their ability to lower plasma triglycerides and in-crease plasma HDL levels [63]. Although these effects are common to all fibrates, it has been shown that the various fibrates display different clinical efficacy towards the various lipoprotein classes [63], which likely originates in differential effects on hepatic gene expression. According-ly, Frederiksen et al. [64] studied the link between changes in gene expression elicited by various PPARα agonists, and the concomitant physiological changes, measured as plasma lipid profiles. In an animal model of dyslipidemia, treatment with different PPARα agonists caused a significant reduction in plasma total cholesterol and triglyceride level for all compounds studied [64]. In ad-dition, four of the seven compounds showed an increase in plasma HDL concentration. Microar-ray analysis of liver samples led to the identification of a set of 19 genes that was regulated in at least three of the seven compounds, and that was capable of predicting the triglyceride-lowering activity of each of these drugs. Many of the genes in this set were involved in lipid metabolism and have previously been shown to be regulated by PPARα. A good linear correlation was found between the quantitative regulation of gene expression, measured as fold-changes, and the trans-activation efficacy of the specific drugs assayed in vitro.
A different approach to study PPARα function is to make a distinction between transcriptional responses in different species, triggered by specific PPARα activation. Aiming at comparing ex-pressional responses in humans and rodents using Affymetrix GeneChips, primary human, rat and mouse hepatocytes were treated with clofibric acid for 72hrs [65]. For human hepatocytes the difference between donors gave rise to a larger variation on transcriptional profile than the effect of clofibrate treatment. In mice or rats the inter-animal variability of the response towards PPARα activation was relatively low. The differentially expressed genes in each of the species could be grouped into three main pathways: fatty acid transport and metabolism, xenobiotic metabolism, and cell/organelle proliferation and cell death. Whereas genes of the cytosolic, microsomal, and mitochondrial pathways involved in fatty acid transport and metabolism were upregulated across all species, genes of the peroxisomal pathway were only upregulated in rodents. The induction of the hepatocyte nuclear factor-1α by clofibrate was observed exclusively in humans, suggesting that this transcription factor may play a key role in the regulation of fatty acid metabolism in hu-man liver and possibly as well as in the non-responsiveness of huhu-man liver to clofibrate-induced regulation of cell proliferation and apoptosis. This is functionally in line with the well-known observed toxic side-effects of artificial PPARα ligands, such as peroxisome proliferation, hepa-tomegaly and hepatocarcinoma, occurring in rodents only [66], and seem to be dependent on intrinsic differences between human and rodent PPARα [67, 68].
Another study investigated the transcriptional effect in primate livers of a four and ten days treat-ment with ciprofibrate at various concentrations (3, 30, 150, 400mg/kg/day) [69]. Pathway analy-sis of the Affymetrix GeneChip data revealed that although fatty acid metabolism was the most
25
upregulated process in primates, the magnitude in terms of fold-induction in e.g. the ß-oxidation pathway was substantially greater in rodent liver. A pathway very strongly downregulated was the complement and coagulation pathway. This correlated with human clinical studies in which fibrates have been shown to have an effect on coagulation and fibrinolysis, and reduced human plasma fibrinogen levels by 12–15% [69]. Unlike the ß-oxidation genes, the magnitude of the response in the primate and the rat liver appear similar, in that the downregulation in both species is modest, often not exceeding two-fold. Moreover, it was found that genes related to ribosome and proteasome biosynthesis were significantly upregulated upon ciprofibrate treatment, whereas a number of key regulatory genes, including members of the Jun oncogene, c-myc proto-onco-gene, and nuclear factor kappa B families were downregulated. The latter result may different in rodents, as Jun oncogene and c-myc proto-oncogene are reported to be upregulated after PPARα agonist treatment. Although data from fenofibrate-treated animals were not presented, the authors reported that the transcriptional response for ciprofibrate was more robust than fenofibrate, and that the fenofibrate dataset appeared to be largely a subset of the ciprofibrate dataset [69].
Small intestine
Even though the small intestine expresses PPARα at high level and is frequently exposed to high levels of PPARα agonists via the diet, the role of PPARα in this organ was not investigated until recently [70]. Gene expression profiling on Affymetrix GeneChips of small intestines from wild-type and PPARα-null mice fed WY14643 for five days revealed that in addition to genes involved in fatty acid and triglyceride metabolism, transcription factors and enzymes connected to sterol and bile acid metabolism, including farnesoid x receptor and sterol regulatory element binding factor (SREBP)-1, were induced. In contrast, genes involved in cell cycle and differentiation, apoptosis, and host defense were repressed by PPARα activation, which morphologically resulted in a 22% increase in villus height and a 34% increase in villus area of wild-type animals treated with WY14643 [70]. Besides providing a comprehensive overview on PPARα-dependent gene regulation in small intestine, this study also pointed towards organ-specific functions of PPARα.
26 Research question Model 1 Species 2 Strain 3 Platform 4 Design Conclusion GEO id 5 Year Reference PP AR α-Gastrointestinal tract Examine the expression of PP AR α in response to ambient hypoxia CL H T84 AF T84 epithelial-like cells were exposed to ambient hypoxia T84cells for 6 or 18h. Expression of PP AR α was time-dependently downregulated. Subsequent studies revealed that the PP AR α gene contains a DNA consensus motif
for the hypoxia-inducible factor 1 (HIF-1).
2001 [82] Study PP AR α function in
the small intestine
IV M WT(SV129) PPAR α-null AF , MOE430A Mice were fed control diet, or diet supplemented with 0.1% WY14643 for 5 days. PP AR α induced enzymes in lipid metabolism; repressed host defense, cell growth and apoptosis, enhanced cell dif ferentiation, resulting in an increased number of mature absorptive enterocytes. PP AR α not only governs diverse aspects of lipid metabolism but may also play a
major role in maintaining epithelial integrity
. GSE5475 2007 [70] PP AR α-Skeletal muscle Characterize the pattern of gene regulatory events in animals specifically overexpressing PP AR α in muscle. IV M WT , mice rexpressing PPAR α in muscle (MCK- PPAR α), PP AR α-null mice AF , U34A Expression profi les
in gastrocnemius muscle were compared between WT
, MCK-PP AR α, and PP AR α-null mice MCK-PP AR α mice developed glucose intolerance despite being protected from diet-induced obesity . Conversely , PP AR α-null mice were protected from diet-induced insulin resistance. In skeletal muscle, MCK-PP AR α mice exhibited increased fatty acid oxidation rates, diminished AMP-activated protein kinase activity , and reduced insulin-stimulated glucose uptake without alterations in the phosphorylation status of key insulin-signaling proteins. Pharmacologic inhibition of fatty acid oxidation or mitochondrial respiratory coupling prevented the ef fects of PP AR α on glucose homeostasis. 2005 [72] Identify the fiber-type – selective nature of PP AR α activation in soleus (type I) and quadriceps femoris (type II) muscle fibers (toxicogenomics). IV R CDIGS AG Rats were dosed daily for 2, 3, 4, 5, 6, 9, and 16 days with fenofibrate (400 mg/ kg/day), WY14643 (100 mg/kg/day), bezafibrate (250 mg/ kg/day , rosiglitazone (100 mg/kg/day), and all-trans retinoic acid (25 mg/kg/day). A PP AR α activation signature was identifi ed that was evident in type I, but not type II, skeletal muscle fibers. The fiber-type–selective nature of this response was consistent with increased fatty acid uptake and ß-oxidation, which in muscle is reported to be associated with a lowered concentration in plasma triglycerides and increased insulin sensitivity . Adverse interactions are not likely to occur at the cellular level. The mechanism by which fibrates may produce
muscle toxicity remains elusive.
GSE5100
2006
[73]
Table 1 (continued): Papers that used micr
oarray technology to investigate (parts) of PP
27
PP
AR
α-White adipose tissue
Study whether adipose tissue-remodeling
induced
by a
δ3-adrenergic
receptor agonist is mediated through PP
AR α. IV M WT(SV129), PPAR α-null AF , U74v2 Mice received via osmotic minipumps vehicle (control) or agonists at a rate of 3 nmol/h for 1, 3, or 6 days. For 8 h of treatment, mice were injected intraperito -neally two times with 30 nmol agonist at 4-h intervals. The agonist induced PP AR α, indicating that PP AR α can be a key regulator of inflammatory signaling during lipolytic stress and that expanding ß -oxidation in adipocytes could be an ef fective means of limiting W AT inflammation and its impact
on systemic insulin sensitivity
. GSE2131 2005 [83] 1: Model: CL=Cell line, IV= in vivo, PC= primary isolated cells, 2: Species: H=Human, M=Mouse, R=Rat, P=Primate, 3: W=W istar , SD=Sprague-Dawley , =Fish -er , CDIGS= Charles River Laboratory CD (SD) IGS BR rats, 4: AF=Af fymetrix, AG=Agilent, SA=Sigma-Aldrich, O=Operon, I=Incyte, AB=Amersham Biosciences, CT=Clontech, CG=Compugen, HM=home made, RG=Research Genetics, TB=T oxBlot, DSB= Display Systems Biotech, SupA=SuperArray , TIGR=The Institute for
Genomic Research 5: GEO=Gene Expression Omnibus, http://www
.ncbi.nlm.nih.gov/geo/, 6: http://www
.jbc.or
28
White adipose tissue
To our knowledge, there are no published reports that have employed specific agonists or knock-out mice to study PPARα function in adipose tissue. However, despite the fact that expression of PPARα in white adipose tissue is much lower compared to PPARγ, evidence abounds that PPARα may also influence adipose tissue remodeling. Affymetrix GeneChip analysis revealed that chronic activation of β3-adrenergic signaling expanded the oxidative capacity of adipocytes in white adipose tissue by inducing mitochondrial biogenesis and by upregulating PPARα and genes involved in fatty acid oxidation and mitochondrial electron transport activity [71]. Analy-sis of gene expression patterns also indicated that inflammation and adipocyte-specific gene ex-pression were reciprocally related over time. Combined, the data suggested that upregulation of catabolic activity is critical to suppressing inflammation and restoring phenotypic expression in adipocytes during continuous β3-adrenergic stimulation, which may be mediated by PPARα. Skeletal muscle
The potential role of PPARα in the development of muscle insulin-resistance was explored using a gain- and loss-of-function approach combined with global analysis of gene expression [72]. Transgenic mice that overexpress PPARα in skeletal muscle were used to generate a PPARα-dependent transcriptome signature via Affymetrix GeneChip analysis. A prominent metabolic reprogramming in muscle fibers was observed, characterized by a switch from glucose utilization to fatty acid oxidation pathways, leading to muscle glucose intolerance and insulin resistance. Reciprocal outcomes from studies with PPARα-null mice corroborated these observations, sug-gesting that by altering fuel sources, PPARα has a major impact on skeletal muscle insulin resis-tance. In light of the fact that fibrates have been implicated in skeletal muscle toxicity, the effect of PPARα agonists on skeletal muscle gene transcription was studied in rats using Agilent arrays [73]. A separate analysis was performed for a muscle containing predominantly slow oxidative type I fibers and a muscle containing mainly fast glycolytic type II fibers. By comparing the tran-scriptional responses of different PPAR agonists using an advanced statistical analytical strategy, a PPARα activation signature was identified that was specific for type I, but not type II, skeletal muscle fibers. The fiber-type–selective nature of this response was consistent with increased fatty acid uptake and β-oxidation, yet failed to reveal any obvious off-target pathways that may drive the reported adverse effects.
29
Conclusions
Based on the published studies it is clear that transcriptomics has been primarily used to study two aspects of PPARα activation. One is the characterization of the physiological functions of PPARα in tissues by means of identification of PPARα-regulated genes and processes. However, almost all of these papers used transcriptome analysis merely as a screening tool to identify and further investigate a single gene or pathway other than fatty acid metabolism, thereby avoiding analyses of the whole genome. This is likely because there is an inherent bias among researchers and journal editors in favor of a focused approach that involves extensive functional validation of the data. While this type of approach has proven its effectiveness, exemplified by the findings that PPARα regulates hepatic glycerol and amino acid metabolism, by disregarding the bulk of the data the broader potential of the transcriptome analysis remains unexplored. Without down-playing the importance of proving the functional relevance of PPARα-dependent gene regulation, it should be acknowledged that transcriptome analysis can be valuable beyond study of functional significance. For example, while transcriptome analysis does not provide information about the precise molecular mechanisms underlying SPPARMs, it is probably the most effective tool to substantiate the concept in terms of differential gene regulation. In addition, it is difficult to imagine a method more suitable to systematically compare organ specific PPARα-dependent gene regulation. Numerous other useful applications can be envisioned that can rely or benefit from transcriptomics, including ligand screening, mutational analysis, and, in combination with ChIP on CHIP analysis, a whole genome search for PPAR target genes.
The other bulk of papers focused on the toxicological aspects of the peroxisome-proliferator class of compounds to explore the usefulness of transcriptomics for toxicogenomics research, a discipline aiming to identify and characterize mechanisms of action of known and suspected toxicants. Consequently, these studies did not thoroughly address the physiological functions of PPARα. In addition, an ideally performed genome-wide analysis of PPARα function should be a balanced combination of organ-specific gene knockouts, highly specific PPAR-agonists, and whole genome arrays. Unfortunately, examination of the literature reveals that this has rarely been the case. This is mostly because of a lack of (organ-specific) PPARα-null models, the lack of data mining bioinformatics tools but also the high prices of arrays.
Nonetheless, the application of transcriptomics has clearly provided new insights on PPARα func-tion. Although the most obvious ‘core-functions’ of the PPARα isoform was identified by targeted approaches shortly after its discovery, array studies revealed, for example, that this nuclear recep-tor modulates the innate immune system of the small intestine.
30
It is expected that the exhaustive characterization and comparison of organ-specific PPAR transcriptomes will ultimately allow a comprehensive description of the whole body biology that is under control of PPARs, and finally their physiological relevance. Crucial for this is the proper and accessible storage of data allowing additional analyses when improved algorithms and bioinformatics tools become available.
Finally, only a few studies have investigated functions of PPARα using natural agonists, i.e. eicosanoids, unsaturated as well as long-chain fatty acids, and their activated derivatives (acyl-CoA esters). Instead, specific pharmacological PPARα-agonists have been employed. As nutrition researchers, we have to demonstrate beyond doubt that transcriptomics studies performed with synthetic agonists are of relevance for our understanding of nutrient-mediated gene regulation. This is one of the fascinating research topics in the field of nutrigenomics. Notwithstanding, at present the use of synthetic agonists is the most practical and effective strategy in nutrigenomics research to magnify the subtle effects of nutrients.
in murine small intestine
Meike Bünger, Heleen van den Bosch, Jolanda van der Meijde, Sander Kersten, Guido Hooiveld, Michael Müller
Published in Physiological Genomics, 2007 July 18;30(2):192-204 PMID: 17426115
34
Abstract
The peroxisome proliferator-activated receptor alpha is a fatty acid-activated transcription factor that governs a variety of biological processes. Little is known about the role of PPARα in the small intestine. Since this organ is frequently exposed to high levels of PPARα ligands via the diet, we set out to characterize the function of PPARα in small intestine using functional genomics experiments and bioinformatics tools.
PPARα was expressed at high levels in both human and murine small intestine. Detailed analyses showed that PPARα was expressed highest in villus cells of proximal jejunum. Microarray analyses of total tissue samples revealed, that in addition to genes involved in fatty acid and triacylglycerol metabolism, transcription factors and enzymes connected to sterol and bile acid metabolism, including FXR and SREBP1, were specifically induced. In contrast, genes involved in cell cycle and differentiation, apoptosis, and host defense were repressed by PPARα activation. Additional analyses showed that intestinal PPARα dependent gene regulation occurred in villus cells. Functional implications of array results were corroborated by morphometric data. The repression of genes involved in proliferation and apoptosis was accompanied by a 22% increase in villus height, and a 34% increase in villus area of wild-type animals treated with WY14643. This is the first report providing a comprehensive overview of processes under control of PPARα in the small intestine. We show that PPARα is an important transcriptional regulator in small intestine, which may be of importance for the development of novel foods and therapies for obesity and inflammatory bowel diseases.
35
Introduction
The peroxisome proliferator-activated receptor alpha (PPARα) is a ligand-activated transcription factor with diverse functions and is activated by a variety of synthetic compounds, including the lipid lowering fibrate drugs [53, 54]. High affinity natural ligands include eicosanoids, unsaturated as well as long-chain fatty acids, and their activated derivatives (acyl-CoA esters) [84-87]. In analogy with other nuclear receptors, PPARα forms obligate heterodimers with the retinoid X receptor and stimulates gene expression by binding to peroxisome proliferator response ele-ments (PPREs) located in the regulatory domain of genes [53]. PPARα is expressed in a variety of tissues including the small intestine [88, 89], however, its function has been almost exclusively studied in liver.
In liver PPARα is critical for the coordinate transcriptional activation of genes involved in lipid catabolism, including cellular fatty acid uptake and activation, mitochondrial β-oxidation, per-oxisomal fatty acid oxidation, ketone body synthesis, fatty acid elongation and desaturation, and apolipoprotein synthesis [53, 54]. In addition, PPARα is an important regulator of the hepatic acute phase response.
While the function of PPARα in liver is well studied, little is known about PPARα and PPARα target genes in non-hepatic tissues. This is especially true with respect to the role of PPARα in the small intestine, which has only been addressed in few studies [90, 91]. Knowledge on the regu-latory and physiological function of PPARα in the small intestine is of particular interest, since the average Western diet contains a high amount of triacylglycerols [92] that are hydrolyzed to monoacylglycerol and free fatty acids before entering the enterocyte [93]. Consequently the small intestine is frequently exposed to high levels of PPARα ligands.
Therefore we set out to determine the role of PPARα in the small intestine. We first analyzed in detail the expression of PPARα throughout the small intestine and then evaluated the outcome of specific PPARα activation on small intestinal gene expression using microarrays and bioinfor-matics tools.
This allowed the genome-wide identification of intestinal PPARα target genes and corresponding processes. We conclude that PPARα plays an important role in the regulation of intestinal func-tion by governing diverse processes ranging from numerous metabolic pathways to the control of apoptosis and cell cycle.
36
Materials and Methods
AnimalsPure bred wild-type (129S1/SvImJ) and PPARα-null (129S4/SvJae) mice [94] were purchased from Jackson Laboratories (Bar Harbor, ME) and bred at the animal facility of Wageningen University. Mice were housed in a light- and temperature-controlled facility and had free access to water and standard laboratory chow (RMH-B, Hope Farms, Woerden, the Netherlands). All animal studies were approved by the Local Committee for Care and Use of Laboratory Animals. Experimental design and tissue handling
Four independent studies were performed. In all studies 4-5 month old male wild-type and PPARα-null mice were used. Study A: Mice were fed chow, or chow supplemented with 0.1% WY14643 (Chemsyn, Lenexa, KA) for five days (n=6 mice per group). On the sixth day, mice were anaesthetized with a mixture of isofluorane (1.5%), nitrous oxide (70%) and oxygen (30%). Small intestines were excised and flushed with ice-cold PBS and all subsequent tissue handlings were performed on ice. Remaining fat and pancreatic tissue was carefully removed, and RNA was isolated from the complete full-length small intestine for microarray analysis. Study B: The above described experiment was repeated (n=3 mice per group), except that after removal the small intestine was divided into 10 equal parts to study gene expression along the proximal-distal axis. Study C: Study A was repeated (n=3-4 mice per group), except that after removal the small intestine was inverted on a 0.75 mm-diameter rod, washed in ice-cold PBS, and divided into segments of 1 cm. Before continuing with the cell isolation protocol, segments of all animals within each experimental group were pooled. Fractions enriched in crypt or villus cells were isolated as described by Flint et al [95]. This isolation protocol was repeated a week later for the control group. Cell fractions were used for RNA isolation. Study D: The feeding experiment was repeated as described, except that in addition to WY14643 mice were fed chow supplemented with fenofibrate (0.1% w/w) (Sigma, St. Louis, MO) for five days (n=5 mice per group). RNA was isolated from the complete full-length small intestine for quantitative reverse-transcription PCR analysis.
RNA isolation and quality control
Total RNA was isolated from small intestinal samples using TRIzol reagent (Invitrogen, Breda, the Netherlands) according to the manufacturer’s instructions. RNA was treated with DNAse and purified using the SV total RNA isolation system (Promega, Leiden, the Netherlands). Concentrations and purity of RNA samples were determined on a NanoDrop ND-1000 spectrophotometer (Isogen, Maarssen, the Netherlands). RNA integrity was checked on an Agilent 2100 bioanalyzer (Agilent Technologies, Amsterdam, the Netherlands) with 6000 Nano Chips according to the manufacturer’s instructions. RNA was judged as suitable for array hybridization only if samples exhibited intact bands corresponding to the 18S and 28S ribosomal RNA subunits, and displayed no chromosomal peaks or RNA degradation products. Total RNA from human tissues (FirstChoice Human Total RNA Survey Panel) was obtained from Ambion (Austin, TX). Each tissue pool comprises RNA from three or four donors.
37
Affymetrix GeneChip oligoarray hybridization and scanning
For microarray analyses, we used RNA isolated from the full-length small intestine. RNA was hybridized on an Affymetrix GeneChip Mouse Genome 430A array. This array detects 22,626 transcripts that represent approximately 13,700 known genes. For each experimental group, three biological replicated were hybridized, thus in total 12 arrays were used. Detailed methods for the labeling and subsequent hybridizations to the arrays are described in the eukaryotic section of the GeneChip Expression Analysis Technical Manual, Revision 3, from Affymetrix (Santa Clara, CA). Array data have been submitted to the Gene Expression Omnibus, accession number GSE5475.
Analyses and functional interpretation of microarray data
Scans of the Affymetrix arrays were processed using packages from the Bioconductor project [96]. Expression levels of probe sets were summarized using GCRMA [97], where after differentially expressed probe sets were identified using Limma [98]. P-values were corrected for multiple testing using a false discovery rate method [99]. Probe sets that satisfied the criterion of FDR < 5% (q-value < 0.05) were considered to be significantly regulated. Of these, probe sets that were also >1.5 fold changed in wild-type mice upon WY14643 treatment, but were not changed in treated PPARα-knockout mice, were designated PPARα regulated. Three complementary methods were applied to relate changes in gene expression to functional changes. One method is based on overrepresentation of Gene Ontology (GO) terms [100]. Another approach, gene set enrichment analysis (GSEA), takes into account the broader context in which gene products function, namely in physically interacting networks, such as biochemical, metabolic or signal transduction routes [39]. Both applied methods have the advantage that it is unbiased, because no gene selection step is used, and a score is computed based on all genes in a GO term or gene set. In addition, biological interaction networks among PPARα regulated genes were identified using Ingenuity Pathways Analysis (IPA) (Ingenuity Systems, Redwood City, CA). Detailed descriptions of the applied methods are available in the supplemental text (supplemental_1).
Quantitative reverse-transcription polymerase chain reaction
Single-stranded complementary DNA (cDNA) was synthesized from 1 µg total RNA using the reverse-transcription system from Promega (Leiden, the Netherlands) according to the supplier’s protocol. Quantitative reverse-transcription polymerase chain reaction (qRT-PCR) was performed on a MyIQ thermal cycler (BioRad, Veenendaal, the Netherlands) using Platinum Taq DNA polymerase (Invitrogen, Breda, the Netherlands) and SYBR green (Molecular Probes, Leiden, the Netherlands). Most of the primer sequences were obtained from the PrimerBank at Harvard University [101]. Primer sequences are listed in table 1 of the supplemental data (supplemental_ 3). Samples were analyzed in duplicate and standardized to either cyclophilin or 18S expression. Expression levels in isolated villus cells were standardized to villin.
38
Histology
For histology studies a fifth, independent experiment was performed, exactly as described in study A. After removal, intestines were divided into 3 equal parts, which are referred to as duodenum, jejunum and ileum, respectively. Each section was prepared using a ‘Swiss role’ technique [102] to evaluate the entire longitudinal section on one slide. Tissues were fixed by immersion in 4% PBS-buffered formaldehyde, processed in an automatic tissue processor, embedded in paraffin, sectioned at 5µm, and stained with haematoxylin and eosin (H&E). Sections were examined on a CKX41 microscope (Olympus, Zoeterwoude, the Netherlands), equipped with calibrated DP-software, version 3.2 (Olympus).
This software was used to measure villus height, crypt depth and villus area of 50 villi per section for each animal. Statistical analysis between groups was performed using ANOVA, followed by the Least Significant difference (LSD) post-hoc test.
39
Results
PPARα expression in human and murine small intestine
To ascertain whether PPARα may be functionally relevant in the small intestine, the expression of PPARα was measured in twenty human tissues by quantitative RT-PCR (qRT-PCR).
In humans the highest expression levels of PPARα were observed in kidney, followed by heart, small intestine and liver (Figure 1A). In mice, the expression of PPARα was slightly higher in liver compared to small intestine of SV129 mice, whereas RXRα expression was comparable between both tissues (Figure 1B). These data suggest that PPARα may be functionally relevant in the small intestine.
Figure 1: PPARα is ex-pressed at high levels in human and murine small intestine.
(A) Expression levels of
PPARα in various human tissues. The top 10 out of the 20 tissues analyzed are presented, PPARα expres-sion in kidney was arbi-trarily set to 100%. (B) Murine PPARα and RXR expression levels in small intestine and liver. Data are presented as mean ± standard deviation, n=5. The small intestine was ar-bitrarily set to 100%. For both human and mouse samples, expression levels were standardized to 18S.
