HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) An erratum has been published All Versions of this Article: M300239-JLR200v1 45/3/592    most recent Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Frederiksen, K. S. Articles by Fleckner, J. Search for Related Content PubMed PubMed Citation Articles by Frederiksen, K. S. Articles by Fleckner, J. Social Bookmarking                      What's this?

Journal of Lipid Research, Vol. 45, 592-601, March 2004
Copyright © 2004 by American Society for Biochemistry and Molecular Biology

Methods

Prediction of PPAR- ligand-mediated physiological changes using gene expression profiles

Klaus Stensgaard Frederiksen^*, Erik Max Wulff, Per Sauerberg, John Patrick Mogensen, Lone Jeppesen and Jan Fleckner^*

^* Department of Molecular Genetics, Novo Nordisk A/S, DK-2880 Bagsværd, Denmark
Departments of Pharmacological Research 2, Novo Nordisk Park, DK-2760 Måløv, Denmark
Medicinal Chemistry Research III, Novo Nordisk Park, DK-2760 Måløv, Denmark

Published, JLR Papers in Press, December 16, 2003. DOI 10.1194/jlr.M300239-JLR200

To whom correspondence should be addressed. e-mail: ksf{at}novonordisk.com

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Peroxisome proliferator-activated receptor (PPAR)- controls the transcription of a variety of genes involved in lipid metabolism and is the target receptor for the hypolipidemic drug class of fibrates. In the present study, the molecular and physiological effects of seven different PPAR-activating drugs have been examined in a rodent model of dyslipidemia. The drugs examined were selected to display varying potencies and efficacies toward PPAR-. To help elucidate the link between the gene regulation elicited by PPAR- ligands and the concomitant physiological changes, we have used cDNA microarray analysis to identify smaller gene sets that are predictive of the function of these ligands.

A number of genes showed strong correlations to the relative PPAR- efficacy of the drugs. Furthermore, using multivariate analysis, a strong relationship between the drug-induced triglyceride lowering and the transcriptional profiles of the different drugs could be found.

Abbreviations: ACBP, acyl-CoA binding protein (diazepam binding inhibitor); apoC-III, apolipoprotein C-III; bifunctional enzyme, peroxisomal enoyl-CoA:hydrotase-3-hydroxyacyl-CoA bifunctional enzyme; Cyp4A10, cytochrome P450 4A10; HCD, high-cholesterol diet; PLS, partial least-squares projection to latent structures; PPAR, peroxisome proliferator-activated receptor

Supplementary key words dyslipidemia • pharmacodynamics • peroxisome proliferator-activated receptor- in vivo activation • transcriptional profiling

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Peroxisome proliferator-activated receptor (PPAR)- is recognized as the target receptor for fibrates (1), which are used in the treatment of dyslipidemic patients. Fibrates are recognized to exert their hypolipidemic action by altering lipid metabolism at multiple levels, and PPAR- is an important regulator of both intracellular and extracellular lipid metabolism. At the intracellular level, PPAR- regulates a number of genes that encode enzymes involved in the oxidation of fatty acids. These include upregulations of acyl-CoA oxidase (2, 3) the rate-limiting enzyme of peroxisomal ß-oxidation, medium-chain acyl-CoA dehydrogenase (4), a central enzyme of mitochondrial ß-oxidation, and cytochrome P450 genes (5) involved in microsomal -hydroxylation of fatty acids. The induction of fatty acid oxidation, combined with the upregulation of the fatty acid transport protein (6), causes a shift in hepatic fatty acid metabolism with decreased triglyceride synthesis and increased catabolism. These intracellular changes are accompanied by regulation of extracellular lipid metabolism and transport. Induction of the lipoprotein lipase by PPAR- (7) increases the lipolysis of triglycerides in chylomicrons and VLDL particles. This lipolysis generates precursors to HDL particles. The expression of the two major apolipoproteins of HDL particles, apolipoproteins A-I and A-II, is also upregulated by PPAR- (8, 9). These events are important factors in fibrate-induced increases of HDL levels. The increased HDL levels may, along with increased cholesterol efflux from peripheral cells through upregulation of the ATP binding cassette transporter A1 transport protein (10), facilitate the reverse cholesterol transport from peripheral tissues back to the liver. Furthermore, apolipoprotein C-III (apoC-III), a major lipoprotein of chylomicrons and VLDL particles, is transcriptionally downregulated by PPAR- (11).

In the present study, rats fed a high-cholesterol diet (HCD) were used as a dyslipidemic model (12) to study the physiological and transcriptional changes occurring as the result of treatment with various PPAR- activators. Seven different PPAR- activators with varying efficacies were selected and tested in the rat model. Using cDNA microarrays, the induced transcriptional changes in the liver were examined. The transcriptional changes of genes encoding apoC-III and peroxisomal enoyl-CoA:hydrotase-3-hydroxyacyl-CoA bifunctional enzyme (bifunctional enzyme) correlated very well to the in vitro-measured PPAR- efficacies of the drugs. Thus, these genes may be useful as markers for the in vivo efficacy of PPAR- activators. Furthermore, using multivariate analysis, we identified a set of 19 genes capable of predicting the triglyceride-lowering capacity of each of the drugs.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
DNA microarray analysis
All cDNA inserts were PCR-amplified using the T7 and M13 reverse primers. PCR products (100 µl each) were purified on the Qiagen 96-well PCR purification system and subsequently checked for size and purity on agarose gels. The purified PCR products were mixed 1:1 in DMSO (Sigma, St. Louis, MO) and rearranged in 384-well plates. cDNA targets were spotted in duplicate onto silanized glass slides (Amersham Biosciences, Buckinghamshire, UK) at 55% relative humidity using the Molecular Dynamics Gen. III microarray spotter. All slides were UV cross-linked (50 mJ) in a Stratagene UV cross-linker. Cy-3- and Cy-5-labeled single-stranded cDNA probes were prepared from 15 µg of total RNA from the test and control tissues, respectively. The RNA was incubated with 1 µg of anchored oligo-dT primer at 70°C for 10 min and subsequently chilled on ice for 30 s The primer-annealed RNA was incubated with reaction mixture [1x Superscript II buffer (Life Technologies, Taastrup, Denmark), 10 mM DTT, 200 µM deoxy-guanosine, -adenosine, -thymidine triphosphate, 100 µM dCTP, 100 µM Cy-3/Cy-5-dCTP (Amersham Biosciences), and 200 U of Superscript II reverse transcriptase (Life Technologies)] at 42°C for 2.5 h in a final volume of 20 µl. The RNA template was removed by alkaline denaturation (incubation with 2 µl of 2.5 M NaOH at 37°C for 15 min). Labeled probes were purified on GFX spin columns (Pharmacia Biotech, Uppsala, Sweden). The Cy-3- and Cy-5-labeled cDNAs were mixed 1:1 in hybridization buffer version 2 (Amersham Biosciences) and subsequently 1:1 in formamide (Sigma). The mixed probes were injected into an automated slide processor (Molecular Dynamics, Sunnyvale, CA) and hybridized at 42°C for 16 h. After hybridization, the slide was washed in 1x SSC, 0.2% SDS at 55°C for 10 min and subsequently in 0.1x SSC, 0.2% SDS at 55°C for 10 min. The slide was scanned in a Gen. III microarray laser scanner (Molecular Dynamics). Image analyses were performed with ArrayVision (Imaging Research). All ratios were calculated using background-subtracted, artifact-removed median densities. The Cy-3 and Cy-5 signals were normalized to the spot intensities of all of the spots in each channel.

Northern blots
Total RNA was isolated from tissues using the Trizol reagent according to the instructions of the manufacturer, but with two additional phenol/chloroform extractions. An amount of 20 µg of total RNA was separated on a denaturing gel containing 1% agarose, 20 mM MOPS, 5 mM NaOAc, 6% formaldehyde, and 1 mM EDTA. The size-fractionated RNA was transferred to a Hybond N⁺ membrane (Amersham Biosciences) using capillary blotting and was subsequently immobilized by UV cross-linking. cDNA fragments were labeled using the Prime It kit (Stratagene) and [-³²P]dATP (3000 Ci/mmol; Amersham Biosciences). For hybridization and prehybridization, the Express Hyb buffer (Clontech) was used. Signals were detected using a PhosphorImager (Molecular Dynamics) and quantitated with ImageQuant software (Molecular Dynamics).

Quantitative RT-PCR
cDNA was prepared from 1 µg of total RNA from each of the treatment groups using random primers and TaqMan Reverse Transcription reagents (Applied Biosystems, Foster City, CA) according to the manufacturer's instructions. Quantitative PCR was performed on three samples from each treatment group (10-fold dilutions of cDNA) using TaqMan PCR core reagents (Applied Biosystems) on an ABI PRISM^® 7000 Sequence Detection System. Primers and 6-carboxyfluorescein-labeled-probes for lipoprotein lipase (LPL), fatty acid transport protein (FATP), acyl-CoA oxidase, and 18S rRNA were ordered as Assays-on-Demand (Applied Biosystems). Probe sequences for these assays were as follows: LPL (ATCCATGGATGGACGGTGACAGGAA; assay Rn00561482_m1), FATP (TGTCAAATATAATTGCACGGTAGTG; assay Rn00585821_m1), acyl-CoA oxidase (TGCTGCAGACAGCCAGGTTCTTGAT; assay Rn00569216_m1), and 18S rRNA (TGGAGGGCAAGTCTGGTGCCAGCAG; assay HS99999901_s1). Data were analyzed using ABI Prism 7000 SDS software (version 1.0; Applied Biosystems), and expression levels for LPL, FATP, and acyl-CoA oxidase were normalized to the 18S rRNA levels.

In vitro PPAR activation assay
HEK293 cells were grown in DMEM plus 10% FCS. Cells were seeded on 96-well plates the day before transfection to give a confluence of 50–80% at transfection. A total of 0.8 µg of DNA containing 0.64 µg of pM1/LBD, 0.1 µg of pCMVßGal, 0.08 µg of pGL2(Gal4)₅, and 0.02 µg of pADVANTAGE was transfected per well using FuGene transfection reagent according to the manufacturer's instructions (Roche). Cells were allowed to express protein for 48 h followed by the addition of compound.

Human PPAR-, -, and - cDNAs were obtained by PCR amplification using cDNA synthesized by reverse transcription of mRNA from human liver, adipose tissue, and placenta, respectively. Amplified cDNAs were cloned into pCR2.1 and sequenced. The ligand binding domain (LBD) of each PPAR isoform was generated by PCR (PPAR-, amino acid 167 to the C terminus; PPAR-, amino acid 165 to the C terminus; PPAR-, amino acid 128 to the C terminus) and fused to the DNA binding domain of the yeast transcription factor GAL4 by subcloning fragments in frame into the vector pM1 (13), generating the plasmids pM1LBD, pM1LBD, and pM1LBD. Ensuing fusions were verified by sequencing. The reporter was constructed by inserting an oligonucleotide encoding five repeats of the GAL4 recognition sequence [5x CGGAGTACTGTCCTCCG(AG)] (14) into the vector pGL2 promotor (Promega, Madison, WI), generating the plasmid pGL2(GAL4)₅. pCMVß-Gal was purchased from Clontech, and pADVANTAGE was purchased from Promega.

All PPAR- ligands were dissolved in DMSO and diluted 1:1,000 upon addition to the cells. Compounds were tested in quadruple in concentrations ranging from 0.001 to 300 µM. Cells were treated with compound for 24 h followed by luciferase assay. Each compound was tested in at least three separate experiments.

Medium including test compound was aspirated, and 100 µl of PBS including 1 mM Mg²⁺ and Ca²⁺ was added to each well. The luciferase assay was performed using the LucLite kit according to the manufacturer's instructions (Packard Instruments). Light emission was quantified by counting on a Packard LumiCounter. To measure ß-galactosidase activity, a 25 µl supernatant from each transfection lysate was transferred to a new microplate. ß-Galactosidase assays were performed on the microwell plates using a kit from Promega and read in a Labsystems Ascent Multiscan reader. The ß-galactosidase data were used to normalize transfection efficiency and cell growth for the luciferase data.

The activity of a compound is calculated as fold induction compared with an untreated sample. For each compound, the efficacy (maximal activity) is given as relative activity compared with Wy-14643 for PPAR-, rosiglitazone for PPAR-, and carbacyclin for PPAR-. The EC₅₀ is the concentration giving 50% of maximal observed activity. EC₅₀ values were calculated via nonlinear regression using GraphPad PRISM 3.02 (GraphPad Software, San Diego, CA). The results are expressed as means ± SD.

PPAR activation in vivo
Male Sprague Dawley rats (Crl:CD BR; Charles River, Sulzfeld, Germany), 6 weeks of age, 280 g body weight, were fed a HCD (1.25% cholesterol; C 13002; Research Diets, Inc.) for 10 days. The cholesterol feeding induced a severe hypercholesterolemia and modest hypertriglyceridemia, whereas serum glucose and insulin remained within the range of normal chow-fed rats. Upon 6 days of feeding, HCD animals were administrated vehicle (1 ml/kg), NNC 61-3058 (10 mg/kg), NNC 61-4424 (10 mg/kg), NNC 61-4706 (10 mg/kg), NNC 61-4718 (10 mg/kg), fenofibrate (300 mg/kg), rosiglitazone (30 mg/kg), or Wy-14643 (10 mg/kg) orally daily for the subsequent 4 days. These doses have previously been determined to induce maximal efficacy in our laboratory. A group of rats fed normal chow was included in parallel. A total of six animals were allocated per group. Nonfasted blood samples were collected from the retro-orbital sinus into plain tubes. Total serum cholesterol, circulating concentrations of HDL-cholesterol, and serum triglycerides were measured after 4 days of treatment on the last day on HCD. Liver weight was measured at the end of the study. Body weight and 24 h food intake were measured on day 10. All serum analyses were performed on a Hitachi 912 autoanalyzer (Roche). Data are presented as means with variations of the mean. Fenofibrate and carbacyclin was purchased from Sigma, and Wy-14643 (C1323) was purchased from Tokyo Kasai Kogyo Co., Ltd. (Toshima, Tokyo, Japan). The compounds rosiglitazone (15), NNC 61-4424 (16), NNC 61-3058 (17), NNC 61-4706 (18), and NNC 61-4718 (17) were all synthesized according to published procedures.

Partial least-squares projection to latent structures analysis
The partial least-squares projection to latent structures (PLS) analysis was performed with Simca-P 9 software (Umetrics, Umea, Sweden). The normalized signal intensities from the 19 genes listed in Table 2 in both color combinations (Cy-3 and Cy-5) were used as single observations (predictor variables). All variables were centered and scaled to unit variance, and log transformed, before the PLS analysis was performed.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

View larger version (19K): [in this window] [in a new window]   Fig. 1. Chemical structures of the tested compounds: rosiglitazone, Wy-14643, fenofibrate, NNC 61-4424, NNC 61-3058, NNC 61-4718, and NNC 61-4706.

View larger version (26K): [in this window] [in a new window]   Fig. 2. High-cholesterol diet (HCD) effects on plasma lipid parameters. Percentage changes in plasma lipid parameters (HCD-fed rats relative to chow-fed controls).

View larger version (23K): [in this window] [in a new window]   Fig. 3. Pharmacodynamics of the tested compounds. Changes in plasma lipid parameters in Sprague Dawley rats fed a HCD after 4 days of oral treatment. All changes are relative to vehicle-treated rats fed a HCD.

Identification of biomarker genes
To investigate whether the quantitative regulation of the genes listed in Table 2 reflected the in vitro efficacies of the drugs, the fold change of each gene was plotted as a function of the relative efficacy. For the genes encoding peroxisomal bifunctional enzyme, cytochrome P450 4A10 (Cyp4A10), diazepam binding inhibitor [acyl-CoA binding protein (ACBP)], and apoC-III, very good linear correlations between the expression changes and the relative in vitro efficacies of the ligands were found. Linear regression coefficients for these four genes are 0.8 or greater. To confirm the findings of the microarray experiments and the correlation to the in vitro efficacies, Northern blot analysis were done on these regulated genes. The Northern blot data for apoC-III and bifunctional enzyme confirmed the microarray data and the correlations to the relative in vitro efficacies (Fig. 4) . Cyp4A10 and ACBP were also clearly shown to be regulated based on the Northern blot data, but the correlation to relative efficacies was weaker (r² values between 0.5 and 0.6) than the correlation found based on the microarray data. Intensities from the Northern blots were normalized to the signal intensities for ribosomal phosphoprotein 36B4, which has previously been shown not to be regulated by PPAR- agonists in rat liver (6).

View larger version (32K): [in this window] [in a new window]   Fig. 4. Relationship between regulated target genes and the relative drug peroxisome proliferator-activated receptor- (PPAR-) efficacy. A: Comparison of quantitative expression levels as measured in microarrays and Northern blot analysis and the relative PPAR- efficacies of the tested compounds using linear regression analysis. Fold regulations as calculated in Table 2 were used for the microarray data, whereas signal intensities for the genes in Northern blots were collected using a PhosphorImager and normalized to the signals for ribosomal phosphoprotein 36B4. For apolipoprotein C-III (apoC-III), the log fold change was used because this gene is downregulated and its expression ratios lie between 1 and 0. B: Northern blots showing the expression of peroxisomal enoyl-CoA:hydrotase-3-hydroxyacyl-CoA bifunctional enzyme (bifunctional enzyme), apoC-III, and ribosomal phosphoprotein 36B4.

View larger version (9K): [in this window] [in a new window]   Fig. 5. Relationship between the degree of downregulation observed for apoC-III and the change in HDL-cholesterol. Fold changes for apoC-III were calculated as described in Table 2. Because apoC-III is downregulated (i.e., fold regulations are in the interval between 1 and 0), these values were plotted as log-transformed to expand the scale. HDL ratios were calculated as the HDL levels observed in treated animals (HCD-fed) versus those observed in vehicle-treated animals (HCD-fed).

In an attempt to describe the observed pharmacodynamic changes as observed in vivo (Fig. 3) by using all of the found transcriptional changes, multivariate data analysis was invoked. PLS uses a linear multivariate model to relate two data matrices (X) and (Y) (20). Using PLS analysis, all of the regulated gene expression changes (Table 2) were defined as predictor variables (X) and the pharmacological parameters were defined as dependent variables (Y). PLS, unlike multiple linear regression, can analyze highly correlated X variables. Here, rather than using ratios, the normalized signal intensities for the given genes from treated and control animals in both color combinations (Cy-3 and Cy-5) were used as single observations. All variables were centered and scaled to unit variance and log transformed.

By using PLS modeling, it was possible to build a model to predict the PPAR--mediated change in triglyceride with good confidence, and this was the pharmacodynamic property that was predicted the best (Fig. 6) . When validating the PLS model for triglyceride change by permuting the data, the R ² value, describing the relation between the observed and predicted variables, decreased from 0.88 to 0.18. This strongly indicates that the model for triglyceride lowering is valid and not overfitted. In contrast, using the same set of gene expression data, it was not possible to generate models that could predict either total cholesterol changes or HDL changes. In the latter case, this was predicted just as well simply by following the expression level of apoC-III.

View larger version (9K): [in this window] [in a new window]   Fig. 6. Generation of a model to predict triglyceride (TG) changes from multivariate gene expression data. Partial least-squares projection to latent structures regression analysis of the change in triglyceride used the observed transcriptional changes listed in Table 2. The four data points shown for each compound represent the normalized signal intensities from the Cy-3- and Cy-5-labeled probe preparations. On the x axis, the values for the predicted triglyceride reduction based on gene expression data are plotted. On the y axis, the actual data for the triglyceride reduction are plotted.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

All of the studied PPAR- agonists had expectedly a lowering effect on the plasma levels of triglyceride and total cholesterol (Fig. 3). The reduction in plasma levels of triglyceride ranged from 11% for rosiglitazone to 74% for NNC 61-4424. Lowering of triglycerides is the hallmark of fibrates in both humans and animal models, and drugs such as gemfibrozil (21), ciprofibrate (22), bezafibrate, and fenofibrate (23) have been reported to reduce plasma triglycerides in the range of 40–60% in either normal chow- or cholesterol-fed rats. Total plasma cholesterol is also a parameter that has been observed to be reduced by fibrates in rats (24). In the present study, rather large reductions in plasma total cholesterol were observed, with Wy-14653 (60% reduction) and fenofibrate (71% reduction) as the most efficacious drugs. These sizable reductions can most likely be explained by the very large diet-induced increase in total cholesterol. The drug-induced increase in HDL-cholesterol is also very diet-dependent in rats. In chow-fed rats, drugs such as bezafibrate, clofibrate, fenofibrate, and gemfibrozil have shown no HDL-increasing effect (25). But in rats fed a HCD, fenofibrate, ciprofibrate, and especially gemfibrozil have been shown to increase HDL-cholesterol (26, 27). In the present study, an increase in HDL-cholesterol was observed for fenofibrate, NNC 61-3058, and Wy-14653. Surprisingly, an increase in HDL-cholesterol (62%) was also observed for rosiglitazone. For rosiglitazone, this effect seems to be related to other factors than activation of PPAR-, because NNC 61-4706, NNC 61-4718, and NNC 61-4424, which are more efficacious PPAR- activators, do not increase HDL but actually marginally reduce HDL in the range of 9–16%. The rosiglitazone-mediated upregulation of HDL levels, however, does raise the question of whether the HDL upregulation mediated by the PPAR- agonists is an effect solely mediated through PPAR- activation.

Many of the genes that were found to be regulated (Table 2) are involved in lipid metabolism and have previously been shown to be regulated by PPAR agonists. These include bifunctional enzyme (peroxisomal ß-oxidation) (28), Cyp4A10 (microsomal -hydroxylation) (29), acetyl-CoA C-acetyltransferase (mitochondrial ß-oxidation) (30), ß-ketothiolase (mitochondrial ß-oxidation), long-chain acyl-CoA dehydrogenase (mitochondrial ß-oxidation) (31), long-chain fatty acyl-CoA synthetase (fatty acid activation) (32), ACBP (fatty acid compartmentalization) (33), apoC-III (lipoprotein metabolism) (11), HMG-CoA synthase (ketone body synthesis) (34), and malic enzyme (NADPH supply for fatty acid synthesis) (35). It has also previously been shown that the liver GLUT-2 promotor contains a functional PPAR response element (36). In experiments in which animals were dosed for 10 days with Wy-14643 and NNC 61-4706 (Table 2), gene regulations were found to be very similar to those observed in animals dosed for 4 days. Data indicate a slight tendency toward larger gene expression changes after 4 days of dosing for a number of genes, which might reflect minor transcriptional feedback regulation during longer treatment protocols. In addition to the compound-mediated transcriptional effects, the HCD used may also have influenced the expression profiles of a number of the genes listed in Table 2. Recently, a microarray study of the transcriptional responses to a HCD reported 69 genes as regulated in livers of mice (37). Genes significantly downregulated by the cholesterol component of the diet included malic enzyme, ACBP, and HMG-CoA synthase. These were also regulated in the present study (Table 2) and are as described above known PPAR target genes. Thus, the regulation of these genes was most likely affected by both the ligand-mediated decrease of total cholesterol (Fig. 3) and the direct PPAR--mediated transcriptional activation. This emphasizes the importance of the choice of diet for gene expression studies and suggests that predictive gene sets may be most accurate when used in animal models similar to the one in which they were identified.

When searching for possible correlations between the observed in vitro PPAR- efficacies for each compound and the expression profiles of the regulated genes, a number of very clear relationships were identified (Fig. 4A). The best linear regressions, based on data from both Northern blotting and microarrays, were found for apoC-III and bifunctional enzyme. This is to our knowledge the first time that a relationship between the gene expression profiles from a series of different drug treatments and the efficacy of the included compounds has been reported. These genes may be useful in the in vivo evaluation of new PPAR--activating compounds.

Even though expression profiles of single genes can correlate to different pharmacodynamic properties (Fig. 5), models that take multiple variables (e.g., gene expression profiles) into account may provide a much more robust prediction of the physiological response of interest. This may especially be the case for drugs with multiple target points, such as fibrates and thiazolidinediones. Therefore, we used a multivariate model called PLS and defined the expression profiles of the regulated genes (Table 2) as predictor (X) variables and the pharmacodynamic parameters as dependent (Y) variables. The pharmacodynamic property, which was best predicted using the hybridization signals from the regulated genes, was triglyceride lowering (Fig. 6). When plotting the observed triglyceride lowering against the values predicted by the PLS model, a very good correlation was observed (R ² = 0.88). A very large fraction of the genes used as predictor variables do indeed encode enzymes involved in triglyceride metabolism. In comparison to the very good prediction of triglyceride change, the prediction of the total cholesterol and the HDL level changes were weaker. When plotting the predicted values against the observed values for these parameters, the R² values were 0.6 for the change in plasma total cholesterol and 0.2 for the change in HDL levels. It is very important not to "overfit" PLS models, and any model must always be validated, for example, by permuting the data that should cause an observed correlation between predicted and observed variables to disappear.

In the present study, seven different PPAR- activators with varying efficacies were dosed in rats fed a HCD. Our observations have shown that comparative gene expression profiling in vivo can reveal useful biomarkers that correlate very well to both the receptor-activating properties of the drugs found in vitro and the pharmacodynamic properties observed in vivo. For the prediction of pharmacodynamic parameters, the use of multivariate models such as PLS seems to be a very useful way of interpreting complex gene expression data in a physiological context. Using the information from 19 regulated genes, a very good prediction of the triglyceride-lowering effect of the included drugs could be made. The data from the present study will be useful when evaluating the potential of new PPAR- activators in rats and demonstrate the applicability of gene expression profiling in the characterization of potentially new drug candidates.

	ACKNOWLEDGMENTS

 
The authors thank Lene Hansen, Tine Pedersen, Anette Seneca, and Rikke Burgdorf for expert technical assistance, Dr. Vladka Panajotova for performing the animal studies, and Drs. Esper Boel and Johan Selmer for critical comments and review of the manuscript.

Manuscript received June 6, 2003 and in revised form November 6, 2003 and in re-revised form December 1, 2003.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

1. Issemann, I., and S. Green. 1990. Activation of a member of the steroid hormone receptor superfamily by peroxisome proliferators. Nature. 347: 645–650.[CrossRef][Medline]

2. Dreyer, C., G. Krey, H. Keller, F. Givel, G. Helftenbein, and W. Wahli. 1992. Control of the peroxisomal beta-oxidation pathway by a novel family of nuclear hormone receptors. Cell. 68: 879–887.[CrossRef][Medline]

3. Tugwood, J. D., I. Issemann, R. G. Anderson, K. R. Bundell, W. L. McPheat, and S. Green. 1992. The mouse peroxisome proliferator activated receptor recognizes a response element in the 5' flanking sequence of the rat acyl CoA oxidase gene. EMBO J. 11: 433–439.[Medline]

4. Gulick, T., S. Cresci, T. Caira, D. D. Moore, and D. P. Kelly. 1994. The peroxisome proliferator-activated receptor regulates mitochondrial fatty acid oxidative enzyme gene expression. Proc. Natl. Acad. Sci. USA. 91: 11012–11016.[Abstract/Free Full Text]

5. Muerhoff, A. S., K. J. Griffin, and E. F. Johnson. 1992. The peroxisome proliferator-activated receptor mediates the induction of CYP4A6, a cytochrome P450 fatty acid omega-hydroxylase, by clofibric acid. J. Biol. Chem. 267: 19051–19053.[Abstract/Free Full Text]

6. Martin, G., K. Schoonjans, A. M. Lefebvre, B. Staels, and J. Auwerx. 1997. Coordinate regulation of the expression of the fatty acid transport protein and acyl-CoA synthetase genes by PPARalpha and PPARgamma activators. J. Biol. Chem. 272: 28210–28217.[Abstract/Free Full Text]

7. Schoonjans, K., J. Peinado-Onsurbe, A. M. Lefebvre, R. A. Heyman, M. Briggs, S. Deeb, B. Staels, and J. Auwerx. 1996. PPAR alpha and PPAR activators direct a distinct tissue-specific transcriptional response via a PPRE in the lipoprotein lipase gene. EMBO J. 15: 5336–5348.[Medline]

8. Vu, D. N., K. Schoonjans, B. Laine, J. C. Fruchart, J. Auwerx, and B. Staels. 1994. Negative regulation of the human apolipoprotein A-I promoter by fibrates can be attenuated by the interaction of the peroxisome proliferator-activated receptor with its response element. J. Biol. Chem. 269: 31012–31018.[Abstract/Free Full Text]

9. Vu, D. N., K. Schoonjans, V. Kosykh, J. Dallongeville, J. C. Fruchart, B. Staels, and J. Auwerx. 1995. Fibrates increase human apolipoprotein A-II expression through activation of the peroxisome proliferator-activated receptor. J. Clin. Invest. 96: 741–750.

10. Chinetti, G., S. Lestavel, V. Bocher, A. T. Remaley, B. Neve, I. P. Torra, E. Teissier, A. Minnich, M. Jaye, N. Duverger, H. B. Brewer, J. C. Fruchart, V. Clavey, and B. Staels. 2001. PPAR-alpha and PPAR-gamma activators induce cholesterol removal from human macrophage foam cells through stimulation of the ABCA1 pathway. Nat. Med. 7: 53–58.[CrossRef][Medline]

11. Staels, B., D. N. Vu, V. A. Kosykh, R. Saladin, J. C. Fruchart, J. Dallongeville, and J. Auwerx. 1995. Fibrates downregulate apolipoprotein C-III expression independent of induction of peroxisomal acyl coenzyme A oxidase. A potential mechanism for the hypolipidemic action of fibrates. J. Clin. Invest. 95: 705–712.

12. Paigen, B., A. Morrow, C. Brandon, D. Mitchell, and P. Holmes. 1985. Variation in susceptibility to atherosclerosis among inbred strains of mice. Atherosclerosis. 57: 65–74.[CrossRef][Medline]

13. Sadowski, I., B. Bell, P. Broad, and M. Hollis. 1992. GAL4 fusion vectors for expression in yeast or mammalian cells. Gene. 118: 137–141.[CrossRef][Medline]

14. Webster, T. D., and R. C. Dickson. 1988. Nucleotide sequence of the galactose gene cluster of Kluyveromyces lactis. Nucleic Acids Res. 16: 8192–8194.[Free Full Text]

15. Cantello, B. C. C., M. A. Cawthorne, D. Haigh, R. M. Hindley, S. A. Smith, and P. L. Thurlby. 1994. The synthesis of brl-49653—a novel and potent antihyperglycemic agent. Bioorg. Med. 4: 1181–1184.[CrossRef]

16. Sauerberg, P., I. Pettersson, L. Jeppesen, P. S. Bury, J. P. Mogensen, K. Wassermann, C. L. Brand, J. Sturis, H. F. Woldike, J. Fleckner, A. S. T. Andersen, S. B. Mortensen, L. A. Svensson, H. B. Rasmussen, S. V. Lehmann, Z. Polivka, K. Sindelar, V. Panajotova, L. Ynddal, and E. M. Wulff. 2002. Novel tricyclic-alpha-alkyloxyphenylpropionic acids: dual PPAR alpha/gamma agonists with hypolipidemic and antidiabetic activity. J. Med. Chem. 45: 789–804.[CrossRef][Medline]

17. Mogensen, J. P., P. Sauerberg, P. S. Bury, L. Jeppesen, and I. Pettersson. 2001. Propionic acid derivatives and their use in the treatment of diabetes and obesity. US patent WO 01/55085 A1.

18. Mogensen, J. P., P. Sauerberg, P. S. Bury, L. Jeppesen, and I. Pettersson. 2001. Alkynylsubstituted propionic acid derivatives and their use against diabetes and obesity. US patent WO 01/55086 A1.

19. Albrektsen, T., K. S. Frederiksen, W. E. Holmes, E. Boel, K. Taylor, and J. Fleckner. 2002. Novel genes regulated by the insulin sensitizer rosiglitazone during adipocyte differentiation. Diabetes. 51: 1042–1051.[Abstract/Free Full Text]

20. Wold, S., M. Sjoström, and L. Eriksson. 2001. PLS-regression: a basic tool of chemometrics. Chemometr. Intell. Lab. Syst. 58: 109–130.

21. Krause, B. R., R. Bousley, K. Kieft, D. Robertson, R. Stanfield, E. Urda, and R. S. Newton. 1994. Comparison of lifibrol to other lipid-regulating agents in experimental animals. Pharmacol. Res. 29: 345–357.[CrossRef][Medline]

22. Cignarella, A., M. Nastasi, E. Cavalli, and L. Puglisi. 1996. Novel lipid-lowering properties of Vaccinium myrtillus L. leaves, a traditional antidiabetic treatment, in several models of rat dyslipidemia: a comparison with ciprofibrate. Thromb. Res. 84: 311–322.[CrossRef][Medline]

23. Naderali, E. K., L. C. Pickavance, J. P. H. Wilding, P. J. Doyle, and G. Williams. 2001. Troglitazone corrects metabolic changes but not vascular dysfunction in dietary-obese rats. Eur. J. Pharmacol. 416: 133–139.[CrossRef][Medline]

24. Krause, B. R., and H. M. G. Princen. 1998. Lack of predictability of classical animal models for hypolipidemic activity: a good time for mice? Atherosclerosis. 140: 15–24.[CrossRef][Medline]

25. Haubenwallner, S., A. D. Essenburg, B. C. Barnett, M. E. Pape, R. B. DeMattos, B. R. Krause, L. L. Minton, B. J. Auerbach, R. S. Newton, T. Leff, and C. L. Bisgaier. 1995. Hypolipidemic activity of select fibrates correlates to changes in hepatic apolipoprotein C-III expression: a potential physiologic basis for their mode of action. J. Lipid Res. 36: 2541–2551.[Abstract]

26. Krause, B. R., and R. S. Newton. 1986. Gemfibrozil increases both apo A-I and apo E concentrations. Comparison to other lipid regulators in cholesterol-fed rats. Atherosclerosis. 59: 95–98.[CrossRef][Medline]

27. Petit, D., M. T. Bonnefis, C. Rey, and R. Infante. 1988. Effects of ciprofibrate and fenofibrate on liver lipids and lipoprotein synthesis in normo- and hyperlipidemic rats. Atherosclerosis. 74: 215–225.[CrossRef][Medline]

28. Bardot, O., T. C. Aldridge, N. Latruffe, and S. Green. 1993. PPAR-RXR heterodimer activates a peroxisome proliferator response element upstream of the bifunctional enzyme gene. Biochem. Biophys. Res. Commun. 192: 37–45.[CrossRef][Medline]

29. Bell, D. R., N. J. Plant, C. G. Rider, L. Na, S. Brown, I. Ateitalla, S. K. Acharya, M. H. Davies, E. Elias, and N. A. Jenkins. 1993. Species-specific induction of cytochrome P-450 4A RNAs: PCR cloning of partial guinea-pig, human and mouse CYP4A cDNAs. Biochem. J. 294: 173–180.

30. Gould-Rothberg, B. E., S. S. Sundseth, V. A. DiPippo, P. J. Brown, D. A. Winegar, W. K. Gottshalk, S. G. Shenoy, and J. M. Rothberg. 2001. The characterization of PPAR-alpha ligand drug action in an in vivo model by comprehensive differential gene expression profiling. Funct. Integr. Genomics. 1: 294–304.[CrossRef][Medline]

31. Ouali, F., F. Djouadi, C. Merlet-Benichou, and J. Bastin. 1998. Dietary lipids regulate beta-oxidation enzyme gene expression in the developing rat kidney. Am. J. Physiol. 44: F777–F784.

32. Schoonjans, K., M. Watanabe, H. Suzuki, A. Mahfoudi, G. Krey, W. Wahli, P. Grimaldi, B. Staels, T. Yamamoto, and J. Auwerx. 1995. Induction of the acyl-coenzyme A synthetase gene by fibrates and fatty acids is mediated by a peroxisome proliferator response element in the C promoter. J. Biol. Chem. 270: 19269–19276.[Abstract/Free Full Text]

33. Elholm, M., G. Bjerking, J. Knudsen, K. Kristiansen, and S. Mandrup. 1996. Regulatory elements in the promoter region of the rat gene encoding the acyl-CoA-binding protein. Gene. 173: 233–238.[CrossRef][Medline]

34. Rodriguez, J. C., G. Gil-Gomez, F. G. Hegardt, and D. Haro. 1994. Peroxisome proliferator-activated receptor mediates induction of the mitochondrial 3-hydroxy-3-methylglutaryl-CoA synthase gene by fatty acids. J. Biol. Chem. 269: 18767–18772.[Abstract/Free Full Text]

35. Castelein, H., T. Gulick, P. E. Declercq, G. P. Mannaerts, D. D. Moore, and M. I. Baes. 1994. The peroxisome proliferator activated receptor regulates malic enzyme gene-expression. J. Biol. Chem. 269: 26754–26758.[Abstract/Free Full Text]

36. Kim, H., J. Kim, S. Kim, J. Y. Cha, K. S. Kim, and Y. Ahn. 2000. Identification and functional characterization of the peroxisomal proliferator response element in rat GLUT2 promoter. Diabetes. 49: 1517–1524.[Abstract]

37. Maxwell, K. N., R. E. Soccio, E. M. Duncan, E. Sehayek, and J. L. Breslow. 2003. Novel putative SREBP and LXR target genes identified by microarray analysis in liver of cholesterol-fed mice. J. Lipid Res. 44: 2109–2119.[Abstract/Free Full Text]

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				P. J. Bajwa, A. Alioua, J. W. Lee, D. S. Straus, L. Toro, and C. Lytle Fenofibrate inhibits intestinal Cl secretion by blocking basolateral KCNQ1 K+ channels Am J Physiol Gastrointest Liver Physiol, December 1, 2007; 293(6): G1288 - G1299. [Abstract] [Full Text] [PDF]

				S. Luci, B. Giemsa, H. Kluge, and K. Eder Clofibrate causes an upregulation of PPAR-{alpha} target genes but does not alter expression of SREBP target genes in liver and adipose tissue of pigs Am J Physiol Regulatory Integrative Comp Physiol, July 1, 2007; 293(1): R70 - R77. [Abstract] [Full Text] [PDF]

				J. D. Brown and J. Plutzky Peroxisome Proliferator Activated Receptors as Transcriptional Nodal Points and Therapeutic Targets Circulation, January 30, 2007; 115(4): 518 - 533. [Abstract] [Full Text] [PDF]

				R. Ringseis, A. Muschick, and K. Eder Dietary Oxidized Fat Prevents Ethanol-Induced Triacylglycerol Accumulation and Increases Expression of PPAR{alpha} Target Genes in Rat Liver J. Nutr., January 1, 2007; 137(1): 77 - 83. [Abstract] [Full Text] [PDF]

				T. Degenhardt, M. Matilainen, K.-H. Herzig, T. W. Dunlop, and C. Carlberg The Insulin-like Growth Factor-binding Protein 1 Gene Is a Primary Target of Peroxisome Proliferator-activated Receptors J. Biol. Chem., December 22, 2006; 281(51): 39607 - 39619. [Abstract] [Full Text] [PDF]

				S. Luci, H. Kluge, F. Hirche, and K. Eder Clofibrate Increases Hepatic Triiodothyronine (T3)- and Thyroxine (T4)-Glucuronosyltransferase Activities and Lowers Plasma T3 and T4 Concentrations in Pigs Drug Metab. Dispos., November 1, 2006; 34(11): 1887 - 1892. [Abstract] [Full Text] [PDF]

				N. F. Cariello, E. H. Romach, H. M. Colton, H. Ni, L. Yoon, J. G. Falls, W. Casey, D. Creech, S. P. Anderson, G. R. Benavides, et al. Gene Expression Profiling of the PPAR-alpha Agonist Ciprofibrate in the Cynomolgus Monkey Liver Toxicol. Sci., November 1, 2005; 88(1): 250 - 264. [Abstract] [Full Text] [PDF]

				F. Molnar, M. Matilainen, and C. Carlberg Structural Determinants of the Agonist-independent Association of Human Peroxisome Proliferator-activated Receptors with Coactivators J. Biol. Chem., July 15, 2005; 280(28): 26543 - 26556. [Abstract] [Full Text] [PDF]

				K. Kitajka, A. J. Sinclair, R. S. Weisinger, H. S. Weisinger, M. Mathai, A. P. Jayasooriya, J. E. Halver, and L. G. Puskas Effects of dietary omega-3 polyunsaturated fatty acids on brain gene expression PNAS, July 27, 2004; 101(30): 10931 - 10936. [Abstract] [Full Text] [PDF]

This Article Abstract Full Text (PDF) An erratum has been published All Versions of this Article: M300239-JLR200v1 45/3/592    most recent Alert me when this article is cited Alert me if a correction is posted Citation Map Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to cit