Advertisement
Research ArticleArticle

Effects on Gene Expression in Rat Liver after Administration of RXR Agonists: UAB30, 4-Methyl-UAB30, and Targretin (Bexarotene)

Peter T. Vedell, Yan Lu, Clinton J. Grubbs, Yuxin Yin, Hui Jiang, Kirby I. Bland, Donald D. Muccio, Dusica Cvetkovic, Ming You and Ronald Lubet
Molecular Pharmacology March 2013, 83 (3) 698-708; DOI: https://doi.org/10.1124/mol.112.082404

Abstract

Examination of three retinoid X receptor (RXR) agonists [Targretin (TRG), UAB30, and 4-methyl-UAB30 (4-Me-UAB30)] showed that all inhibited mammary cancer in rodents and two (TRG and 4-Me-UAB30) strikingly increased serum triglyceride levels. Agents were administered in diets to female Sprague-Dawley rats. Liver RNA was isolated and microarrayed on the Affymetrix GeneChip Rat Exon 1.0 ST array. Statistical tests identified genes that exhibited differential expression and fell into groups, or modules, with differential expression among agonists. Genes in specific modules were changed by one, two, or all three agonists. An interactome analysis assessed the effects on genes that heterodimerize with known nuclear receptors. For proliferator-activated receptor α/RXR-activated genes, the strongest response was TRG > 4-Me-UAB30 > UAB30. Many liver X receptor/RXR-related genes (e.g., Scd-1 and Srebf1, which are associated with increased triglycerides) were highly expressed in TRG and 4-Me-UAB30- but not UAB30-treated livers. Minimal expression changes were associated with retinoic acid receptor or vitamin D receptor heterodimers by any of the agonists. UAB30 unexpectedly and uniquely activated genes associated with the aryl hydrocarbon hydroxylase (Ah) receptor (Cyp1a1, Cyp1a2, Cyp1b1, and Nqo1). Based on the Ah receptor activation, UAB30 was tested for its ability to prevent dimethylbenzanthracene (DMBA)-induced mammary cancers, presumably by inhibiting DMBA activation, and was highly effective. Gene expression changes were determined by reverse transcriptase-polymerase chain reaction in rat livers treated with Targretin for 2.3, 7, and 21 days. These showed similar gene expression changes at all three time points, arguing some steady-state effect. Different patterns of gene expression among the agonists provided insight into molecular differences and allowed one to predict certain physiologic consequences of agonist treatment.

Introduction

Retinoid X receptor (RXR) agonists form heterodimers with the widest range of nuclear receptors, including the peroxisome proliferator activated receptors (PPARs), retinoic acid receptors (RARs), liver X receptors (LXRs), thyroxine receptors, constitutive androstenedione receptor (CAR), vitamin D (VDR) receptor, pregnane X receptor (PXR), and farnesoid X receptor (FXR) (Szanto et al., 2004; Sussman and DeLera, 2005). The resulting heterodimers become transcriptional modulators of a wide variety of genes. There has been significant interest in this class of compounds in cancer because an RXR agonist [Targretin (TRG)] is approved clinically for the treatment of cutaneous T-cell lymphoma (Lansigan and Foss 2010) and has demonstrated clinical efficacy in non–small-cell lung cancer (Dragnev et al., 2011; Kim et al., 2011). In addition, this class of compounds has shown efficacy in a variety of mammary and lung cancer models in rodents (Pereira et al., 2006; Wang et al., 2006b, 2009; Liby, et al., 2007; Zhang et al., 2011), and more recently in humans (Dragnev et al., 2011; Kim et al., 2011). However, the elevation of triglycerides levels by 9-cis-RA, TRG, and various retinoids has been known for many years and is a major concern in the use of these agents, particularly in a cancer-prevention setting (Grubbs et al., 2006; Kolesar et al., 2010). We previously designed selective RXR agonists based on 9-cis-RA that are effective agents with lower toxicity and with varying ability to inhibit mammary cancer formation (Muccio et al., 1998; Atigadda et al., 2003;.Grubbs et al., 2006). One of these analogs (UAB30) did not increase serum triglycerides in rodents (Grubbs et al., 2006) and humans (Kolesar et al., 2010) but was nevertheless effective in preventing mammary cancers in rats.

In the present studies, the ability of three RXR agonists [TRG, UAB30, and 4-Me-UAB30 (4-methyl-UAB30)] to induce gene expression in the liver of treated rats was examined. We aimed to observe the effects of these agonists on genes known to be modulated by agonists for specific receptors, including PPARα, CAR, LXR, and others. Surprisingly, it was found that although each of these compounds binds to the purified RXR receptor and activates gene expression in an RXR α luciferase construct in cell culture, there was limited overlap in the gene expressions modulated. In addition, the effects of these agents on genes associated with increased triglycerides levels (Scd1 and Srebf1) were evaluated. Agonists that increased triglyceride levels (TRG, 4Me-UAB-30) increased expression of the LXR-modulated genes Scd1 and Srebf1), whereas UAB-30 did not. Interestingly, we found that UAB-30 modulated genes associated with the Ah receptor and blocked initiation of mammary tumors by dimethylbenzanthracene. In addition, we examined the ability of TRG to induce 10 specific genes in livers after 2.3, 7, and 21 days of treatment and found similar results at all three time points. This implies that we had reached some steady state regarding gene expression. Nevertheless, the most surprising result was the limited gene expression overlap of the RXR agonists.

Materials and Methods

Animals and Liver Collection for RNA Analysis.

The present experiments were similar to our previously published studies examining the efficacy of these agonists in a rat mammary cancer model (Grubbs et al., 2006). In brief, agonists were administered in Teklad mash diet (4%) to female Sprague-Dawley rats obtained from Harlan Sprague-Dawley, Inc. (Madison, WI). The dietary supplements [TRG (150 ppm in diet), 4-Me-UAB30 (200 ppm in diet), and UAB30 (200 ppm in diet)] were obtained from Eisai, Inc. (TRG) or was synthesized at the University of Alabama at Birmingham (4-Me-UAB30, UAB30) (Atigadda et al., 2003). In brief, rats were obtained at 28 days of age and housed in polycarbonate cages (five per cage) in a room lighted 12 hours per day and maintained at 22°C ± 2°C. The dietary supplements were administered for 7 days. The animals were euthanized by CO2, and the livers were rapidly excised and frozen in liquid N2. Until assayed, the livers were kept at −85°C. A second set of rats were administered TRG (150 ppm) only to assess whether the gene changes observed at 7 days were also seen at other times (2.3 days, 7 days, and 21 days). Otherwise animals and livers were generated as performed with the 7-day samples.

RNA Isolation and Microarray Experiment.

Total RNA were isolated by Trizol (Invitrogen, Carlsbad, CA) and purified using the RNeasy Mini Kit and RNase-free DNase Set (Qiagen, Valencia, CA) according to the manufacturers’ protocols. One microgram of each RNA sample was processed as prescribed by the Affymetrix GeneChip Whole Transcript Sense Target Labeling Assay. The GeneChip WT cDNA Synthesis Kit, cDNA Amplification Kit, and Terminal Labeling Kit (Affymetrix, Inc., Santa Clara, CA) were used for target preparation. A total of 8 μg of cRNA were input into the second-cycle cDNA reaction. Hybridization cocktails containing 3–4 μg of fragmented, end-labeled cDNA were prepared and applied to GeneChip Rat Exon 1.0 ST arrays. The arrays were hybridized for 16 h. A wash and stain script (precommercial FS450_0001 script) was applied using the MES_EukGE-WS2v5_450-DEV fluidics station. Arrays were scanned using the Affymetrix GCS 3000 7G and GeneChip Operating Software v. 1.3 to produce Affymetrix CEL-formatted files which contain quantitative measures for probe hybridization.

Normalization and Probe Filtering.

The probe-level data were contained Affymetrix GeneChip Rat Exon 1.0 ST Array probe set CEL files. They were read into the R statistical software using the R/xps package [Stratowa, 2012 (http://www.bioconductor.org/packages/2.11/bioc/html/xps.html)]. We performed our analysis on the complete set of probe sets that map to a unique, transcriptionally active locus in the rat genome (which includes probe sets classified as “core,” “extended,” and “full” by the chip manufacturer). This set contains approximately 728,247 exons that target 18,192 unique genes. The data were normalized by the Robust Multichip Average quantile normalization method (Bolstad et al., 2003). To improve the sensitivity of downstream analysis, we further filtered the probe sets, retaining only those with an intensity that falls above or in the extreme top of the observed intensity distribution for the antigenomic or negative control, probes after adjustment for guanine/cytosine base content in at least one of the 20 samples (detection above background P value < 0.05). After this filtering, 421,360 exons targeting 16,838 genes remained. Thus, there is evidence of expression in the exonic regions targeted by the probe sets that remain.

Identifying Transcripts Differentially Expressed.

Analysis of variance methods, performed with the R/maanova package (Wu et al., 2002), were used to identify probe sets statistically with differential expression associated with treatment. The factorial experimental design includes one factor (“treatment”) with four levels (4-Me-UAB30, TRG, UAB30, and control) and four samples per factor level, each from a different individual sample. For each probe set g, treatment i, and replicate k, a linear model for the log-transformed expression measure, yijkg, can be formulated as a sum of components that contribute to the overall intensity value:

Embedded Image(1)

where μg is the mean intensity over all 20 samples, αijg is the effect of the factor level corresponding to treatment i (I = 1, 2, 3, 4, 5), and εijkg is the residual error.

To identify transcripts with the highest probability of differential expression between any two experimental groups, we applied the omnibus or general F test. Using R/maanova, we performed a modified general F-test (called Fs), which incorporates shrinkage estimates of residual variance and increases the power to detect differential expression compared with an ordinary F-test (Cui et al., 2005). P values were calculated by permuting model residuals. Using permutation P values rather than tabular P values is more robust to departures from model assumptions such as the underlying statistical distribution. As an adjustment for multiple testing, we used the Benjamin-Hochberg transformation of the P values to estimate false discovery rate (Benjamini and Hochberg, 1995). Transcripts with differences between strains were identified as those with estimated false discovery rates (FDRs) less than 0.01 (i.e., FDR < 0.01). In general, theory suggests that using an FDR threshold of n implies that only about 100 × n out of 100 genes are not actually differential expressions. We also performed F tests for the four contrasts defined by comparing each treatment with the control group and used an analogous approach for detecting differential expression.

Weighted Gene Coexpression Network Analysis.

The coexpression patterns of differentially expressed genes (according to our general F-test) were analyzed by weighted gene coexpression network analysis (Ravasz et al., 2002; Zhang and Horvath, 2005). For each pair of probe sets, the Pearson correlation coefficient, r, was transformed to a distance, or dissimilarity metric metric, d, on the unit scale where a correlation of 1 has a dissimilarity of 0 and a correlation of −1 has a dissimilarity of 1. The probe sets were given a weighted connection, d6, and then the topographical overlap transformation of these connections was computed using the R/WGCNA package (Langfelder and Horvath, 2008). For a pair of probe sets, the topographical overlap measures the similarity of their distributions of distances to all other genes. Transcript abundance profiles were hierarchically clustered modules obtained by the hybrid dynamic dendrogram cutting method of the R/dynamicTree Cut package. Modules are referenced by a color index. The expression patterns and module assignments of the treatment genes were displayed via a heat map generated using Java Tree View (Saldanha, 2004).

Gene-Set Enrichment and Pathway Analysis.

Gene-set enrichment and pathway analysis was performed in the GeneGo Metacore software using the Enrichment Workflow and the Interactome Analysis Workflow [Thomson Reuters, 2010 (http://thomasonreuters.com/content/science/pdf/ls/metacore-cfs-en.pdf)]. We applied these workflows for various subsets of differentially expressed genes as described in the Results section. For each application of the Enrichment Workflow, we reported the GeneGo Pathway Maps, GeneGo Process Networks, GeneGo Diseases, GeneGo Metabolic Networks (Endogenous), and Top GO Processes as Supplemental Tables. For each application of the Interactome Analysis Workflow, we reported the most overconnected proteins by functional category as a supplemental table where the categories are transcription factors, receptors, ligands, kinases, proteases, phosphases, enzymes, and others. The GeneGo Metasearch database was used to identify the known targets of the selected nuclear receptor-related targets, except in the cases of CAR/RXRα and PXR/RXRα. The known targets of these RXR-containing complexes were obtained from the study presented by Tolson and Wang (2010).

Gene ID Conversion.

The gene ID conversion tools of the DAVID Bioinformatics Resource (http://david.abcc.ncifcrf.gov/) (Dennis et al., 2003) and the RGD (http://rgd.mcw.edu) (Twigger et al., 2007) were used to identify current IDs where needed.

Reverse Transcriptase-Polymerase Chain Reaction of Livers Exposed to Targretin for Varying Time Periods.

To determine whether the gene changes observed at 7 days are similarly observed at other times, we treated groups of rats (n = 5) with TRG (150 ppm) for 2.3, 7, and 21 days. Livers were isolated and RNA isolated as already described. Quantitative RT-PCR (reverse transcriptase-polymerase chain reaction) of the isolated livers was performed. For 10 genes that we found to have upregulated expression in the livers of rats treated with TRG relative to controls in the original experiment, we measured the relative expressions by quantitative RT-PCR. Total RNAs were isolated and purified using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s protocols. One microgram of total RNA per sample was converted to cDNA using Script cDNA synthesis kit (Bio-Rad, Hercules, CA). Primers for quantitative RT-PCR analysis were designed by using Primer5 software (Supplemental Table 1). Amplification of each target was performed using a SsoFast EvaGreen Supermix (Bio-Rad) with 500 nM of each primer and 10 ng of cDNA per reaction. Each real-time assay was done in duplicate on a BioRad CFX96 thermal cycler, and data were collected and analyzed with BioRad CFX software (Version 2.1). The internal control gene GAPDH and target genes were amplified with equal efficiencies. The fold change of target gene expression was calculated as 2-ΔΔCT (ΔΔCT = ΔCT treated − ΔCT control), and all data are presented as mean ± S.E. of relative gene expression.

Prevention of Dimethylbenzanthracene-Induced Mammary Tumors.

This model examines the ability of an agent to block the initiation of mammary carcinogenesis (positive) in rats. We have published multiple articles using this model and recently showed that 5, 6 benzoflavone was highly effective (Grubbs et al., 1995; Lubet et al., 2011). In brief, female Sprague-Dawley rats were administered UAB30 (200 ppm), 5, 6 benzoflavone (400 ppm), or control diet from 43 days of age until 57 days of age. After 57 days of age, all rats were placed on a control diet. At 50 days of age, rats were given a single intragastric dosing of DMBA (dimethylbenzanthracene; 12 mg). Rats were then monitored for an additional 120 days for the development of mammary cancers.

Results

Microarray Experiment.

A microarray experiment was performed employing the Affymetrix GeneChip Rat Exon 1.0 ST Array platform to determine the short-term effects of three RXR agonists on gene expression in rat liver and to relate them to the effects that different RXR agonists have on serum triglyceride levels. The RXR agonists were administered in the diet for 7 days; rats were then killed, and RNA was isolated as described in Materials and Methods. The analyses were performed across all the probe sets of this array platform that, based on the chip manufacturer’s annotation, map to a unique transcription ally active locus in the rat genome. This set contains more than 700,000 exons that target more than 18,000 unique genes. A filtering step was used to retain only probe sets that show evidence of target gene expression in at least one of the 20 samples, after which about 400,000 exons remained.

Differential Expression Is Less for UAB30 than for Other Agonists.

To assess the statistical significance of variation in between-treatment-group liver gene expression, an empirical Bayes-like general F-test was applied to identify genes that were variable between treatment groups (Cui et al., 2005). Greater differential expression than expected by chance (Supplemental Fig. 1) was found. For example, there were 777 exons of 185 unique genes that were differentially expressed based on highly stringent permutation-based P values after adjustment by a false-discovery-rate–based multiple-test correction (i.e., FDR < 0.005). The pairwise contrasts between the control group and each of the three treatment groups were also tested. The distribution of significance scores (permutation-based P values) show that, estrogen receptor compared with controls, there is greater differential expression for 4-Me-UAB30 and TRG than for UAB30 (Supplemental Fig. 1, b–d). Specifically, at FDR < 0.01, 544 (4-Me-UAB30), 298 (TRG), and 14 (UAB30) exons of 143 (4-Me-UAB30), 76 (TRG), 5 (UAB30) unique genes were differentially expressed. Using the same stringent differential expression criterion among the genes selected by the general F-test criterion that was used, the number of genes differentially expressed between controls and UAB30 was very small (five genes). However, when the FDR criterion was relaxed to FDR < 0.25, it was found that 145 exons for 25 unique genes were differentially expressed for UAB30. Of these 25, 15 genes were represented in the set of genes selected by the empirical Bayes general F-test criterion. Using this less stringent criterion, we observed 497 genes whose expression was modulated by 4-Me-UAB30 and 210 genes whose expression was modulated by TRG. Thus, there is evidence for differential expression between UAB30 and controls, but the number of genes and the quantity of differential expression are small relative to the differential expression between the other two RXR agonists and controls.

Genes Fall into Coexpression Network Modules that are Biologically Enriched.

The genes detected as differentially expressed by the general F-test, in general, may exhibit multiple patterns of coexpression across the samples. Weighted gene coexpression network analysis was used to cluster those genes into modules with correlated patterns of variation (Fig. 1). Eight modules containing up to 62 genes were found. We refer to the modules using color indices (genes per module: blue, 62; brown, 29; green, 27; magenta, 8; pink, > 3; red, 20; turquoise, 29; yellow 34). Three modules in which TRG-treated rats were upregulated were enriched with target genes for PPARα/RXRα (green, yellow, brown). One of those modules (brown) also exhibited upregulation with 4-Me-UAB30 and was enriched with target genes known to be activated by the PPARα/RXRα heterodimer complex (six genes, P < 1E-07, e.g., Acaa1, Ehhadh Mdh1), LXRα/RXRα (three genes, P < 1E-04, Scd1, Acaca, Cd36), and TRα/RXRα (three genes, P < 1E-04, e.g., Slc2a5, Mdh1). There was also some enrichment (P < 0.002) for CAR/RXRα (two genes), PPARγ/RXRα (three genes), nuclear SREBP1 (five genes), MLX (two genes including Thrsp), and c-Myc (10 genes). The key gene of the nuclear SREBP1 pathway, Srebf1, also had a pattern of expression that highly correlated with the pattern of this module (Fig. 1). UAB30 showed relatively little or no response for genes of this module. Another module (turquoise) exhibited downregulation for 4-Me-UAB30 and, to a lesser extent, TRG. This module was enriched for target genes of FXR and FXR-regulated cholesterol and bile acids cellular transport pathways (Supplemental Table 2). Downregulation of this pathway is associated with higher triglycerides levels in rats (Evans et al., 2009). Again, UAB30 showed relatively little or no response for genes of that module. Thus, two modules (brown and turquoise) had common patterns of expression in both TRG- and 4-Me-UAB30-treated rats that are associated with higher triglycerides levels.

Fig. 1.
Fig. 1.

Heat map with coexpression module assignments from clustering of variable genes. A heat map is shown for the differentially expressed genes (based on the general F-test). Samples are hierarchically clustered based on Euclidean distance, and exons are hierarchically clustered using the topographical overlap dissimilarity distance between probe sets, which is a correlation-based measure used for clustering. Treatment groups exhibited substantial separation in the sample dendrogram. Expression values are centered and scaled within each exon (i.e., row). Low expression is indicated by white, and high expression is indicated by black. Coexpression modules define highly connected sets of correlated expression profiles as indicated by color indices in the vertical margins of the maps. The modules are characterized in Supplemental Table 2.

Two modules (yellow and green) exhibited increased expression after TRG treatment, whereas expression of these genes was slightly reduced by 4-Me-UAB30 (green). Both these modules were enriched with genes that have been previously shown to be activated by PPARα/RXRα or by PPARα (not necessarily dimerized with RXRα). Thus, there appears to be a greater PPARα- related transcriptional activation for TRG than for the other RXR-agonists.

The largest module (blue, 60 genes) contained several genes that are activated by PPARγ and RXRα (Cyp26a1, Cyp26b1, Cyp26c1, Orm1, Jak3) or PPARγ (not necessarily dimerized with RXRα [Egr1 (early growth response 1), Akt 1]. The genes Cyp26a1, Cyp26b1, Cyp 26c1 as well as Cyp2c13, Cyp4a8 are cytochrome P450 genes, all of which are involved in retinol metabolism. Ugt 2a3, also in this module, is similarly involved in retinol metabolism. The Egr1 protein interacts with many other proteins. In the context of this experiment, these annotations suggested that there was greater PPARγ/RXRα activity for 4-Me-UAB30-treated rats than for the other RXR agonists. The expression for one of these genes (Egr1) was also increased by UAB30, albeit to a lesser extent than by 4-Me-UAB30.

Expression Patterns for Genes Regulated by RXR-Containing Complexes Are Diverse.

The interactome analysis of the gene modules revealed that several modules were highly enriched with targets of RXR-containing complexes. However, because a highly stringent criterion for determining the differentially expressed genes was used, the analysis does not indicate whether genes that are regulated by a particular RXR complex are all affected similarly or if a particular RXR agonist affects particular targets. To address this, the MetaCore database [Thomson Reuters, 2010 (http://thomasonreuters.com/content/science/pdf/ls/metacore-cfs-en.pdf)] was queried separately for each of 31 transcription factors that are RXR-containing complexes or are components of these complexes. We identified genes annotated in the rat liver to be under transcriptional regulation or whose expression was otherwise influenced by the transcription factors (Supplemental Table 3). Then, for each gene set, the expression patterns observed for all exons of those genes were examined. It was observed that no set showed a uniform response across all genes. However, the target genes of a number of proteins or protein complexes included subsets of genes with similar patterns with relative uniformity within experimental treatment group. For example, the strongest signal observed among the genes annotated to be activated by LXRα/RXRα was a group of genes that exhibited higher expression for TRG and 4-Me-UAB30 than for the other groups (Fig. 2; Table 1), including brown module genes such as Acaca and Cd36. Among genes annotated to be activated by PPARα/RXRα was a group that exhibited higher expression for TRG than for the other three groups (e.g., yellow or green module genes Acadm, Gpd1, Me1, and Ephx1). There were, however, some genes that also showed high expression in the 4-Me-UAB30 samples (Fig. 3; Table 2) (such as Ech1 and Echdc1). Several genes with higher expression in 4-Me-UAB30-treated rats were activated by PPARγ/RXRα or PPARγ (e.g., Lrp1, Aldh2, blue module genes, Cyp26a1, Cyp26b1, Cyp26c1, Orm1, Jak3, Egr1, and Akt 1) (Fig. 1). Among genes annotated to be activated by PXR/RXRα, Sult1b1, and various glutathione S-transferases (e.g., Gsta3, Gsta3, Gstp1, Gstm4) were upregulated by 4-Me-UAB30 (Fig. 4; Table 3). Ces6, Ces21, Cyp3a2, and Cyp3a23 were upregulated by TRG (Fig. 4; Table 3). Akrlc14, Ugt1a1, Ugt1a6, and other glutathione S-transferases (Gstm1, Gstm3, Gstm6l) were upregulated by UAB30 (Fig. 4; Table 3). There was increased expression among genes annotated to be activated by FXR/RXRα in the TRG samples (Abcb11, Ppargc1a, Slc27a5) Supplemental Table 4, Supplemental Figure 4). By contrast, none of the studied RXR-agonists increased genes annotated to be activated by RARα/RXRα (Supplemental Fig. 2), nor did they significantly activate genes annotated to be activated by VDR/RXRα (Supplemental Fig. 3).

Fig. 2.
Fig. 2.

Genes activated by LXRα/RXRα. A heat map is shown for the exons of genes activated by LXRα/RXRα in the rat liver. Samples are hierarchically clustered based on Euclidean distance, and exons are hierarchically clustered using the topographical overlap dissimilarity distance between probe sets, which is a correlation-based measure used for clustering. Low expression is indicated by black, and high expression is indicated by white. Increased expression was observed for 4-Me-UAB30 and TRG on a large block of exons. Expression values are centered and scaled within each exon (i.e., row).

View this table:
TABLE 1

Genes activated by liver X receptors (LXRs)α/retinoid X receptors (RXRs)α

For selected genes known to be activated by LXRα/RXRα, a fold change (FC) is shown for each treatment group relative to the control group. These are the fold changes for the exon of that gene with the smallest permutation P value when an empirical Bayes overall F-test is applied. The corresponding unadjusted and false discovery rate (FDR)-adjusted permutation P values are shown.

Fig. 3.
Fig. 3.

Genes activated by PPARα/RXRα. A heat map is shown for the exons of genes activated by PPARα/RXRα in the rat liver. Samples are hierarchically clustered based on Euclidean distance, and exons are hierarchically clustered using the topographical overlap dissimilarity distance between probe sets, which is a correlation-based measure used for clustering. The most dominant signal is increased expression for the TRG group. There are also some genes with increased expression in the 4-Me-UAB30 treated group.

View this table:
TABLE 2

Genes activated by peroxisome proliferator-activated receptors (PPARs)α/retinoid X receptor (RXRs)α

For selected genes known to be activated by PPARα/RXRα, a fold change (FC) is shown for each treatment group relative to the control group. These are the fold changes for the exon of that gene with the smallest permutation P value when an empirical Bayes overall F-test is applied. The corresponding unadjusted and false discovery rate (FDR)-adjusted permutation P values are shown.

Fig. 4.
Fig. 4.

Genes activated by PXR/RXRα. A heat map is shown for the exons of genes activated by PXR/RXRα in the rat liver. Samples are hierarchically clustered based on Euclidean distance, and exons are hierarchically clustered using the topographical overlap dissimilarity distance between probe sets, which is a correlation-based measure used for clustering. There are distinct sets of genes with increased expression for TRG and 4-Me-UAB30.

View this table:
TABLE 3

Genes activated by pregnane X receptor (PXR)/ retinoid X receptor (RXR)α

For selected genes known to be activated by PXR/RXRα, a fold change (FC) is shown for each treatment group relative to the control group. These are the fold changes for the exon of that gene with the smallest permutation P value when an empirical Bayes overall F-test is applied. The corresponding unadjusted and false discovery rate (FDR)-adjusted permutation P values are shown.

Enrichment for Genes Specifically Altered by UAB30 Is Different.

One small module of genes selected by the general F-test criterion (pink) exhibited a strong response for UAB30 but not for the other agonists. The genes of this module were contained in the set of genes differentially expressed between UAB30 and controls using the slightly more relaxed criterion FDR < 0.25. The genes with increased expression in UAB30 (Fig. 4; Table 4) had overrepresentation of targets not for RXR-containing complexes, e.g., CAR-RXR complex (Supplemental Figure 5) but rather AHR, NRF2, and HNF1α (Table 4). The genes with increased expression in UAB30 included at least five genes whose protein products are associated with Ah receptor (aryl hydrocarbon receptor) activation [Cyp1a1, Cyp1a2, and Cyp1b1, Ugt1a6 and Nqo1 (NADH dehydrogenase, quinone1)]. Two of the three known rat orthologs to the four human genes primarily involved in UAB30 metabolism (Gorman et al., 2007) were upregulated (LOC293989: FC = 1.8, P = 0.001, Cyp2c79: FC = 1.3, P = 0.03) in UAB30 samples, whereas the other was not (Ugt1a9: FC = −1.05, P = 0.86). However, another UDP-glucuronosyltransferase (Ugt1a6) was up over 3-fold in the UAB30 samples. In toto, the pathways and the interactions perturbed by UAB30 were substantially different from those for the other RXR agonists and do not include categories commonly associated with triglyceride metabolism. (Fig.5)

View this table:
TABLE 4

Genes differentially expressed between UAB30 and controls [false discovery rate (FDR) < 0.25]

For selected genes upregulated on UAB30 relative to controls, a fold change (FC) is shown for each treatment group relative to the control group. These are the fold changes for the exon of that gene with the smallest permutation P value when an empirical Bayes overall F-test is applied. The corresponding unadjusted and FDR-adjusted permutation P values are shown. Note: Two other genes that satisfied the criteria for increased expression by UAB30 and whose homologs were targets of AHR in Homo sapiens were Per1 and Bhlhe40.

Fig. 5.
Fig. 5.

Genes differentially expressed between UAB30 and controls (FDR < 0.25). A heat map is shown for the exons of differentially expressed genes between UAB30 and controls (FDR < 0.25). Samples are hierarchically clustered based on Euclidean distance, and exons are hierarchically clustered using the topographical overlap dissimilarity distance between probe sets, which is a correlation-based measure used for clustering.

Comparison with Similar Studies Performed with RT-PCR.

Based on our previous data with TRG (Wang, et al., 2006a), we had performed RT-PCR for a more limited number of genes in livers treated with TRG or UAB30. The results of the present arrays were compared with the prior quantitative RT-PCR findings. A substantial concordance was observed. All fold change directions were the same between the two experiments for these genes (Supplemental Table 4). Furthermore, for eight genes for which PCR was performed and for which significant differential expression was observed, a high degree of correlation in fold change (r = 0.73, P = 0.03; Supplemental Fig. 6) was found. We performed a second RT-PCR experiment with TRG to determine whether the gene expression changes observed at 7 days in the microarray experiment were similarly observed in livers of animals treated for different time periods (2 1/3, 7, and 21 days). This experiment examined 10 genes that showed increased gene expression in the initial 7-day microarray experiment on independent samples. In the follow-up quantitative RT-PCR experiment, we found the average fold changes for all the TRG-treated time-point–defined groups were positive for each gene (Table 5). The fold increases ranged from 1.4-fold to 9.6-fold (2.3 days: 1.8- to 7.7-fold increase; 7 days: 1.4- to 7.0-fold increase; 21 days: 1.8- to 9.6-fold increase) (Table 5). The difference compared with controls after 21 days is significant for each gene (P < 0.05). The difference compared with controls after 2.5 days and 7 days is significant for all genes at P < 0.15 and for most genes at P < 0.05 (2.5-day group: 9/10; 7-day group: 6/10) (Table 5). In summary, these genes invariably show substantial upregulation at each of three time points ranging from less than 3 days to 3 weeks, and almost all observed increases are statistically significant, even with just five individual rats per group.

View this table:
TABLE 5

Fold change (FC) 9FC0 and P values for Targretin time-series reverse transcriptase (RT)-polymerase chain reaction (PCR) experiment

The FC and P values for the follow up Targretin time-series RT-PCR experiment are shown. The fold changes are positive for all genes at all time points. The difference compared with controls after 21 days is significant for all genes (P < 0.05). The difference compared with controls after 2.5 days and 7 days is significant for all genes at P < 0.15 and for most genes at P < 0.05 (2.3 days: 9/10; 7 days: 6/10).

Effects of UAB30 and 5,6 Benzoflavone on DMBA-Induced Rat Mammary Tumors.

Sprague-Dawley rats were administered UAB30 (200 ppm) or 5,6 benzoflavone (400 ppm) in the diet beginning at age 43 days and continuing until 57 days of age. At 50 days of age, rats were administered a single oral dose of DMBA. Treatment with either UAB30 or 5, 6 benzoflavone (a known Ah receptor agonist) reduced the incidence of rat mammary tumors by more than 85% (Fig. 6).

Fig. 6.
Fig. 6.

Effect of UAB30 on mammary tumorigenesis. UAB30 and the positive control 5,6 benzoflavone were administered to female Sprague-Dawley rats for 1 week before and 1 week after DMBA treatment at 50 days of age. The study was terminated 120 days after the rats received the carcinogen.

Discussion

The effects on gene expression of three known RXR agonists were determined. 4-Methyl-UAB30 and UAB30 were designed based on empirical structure-activity relationships and computer modeling. We have found in transactivation assays (Muccio et al., 1998) that although TRG and 4-Me-UAB30 have similar binding affinities (∼25 nM), the affinity for UAB30 is roughly 120 nM. None of the agents significantly bound to RAR at 2000 nM (Muccio et al., 1998). Previously, we showed that these agents prevented ER+ mammary cancer (Grubbs et al., 2006). TRG and 4-Me-UAB30 reduced ER+ mammary cancers by ∼80%, whereas UAB30 inhibited roughly 55–65%. However, TRG and 4-Me-UAB30 increased serum triglycerides by 300–400%, whereas UAB30 had minimal effects on triglycerides (Grubbs et al., 2006). We expected rexinoids to alter the interactome involving the PPARα/RXRα complex since a number of these genes associated with peroxisome proliferation were highly induced by TRG and 9cisRA (Wang et al., 2006a). Also, activation of PPARα was associated with hepatocytomegaly and liver tumor promotion in rodents (Viswakarma et al., 2010), and we observed hepatocytomegaly with both TRG and 4-Me-UAB30. TRG increased a wide variety of PPARα-related genes, whereas 4-Me-UAB30 and UAB30 modulated a more select group (Fig. 3; Table 2). The large differences in gene expression between UAB30 and 4-Me-UAB30 were surprising since they differ by a single methyl group. Among the limited number of genes altered by UAB30, at least five genes were associated with the activation of the Ah receptor [Cyp1a1, Cyp1a2, Cyp1b1, Ugt1a6, and Nqo1 (quinone reductase 1)] (Table 4). We showed that these genes were highly induced by the Ah receptor agonist 5,6-benzoflavone (Lubet et al., 2009). UAB30 clearly activated Ah receptor related genes (Table 4), and our prior findings showed that compounds that activate this receptor inhibit DMBA-induced mammary cancer and aflatoxin-induced liver cancer (Grubbs et al., 1995; Lubet et al., 2011). Therefore, we tested UAB30 (200 ppm) in the DMBA-induced mammary model, which reduced cancers by >80%, as much as a suboptimal dose (400 ppm) of 5,6 benzoflavone (Fig. 6). Thus, the array data alone allowed us to predict a significant biologic outcome based strictly on the gene expression data. At least two additional questions might arise regarding the results with DMBA. First, is activation of the Ah receptor the only way to alter DMBA metabolism and DMBA-induced initiation? In fact, there appear to be two clearly documented mechanisms for accomplishing this goal: induction of the Ah receptor, which alters various cytochromes and certain phase II enzymes, and induction of the antioxidant response element, which highly induces a wide variety of phase II drug-metabolizing enzymes. We and others have discussed these two major alternative pathways. In fact, none of the three RXR agonists significantly induced the ARE-related genes [GST Pi, aldo keto reductase 3A7 (aflatociol)]. Only UAB30 induced the Ah-related genes. We previously tested TRG for the ability to inhibit DMBA metabolism and found it to be ineffective (unpublished data). We have not tested 4-Me-UAB30. The induction of Ah receptor–mediated genes by UAB30 raises multiple questions. First, will UAB30 block initiation by other carcinogens (e.g., aflatoxin and aromatic amines that are blocked by 5,6-benzoflavone (discussed in Lubet et al., 2009). Second, how UAB30 activates the Ah receptor is unclear. One would not expect UAB30 to be an effective Ah receptor agonist since binding to the receptor requires a highly planar molecule and the low-energy confirmation is L-shaped to fit into the RXR binding pocket (Muccio, unpublished data). Furthermore, 4-Me-UAB30, which differs by a single methyl group, only minimally affected these genes. We have shown that UAB30 inhibits DMBA initiation of tumorigenesis and TRG does not. We previously demonstrated that all three agents inhibit the progression stage of tumor development in ER+ rat tumors and the progression stage of ER- tumors in MMTV-Neu/P53 transgenic mice (unpublished data). This clear overlap of response to progression in both ER+ and ER- rodent models leads us to expect a more similar gene response than we observed.

TRG and 4-Me-UAB30 altered expression of a wide number of genes associated with specific nuclear receptor-RXR heterodimers [e.g., PPARα/RXRα, PXR/RXRα, LXRα/RXRα (Figs. 2–4; Tables 1–3)] but had limited effect on other RXR heterodimers (e.g., vitamin D receptor/RXR, VDR/RXR, RAR/RXR) (Supplemental Figs. 2 and 3). Thus, the RXR agonists do not stimulate Tran activation of genes for VDR/RXR and RAR/RXR. These heterodimers require the presence of VDR or RAR agonists, respectively (nonpermissive or conditional permissive). In contrast, genes modulated by the heterodimers PPARα/ RXRα respond to either PPARα agonists or to RXR agonists (Wang et al., 2006a). The combinations of agonists for both nuclear receptors often lead to synergistic gene activation (permissive). This study also revealed different responses to the different RXR agonists even among genes that are annotated to be activated by the same nuclear receptor heterodimer (e.g., PPARα/RXRα, LXRα/RXRα) (Figs. 2–4; Tables 1–3). Some of the diversity may be due to the complexities of transcriptional control. Coregulators (coactivators/corepressors) may be affected by the presence of specific RXR agonists and conformation changes in the protein heterodimer complex.

Finally, we determined whether gene expression changes in the liver might help explain the differential effects on triglycerides levels (TRG and 4-Me-UAB30 increased; UAB30 had minimal effect). Previous studies with RXR agonists showed that after TRG treatment, LXRα/RXRα modulates certain genes, including Scd1 and Srebf1, that are associated with increased triglycerides. In the current study, these genes were increased by both TRG and 4-Me-UAB30 but were minimally affected by UAB30. Llalloyer et al. (2009) argued that the TRG-increased serum triglycerides in mice were due to activation of the LXRα/RXRα heterodimers by increasing levels of Scd1 and Srebf1. Our study supports this finding. In addition to Scd1 and Srebf1, other potential triglyceride-related genes were induced by TRG and 4-Me-UAB30 but not UAB30. LXRα/RXRα (brown module) controls the expression of Acaca, Acaa1, and Cd36. Acaca and CD 36, all directly involved in the metabolism of triglycerides. The Srebf1 protein, SREBP1, regulates transcription of the brown module genes Pdk1, Mdh1, and Idh3B (Dennis et al., 2003). Other brown module genes that were increased both by TRG and 4-Me-UAB30 were Acsm3, Acsm5, Ces6, Crot, and Decr2. These genes are involved in lipid and fatty acid metabolism, whereas brown module genes Dlat, Pdhb, and Mdh1 are involved in acetyl-CoA metabolism (Dennis et al., 2003). Fatty acid metabolism and acetyl-CoA metabolism are closely related to triglyceride metabolism, and the brown module genes Pdlim1, Acsm3, Acsm5, Cyp2c22, Nupr1, Scd1, and Vwce are all found within triglyceride quantitative trait loci regions of the rat genome on chromosome 1 (Twigger et al., 2007). Thus, there appears to be a wide variety of genes beyond Scd1 and Srebf1 that may contribute to the triglyceride effects.

One question that often arises is how meaningful are the results observed by arrays. This has really multiple components, including the following: Are the RT-PCR results reproducible, and are they consistent over time? The first study was performed in some overall view by comparing a series of RT-PCR determinations on a different set of similarly treated samples; results were obtained by array analysis, and we observed strong overall agreement (Supplementary Table 4). The second question was whether results obtained at 7 days of treatment were similar at other treatment times. As shown in Table 5 for 10 genes, expression changes observed at 7 days were similarly observed after 2.3 days or 21 days of treatment. Although absolute numbers may have varied, the same gene changes were seen and the magnitude of the changes was similar.

In conclusion, three RXR-selective agonists modulated gene expression differently in rat liver, even among those genes annotated to be activated by the same RXR heterodimer. One of the clearest implications of this study is that each RXR-agonist must be fully characterized for their gene expression profiles to understand their physiologic effects fully. In our study, a pattern of increased expression was observed for LXRα/RXRα complex target genes related to increased triglycerides levels for two RXR-selective agonists (4-Me-UAB30 and TRG), but not for UAB30, which did not increase triglycerides. However, we identified many additional genes that are associated with this triglyceride effect, which is a dose-limiting toxicity of many retinoids. The reduced triglyceride effect for UAB30, which was reflected in minimal gene expression changes, makes it a promising drug for cancer prevention, which requires chronic administration of drugs to high risk, but otherwise healthy, populations. These studies showed that UAB30 activated Ah receptor genes, which predicted the efficacy of this agent in the DMBA mammary cancer model. This was presumably due to inhibition of DMBA initiation and is an additional chemoprevention mechanism for this specific agent. Thus, the gene data alone allow one to make predictions about significant physiologic effects. Also, it was found that observing gene changes at 7 days was reflective of similar changes at 2.3 and 21 days, implying a steady state.

Acknowledgments

The authors thank Ricardo Verdugo for helpful exchanges regarding the exon array data analysis. 4-Methyl-UAB30 and UAB30 were synthesized and purified by Dr. Reddy Atigadda in the Department of Chemistry, University of Alabama at Birmingham.

Authorship Contributions

Participated in research design: Vedell, Grubbs, Muccio, Cvetkovic, You, Lubet.

Contributed new reagents or analytic tools: Vedell, Lu, Yin, Jiang, Muccio, You.

Performed data analysis: Vedell, Lu, Yin, Jiang, Cvetkovic, You.

Wrote or contributed to the writing of the manuscript: Grubbs, Muccio, Lubet, Bland.

Footnotes

  • This research was supported by the National Institutes of Health National Cancer Institute [Grant P50 CA089019 and contract number HHSN261200433001C]. This research was also supported in part by the Intramural Research Program of the National Institutes of Health (National Cancer Institute).

  • dx.doi.org/10.1124/mol.112.082404.

  • ↵Embedded ImageThis article has supplemental material available at molpharm.aspetjournals.org.

Abbreviations

Ah receptor
aryl hydrocarbon hydroxylase receptor
CAR
constitutive androstenedione receptor
DMBA
dimethylbenzanthracene
Egr1
early growth response 1
FDR
false discovery rate
FXR
farnesoid X receptor
LXR
liver X receptor
PPAR
peroxisome proliferator-activated receptor
PXR
pregnane X receptor
RAR
retinoic acid receptor
RT-PCR
reverse transcriptase-polymerase chain reaction
RXR
retinoid X receptor
TRG
Targretin
4-Me-UAB30
4-methyl-UAB30
VDR
vitamin D receptor
  • Received September 17, 2012.
  • Accepted January 4, 2013.

References

  1. ↵
    1. Atigadda VR,
    2. Vines KK,
    3. Grubbs CJ,
    4. Hill DL,
    5. Beenken SL,
    6. Bland KI,
    7. Brouillette WJ,
    8. Muccio DD
    (2003) Conformationally defined retinoic acid analogues. 5. Large-scale synthesis and mammary cancer chemopreventive activity for (2E,4E,6Z,8E)-8-(3′,4′-dihydro-1′(2’H)-naphthalen-1′-ylidene)-3,7-dimethyl-2,4,6-octatrienoic acid (9cUAB30). J Med Chem 46:3766–3769.
    OpenUrlCrossRefPubMed
  2. ↵
    1. Benjamini Y,
    2. Hochberg Y
    (1995) Controlling the false data rate: a practical and powerful approach to multiple testing. J R Statist Soc Ser B (Methodological) 57:289–300.
    OpenUrl
  3. ↵
    1. Bolstad BM,
    2. Irizarry RA,
    3. Astrand M,
    4. Speed TP
    (2003) A comparison of normalization methods for high density oligonucleotide array data based on variance and bias. Bioinformatics 19:185–193.
    OpenUrlAbstract/FREE Full Text
  4. ↵
    1. Cui X,
    2. Hwang JT,
    3. Qiu J,
    4. Blades NJ,
    5. Churchill GA
    (2005) Improved statistical tests for differential gene expression by shrinking variance components estimates. Biostatistics 6:59–75.
    OpenUrlAbstract
  5. ↵
    1. Dennis G Jr.,
    2. Sherman BT,
    3. Hosack DA,
    4. Yang J,
    5. Gao W,
    6. Lane HC,
    7. Lempicki RA
    (2003) DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol 4:P3.
    OpenUrlCrossRefPubMed
  6. ↵
    1. Dragnev KH,
    2. Ma T,
    3. Cyrus J,
    4. Galimberti F,
    5. Memoli V,
    6. Busch AM,
    7. Tsongalis GJ,
    8. Seltzer M,
    9. Johnstone D,
    10. Erkmen CP,
    11. et al.
    (2011) Bexarotene plus erlotinib suppress lung carcinogenesis independent of KRAS mutations in two clinical trials and transgenic models. Cancer Prev Res (Phila) 4:818–828.
    OpenUrlAbstract/FREE Full Text
  7. ↵
    1. Evans MJ,
    2. Mahaney PE,
    3. Borges-Marcucci L,
    4. Lai K,
    5. Wang S,
    6. Krueger JA,
    7. Gardell SJ,
    8. Huard C,
    9. Martinez R,
    10. Vlasuk GP,
    11. et al.
    (2009) A synthetic farnesoid X receptor (FXR) agonist promotes cholesterol lowering in models of dyslipidemia. Am J Physiol Gastrointest Liver Physiol 296:G543–G552.
    OpenUrlAbstract/FREE Full Text
  8. ↵
    1. Gorman GS,
    2. Coward L,
    3. Kerstner-Wood C,
    4. Cork L,
    5. Kapetanovic IM,
    6. Brouillette WJ,
    7. Muccio DD
    (2007) In vitro metabolic characterization, phenotyping, and kinetic studies of 9cUAB30, a retinoid X receptor-specific retinoid. Drug Metab Dispos 35:1157–1164.
    OpenUrlAbstract/FREE Full Text
  9. ↵
    1. Grubbs CJ,
    2. Steele VE,
    3. Casebolt T,
    4. Juliana MM,
    5. Eto I,
    6. Whitaker LM,
    7. Dragnev KH,
    8. Kelloff GJ,
    9. Lubet RL
    (1995) Chemoprevention of chemically-induced mammary carcinogenesis by indole-3-carbinol. Anticancer Res 15:709–716.
    OpenUrlPubMed
  10. ↵
    1. Grubbs CJ,
    2. Lubet RA,
    3. Atigadda VR,
    4. Christov K,
    5. Deshpande AM,
    6. Tirmal V,
    7. Xia G,
    8. Bland KI,
    9. Eto I,
    10. Brouillette WJ,
    11. et al.
    (2006) Efficacy of new retinoids in the prevention of mammary cancers and correlations with short-term biomarkers. Carcinogenesis 27:1232–1239.
    OpenUrlAbstract/FREE Full Text
  11. ↵
    1. Kim ES,
    2. Herbst RS,
    3. Wistuba II,
    4. Lee JJ,
    5. Blumenschein GR Jr.,
    6. Tsao A,
    7. Stewart DJ,
    8. Hicks ME,
    9. Erasmus J Jr.,
    10. Gupta S,
    11. et al.
    (2011) The BATTLE trial: personalizing therapy for lung cancer. Cancer Discov 1:44–53.
    OpenUrlAbstract/FREE Full Text
  12. ↵
    1. Kolesar JM,
    2. Hoel R,
    3. Pomplun M,
    4. Havighurst T,
    5. Stublaski J,
    6. Wollmer B,
    7. Krontiras H,
    8. Brouillette W,
    9. Muccio D,
    10. Kim K,
    11. et al.
    (2010) A pilot, first-in-human, pharmacokinetic study of 9cUAB30 in healthy volunteers. Cancer Prev Res (Phila) 3:1565–1570.
    OpenUrlAbstract/FREE Full Text
    1. Lalloyer F,
    2. Pedersen TA,
    3. Gross B,
    4. Lestavel S,
    5. Yous S,
    6. Vallez E,
    7. Gustafsson JA,
    8. Mandrup S,
    9. Fiévet C,
    10. Staels B,
    11. et al.
    (2009) Rexinoid bexarotene modulates triglyceride but not cholesterol metabolism via gene-specific permissivity of the RXR/LXR heterodimer in the liver. Arterioscler Thromb Vasc Biol 29:1488–1495.
    OpenUrlAbstract/FREE Full Text
  14. ↵
    1. Langfelder P,
    2. Horvath S
    (2008) WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics 9:559572.
    OpenUrlCrossRefPubMed
  15. ↵
    1. Lansigan F,
    2. Foss FM
    (2010) Current and emerging treatment strategies for cutaneous T-cell lymphoma. Drugs 70:273–286.
    OpenUrlCrossRefPubMed
  16. ↵
    1. Liby K,
    2. Royce DB,
    3. Risingsong R,
    4. Williams CR,
    5. Wood MD,
    6. Chandraratna RA,
    7. Sporn MB
    (2007) A new rexinoid, NRX194204, prevents carcinogenesis in both the lung and mammary gland. Clin Cancer Res 13:6237–6243.
    OpenUrlAbstract/FREE Full Text
  17. ↵
    1. Lubet RA,
    2. Heckman BM,
    3. De Flora SL,
    4. Steele VE,
    5. Crowell JA,
    6. Juliana MM,
    7. Grubbs CJ
    (2011) Effects of 5,6-benzoflavone, indole-3-carbinol (I3C) and diindolylmethane (DIM) on chemically-induced mammary carcinogenesis: is DIM a substitute for I3C? Oncol Rep 26:731–736.
    OpenUrlPubMed
  18. ↵
    1. Lubet RA,
    2. Yao R,
    3. Grubbs CJ,
    4. You M,
    5. Wang Y
    (2009) Induced expression of drug metabolizing enzymes by preventive agents: role of the antioxidant response element. Chem Biol Interact 182:22–28.
    OpenUrlCrossRefPubMed
  19. ↵
    1. Muccio DD,
    2. Brouillette WJ,
    3. Breitman TR,
    4. Taimi M,
    5. Emanuel PD,
    6. Zhang X,
    7. Chen G,
    8. Sani BP,
    9. Venepally P,
    10. Reddy L,
    11. et al.
    (1998) Conformationally defined retinoic acid analogues. 4. Potential new agents for acute promyelocytic and juvenile myelomonocytic leukemias. J Med Chem 41:1679–1687.
    OpenUrlCrossRefPubMed
  20. ↵
    1. Pereira MA,
    2. Kramer PM,
    3. Nines R,
    4. Liu Y,
    5. Alyaqoub FS,
    6. Gunning WT,
    7. Steele VE,
    8. Lubet RA
    (2006) Prevention of mouse lung tumors by targretin. Int J Cancer 118:2359–2362.
    OpenUrlCrossRefPubMed
  21. ↵
    1. Ravasz E,
    2. Somera AL,
    3. Mongru DA,
    4. Oltvai ZN,
    5. Barabasi AL
    (2002) Hierarchical organization of modularity in metabolic networks. Science 297:1151–1155.
    OpenUrlAbstract/FREE Full Text
  22. ↵
    1. Saldanha AJ
    (2004) Java Treeview—extensible visualization of microarray data. Bioinformatics 20:3246–3248.
    OpenUrlAbstract/FREE Full Text
  23. ↵
    1. Sussman F,
    2. de Lera AR
    (2005) Ligand recognition by RAR and RXR receptors: binding and selectivity. J Med Chem 48:6212–6219.
    OpenUrlCrossRefPubMed
  24. ↵
    1. Szanto A,
    2. Narkar V,
    3. Shen Q,
    4. Uray IP,
    5. Davies PJ,
    6. Nagy L
    (2004) Retinoid X receptors: X-ploring their (patho)physiological functions. Cell Death Differ 11(Suppl 2):S126–S143.
    OpenUrlCrossRefPubMed
  25. ↵
    1. Tolson AH,
    2. Wang H
    (2010) Regulation of drug-metabolizing enzymes by xenobiotic receptors: PXR and CAR. Adv Drug Deliv Rev 62:1238–1249.
    OpenUrlCrossRefPubMed
  26. ↵
    1. Twigger SN,
    2. Shimoyama M,
    3. Bromberg S,
    4. Kwitek AE,
    5. Jacob HJ,
    6. RGD Team
    (2007) The Rat Genome Database, update 2007—easing the path from disease to data and back again. Nucleic Acids Res 35 (Database issue):D658–D662.
    OpenUrlAbstract/FREE Full Text
  27. ↵
    Viswakarma N, Jia Y, Bai L, Vluggens A, Borensztajn J, Xu J, Reddy JK. Coactivators in PPAR-regulated gene expression. PPAR Res pii.250126 Epub 2010 Aug. 5.
  28. ↵
    1. Wang Y,
    2. Wen W,
    3. Yi Y,
    4. Zhang Z,
    5. Lubet RA,
    6. You M
    (2009) Preventive effects of bexarotene and budesonide in a genetically engineered mouse model of small cell lung cancer. Cancer Prev Res (Phila) 2:1059–1064.
    OpenUrlAbstract/FREE Full Text
  29. ↵
    1. Wang Y,
    2. Yao R,
    3. Maciag A,
    4. Grubbs CJ,
    5. Lubet RA,
    6. You M
    (2006a) Organ-specific expression profiles of rat mammary gland, liver, and lung tissues treated with targretin, 9-cis retinoic acid, and 4-hydroxyphenylretinamide. Mol Cancer Ther 5:1060–1072.
    OpenUrlAbstract/FREE Full Text
    1. Wang Y,
    2. Zhang Z,
    3. Yao R,
    4. Jia D,
    5. Wang D,
    6. Lubet RA,
    7. You M
    (2006b) Prevention of lung cancer progression by bexarotene in mouse models. Oncogene 25:1320–1329.
    OpenUrlCrossRefPubMed
  31. ↵
    1. Wu H,
    2. Kerr K,
    3. Churchill GA
    (2002) MAANOVA: a software package for the analysis of spotted cDNA microarray experiments, in The Analysis of Gene Expression Data (Parmigiani G, Garrett ES, Irizarry RA, and Zeger SL eds) pp 313341, Springer, New York.
  32. ↵
    1. Zhang B,
    2. Horvath S
    (2005) A general framework for weighted gene co-expression network analysis. Stat Appl Genet Mol Biol 4:15446115.
    OpenUrl
  33. ↵
    1. Zhang Q,
    2. Pan J,
    3. Zhang J,
    4. Liu P,
    5. Chen R,
    6. Chen DR,
    7. Lubet R,
    8. Wang Y,
    9. You M
    (2011) Aerosolized bexarotene inhibits lung tumorigenesis without increasing plasma triglyceride and cholesterol levels in mice. Cancer Prev Res (Phila) 4:270–276.
    OpenUrlAbstract/FREE Full Text
View Abstract
Back to top

In this issue

Download PDF
Article Alerts
Email Article
Citation Tools
Research ArticleArticle

Effects of RXR Agonists on Gene Expression

Peter T. Vedell, Yan Lu, Clinton J. Grubbs, Yuxin Yin, Hui Jiang, Kirby I. Bland, Donald D. Muccio, Dusica Cvetkovic, Ming You and Ronald Lubet
Molecular Pharmacology March 1, 2013, 83 (3) 698-708; DOI: https://doi.org/10.1124/mol.112.082404
del.icio.us logo Digg logo Reddit logo Twitter logo CiteULike logo Facebook logo Google logo Mendeley logo

Jump to section