Genetic profiling defines the xenobiotic gene network controlled by the nuclear receptor pregnane X receptor

Please download to get full document.

View again

of 15
15 views
PDF
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Document Description
Genetic profiling defines the xenobiotic gene network controlled by the nuclear receptor pregnane X receptor
Document Share
Document Tags
Document Transcript
  Genetic Profiling Defines the Xenobiotic GeneNetwork Controlled by the Nuclear ReceptorPregnane X Receptor JOHN M. ROSENFELD, REYNALDO VARGAS, J R ., WEN XIE,  AND  RONALD M. EVANS The Salk Institute for Biological Studies (J.M.R., R.M.E.), Gene Expression Laboratory, Howard Hughes Medical Institute (R.M.E.), La Jolla, California 90237; and Center for Pharmacogenetics(W.X.), Department of Pharmaceutical Sciences, University of Pittsburgh, Pittsburgh,Pennsylvania 15213 The orphan nuclear receptor pregnane X receptor(PXR) is essential for the transcriptional regulationof hepatic xenobiotic enzymes including the cyto-chrome 3A isoenzymes. These enzymes are cen-tral to the catabolism and clearance of most en-dogenous sterol metabolites (endobiotics) and avast diversity of foreign compounds (xenobiotics)including pharmaceuticals, pesticides, and toxinsencounteredthroughdietandenvironmentalexpo-sure. To explore a broader role of PXR in the mam-malian xenobiotic response, we have conducted aunique microarray gene profiling analysis on liversamples derived from PXR knockout mice andmice expressing a constitutively active variant, VP-hPXR. This genetically guided expression analysisenables targeting and restriction of the PXR re-sponse to liver, and is devoid of side effects result-ing from drugs and their metabolites. As with phar-macological studies, receptor-dependent genesinclude both phase I and phase II metabolic en-zymes,aswellascertaindrugandaniontransport-ers as principal PXR targets. Moreover, compara-tive analysis of data from both genetic andpharmacological arrays reveals a core networkthat represents a genetic description of the xeno-biotic response. (   Molecular Endocrinology   17:1268–1282, 2003) T HE INDUCTION OF cytochrome P450 (CYP) andother phase I monooxygenases by endo- and xe-nobiotic compounds defines the prototypical primaryhepatic response in rodents and humans (1). Thesexenobiotics include rifampin, phenobarbital, natural orsyntheticsteroids,bileacids,industrialpesticides,andnumerous classes of chemical inducers. This hepaticresponse provides an unusual biochemical and mo-lecular interface between mammalian physiology,pharmacology, and toxicology. The nuclear receptorspregnane X receptor (PXR) and constitutive andro-stane receptor (CAR) were srcinally established asxenosensors that regulate the genes encoding theCYP3A and CYP2B oxidative enzymes, respectively(2–4). The biochemical activities of these enzymes areprimarily responsible for the metabolic inactivationand subsequent clearance of most pharmaceuticalsincluding statins, thiazolidinediones, and chemother-apeutic agents. PXR, a broad specificity nuclear re-ceptor,canbeactivatedinaligand-dependentfashionby numerous structurally diverse endo- and xenobiot-ics (5). CAR, while exhibiting constitutive transcrip-tional activity in cultured cells, is activated  in vivo by cytoplasmic to nuclear translocation in responseto treatment with phenobarbital, or the derivative1,4-bis[2-(3,5-dichloropyridiloxy)]benzene (TCPOBOP)(6, 7). The activity of CAR may also be affected by thepresence of endogenous steroid metabolites (8, 9). Inaddition to distinct differences in the activation ofthese receptors, they bind distinct but partially over-lapping consensus DNA response elements. However,the true extent of the overlap of genomic domainscontrolled by PXR  vs.  CAR has not been systemicallyexamined.To characterize the genetic and physiological do-mains of these receptors, we and others have gener-ated transgenic knockout (KO) mice for both PXR andCAR (10–13). As expected, PXR null animals arelargely deficient in responding to PXR specific induc-ers such as pregnenolone carbonitrile (PCN) or dexa-methasone (DEX) treatment. CAR-specific and PXR/ CAR dual agonists retain their CAR-mediated effectsin PXR null animals. We have also “humanized” thesemice by introducing transgenes to direct hepatic ex-pression of either the full-length human PXR, or atranscriptionally active variant, termed VP-hPXR. In  Abbreviations: ABC, ATP binding cassette; CAR, constitu-tive androstane receptor; CYP, cytochrome P450; DEX,dexamethasone; EST, expressed sequence tag; FMOC5, fla-vin containing containing monooxygenase 5; FXR, farnesoid X receptor; GST, glutathione- S -transferase; hPXR, humanPXR; HSD, hydroxysteroid dehydrogenase; KO, knockout;mPXR, mouse PXR; PAPSS2, 3  -phosphoadenosine 5  -phosphosulfate synthase 2; PB, phenobarbital; PCN, preg-nenolone carbonitrile; PPAR  , peroxisome proliferator-acti-vated receptor   ; PXR, pregnane X receptor; SDS, sodiumdodecyl sulfate; SSC, saline sodium citrate; STa2, sulfo-transferasea2;TCPOBOP,1,4-bis[2-(3,5-dichloropyridiloxy)]-benzene;UGT,uridinediphosphate-glucuronosyltransferase; VDR, vitamin D receptor; VP16, viral protein 16; VP-hPXR,transcriptionally active variant of hPXR. 0888-8809/03/$15.00/0 Molecular Endocrinology 17(7):1268–1282 Printed in U.S.A.  Copyright © 2003 by The Endocrine Societydoi: 10.1210/me.2002-0421 1268  the VP-hPXR mice, prototypical PXR target genes areconstitutively up-regulated (11). These animals pro-vide invaluable tools to identify and characterize thePXR-specific genetic network. Given the difficulty andvariability in producing primary cultured hepatocytescombined with the nonhepatic properties of immortal-ized liver cell lines such as HepG2, the  in vivo  assess-ment of activation of the receptor provides a physio-logically relevant approach to identifying xenobiotictarget genes. Furthermore, use of the constitutivelyactivated VP-hPXR transgene avoids employment ofdrug activation, which harbors many variables includ-ing production of metabolites, frequency of dosing,route of delivery, effective concentration of drugs, re-ceptor independent actions of drugs, hepatic toxicity(   i.e.  PCN), and pan-receptor activity of certain com-pounds, such as DEX. For example, various bile acidmetabolites can directly activate at least three nuclearreceptors, farnesoid X receptor (FXR)/bile acid recep-tor, PXR, and the vitamin D receptor (VDR), all of whichare expressed in the enterohepatic axis (10, 14 – 18).Use of the genetically activated receptor avoidsmany of the above problems to create a pharmaco-logically unbiased profiling data set. As expected, theresulting expression data identify broad componentsof xenobiotic phase I and II response genes, as well asnovel targets that may facilitate phase III endo andxenobiotic elimination. Comparison of the geneticallyderived data to chemical agonist profiles identifiesboth new putative target genes, as well as subtledifferences in regulation of previously known PXR andCAR gene targets. Our data set also considerablyilluminates the roadmap of PXR and CAR specific andoverlapping gene targets, further defining the xenobi-otic metabolic network. RESULTSExpression Profiling of VP-hPXR Gain-of-Function Mice To search for genes affected by PXR activation, wepooled total hepatic RNA from age-matched adultmale nontransgenic or VP-hPXR transgenic mice inthe mouse PXR (mPXR)   /    or PXR null background(Fig. 1A) (11). Two color cDNA microarray compari-sons on a collection of approximately 8700 sequenceverified IMAGE clones were performed, using fluoror-eversal confirmation to select differentially expressedclones (see  Materials and Methods  ). Comparisonswere performed with first-strand cDNA probes gener-ated from nontransgenic  vs.  albumin driven VP-hPXRtransgenic liver mRNA samples, as well as betweenliver mRNAs from PXR null mice in the absence  vs. presence of VP-hPXR transgene.Expression analysis of VP-hPXR in the presence ofendogenous mPXR (wild type) produced 168 differen-tially ESTs, with a roughly equal distribution of posi-tively and negatively affected genes across four spotmeasurements (97 up and 71 down) (Fig. 1, B and C).Based upon densitometry of predicted genes that val-idate by Northern blot, true fold changes are oftengreater than the averaged fluororeversed data, allow-ing us to consider genes as low as 1.3-fold (as long asthey are confirmed by flurophor reversal). Utilizing thisfiltering criteria, more than 90% of clones identified asdifferentially expressed using gene calls from flouro-reversal-confirmed data can be both detected anddirectionally confirmed by Northern hybridization (datanot shown). In the mPXR null background, 146 genetags were differentially expressed due to VP-hPXRpresence (56 up and 90 down). Based on GenBankexpressed sequence tag (EST) accession numbers, 43tags were differentially expressed in both PXR   /   and PXR   /    backgrounds. Approximately 14 addi-tional redundant ESTs were detected in both compar-isons when clone identities were assigned, resultingin a true overlap of 57 tags (see the first table inthe supplemental data published on The EndocrineSociety ’ s Journals Online web site at http://mend.endojournals.org; and see Fig. 1B). The directionalrelationships among all of these genes is shown sche-matically in Fig. 1C, with each subpopulation (7) ofpredicted gene hits assigned a class or cluster lettera – f in Fig. 1D. Class a and e genes are affected by VP-hPXR expression in both the presence and ab-sence of endogenous mPXR, and are predictivelymore reliable due to repetition of detection (   i.e.  in-crease or decrease in eight of eight spot measure-ments), whereas classes b, c, f, and g report as dif-ferentially expressed in one comparison or the other,but not both. Figure 1E illustrates hierarchal clusteringof the differentially expressed tags from both compar-isons, based upon averaged fold differences reportedfor each individual comparison, and the position ofgenes within these experiment clusters. There are 17tags of the d class that are of uncertain significanceduetothelackofdirectionalconsistency,increasinginresponse to VP-hPXR in the presence of mPXR butdecreasing in its absence.The predictive quality of the array data was vali-dated by Northern hybridization and densitometry ofnormalized blots. As an example of the reliability of ourcDNA microarray gene profiling system, 12 of 13(92%) of the genes from class  “ a ”  were validated byNorthern blot analysis of samples from both compar-isons (see Table 1 and Figs. 2 and 3, data not shown).Because of the stringency of our thresholding andselection criteria, we have observed that our low falsepositive rate of detection results in an obligate in-crease in the false negative rate. For example, severalgenes that reportedly increased by virtue of VP-hPXRexpression in the presence of mPXR (wild type) werenot detected as increases in the companion data setobtained in the absence of mPXR. However, whenanalyzed by Northern blot hybridization, these geneswere similarly affected in both genetic backgrounds(  e.g.  CYP3A11). Thus, the differences between mi-croarray reporting of these RNA samples do not nec- Rosenfeld  et al  .  •  Xenobiotic Regulation by PXR Mol Endocrinol, July 2003, 17(7):1268 – 1282  1269  Fig. 1.  cDNA Expression Profiling of Wild-Type (WT) and PXRKO Albumin (Alb)-VP-hPXR Livers A, Northern blot analysis of hepatic total RNA from WT and PXRKO transgenics. Blots were hybridized with mPXR and VP-hPXRprobesandarecontrolledforRNAloadingbyhybridizationto36B4ribosomalmessage.B,OverlapofdifferentiallyESTsbetween WT and PXRKO animals expressing the VP-hPXR transgene. ESTs that passed conservative criteria for differential 1270  Mol Endocrinol, July 2003, 17(7):1268 – 1282 Rosenfeld  et al  .  •  Xenobiotic Regulation by PXR  essarily indicate differential response to VP-hPXR inthe absence or presence of endogenous mPXR. Ta-bles 1 and 2 describe a subset of genes increased anddecreased (respectively) by VP-hPXR expression, withgenes that have been validated by Northern hybridiza-tion shown in boldface type. The entirety of fluoro-reverse filtered differential expression data and fullarray content are supplied in the first and secondtables in the supplemental data, respectively, and theraw custom cDNA array data before fluororeversalcomparison and filtering is supplied in the remainingtables in the supplemental data. Regulation of Phase I CYPs  As expected, transcription of several CYP geneswas increased in the VP-hPXR transgenic mice.These include CYP2B10, CYP3A11, and two occur-rences of CYP20 (Table 1). CYP20, an uncharacter-ized mouse homolog of human CYP-M, is weaklysimilar on the protein level to mouse CYP4V3 andCYP3A13. Despite the high degree of similarity be-tween mouse CYP3A11 and CYP3A13 (71% aminoacid identity, 76% DNA identity throughout codingregion), the CYP20 transcript has no significant nu-cleotide similarity to either of these transcripts. BothCYP20 and CYP2B10, a prototypic CAR target genethat is also responsive to PXR activators, were de-tected as a class  “ a ”  genes (4, 19). Examples of therelative differences among these and other CYPsknown to be regulated by PXR are compared byNorthern analysis in Fig. 2A.Surprisingly, several distinct CYPs reported as de-creases when VP-hPXR was present (Table 2). Theseinclude duplicate ESTs encoding CYP4A10 andmouse CYP4V3, as well as a distinct CYP2C44 iso-form. The CYP4A family is involved in mitochondrialand peroxisomal  - and   -oxidation of long chain fattyacids (including steroids and some xenochemicals),and independently, is a known target of peroxisomeproliferator-activated receptor    (PPAR   ) in the liver(20 – 23). CYP2C44 is most similar to human CYP2C19(62% amino acid identity), and CYP4V3 is most similarto Drosophila 4C3 (48% amino acid identity). Previousreports have shown the CYP2C19 is strongly inhibitedby omeprazole treatment, a known CYP3A4 inducerby virtue of PXR activation (24, 25). The CYP4C/4Fenzymes have been implicated in pesticide and leu-kotriene metabolism, and the mouse CYP4F is also aPPAR  target (26, 27). Two other PPAR  target genes,acyl coenzyme A oxidase (decrease) and carnitine-palmitoyl transferase I (increase), were also affectedby VP-hPXR expression (Tables 1 and 2). This meta-bolic antagonism or cross-regulation has also beenobserved in several fibrate and bile acid feeding ex-periments involving PPAR   and presumptive FXR ac-tivation, although both the molecular mechanisms andphysiologicalrelevanceinvolvingthisphenomenonareunclear (28, 29). Esterases and Other Phase I Enzymes Affectedby VP-hPXR Expression In addition to the regulation of CYP enzyme genes, anumber of carboxyesterase like ESTs including car-boxylesterase 3 (triacylglycerol hydrolase), as well astwo carboxyesterase 2 encoding ESTs, were up-regulated (Table 1 and Fig. 2B). These two esteraseproteinsare44%identicalthroughouttheirlength,andthe representative EST sequences exhibit very shortregions of weak nucleotide similarity. Esterases suchas these are involved in the activation of certain pro-drugs as well as metabolism of natural substrates, andat least one related family member (egasyn, esterase22) is inducible by androgens (30). Interestingly, the  - D -glucoronidase that associates with egasyn in themicrosome is also induced in our data set and is alsoinducible by androgens (see supplemental data andRef. 31). While the molecular details of this regulationare unknown, androgen induction of these genescould be exerted either by the androgen receptor di-rectly (present in all samples, data not shown), orpossibly by androgen or androgen metabolite activa-tion of PXR. This is relevant because androgens arealso substrates for the CYP enzymes discussedabove, and we have observed that androstenedione isan efficacious PXR activator (data not shown). Theexpression of a fourth esterase, male specific esterase31 was decreased in both comparisons (see Table 2and Ref. 32). This enzyme is most similar to carboxy-lesterase 2 (43% amino acid identity). Thus, constitu-tive hPXR activation has both positive and negativeaffects on the regulation of multiple classes of phase Ienzymes. A number of dehydogenases were also inducedby VP-hPXR expression. These include alcohol de- expression with fluorophor-reversal are presented. Forty-three ESTs were identified by identical GenBank accession numbers,whereas an additional 14 were included based on overlapping/redundant clone content irrespective of accession number. C andD,Distributionanddirectionalityof271uniqueVP-hPXRdifferentiallyexpressedgenesresultsin7classesofgenebehavior.ESTsfrom B were sorted by fold difference of VP-hPXR expressing livers  vs.  control (positive and negative) in both WT and PXRKOgenetic backgrounds. The relationships and overlapping behavior between these ESTs across both experiments is illustrated inC, with classes of genes assigned a letter a – f in D. E, Hierarchal clustering of 314 identified ESTs in the microarray comparisonsof WT    /    VP-hPXR and PXRKO    /    VP-hPXR. Clustering was performed using Genespring 4.2 based upon averaged folddifference values for all ESTs validated by fluorophor-reversal and thresholding at a scaled value of 1.3-fold. Genes shown in  red  are increased in the presence of VP-hPXR, and genes shown in  green  are decreased, with intensity of color associated withmagnitude of fold difference. The positions of genes from classes a – f are shown with  brackets , and the number of ESTs in eachclass is shown in  parentheses .Rosenfeld  et al  .  •  Xenobiotic Regulation by PXR Mol Endocrinol, July 2003, 17(7):1268 – 1282  1271  hydrogenase 3A2 [ADH3A2, or fatty aldehyde dehy-drogenase (Aldh4)] and aldehyde dehydrogenase1a7 (Aldh1a7) (Table 1). Aldh1a7 is phenobarbitalinducible and is also known to isomerize retinalde-hyde to generate retinoids (33, 34). ADH3A2 is amicrosomal oxidoreductase known to be inducibleby dioxin and clofibrate (35) (Fig. 2B). Phase I en-zymes negatively affected by PXR activation includeflavin containing monooxygenase 5 (FMOC5), aswell as multiple occurrences of hydroxysteroid de-hydrogenase-4,   5  -3-   (HSD3  4) and hydroxy-steroid dehydrogenase-1,   5  -3-   (HSD3  1)(Table 2). The 3  HSDs consist of a multigene familyof enzymes involved in the conversion of preg-nenolone and its metabolites to progesterones andandrostenedione. In addition to their sexually dimor-phic expression in the liver and kidney, isoforms ofthis enzyme are likely to be involved in the produc-tion of androstane metabolites that may function asPXR and/or CAR ligands (36, 37). Regulation of Phase II Enzymes In addition to the phase I targets, a number of phase IIconjugating enzymes are also regulated in response to VP-hPXR expression (Table 1 and Fig. 3A). For ex- Table 1.  Genes Up-Regulated by Hepatic VP-hPXR Expression Fold DifferenceClass Accession No. DescriptionPXR   /    PXR   /   1.53 C AA122658 6-Phosphogluconate dehydrogenase, decarboxylating 2.25 1.64 A AA238875 Alcohol dehydrogenase 3A2 (ADH3A2) 2.26 B AA122814 Aldehyde dehydrogenase family 1, subfamily A7 (AldH1a7) 2.47 1.42 A AA203871 A-raf1.42 B AA059837 ATP-binding cassette B9 (ABCB9) similar 1.65 C AA000856   -glucuronidase structural 2.09 1.65 A AA265270 BI652730 1.79 B AA168956 Calcium binding protein, intestinal 14.16 8.68 A AA237173 Carboxylesterase 22.66 2.47 A AA271522 Carboxylesterase 21.84 B AA509566 Carboxylesterase 3 1.44 B AA050178 Carnitine palmitoyl transferase I (CPT1)1.39 C AA271223 CCAAT/enhancer binding protein (C/EBP),   1.93   1.43 D AA108600 Coproporphyrinogen oxidase1.43 B AA050944 Coproporphyrinogen oxidase similar1.48 C W88005 Cyclin-dependent kinase inhibitor 1A (P21)1.49   1.48 D AA108457 CYP20 8.68 3.58 A W12874 CYP2B104.12 B AA268120 CYP3A11 1.67 B AA096870 Cystathionine     lyase 1.79 1.79 A AA120757 Ectonucleoside triphosphate diphosphohydrolase 5 (PCPH)1.65 C AA271043 Ectonucleoside triphosphate diphosphohydrolase 5 (PCPH)1.48 B AA116513 Fatty acid synthase2.25 2.71 A W54349 GSTA44.39 13.37 A AA250402 Huntingtin1.76 B AA061468 Hydroxymethylglutaryl-CoA synthase, cytoplasmic3.59 B AA197454 INSIG22.83 B AA444946 INSIG22.14 B W98208 Latent TGF-B BP1 (LTBP-1) related 1.87 C AA087193 Lipocalin 2 1.88 1.77 A W13053 Microsomal triglyceride transfer protein (MTTP)2.29 1.53 A W18519 PAPSS22.40 B AA244536 PAPSS23.46 C AA097421 Progesterone membrane binding protein (PMBP) like 2.09 C W14332 PTH synthase/DCOH (TCF1) 1.39 B W90961 Putative transmembrane protein (PTG)9.92 45.24 A AA230451 S100 calcium binding protein A8 (MRP8, calgranulin A) 6.05 16.72 A AA253700 Serine (or cysteine) proteinase inhibitor, clade D, member 1 4.36 B AA269437 Sulfotransferase, hydroxysteroid preferring 2 (STa2)1.98 B AA220699 Transcobalamin 2 2.09 C W34612 Transglutaminase 2, C polypeptide Averaged fold differences for flurophor-reversal confirmed genes is shown, as well as class identifications from Fig. 1D. Genesin  bold   have been verified by normalized Northern blot analysis and densitometry (Figs. 2 and 3, and data not shown). PTH,6-Pyruvoyl-tetrahydropterin; DCOH, dimerization cofactor of hepatocyte nuclear factor 1  ; TCF1, transcription factor 1. 1272  Mol Endocrinol, July 2003, 17(7):1268 – 1282 Rosenfeld  et al  .  •  Xenobiotic Regulation by PXR
Similar documents
View more...
Search Related
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks