Toxicology and Applied Pharmacology 234 (2009) 89–97

Contents lists available at ScienceDirect

Toxicology and Applied Pharmacology

j ourna l homepage: www.e lsev ie r.com/ locate /ytaap

Estrogenic status modulates the effect of soy on hepatic responses to7,12-dimethylbenz(a)anthracene (DMBA)

Rohit Singhal a, Thomas M. Badger b,c, Martin J. Ronis a,c,⁎a Department of Pharmacology and Toxicology, University of Arkansas for Medical Sciences, Little Rock, AR-72205, USAb Department of Physiology and Biophysics, University of Arkansas for Medical Sciences, Little Rock, AR-72205, USAc Arkansas Children's Nutrition Center, Little Rock, AR-72202, USA

Abbreviations: AhR, aryl hydrocarbon receptor; E2, eCYP1A1, cytochrome P-450-1A1; DMBA, 7,12-dimethovariectomized; PAHs, polycyclic aromatic hydrocarbosoy protein isolate; XRE, xenobiotic response element.⁎ Corresponding author. Arkansas Children's Nutriti

Marshall Street, Little Rock, AR 72202, USA. Fax: +1 501E-mail address: ronismartinj@uams.edu (M.J. Ronis)

0041-008X/$ – see front matter © 2008 Elsevier Inc. Aldoi:10.1016/j.taap.2008.09.027

a b s t r a c t

a r t i c l e i n f o

Article history:

We examined the influenc Received 15 May 2008Revised 29 September 2008Accepted 29 September 2008Available online 15 October 2008

Keywords:Aryl hydrocarbon receptorAhR, SoyMicroarray

e of estradiol (E2) status and soy protein isolate (SPI) intake on the hepaticresponses altered by 7,12-dimethylbenz(a)anthracene (DMBA, a polycyclic aromatic hydrocarbon [PAH]).Sprague–Dawley rats were ovariectomized (OVX) at PND50 and infused with E2 or vehicle for 14 d andgavaged with 50 mg/kg DMBA or vehicle 24 h before sacrifice at PND64. Rats were fed an AIN-93G diet madewith SPI or casein as sole protein source throughout the study. Basal AhR protein levels were reduced(Pb0.05) by SPI feeding irrespective of the E2 status. However, DMBA increased (Pb0.05) AhR-inducedCYP1A1 gene expression in OVX, SPI-fed rats, but reduced (Pb0.05) CYP1A1 in OVX+E2, SPI-fed rats.Chromatin-immunoprecipitation demonstrated lower (Pb0.05) DMBA-mediated recruitment of estrogenreceptor alpha to the CYP1A1 promoter by SPI feeding in the presence of E2, suggesting an estrogen-likeaction of SPI on DMBA-mediated signaling in the absence of E2. Further, microarray analysis (Rat 230-2.0Affymetrix-GeneChip™) revealed 231 genes common to SPI+DMBA and SPI+E2+DMBA (normalized to E2)treatments. AhR-activated genes (CYP1A1, CYP1A2, and NQO1) were down-regulated by SPI+E2+DMBAcompared to SPI+DMBA. Unique interactions among SPI, DMBA and E2 altered the expression profile of 316genes, not observed by either treatment alone. Our data suggest that although E2 status does not effect soy-mediated AhR degradation, it modulates the effects of soy on many genes, including CYP1A1.

© 2008 Elsevier Inc. All rights reserved.

Introduction

Polycyclic aromatic hydrocarbons (PAHs) are ubiquitous environ-mental contaminants produced by the partial combustion of organicsubstances. These compounds bind to and activate the cytosolictranscription factor, aryl hydrocarbon receptor (AhR). Ligand bound,activated AhR translocates to nucleus, heterodimerizes with AhRnuclear translocator (ARNT), and binds to xenobiotic response ele-ments (XRE) in a variety of genes including, cytochrome P450,family 1 (CYP1) members. The AhR-induced CYP1 enzymes are ofparticular interest for their ability to metabolize PAHs into reactivemetabolites which may cause toxicity and generate DNA adducts(Ma and Lu, 2007; Tompkins and Wallace, 2007). Genomic profilingof PAHs- or dioxin-activated hepatic AhR in rodents suggests analteration of genes involved in cellular proliferation, endocrinedisruption, immunomodulation, metabolism, transport and neoplas-

stradiol; ER, estrogen receptor;ylbenz [a] anthracene; OVX,ns; PND, post natal day; SPI,

on Center, Slot 512-20B, 1212364 3161..

l rights reserved.

tic transformation (Vezina et al., 2004; Singhal et al., 2008b; Sato etal., in press; Yoon et al., 2006), suggesting its role in multipletoxicological endpoints.

Contrary to the exposure to PAHs which increases cancer risk,consumption of soy products has been associated with reduced risk ofcancers of different origins including cancers of breast, prostate, colonand liver tissues (Sharp et al., 2005; Badger et al., 2005; Wu et al.,2002; Wu et al., 2008). Previous observations from our lab and othershave shown a reduction in 7,12 dimethylbenz(a)anthracene (DMBA), aPAH, mediated DNA adduct formation (Badger et al., 2005; Upadhyayaand El Bayoumy, 1998) and CYP1A1 inducibility in rats fed with soyprotein isolate (SPI) containing diet (Ronis et al., 2001; Rowlands et al.,2001). In later reports we demonstrated that a SPI-associated factor, asyet unknown, destabilizes basal AhR followed by its degradation byubiquitin-26S proteasome machinery (Singhal et al., 2008a). Thusreduced AhR levels results in reduced DMBA-mediated CYP1A1 induc-tion and DNA adduct formation.

Soy has been reported to possess N137 phytochemicals (Fang et al.,2004) which either alone or in synergism with soy peptides orproteins render multiple mechanisms of disease protection. Many ofthe soy-mediated effects, including reduced mammary and prostatecarcinogenesis (Constantinou et al., 2001; Goetzl et al., 2007) andalleviated post-menopausal symptoms (Marini et al., 2007; Fuhrman

90 R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

et al., 2008), are attributed to estrogenic actions of isoflavones —

genistein and daidzein. They resemble to endogenous estrogens intheir molecular structural and binding affinities to estrogen receptors(ER) α and β (Setchell, 2001). Their actions, however, could beestrogenic, anti-estrogenic, or partial agonists depending upon thetissue, cell type, isoflavone concentration and other conditions, suchas age, estrogenic status, etc. (Hwang et al., 2006; Patisaul et al., 2001;Barkhem et al., 1998). Among various influencing factors, estrogenicstatus plays the most crucial role in determining the efficacy of soyand isoflavone-mediated physiological and pharmacological effects(Adlercreutz et al., 1993; Welshons et al., 2006).

Previously,we reported E2-mediatedpotentiation of AhR-mediatedsignaling (Singhal et al., 2008b). A positive cross-talk between 17β-estradiol (E2)-activated ERα and DMBA activated-AhR, in ovariec-tomized female Sprague–Dawley rats was demonstrated, suggestingthat supplementation with E2, an endogenous estrogen, followedby DMBA treatment results in hepatic transcriptional auto-upregu-lation of AhR which leads to increased Phase I enzymes includingCYP1A1 and CYP1B1. Our finding was in corroboration with epide-miological reports in smokers indicating significantly higher levelsof bulky-DNA adducts and higher susceptibility to lung canceramong females compared to males (Mollerup et al., 1999; Mollerupet al., 2006). The in vivo studies also suggested a reduced 2,3,7,8-tetracholorodibenzo-p-dioxin (TCDD)-induced hepatic DNA damagein ovariectomized rats as compared to control females, suggesting arole of estrogens in dioxin and PAH-mediated carcinogenicity(Tritscher et al., 1996). This indicates that PAH-mediated toxicityis influenced by the endogenous estrogenic status.

In contrast, feeding SPI (which has estrogenic isoflavones asso-ciated with it) down-regulated AhR-mediated signaling by reducingthe basal AhR protein levels, as described above (Singhal et al., 2007;Singhal et al., 2008a). Although the interactions between SPI – DMBA/AhR and E2 – DMBA/AhR have been established, it is unclear how theinteractions between SPI and E2 would affect DMBA/AhR signaling. Inthe present studies, we have shown that effects of DMBA on globalhepatic gene expression are modulated by SPI differently in thepresence or absence of E2. This presents a novel scenario of how theprotection offered by nutrients against diseases could be dependentupon the endogenous factors, in this case estrogenic status.

Methods

Animal care and experiment design. The experiment received priorapproval from the Institutional Animal Care and Use Committee atUAMS. Adult female Sprague–Dawley rats, postnatal day (PND) 30,were obtained from Charles River Laboratories (Wilmington, MA) andwere housed in polycarbonate cages in an environmentally controlledroom with a 12 h light–dark cycle. The rats had ad libitum access todiets made according to the AIN-93G formula (Reeves et al., 1993),except that corn oil replaced soybean oil and the protein source waseither casein (CAS) or soy protein isolate SPI (Singhal et al., 2007). PNDrats were ovariectomized and half the rats were subcutaneouslyinfused either with E2 or polyethylene glycol vehicle (Sigma, St. Louis,MO) using Alzet 2002™ mini-osmotic pumps (Alza corp., Palo Alto,CA) that were filled to release 0.5 μl/h for 14 d to produce an E2 dose of5 μg/kg/d. At PND 64, half of the rats from each group (±E2) wereorally gavaged with 50 mg/kg DMBA or sesame oil vehicle (Sigma,St. Louis, MO) and 24 h later anesthetized with Nembutal (100 mg/kg, ip.) followed by decapitation. Liver and mammary glands werecollected and kept frozen at −80 °C until analysis. All the rats wereovariectomized and the treatment groups were designated as: 1)Control (CAS Control diet, no treatment); CAS+DMBA treatment(DMBA); CAS+E2 treatment (E2); and combination of E2 and DMBA(E2+DMBA). 2) SPI: no treatment (SPI); SPI+DMBA treatment (SPI+DMBA); SPI+E2 treatment (SPI+E2); and combination of SPI, E2 andDMBA (SPI+E2+DMBA).

RNA isolation and quantitative real-time PCR. Total RNAwas isolatedfrom approximately 100 mg of liver or mammary glands using TRIreagent (Molecular Research Center, Cincinnati, OH). Procedures forRNA purification, cDNA synthesis, and primer design are previouslydescribed by Eason et al. (Eason et al., 2004). Corresponding primersequences are as follows: AhR (NM_013149) F 5′-1281-ACG CAC CAAAAG CAA CAC TAG TAG-1305 3′, R 5′ 1381-GTT GGA TCA AGG CAC TCATAA GG 1358-3′; CYP1A1 (NM_012540) F 5′- 520-TCC ATA GCC TCAGAC CCA ACA C-541-3′, R 5′-620-GCC ATC AGC TTC TGG AAC TTG-599-3′; CYP1B1 (NM_012940) F 5′-1485-TAA CCA AAA CGA GCC CTC AAAC-1507-3′, R 5′-1585-GGA GCT TCA TGG ACT CTC TGA GA -1562-3′;and GAPDH (AF106860): F 5′ 800 -TGA GGT GAC CGC ATC TTC TTG-820 3′, R 5′901-TGG TAA CCA GGC GTC CGA TA- 882 3′. mRNA levelswere normalized to that of the GAPDHmRNA to control for input RNA.

Western blotting. Total liver lysate was prepared and westernblotting was performed as described previously (Singhal et al.,2007). Rabbit anti-AhR (the gift of Dr. Richard Pollenz, University ofSouth Florida) and goat anti-GAPDH (Santa Cruz Biotechnology Inc.,Santa Cruz, CA) were used as primary antibodies, followed bycorresponding HRP conjugated secondary antibodies — goat anti-rabbit and bovine anti-goat (Santa Cruz Biotechnology Inc.). The blotswere stripped using Restore™ (Pierce Biotechnology Inc., Rockford, IL)blot stripping buffer and subsequently blocked with 5% milk in TBSTand probed for GAPDH protein as an internal control.

Chromatin immunoprecipitation (ChIP) assay. The ChIP-IT™Enzymatic kit (Active Motif, Carlsbad, CA) was used for ChIP. Theprotocol for in vitro ChIP analysis was modified for tissue samples.Briefly, ∼200 mg liver tissue (200 mg) was finely minced with arazor blade. The protein cross-linking was performed by incubatingthe minced tissue with 20 ml of 37% formalin (Sigma) solution for10 min on a rocking shaker at room temperature. The cross-linkingreaction was stopped by adding 5 ml of glycine-stop fix solution[3 ml of 10× glycine buffer; 3 ml 10× PBS (provided with the kit) and24 ml water] followed by homogenization using a douncehomogenizer in 2 ml of ice cold PBS supplemented with proteaseinhibitor and 100 mM PMSF and the homogenate was passedthrough 100 μm cell strainer (BD Falcon™, Bedford, MA) followed bycentrifugation for 10min at 2500 rpm at 4 °C; and processed further asmentioned in the kit protocol. The chromatin obtained by enzymaticshearingof the nucleiwas ‘pre-cleared’with proteinG agarose beads toreduce non-specific background followed by incubation with 2 μg ofantibodies (Santa Cruz Inc.): a) anti-AhR goat polyclonal antibody (sc-8088) or b) anti-ERα rabbit polyclonal (sc-542) on a rotator at 4 °Covernight. The yield of target region CYP1A1 promoter possessing XREpentanucleotide GCGTG was analyzed by quantitative real-time PCRwith the following primer (IDT Inc., Skokie, IL) sequence CYP1A1-XRE— F 5′- CGC CCT TGCAAAGCT TAAGAC - 3′; R - 5′ TCCCAGTGC TGT CACGCT AG - 3′.

Microarray preparation, normalization and data analysis. Total RNAwas isolated and cleaned, as described previously (Singhal et al.,2008b). First- and second-strand cDNA synthesis, biotin-labeledcRNA synthesis, fragmentation of cRNA and hybridization reactionswere performed using one cycle cDNA synthesis kit (Affymetrix,Santa Clara, CA). Briefly, 8 μg of purified RNA was used to synthesizecDNA. Labeled cRNA was synthesized from cDNA using a GeneChipIVT labeling kit (Affymetrix) according to the manufacturer'sinstructions. Twenty microgram cRNA was then fragmented in asolution of 5× fragmentation buffer and RNase for 35 min. Com-plimentary RNA was hybridized by filling 300 μl volume of theclarified hybridization cocktail to the Rat genome 230 2.0GeneChip™. The cRNA was hybridized for 16 h at 45 °C in thehybridization oven set at 60 rpm. The probe array was washed andstained using Affymetrix kit in GeneChip fluidics station 450 and

Fig. 1. Consumption of diet made with SPI as the sole protein source degrades basalhepatic AhR protein in female Sprague–Dawley rats which is not affected by estrogenicstatus. (A) Basal hepatic AhR protein and mRNA levels as determined by immuno-quantitation of western blots and quantitative real-time RT-PCR, respectively, in thepresence or absence of SPI diet and E2 (B) DMBA-activated hepatic AhR protein andmRNA levels as determined by immunoquantitation of western blots and quantitativereal-time RT-PCR, respectively, in the presence or absence of SPI diet and E2. All rats areovariectomized. Control, casein fed rats; SPI, Soy protein isolate fed rats; E2, 17β-estradiol infused rats; and DMBA, 7, 12 dimethylbenz (a) anthracene gavaged rats. Dataanalyzed by 2-way ANOVA and represent mean±SEM (n=6). Means, normalized byGAPDH levels, with different letter differ significantly (P≤0.05).

91R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

scanned using GeneChip Scanner 3000. For each of the 31,099 geneson the Affymetrix Rat genome 230 2.0 array, the data on induction orrepression values were analyzed using GeneChip Operating Software(GCOS) obtained from Affymetrix.

The data files (.CEL files) containing the probe level intensitieswere processed using the robust multiarray analysis algorithm(GeneSpring 7.3X, Silicon Genetics, Redwood City, CA) for backgroundcorrection (Irizarry et al., 2003). Subsequently, the data weresubjected to normalization by setting measurements b0.01 to 0.01and by per-chip and per-gene normalization using GeneSpring'snormalization algorithms. The normalized data were then subjectedto a series of pair-wise comparisons. Comparisons were madebetween various treatments: DMBA vs. control; E2+DMBA vs. control;SPI+DMBA vs. control; and SPI+E2+DMBA vs. control. The resultinggene lists generated from each pairwise comparisons included onlythe genes that had a fold change value N+1.5 or b−1.5 (i.e.±1.5),Pb0.05 and passed Benjamini and Hochberg false discovery ratemultiple testing correction. A list of differentially expressed genes wasgenerated by combining the gene lists from individual pair-wisecomparisons and subjected to hierarchical clustering. Hierarchicalclustering was performed using the “smooth correlation for distancemeasure” algorithm to identify samples and genes with similarpattern of expression. The top canonical functions, top molecularfunctions and their corresponding median enrichment p-value weredetermined by Ingenuity Analysis Pathway™ (IPA) software.

Statistical analysis. Statistical analysis was performed using SigmaStat software package (Systat Software, Inc., San Jose, CA). Data wereanalyzed by two-way analysis of variance (ANOVA) or Student's t test,followed by Student–Newman–Keuls post hoc analyses. Statisticalsignificance was set at Pb0.05. Differences between treatment groupsin microarray data were analyzed by ‘Fold changes in Volcano Plot’(GeneSpring Software) and changes were considered significant at±1.5 fold and Pb0.05 followed by Benjamini and Hochberg multipletesting correction for false discovery rate.

Results and discussion

The rationale to perform the present study comes from previousreports in which we demonstrated: a) that SPI feeding to Sprague–Dawley rats results in the degradation of the basal AhR protein in theliver of females (Singhal et al., 2007; Singhal et al., 2008a) and males(Ronis et al., 2001) and mammary glands of females (Rowlands et al.,2001), secondary to translocation of AhR to the nucleus (Singhal et al2008a); and b) that E2 supplementation to the OVX rats exacerbatesDMBA-mediated signaling (Singhal et al., 2008b). In the currentstudy, Sprague–Dawley rats were ovariectomized and subcuta-neously infused with E2 or vehicle. The infusion with E2 at thelevel utilized in the current study restores plasma E2 values to theaverage value observed physiologically in intact cycling females(Shankar et al., 2006; Chen et al., 2006). Throughout the study ratswere fed diets made either with or without SPI so that the effects ofSPI on AhR signaling could be evaluated in the context of differentlevels of endogenous estrogens. This experimental paradigm wasdesigned to specifically examine SPI/E2 interactions on DMBA-regulated gene expression. However, one limitation of the currentdesign is that ovariectomy results in removal of many more factorsthan just estrogens. Ovaries make a number of steroid and peptidehormones and thus the data obtained may not be completelyreflective of SPI-interactions with the endogenous endocrine milieuin intact animals. The gene expression data presented here clearlydemonstrate that the estrogenic status modulates the effect of SPI onDMBA-mediated signaling. The mechanism may involve differentinteractions between E2 and phytoestrogen components of SPI.CYP1A1, the most inducible gene by DMBA was studied to investigatemechanism of interactions. The data suggest that a component(s) of

SPI behaves like a selective estrogen receptor modulator (SERM),exhibiting estrogen like responses in the absence of estrogens andanti-estrogenic responses in the presence of estrogens, on CYP1A1induction.

SPI-mediated AhR degradation is not effected by estrogenic status

No effects of SPI or E2 were observed on the basal hepatic AhRmRNA levels. E2 treatment also did not have effects on the AhRprotein levels. However, SPI feeding reduced (Pb0.05) basal hepaticAhR protein levels both in the presence and absence of E2 by 40–60%, Fig. 1A. No interactions between SPI and E2 at the level of AhRwere observed, as determined by 2-way ANOVA. This suggests thatSPI-mediated AhR protein reduction is not affected by estrogenicstatus. These data extend our previous observations demonstratingAhR protein levels to be reduced as a consequence of SPI feeding inboth intact female (Rowlands et al., 2001; Singhal et al., 2007;Singhal et al., 2008a) and male rats (Ronis et al., 2001) as the resultof enhanced proteolytic degradation. We have previously publisheddata showing that there is no effect of DMBA treatment alone onAhR mRNA or protein expression compared to control or othergroups (Singhal et al., 2008b).

Next we evaluated the effect of the interactions between SPI andE2 with DMBA-activated AhR and the mechanism involved in it. Aninteraction between E2 and DMBA (2-way ANOVA) was observedresulting in an increased (Pb0.05) AhR mRNA levels, which conse-quently increased AhR protein levels. Contrary to this, no interaction

Fig. 2. DMBA dependent recruitment of E2 or SPI (in the absence of E2) -activated ERαenhances CYP1A1 gene transcription; while SPI (in the presence of E2) interferes withthe recruitment of E2-activated ERα in CYP1A1 regulatory regions. (A) ChIP assayswere performed with chromatin prepared from rat liver subjected under treatmentsusing anti-AhR or anti-ERα antibodies, as described in Methods section and theresulting DNA was amplified by quantitative real-time PCR using specific primers tothe XRE of CYP1A1. Results shown are mean±SEM (n=3). The recruitment mean valueswere normalized to that of Input DNA and represented as percent DMBA treatment. (B)Real-time PCR amplification of CYP1A1 mRNA, normalized to GAPDH mRNA. Datarepresent mean±SEM (n=6). Means with different letter differ significantly (P≤0.05).

92 R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

of SPI with DMBA was observed in the presence or absence of E2 onAhR mRNA level. However, AhR protein levels were reduced(Pb0.05) by SPI feeding in the presence of DMBA when no E2 waspresent. No affect of SPI on AhR protein was observed after co-treatment with E2 and DMBA, Fig. 1B. Therefore, it appears that thephytoestrogenic components in SPI are not enough to mimic E2-mediated increase in AhR levels. The SPI-mediated AhR degradationand E2-mediated enhancement of AhR transcription in the presenceof DMBA appeared to be balanced, resulting in no change in AhRprotein levels when SPI is fed and both DMBA and E2 are present. Noeffect of either treatment on ERα protein level was observed (datanot shown).

SPI interferes with the recruitment of E2-activated ERα in CYP1A1regulatory regions

Further, we evaluated if the interactions among SPI, E2 and DMBA,modulate AhR-mediated downstream signaling. Binding of transcrip-tion factors, AhR and ERα, to the regulatory regions in the CYP1A1gene was determined by ChIP assay. Only DMBA-treated groups wereselected because no effects of either SPI feeding or E2 treatment orcontrol were observed on the CYP1A1 induction in the absence ofDMBA (data not shown). A DMBA responsive recruitment of AhRprotein was observed at the CYP1A1 regulatory regions between−1104 and −1227 bp upstream of transcription start site, which wasenhanced (Pb0.05) by E2 supplementation. The reason could beincreased AhR transcription and translation following combinedtreatment of DMBA and E2 (Singhal et al., 2008b). However, SPIfeeding reduced the DMBA-mediated AhR recruitment at this regionconsistent with the observation that SPI feeding degrades AhR proteinand interferes with the availability of AhR to act as a transcriptionfactor on CYP1A1 gene, Fig. 2A. Co-treatment of E2 and DMBA resultedin the recruitment of ERα in the same regulatory site as AhR.Interestingly, ERα recruitmentwas also observed in the DMBA-treatedrats fed with SPI in the absence of E2, suggesting an estrogen-likeaction of SPI in the absence of E2. However, in the presence of E2 areduced (Pb0.05) SPI +DMBA-mediated ERα recruitment wasobserved, Fig. 2A. The recruitment of AhR or ERα to CYP1A1 wasanalyzed by DNA amplification, obtained from ChIP assay, usingquantitative real-time PCR and represented as normalized values toinput DNA and percent of DMBA treatment.

The E2 treatment significantly (Pb0.05) increased DMBA-inducedCYP1A1 mRNA expression levels compared with DMBA treatmentalone, as determined by real-time RT-PCR. Similar to the effect of E2on DMBA-mediated CYP1A1 induction, SPI feeding in the absence ofE2 also increased DMBA-induced CYP1A1 induction compared withDMBA alone. However, SPI feeding in the presence of E2 reducedDMBA-mediated CYP1A1 induction compared with E2 alone or SPI-feeding in the absence of E2, Fig. 2B. This could be because ofreduced SPI-mediated recruitment of ERα at the CYP1A1 regulatoryregion in the presence of E2 compared to that in the absence of E2.The exact mechanism underlying these E2-dependent actions of SPI-feeding on CYP1A1 induction is not clear. Ligand-activated, AhR-mediated CYP1A1 induction is orchestrated by dynamic changes in anumber of transcription factors, co-activators and co-repressorsincluding, but not limited to, RNA polymerase-II, NCoA2, RIP140,SMRT, HAT and p160 (Matthews et al., 2005; Harper et al., 2006)which can be affected by any of the SPI-associated components in thepresence of E2. Additionally, as described below, an interactionbetween SPI+DMBA and E2 results in change in the expressionprofile of many genes, some of which are related to transcriptioncontrol. These data suggest that effect of SPI-feeding, in the absenceof endogenous estrogens, on hepatic CYP1A1 is similar to E2treatment while being anti-estrogenic when endogenous estrogensare present. These data also suggest that SPI has dual action on AhRsignaling: 1) a non-estrogenic component of soy degrades AhR and

thereby interferes in the function of AhR to act as a transcriptionfactor; and 2) an estrogenic component of soy recruits ERα to theCYP1A1 promoter region resulting in an increase in DMBA-mediatedCYP1A1 induction.

Modulation of DMBA response on the global hepatic gene expression bySPI is dependent upon endogenous estrogens

Gene lists from comparison of individual treatments againstcontrol were combined as described in Methods section. Thiscombined gene list included 2367 genes. Correlation-based unsu-pervised hierarchical clustering analysis was performed based onthe treatment and gene expression type. The E2 infused groups —

E2+DMBA and SPI+E2+DMBA clustered together; while the non E2groups i.e. DMBA and SPI+DMBA clustered together, suggestinggreater common effect of E2 treatment on DMBA-modulated hepa-tic genes irrespective of the SPI feeding. The heat map, Fig. 3A, wasresolved into 7 sub-clusters: I — DMBA-induced genes repressed bythe interaction between E2 and DMBA; II— DMBA-induced genesrepressed by E2; III— SPI-induced genes repressed by E2; IV—SPIrepressed genes induced by the interaction between SPI and E2; V—DMBA repressed genes induced by E2; VI— DMBA-repressed genesinduced by SPI; and VII— DMBA repressed genes induced by theinteraction between SPI and E2. This demonstrates that there arevarious synergistic and antagonistic interactions among DMBA, E2,and SPI and the effect of SPI on DMBA-modulated genes is

Fig. 3. Estrogenic status modulates effect of SPI on DMBA-mediated global gene expression. (A) Hierarchical cluster analysis of DMBA altered (±1.5, P≤0.05) hepatic genes in thepresence or absence of E2 and SPI. The heat map was divided into 7 clusters: I— DMBA-induced genes repressed by the interaction between E2 and DMBA; II— DMBA-induced genesrepressed by E2; III— SPI-induced genes repressed by E2; IV—SPI repressed genes induced by the interaction between SPI and E2; V— DMBA repressed genes induced by E2; VI—DMBA-repressed genes induced by SPI; and VII— DMBA repressed genes induced by the interaction between SPI and E2. (B) Venn diagram on the SPI+DMBA altered genes(normalized with control altered genes) versus SPI+E2+DMBA altered genes (normalized with control and E2 altered genes). (C) Hierarchical cluster analysis of the SPI+DMBAtreated genes altered by E2 supplementation by≥±1.5 fold compared with SPI+E2+DMBA treatment.

93R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

dependent upon the presence or absence of E2. Further, we com-pared the pattern of global hepatic gene expression profilesmodulated by the SPI+E2+DMBA treatment (2619 genes) withSPI+DMBA treatment (1398 genes). Genes altered by SPI+E2+DMBAwere first compared with E2 to eliminate the genes changed by E2treatment itself; the resultant list of 547 genes was compared withSPI+DMBA treatment modulated 1398 genes. The Venn diagramgenerated by GeneSpring™ software, Fig. 3B, displays that out of1398 genes modulated by SPI+DMBA treatment, expression profileof 231 genes was modulated by E2 supplementation, Interestingly,expression profile of 316 genes was modulated only by theinteraction between SPI+DMBA and E2, not by either treatmentalone.

Among the 231 genes common to SPI+DMBA and SPI+E2+DMBAtreatments, only 30 genes were modified by more than±1.5-fold bySPI+E2+DMBA (normalized to E2) compared to SPI+DMBA (normal-ized to control), as shown in the hierarchical clustering in Fig. 3C.Affymetrix online data resource — NetAffx (http://www.affymetrix.com/analysis/index.affx) and GeneSpring™ software were used toretrieve the biological functions of these genes, which revealed thatthe genes down-regulated by SPI+DMBA treatment after E2 supple-mentation were mainly involved in xenobiotic metabolism. TheCYP1A1 induction was reduced by 6-fold by SPI+E2+DMBA comparedto SPI+DMBA treatment, which is in consonant to the QRTPCR data onCYP1A1 as described in Fig. 2C. Besides CYP1A1, Other genes DMBA-inducible genes belonging to AhR battery (Nebert et al., 2000), such as

Table 1AFunctional characterization and fold changes of SPI+DMBA altered genes modulated by E2 supplementation

Ref_seq Genesymbol

Gene title GO biological process description SPI+DMBAfold change

SPI+E2+DMBAfold change

NM_012540 Cyp1a1 Cytochrome P450, family 1, subfamily a, polypeptide 1 Xenobiotic metabolic process 277 48.03NM_012541 Cyp1a2 Cytochrome P450, family 1, subfamily a, polypeptide 2 Xenobiotic metabolic process 7.71 3.75NM_012942 Cyp7a1 Cytochrome P450, family 7, subfamily a, polypeptide 1 Cholesterol metabolism 6.45 3.14NM_001025623 Vnn1 Vanin 1 Cell motility 4.94 1.91XM_342340 Pla2g12a Phospholipase A2, group XIIA Lipid metabolic process 4.34 2.08NM_013134 Hmgcr 3-hydroxy-3-methylglutaryl-Coenzyme A reductase Lipid metabolic process 3.3 2.14NM_012923 Ccng1 Cyclin G1 Cell cycle 3 4.92XM_214979 Abhd2 Abhydrolase domain containing 2 Response to wounding 2.83 1.57NM_022692 Rab5a RAB5A, member RAS oncogene family Signal transduction 2.44 1.63NM_017235 Hsd17b7 hydroxysteroid (17-beta) dehydrogenase 7 Lipid metabolic process 1.55 2.38XM_233467 Ctpsd Cytidine 5′-triphosphate synthase Response to drug −1.53 1.58NM_053605 Smpd3 Phosphodiesterase 3, neutral Response to stress −1.77 1.57NM_017274 Gpam Glycerol-3-phosphate acyltransferase, mitochondrial Lipid metabolic process −1.79 1.68NM_001034028 Hyou1 Hypoxia up-regulated 1 Response to stress −1.96 1.82NM_053933 Pcdha1 Protocadherin alpha 1 Cell adhesion −2.16 1.65XM_342220 Ect2 ect2 oncogene Positive regulation of IkappaB

kinase/NfkappaB cascade−3.125 −1.73

NM_012703 Thrsp Thyroid hormone responsive protein Lipid metabolic process −3.98 −1.67NM_053698 Cited2 Cbp/p300-interacting transactivator, with Glu/Asp-rich

carboxyterminal domain, 2Signal transduction −4.09 −1.54

NM_019292 Ca3 Carbonic anhydrase 3 Response to oxidative stress −5.23 −8.47NM_001025740 Rrm2 Ribonucleotide reductase M2 DNA replication −6.8 −1.53NM_032074 Irs3 Insulin receptor substrate 3 Signal transduction −8.19 −2.28NM_017332 Fasn Fatty acid synthase Lipid metabolic process −9.9 −3.14

GO, Gene Ontology.

Table 1BTop biological functionsa affected by SPI+DMBA and altered by E2 supplementation

Top canonical pathways p-valueLXR/RXR activation 3.83E−05TR/RXR activation 7.85E−05Aryl hydrocarbon receptor signaling 3.33E−04Sphingolipid metabolism 1.08E−03Xenobiotic metabolism 1.34E−03

Top molecular and cellular functions p-valueLipid metabolism 5.01E−08–2.71E−02Small molecule biochemistry 5.01E−08–2.71E−02Drug metabolism 2.68E−07–1.02E−02Vitamin and mineral metabolism 1.83E−05–9.48E−05Nucleic acid metabolism 2.56E−05–1.59E−02a Top biological functions and enrichment values were determined by Ingenuity

Pathway Analysis tool.

94 R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

CYP1A2; and phase II enzymes — NQO1, GST and UGT1A6 were alsodown-regulated by the interaction between SPI and E2. Importantly,our lab and others have previously displayed that these DMBA-inducible phase II genes are up-regulated by the individual treatmentsof E2 and SPI (Singhal et al., 2008b; Appelt and Reicks, 1997), however,the present data suggest a common mechanism of repression of thesegenes by the combined presence of SPI and E2 compared to theirindividual treatments. The SPI+DMBA induced genes involved in ox-idative stress— carbonic anhydrase 3 (CAR3, 1.6-fold; (Yamamoto et al.,2006) and abhydrolase domain containing 2 (ABHD2, −1.8-fold;(Miyata et al., 2008) were also down-regulated by the E2 supple-mentation. A more detailed classification of selected genes withknown biological functions is represented in Table 1A. Further, IPArevealed top altered canonical and molecular functions. The biologicalfunctions affected by SPI+DMBA treatment— aryl hydrocarbon recep-tor signaling, xenobiotic metabolism, lipid metabolism and drugmetabolism, were among the top functions which were altered by E2supplementation, Table 1B.

Interaction between SPI+DMBA and E2 treatments resulted inthe alteration in 316 genes, expression of which was not altered bySPI+DMBA or E2 treatments alone. Estrogens are known to reduce theexpression of CYP3A18 gene as suggested by its higher expression inmales and in pre-pubertal female rats (Mahnke et al., 1997); theinteraction among SPI, E2 and DMBA, however, resulted in 2-foldinduction of this gene. Another interesting gene altered by the inter-action is immediate early growth response 1 (EGR1), a gene coding fortranscription factor linked with tumor suppression. Expression ofEGR1 gene has been demonstrated to be increased by E2 and phyto-estrogens, genistein, by a non-genomic mechanism (Chen et al., 2004;Singletary and Ellington, 2006). Also, Chen et al. demonstrated an anti-estrogenic action of AhR agonists on EGR1, resulting in its repression(Chen et al., 2001). Here we observed a 2-fold reduction in EGR1expression by the interaction among SPI, DMBA and E2 treatments,while no affect of DMBA, an AhR agonist, or endogenous or exogenousestrogens was observed. Cumulatively, this suggests that interactionsamong these treatments could result completely different responsesrelative to individual treatments. A more detailed classification ofselected genes with known biological functions is represented in

Table 2A. IPA revealed that genes involved in biological functions suchas glutathione metabolism, xenobiotic metabolism by CYPs, and theNRF2-mediated oxidative stress response are significantly enrichedby the interaction of SPI, E2 and DMBA, Table 2B. There is limitedevidence suggesting estrogens repress (Ansell et al., 2005) and soy-associated phytoestrogens activate Nrf2-mediated signaling (Barve etal., 2008) while little is known about their interactive effects on thesame. Other dietary phytochemicals such as indole-3-carbinol(Aggarwal and Ichikawa, 2005) and glucosinolates (Hayes et al.,2008) have also been reported to activate Nrf2 signaling. Interest-ingly, induction of Nrf2 signaling is associated with the oxidativeresponse and coupled to AhR signaling. Targets of AhR and Nrf2 canbe activated independently by AhR agonists and oxidative stress,respectively or by coordinated regulation of AhR/Nrf2. Coupling ofAhR and Nrf2 gene batteries conceivably attenuates accumulation ofCYP-generated ROS and free radicals (Kohle and Bock, 2006; Kohleand Bock, 2007). For example, expression of GstA2 by AhR agonistsappears to be mediated by an antioxidant response element (ARE)mechanism involving Nrf2 recruitment (Kohle and Bock, 2007).Stabilization of Nrf2 protein by oxidative stress, generated by AhR

Table 2AFunctional characterization and fold change of genes affected only in the presence of SPI, E2 and DMBA

Ref_Seq Gene symbol Gene title GO biological process description Foldchange

NM_145084 Retsat All-trans-13,14-dihydroretinol saturase Electron transport 3.531NM_031565 Ces1 Carboxylesterase 1 Xenobiotic metabolic process 3.012NM_030868 Nov Nephroblastoma overexpressed gene Growth regulation 2.881NM_022231 Birc4 Baculoviral IAP repeat-containing 4 Apoptosis 2.627NM_053933 Pcdha1 Protocadherin alpha 4 Cell adhesion 2.319NM_145782 Cyp3a18 Cytochrome P450, 3a18 Electron transport 2.259NM_032614 Txnl2 Thioredoxin-like 2 Electron transport 2.115NM_017013 Gsta2 glutathione-S-transferase, alpha type2 Xenobiotic metabolism 2.082NM_024127 Gadd45a Growth arrest and DNA-damage-inducible 45 alpha Cell cycle 2.07NM_001034028 Hyou1 Hypoxia up-regulated 1 Response to stress 2.067NM_013084 Acadsb Acyl-Coenzyme A dehydrogenase, short/branched chain Electron transport 2.063NM_001009685 Drg1 Developmentally regulated GTP binding protein 1 Transcription 2.015NM_001014125 Pdia5 Protein disulfide isomerase-associated 5 Electron transport 1.954NM_017014 Gstm1 Glutathione S-transferase, mu 1 Glutathione metabolic process 1.939NM_171990 Bpnt1 3′(2′),5′-bisphosphate nucleotidase Transcription 1.915NM_031650 Slco1b2 Solute carrier family 21, member 10 Transport 1.868NM_001009683 Tor3a Torsin family 3, member A Amino acid metabolic process 1.85NM_031660 Arpp19 cAMP-regulated phosphoprotein 19 Signal Transduction 1.828NM_022866 Slc13a3 Solute carrier family 13, member 3 Transport 1.801NM_012796 Gstt2 Glutathione S-transferase, theta 2 Glutathione metabolic process 1.799NM_019238 Fdft1 Farnesyl diphosphate farnesyl transferase 1 Cholesterol metabolism 1.789NM_175761 Hspca Heat shock protein 1, alpha Protein folding 1.785NM_019216 Gdf15 Growth differentiation factor 15 Signal transduction 1.772NM_138840 Tgoln2 Trans-Golgi network protein 1 Golgi to endosome transport 1.767NM_053977 Cdh17 Cadherin 17 Transport 1.747NM_001013105 Fkbp11 FK506 binding protein 11 Protein folding 1.722NM_139102 Dmgdh Dimethylglycine dehydrogenase precursor Electron transport 1.721NM_031147 Cirbp Cold inducible RNA binding protein Response to stress 1.72NM_021653 Dio1 Deiodinase, iodothyronine, type I Thyroid hormone generation 1.703NM_001011901 Hsph1 Heat shock protein 105 Response to stress 1.659NM_017298 Cacna1d Calcium channel, voltage-dependent, L type, alpha 1D subunit Transport 1.627NM_147210 Nr1d2 Nuclear receptor subfamily 1, group D, member 2 Transcription 1.623NM_031722 Tmed2 Coated vesicle membrane protein Transport 1.609NM_017022 Itgb1 Integrin beta 1 Cell cycle 1.606NM_054001 Scarb2 CD36 antigen Cell adhesion 1.584NM_053886 Lman1 Lectin, mannose-binding, 1 Protein folding 1.576NM_022300 Basp1 Brain abundant, membrane attached signal protein 1 Transcription 1.576NM_017359 Rab10 RAB10, member RAS oncogene family Transcription 1.574NM_001015021 Dnajb11 DnaJ (Hsp40) homolog, subfamily B, member 11 Protein folding 1.573NM_001012011 Lig3 Ligase III, DNA, ATP-dependent DNA replication 1.572NM_001008335 Eif4a2 Eukaryotic translation initiation factor 4A2 Translation 1.57NM_001005905 Cct2 TCP1, subunit 2 (beta) Protein folding 1.564NM_057100 Gas6 Growth arrest specific 6 Growth regulation 1.562NM_178091 Insig2 Insulin induced gene 2 Lipid metabolism 1.542NM_001013089 Galt Galactose-1-phosphate uridyl transferase Carbohydrate metabolism 1.523NM_053800 Txn1 Thioredoxin 1 Electron transport 1.516NM_022704 Mbl2 Mannose binding lectin 2, protein C Immune response 1.511NM_012543 Dbp D site albumin promoter binding protein Transcription 1.511NM_017284 Psmb2 Proteasome (prosome, macropain) subunit, beta type 2 Ubiquitin-dependent protein

catabolic process1.51

NM_031120 Ssr3 Signal sequence receptor, gamma Signal transduction 1.501NM_017255 P2ry2 Purinergic receptor P2Y, G-protein coupled 2 Signal transduction −1.5NM_053607 Acsl5 Acyl-CoA synthetase long-chain family member 5 Lipid metabolism −1.51NM_134402 Bzw2 Basic leucine zipper and W2 domains 2 Transcription −1.52NM_057210 Sv2a Synaptic vesicle glycoprotein 2a Transport −1.52NM_012596 Lepr Leptin receptor Cholesterol metabolic process −1.52NM_053398 Gfra3 Glial cell line derived neurotrophic factor family receptor alpha 3 Signal transduction −1.54NM_012571 Got1 Glutamate oxaloacetate transaminase 1 Amino acid metabolic process −1.62NM_022534 Tcn2 Transcobalamin 2 Transport −1.69NM_031741 Slc2a5 Solute carrier family 2, member 5 Carbohydrate metabolism −1.74NM_012736 Gpd2 Glycerol-3-phosphate dehydrogenase 2 Carbohydrate metabolism −1.79NM_012551 Egr1 Early growth response 1 Transcription −1.88NM_001009663 Serpina6 Serine (or cysteine) proteinase inhibitor Translation −1.94NM_001012345 Dgat2 Diacylglycerol O-acyltransferase 2 Lipid Metabolism −2.01NM_001013137 Cxcl14 Chemokine (C-X-C motif) ligand 14 Immune response −2.13NM_017006 G6pdx Glucose-6-phosphate dehydrogenase Carbohydrate metabolism −2.82NM_053962 Sds Serine dehydratase Gluconeogenesis −2.92NM_031641 Sult4a1 Sulfotransferase family 4A, member 1 Lipid metabolism −4.57

GO, Gene Ontology.

95R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

agonists such as DMBA, appears to be critical for an appropriate Nrf2responsive gene battery including GstA2 and thioredoxin. Hence, anenrichment of glutathione and cysteine metabolism, xenobiotic

metabolism by CYPs other than CYP1s, such as CYP3A18, and Nrf2-mediated signaling suggest maintenance of redox homeostasis by thecombined action of SPI, E2 and DMBA.

Table 3Correlation between the expression level of each gene as measured by the Affymetrixchip and RT-PCRa

Gene name Gene Id Correlation

CYP1A1 Cytochrome P450, family 1, subfamily a, polypeptide 1 0.814GSTM3 Glutathione S-transferase, mu 1 0.846FASN fatty acid synthase 0.759GADD45A Growth arrest and DNA-damage-inducible 45 alpha 0.726EGR1 Early growth response 1 0.583RAB5A RAB5A, member RAS oncogene family 0.623AhR Ayl hydrocarbon receptor 0.864CYP1A2 Cytochrome P450, family 1, subfamily a, polypeptide 2 0.986CYP3A18 Cytochrome P450, 3a18 0.916HYUO1 Hypoxia up-regulated 1 0.829SULT4A1 Sulfotransferase family 4A, member 1 0.901IRS3 Insulin receptor substrate 3 0.961HSD17B7 Hydroxysteroid (17-beta) dehydrogenase 7 0.676DBP D Site albumin promoter binding protein 0.782INSIG2 Insulin induced gene 2 0.691SERPINA6 Serine (or cysteine) proteinase inhibitor 0.729HMGCR 3-hydroxy-3-methylglutaryl-Coenzyme A reductase 0.667LEPR Leptin receptor 0.873GAS6 growth arrest specific 6 0.796NR1D2 Nuclear receptor subfamily 1, group D, member 2 0.893CA3 Carbonic anhydrase 3 0.881

a Each (8) treatment group had n=3 for correlation analysis.

96 R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

Verification of microarray responseQuantitative real-timeRT-PCRwas used tovalidate the fold changes

in the transcript levels, observed by microarrays. The pattern of ex-pression of 20 genes, randomly selected from Tables 1A, B or 2A, Bdisplayed similar pattern of treatment mediated changes both inmicroarrays and real-time PCR (supplementary online data). Table 3displays the correlation between microarray and RT-PCR for all 20genes performed on replicate samples, n=3.

Conclusion

This study provides an extensive analysis of the influence of thepresence of estrogens on the effects of SPI-feeding on PAH-mediatedactivation of AhR signaling, particularly CYP1A1 induction. These datamay be important in understanding the effects of soy consumption onPAH toxicities in women with or without normal circulating levels ofendogenous estrogens. First, we demonstrated that presence orabsence of E2 does not have any affect on SPI-mediated AhRdegradation, suggesting that some non-estrogenic component of soymight be responsible for this. Secondly, SPI appears to have a dualaction on AhR-mediated signaling. Despite SPI-mediated AhR degra-dation, estrogenic components of SPI recruited ERα to the CYP1A1promoter coinciding with increased transcription of CYP1A1 gene.Reduction of ERα recruitment to the CYP1A1 promoter and suppressedCYP1A1 induction in the presence of E2, suggest a SERM-like action ofSPI on this gene. Thirdly, SPI signature on global hepatic geneexpression in response to DMBA treatment is dependent upon theestrogenic status. The current ChIP data represent a snapshot oftranscription factor binding at a single time point on a single geneafter DMBA treatment. Further detailed in vitro studies are required toassess the dynamics of these responses in the context of expression ofCYP1A1 and other AhR regulated genes. It is clear that AhR responsivesignaling is attenuated by SPI in the presence of E2. However, theeffects of soy foods are gene-dependent and thus cannot be general-ized. Estrogenic status, as in pre- and post-menopausal women, menand infants, appears to have an important influence on the gene-expression profile (signature) induced by SPI, and this suggests thatconsumption of soy foods may alter (reduce) the cancer risk ofenvironmental pro-carcinogens, such as PAHs.

Conflict of interest statement

Authors have no conflict of interest.

Acknowledgments

The financial assistance of the USDA (CRIS# 6251-51000-005-03S)is gratefully acknowledged. We are thankful to Dr. John Marecki forcarefully reviewing the manuscript.

Table 2BTop biological functionsa affected only in the presence of SPI, E2 and DMBA

Top canonical pathways p-valueGlutathione metabolism 7.45E−05LPS/IL-1 mediated inhibition of RXR function 7.87E−05Metabolism of xenobiotics by CYP450 4.29E−04Cysteine metabolism 4.49E−04NRF2-mediated oxidative stress response 5.72E−04

Top molecular and cellular functions p-valueCell death 2.01E−05–1.91E−02Energy production 7.45E−05–7.45E−05Molecular transport 7.45E−05–2.06E−02Carbohydrate metabolism 9.93E−05–1.76E−02Small molecule biochemistry 9.93E−05–2.06E−02a Top biological functions and enrichment values were determined by Ingenuity

Pathway Analysis tool.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.taap.2008.09.027.

References

Adlercreutz, H., Bannwart, C., Wahala, K., Makela, T., Brunow, G., Hase, T., Arosemena,P.J., Kellis Jr., J.T., Vickery, L.E., 1993. Inhibition of human aromatase by mammalianlignans and isoflavonoid phytoestrogens. J. Steroid Biochem. Mol. Biol. 44, 147–153.

Aggarwal, B.B., Ichikawa, H., 2005. Molecular targets and anticancer potential of indole-3-carbinol and its derivatives. Cell Cycle 4, 1201–1215.

Ansell, P.J., Lo, S.C., Newton, L.G., Espinosa-Nicholas, C., Zhang, D.D., Liu, J.H., Hannink,M., Lubahn, D.B., 2005. Repression of cancer protective genes by 17β-estradiol:ligand-dependent interaction between human Nrf2 and estrogen receptor α. Mol.Cell. Endocrinol. 243, 27–34.

Appelt, L.C., Reicks, M.M.,1997. Soy feeding induces phase II enzymes in rat tissues. Nutr.Cancer 28, 270–275.

Badger, T.M., Ronis, M.J., Simmen, R.C., Simmen, F.A., 2005. Soy protein isolate andprotection against cancer. J. Am. Coll. Nutr. 24, 146S–149S.

Barkhem, T., Carlsson, B., Nilsson, Y., Enmark, E., Gustafsson, J., Nilsson, S., 1998.Differential response of estrogen receptor alpha and estrogen receptor beta topartial estrogen agonists/antagonists. Mol. Pharmacol. 54, 105–112.

Barve, A., Khor, T., Nair, S., Lin, W., Yu, S., Jain, M.R., Chan, J.Y., Kong, A., 2008.Pharmacogenomic profile of soy isoflavone concentrate in the prostate of Nrf2deficient and wild-type mice. J. Pharmaceutical Sci. 97, 4528–4545.

Chen, J.R., Haley, R.L., Hidestrand, M., Shankar, K., Liu, X., Lumpkin, C.K., Simpson, P.M.,Badger, T.M., Ronis, M.J.J., 2006. Estradiol protects against ethanol-induced boneloss by inhibiting up-regulation of receptor activator of nuclear factor-kB ligand inosteoblasts. J. Pharmacol. Exp. Ther. 319, 1182–1190.

Chen, C.C., Lee, W.R., Safe, S., 2004. Egr-1 is activated by 17beta-estradiol in MCF-7 cellsby mitogen-activated protein kinase-dependent phosphorylation of ELK-1. J. CellBiochem. 93, 1063–1074.

Chen, I., Hsieh, T., Thomas, T., Safe, S., 2001. Identification of estrogen-induced genesdownregulated by AhR agonists in MCF-7 breast cancer cells using suppressionsubtractive hybridization. Gene 262, 207–214.

Constantinou, A.I., Lantvit, D., Hawthorne, M., Xu, X., van Breemen, R.B., Pezzuto, J.M.,2001. Chemopreventive effects of soy protein and purified soy isoflavones onDMBA-induced mammary tumors in female Sprague–Dawley rats. Nutr. Cancer 41,75–81.

Eason, R.R., Velarde, M.C., Chatman Jr., L., Till, S.R., Geng, Y., Ferguson, M., Badger, T.M.,Simmen, R.C., 2004. Dietary exposure to whey proteins alters rat mammary glandproliferation, apoptosis, and gene expression during postnatal development. J. Nutr.134, 3370–3377.

Fang, N., Yu, S., Badger, T.M., 2004. Comprehensive phytochemical profile of soy proteinisolate. J. Agric. Food Chem. 52, 4012–4020.

Fuhrman, B.J., Teter, B.E., Barba, M., Byrne, C., Cavalleri, A., Grant, B.J., Horvath, P.J.,Morelli, D., Venturelli, E., Muti, P.C., 2008. Equol status modifies the association ofsoy intake and mammographic density in a sample of postmenopausal women.Cancer Epidemiol. Biomarkers Prev. 17, 33–42.

97R. Singhal et al. / Toxicology and Applied Pharmacology 234 (2009) 89–97

Goetzl, M.A., Van Veldhuizen, P.J., Thrasher, J.B., 2007. Effects of soy phytoestrogens onthe prostate. Prostate Cancer Prostatic. Dis. 10, 216–223.

Harper, P.A., Riddick, D.S., Okey, A.B., 2006. Regulating the regulator: factors that controllevels and activity of the aryl hydrocarbon receptor. Biochem. Pharmacol. 72,267–279.

Hayes, J.D., Kelleher, M.O., Eggleston, I.M., 2008. The cancer chemopreventive actions ofphytochemicals derived from glucosinolates. Eur. J. Nutr. 47, 73–88.

Hwang, C.S., Kwak, H.S., Lim, H.J., Lee, S.H., Kang, Y.S., Choe, T.B., Hur, H.G., Han,K.O., 2006. Isoflavone metabolites and their in vitro dual functions: they can actas an estrogenic agonist or antagonist depending on the estrogen concentration.J. Steroid Biochem. Mol. Biol. 101, 246–253.

Irizarry, R.A., Hobbs, B., Collin, F., Beazer-Barclay, Y.D., Antonellis, K.J., Scherf, U., Speed,T.P., 2003. Exploration, normalization, and summaries of high density oligonu-cleotide array probe level data. Biostatistics 4, 249–264.

Kohle, C., Bock, K.W., 2006. Activation of coupled Ah receptor and Nrf2 gene batteries bydietary phytochemicals in relation to chemoprevention. Biochem. Pharmacol. 72,795–805.

Kohle, C., Bock, K.W., 2007. Coordinate regulation of Phase I and II xenobioticmetabolisms by the Ah receptor and Nrf2. Biochem. Pharmacol. 73, 1835–1862.

Ma, Q., Lu, A.Y., 2007. CYP1A induction and human risk assessment: an evolving tale ofin vitro and in vivo studies. Drug Metab. Dispos. 35, 1009–1016.

Mahnke, A., Strotkamp, D., Roos, P.H., Hanstein, W.G., Chabot, G.G., Nef, P., 1997.Expression and inducibility of cytochrome P450 3A9 (CYP3A9) and othermembers of the CYP3A subfamily in rat liver. Arch. Biochem. Biophys. 337,62–68.

Marini, H., Minutoli, L., Polito, F., Bitto, A., Altavilla, D., Atteritano, M., Gaudio, A.,Mazzaferro, S., Frisina, A., Frisina, N., Lubrano, C., Bonaiuto, M., D'Anna, R., Cannata,M.L., Corrado, F., Adamo, E.B., Wilson, S., Squadrito, F., 2007. Effects of thephytoestrogen genistein on bone metabolism in osteopenic postmenopausalwomen: a randomized trial. Ann. Intern. Med. 146, 839–847.

Matthews, J., Wihlen, B., Thomsen, J., Gustafsson, J.A., 2005. Aryl hydrocarbon receptor-mediated transcription: ligand-dependent recruitment of estrogen receptor alphato 2,3,7,8-tetrachlorodibenzo-p-dioxin-responsive promoters. Mol. Cell Biol. 25,5317–5328.

Miyata, K., Nakayama, M., Mizuta, S., Hokimoto, S., Sugamura, K., Oshima, S., Oike, Y.,Sugiyama, S., Ogawa, H., Yamamura, K., 2008. Elevated mature macrophageexpression of human ABHD2 gene in vulnerable plaque. Biochem. Biophys. Res.Commun. 365, 207–213.

Mollerup, S., Berge, G., Baera, R., Skaug, V., Hewer, A., Phillips, D.H., Stangeland, L.,Haugen, A., 2006. Sex differences in risk of lung cancer: expression of genes in thePAH bioactivation pathway in relation to smoking and bulky DNA adducts. Int. J.Cancer 119, 741–744.

Mollerup, S., Ryberg, D., Hewer, A., Phillips, D.H., Haugen, A., 1999. Sex differences inlung CYP1A1 expression and DNA adduct levels among lung cancer patients. CancerRes. 59, 3317–3320.

Nebert, D.W., Roe, A.L., Dieter, M.Z., Solis, W.A., Yang, Y., Dalton, T.P., 2000. Role of thearomatic hydrocarbon receptor and [Ah] gene battery in the oxidative stressresponse, cell cycle control, and apoptosis. Biochem. Pharmacol. 59, 65–85.

Patisaul, H.B., Dindo, M., Whitten, P.L., Young, L.J., 2001. Soy isoflavone supplementsantagonize reproductive behavior and estrogen receptor alpha- and beta-dependent gene expression in the brain. Endocrinology 142, 2946–2952.

Reeves, P.G., Nielsen, F.H., Fahey Jr., G.C., 1993. AIN-93 purified diets for laboratoryrodents: final report of the American Institute of Nutrition ad hoc writingcommittee on the reformulation of the AIN-76A rodent diet. J. Nutr. 123,1939–1951.

Ronis, M.J., Rowlands, J.C., Hakkak, R., Badger, T.M., 2001. Inducibility of hepatic CYP1Aenzymes by 3-methylcholanthrene and isosafrole differs in male rats fed dietscontaining casein, soy protein isolate or whey from conception to adulthood. J. Nutr.131, 1180–1188.

Rowlands, J.C., He, L., Hakkak, R., Ronis, M.J., Badger, T.M., 2001. Soy and whey proteinsdownregulate DMBA-induced liver and mammary gland CYP1 expression in femalerats. J. Nutr. 131, 3281–3287.

Sato, S., Shirakawa, H., Tomita, S., Ohsaki, Y., Haketa, K., Tooi, O., Santo, N., Tohkin, M.,Furukawa, Y., Gonzalez, F.J., Komai, M., in press. Low-dose dioxins alter geneexpression related to cholesterol biosynthesis, lipogenesis, and glucose metabolismthrough the aryl hydrocarbon receptor-mediated pathway in mouse liver. Toxicol.Appl. Pharmacol.

Setchell, K.D., 2001. Soy isoflavones—benefits and risks from nature's selective estrogenreceptor modulators (SERMs). J. Am. Coll. Nutr. 20, 354S–362S.

Shankar, K., Hidestrand, M., Haley, R.L., Skinner, R.A., Hogue, W., Jo, C.H., Simpson, P.,Lumpkin Jr., C.K., Aronson, J., Badger, T.M., Ronis, M.J.J., 2006. Different mechanismsunderlie ethanol-induced bone loss in cycling and pregnant rats. Endocrinology147, 166–178.

Sharp, G.B., Lagarde, F., Mizuno, T., Sauvaget, C., Fukuhara, T., Allen, N., Suzuki, G.,Tokuoka, S., 2005. Relationship of hepatocellular carcinoma to soya food con-sumption: a cohort-based, case-control study in Japan. Int. J. Cancer 115,290–295.

Singhal, R., Badger, T.M., Ronis, M.J., 2007. Reduction in 7,12-dimethylbenz[a]anthracene-induced hepatic cytochrome-P450 1A1 expression following soyconsumption in female rats is mediated by degradation of the aryl hydrocarbonreceptor. J. Nutr. 137, 19–24.

Singhal, R., Badger, T.M., Ronis, M.J., 2008a. Rats fed soy protein isolate (SPI) haveimpaired hepatic CYP1A1 induction by polycyclic aromatic hydrocarbons as a resultof interference with aryl hydrocarbon receptor signaling. Toxicol. Appl. Pharmacol.227, 275–283.

Singhal, R., Shankar, K., Badger, T.M., Ronis, M.J., 2008b. Estrogenic status modulates arylhydrocarbon receptor-mediated hepatic gene expression and carcinogenicity.Carcinogenesis 29, 227–236.

Singletary, K., Ellington, A., 2006. Genistein suppresses proliferation and MET oncogeneexpression and induces EGR-1 tumor suppressor expression in immortalizedhuman breast epithelial cells. Anticancer Res. 26, 1039–1048.

Tompkins, L.M., Wallace, A.D., 2007. Mechanisms of cytochrome P450 induction.J. Biochem. Mol. Toxicol. 21, 176–181.

Tritscher, A.M., Seacat, A.M., Yager, J.D., Groopman, J.D., Miller, B.D., Bell, D., Sutter, T.R.,Lucier, G.W., 1996. Increased oxidative DNA damage in livers of 2,3,7,8-tetrachlor-odibenzo-p-dioxin treated intact but not ovariectomized rats. Cancer Lett. 98,219–225.

Upadhyaya, P., El Bayoumy, K., 1998. Effect of dietary soy protein isolate, genistein, and1,4-phenylenebis(methylene)selenocyanate on DNA binding of 7,12-dimethylbenz[a]anthracene in mammary glands of CD rats. Oncol. Rep. 5, 1541–1545.

Vezina, C.M., Walker, N.J., Olson, J.R., 2004. Subchronic exposure to TCDD, PeCDF,PCB126, and PCB153: effect on hepatic gene expression. Environ. Health Perspect.112, 1636–1644.

Welshons, W.V., Nagel, S.C., vom Saal, F.S., 2006. Large effects from small exposures. III.Endocrine mechanisms mediating effects of bisphenol A at levels of humanexposure. Endocrinology 147, S56–S69.

Wu, A.H., Wan, P., Hankin, J., Tseng, C.C., Yu, M.C., Pike, M.C., 2002. Adolescent and adultsoy intake and risk of breast cancer in Asian-Americans. Carcinogenesis 23,1491–1496.

Wu, A.H., Yu, M.C., Tseng, C.C., Pike, M.C., 2008. Epidemiology of soy exposures andbreast cancer risk. Br. J. Cancer 98, 9–14.

Yamamoto, T., Kikkawa, R., Yamada, H., Horii, I., 2006. Investigation of proteomicbiomarkers in vivo hepatotoxicity study of rat liver: toxicity differentiation inhepatotoxicants. J. Toxicol. Sci. 31, 49–60.

Yoon, C.Y., Park, M., Kim, B.H., Park, J.Y., Park, M.S., Jeong, Y.K., Kwon, H., Jung, H.K.,Kang, H., Lee, Y.S., Lee, B.J., 2006. Gene expression profile by 2,3,7,8-tetrachlo-rodibenzo-p-dioxin in the liver of wild-type (AhR+/+) and aryl hydrocarbonreceptor-deficient (AhR−/−) mice. J. Vet. Med. Sci. 68, 663–668.