In Utero and Lactational Exposure to Vinclozolin and Genistein Induces Genomic Changes in the Rat Mammary Gland

H EI SHEIKH SAAD and others Genistein and vinclozolin on 216:2 245–263 Research mammary gland

In utero and lactational exposure to vinclozolin and genistein induces genomic changes in the rat mammary gland

H El Sheikh Saad, A Toullec, S Vacher1, M Pocard, I Bieche1 and M Perrot-Applanat Correspondence should be addressed INSERM U965, UFR Me´ decine, Hoˆ pital Lariboisie` re; Universite´ Paris 7, 41 Bd de la chapelle, F-75475 Paris Cedex 10, to M Perrot-Applanat France Email 1Laboratoire d’Oncoge´ ne´ tique, Institut Curie Hoˆ pital Rene´ Huguenin, St-Cloud F-92210, France [email protected]

Abstract

Exposure to low doses of environmental estrogens such as bisphenol A and genistein (G) alters Key Words mammary gland development. The effects of environmental anti-androgens, such as the " Endocrine disruptors fungicide vinclozolin (V), on mammary gland morphogenesis are unknown. We previously " Mammary gland reported that perinatal exposure to G, V, and the GV combination causes histological changes development in the mammary gland during the peripubertal period, suggesting alterations to the " Genistein peripubertal hormone response. We now investigate whether perinatal exposure to these " Anti-androgen vinclozolin compounds alters the gene expression proﬁles of the developing glands to identify the " Steroid receptors ER, AR dysregulated signaling pathways and the underlying mechanisms. G, V, or GV (1 mg/kg body

Journal of Endocrinology weight per day) was added to diet of Wistar rats, from conception to weaning; female offspring mammary glands were collected at postnatal days (PNDs) 35 and 50. Genes displaying differential expression and belonging to different functional categories were validated by quantitative PCR and immunocytochemistry. At PND35, G had little effect; the slight changes noted were in genes related to morphogenesis. The changes following exposure to V concerned the functional categories associated with development (Cldn1, Krt17,andSprr1a), carbohydrate metabolism, and steroidogenesis. The GV mixture upregulated genes (Krt17, Pvalb,andTnni2) involved in muscle development, indicating effects on myoepithelial cells during mammary gland morphogenesis. Importantly, at PND50, cycling females exposed to GV showed an increase in the expression of genes (Csn2, Wap,andElf5) related to differentiation, consistent with the previously reported abnormal lobuloalveolar development previously described. Thus, perinatal exposure to GV alters the mammary gland hormone response

differently at PND35 (puberty) and in animals with established cycles. Journal of Endocrinology (2013) 216, 245–263

Introduction

There is a growing body of evidence that endocrine- (PCBs), dioxins, and pesticides. Environmental chemicals disrupting chemicals (EDCs) alter normal mammary with estrogenic activity (xenoestrogens, such as BPA, and gland morphogenesis. EDCs include various families of the pesticide, dieldrin) and other EDCs not exhibiting synthetic compounds, such as plastics (bisphenol A, also estrogenic behavior (such as dioxin and atrazine) may called BPA), plasticizers (phthalates), polychlorobiphenyls cause abnormal mammary gland development (Markey

http://joe.endocrinology-journals.org Ñ 2013 Society for Endocrinology Published by Bioscientiﬁca Ltd. DOI: 10.1530/JOE-12-0395 Printed in Great Britain Downloaded from Bioscientifica.com at 10/03/2021 10:51:25AM via free access Research H EI SHEIKH SAAD and others Genistein and vinclozolin on 216:2 246 mammary gland

et al. 2001, Munoz-de-Toro et al. 2005, Fenton 2006, Enoch development (Russo & Russo 1978). The primary ducts et al. 2007, Vandenberg et al. 2007) or may increase the risk develop from the epithelial bud and extend into the fat of breast cancer (Durando et al. 2007, Cameron & Foster pad in response to signals from the surrounding mesench- 2009). A number of epidemiological studies have suggested yme. Any interference with this process is likely to alter that the dietary intake of phytoestrogens decreases the risk the formation of the mammary gland. Extensive develop- of breast cancer in humans (Badger et al. 2005, Warri et al. ment of the mammary gland then occurs during puberty, 2008, Wu et al. 2008). Soy consumption during childhood when rising levels of ovarian hormones induce the and adolescence has been consistently linked to a marked formation of highly proliferative terminal end buds reduction in breast cancer risk in Asian and American- (TEBs) at the tip of the developing ducts (Sternlicht et al.

Asians (Korde et al. 2009), in agreement with some 2006). Estradiol (E2) and GH, and their interaction with experimental studies showing that pubertal exposure to the corresponding receptors (ERa and GHR), are required genistein (G), a natural phytoestrogen that is prevalent in for normal duct growth and morphogenesis (Daniel et al. plants (Wang et al. 2006), reduces mammary cancer risk 1987, Korach et al. 1996). Progesterone induces side in rodent models (Warri et al. 2008). Many experimental branching of the ductal epithelium (Lydon et al. 1995,

studies indicate that G which interacts with estrogen Atwood et al. 2000). E2 and progesterone are required for receptors (ERs), may affect mammary gland development the development of alveolar structures at each estrous and the incidence or multiplicity of carcinogen-induced cycle. However, much less is known about the role of mammary tumors. The precise effects of G mainly depend androgens in mammary gland development. Abnormal on the window of exposure (in utero alone, during postnatal mammary gland development and growth retardation days (PNDs) 2–8, or in utero until adult life) (Hilakivi-Clarke have been observed in female mice lacking androgen et al. 1999a,b, Lamartiniere et al. 2002, Pei et al. 2003, Foster receptor (AR) (reduced duct branching at prepubertal et al. 2004, Warri et al. 2008). The effects of environmental stages and lower levels of lobuloalveolar development in chemicals with anti-androgenic activity (such as fungi- adults) (Yeh et al. 2003). The effects of exposure to anti- cides) on the female mammary gland development remain androgenic EDCs on the development of the female unknown. Also, the implication of an association of G and mammary gland are unknown. an anti-androgen EDC in female mammary gland develop- One of the first chemicals to be identified as an anti- ment and mammary tumorigenesis has not been explored. androgen was the dicarboximide fungicide vinclozolin (V) The window of exposure appears to be a critical factor (Gray et al. 1994, Kelce et al. 1997). This chemical has Journal of Endocrinology in the modulation of mammary gland development, as is been extensively used to treat fruit and vegetables crop also reported in epidemiological studies of women born to and has been recognized as a contaminant of human mothers treated with the synthetic estrogen diethylstil- foods. V metabolism generates two major anti-androgenic bestrol during pregnancy (Hoover et al. 2011). Adverse compounds, M1 and M2, which may also have non- health outcomes including an increase in breast cancer androgenic functions (Molina-Molina et al. 2006). risk were recently reported in these women (Hoover et al. When administered in utero or during the early postnatal 2011). In recent years, the impact of early exposure to period, V induces malformations, alterations in reproduc- EDCs has been increasingly analyzed in animal models tive function, and sexual differentiation of male rats (Gray (Markey et al. 2001, Munoz-de-Toro et al. 2005, Fenton et al. 1999, Monosson et al. 1999). Anti-androgens 2006, Durando et al. 2007, Vandenberg et al. 2007). including V have been shown to alter the development It was also recently shown that the biological actions of male rat mammary gland. Nipple retention has been of one chemical may be influenced by the presence of identified as a sensitive endpoint for mammary gland another. However, few studies have considered the development in male rats treated with anti-androgen effectsofmixturesofEDCsonmammaryglands compounds (Christiansen et al. 2009), suggesting that (Foster et al. 2004, Wang et al. 2006). One way of studying androgens may play an important role in very early sex such mixtures involves using combinations of EDCs differentiation of the mammary gland. Gestational dietary from the same or different families (Wang et al. 2006, exposure to low-dose V and G has also been shown to have Christiansen et al. 2009, Eustache et al. 2009, Lehraiki et al. antagonistic effects on rat fertility and on fetal germ cell 2011), taking into account the hormonal environment development (Eustache et al. 2009, Lehraiki et al. 2011). of normal development. We recently reported that exposure to V alters the Mammary gland development in rats begins peripubertal development of the female rat mammary 6–7 days before birth, about days 12–14 into embryonic gland (El Sheikh Saad et al. 2011). Increases in epithelial

branching and focal branching defects were observed mating under conditions of 22 8C, 55% humidity, and at puberty (PND35) in the V-exposed group. We also a 12 h light:12 h darkness cycle. Cages and bottles were reported that mammary glands from rats exposed to GV made of polypropylene to avoid any contamination with or G displayed higher levels of epithelial branching and BPA or phthalates, and the water was filtered through proliferation, larger TEBs, and ductal hyperplasia than charcoal to eliminate any pesticide or active endocrine untreated rats (El Sheikh Saad et al. 2011). In animals contaminant. The rats were fed a purified diet (phyto- with an established estrus cycle (PND50), exposure to estrogen-free, INRA) and water ad libitum, as described GV and to a lesser extent V induces the formation of previously (Eustache et al. 2009). At gestational day (GD) abnormal hyperplasic alveolar structures. We therefore 1, determined by the presence of intravaginal sperm hypothesized that V interferes with endocrine signaling plug, the dams were randomly divided into groups of 15. during mammary gland development (El Sheikh Saad et al. From GD1 until weaning (PND21), each pregnant female 2011). If this proves to be the case, it will be important to underwent oral gavage once daily with the appropriate identify the role of this compound in hormonal signaling preparations. This study refers to the same pools of animals pathways during development. In this study, we aimed and to the same chemicals as described in El Sheikh Saad to identify the signaling pathways modified by exposure et al. (2011), as they came from the same experiments. in utero and during lactation to G, V, or GV, at various Molecules (G or V) were administered at 1 mg/kg body stages of development. We found that low doses of weight per day alone or in combination (GV). They were V and GV induced changes in genes related to develop- dissolved in corn oil in order to administrate the one ment, morphogenesis, differentiation, and metabolism 2 ml/kg-day dose. Control animals received the corn oil indicating alterations in the development of mammary vehicle (2 ml/kg body weight per day). Purity was glands, which depend on the compound or mixture controlled weekly as described previously (El Sheikh Saad used and the period of development considered. The et al. 2011). The G dose resembled a soy-based diet intake implications of our findings toward mammary gland (El Sheikh Saad et al. 2011). The V dose was higher than tumorigenesis and relevance to breast cancer risk are human food contamination levels but it was lower than discussed and require further studies. the no observable adverse effect level (NOAEL) combining chronic toxicity, carcinogenicity, and reproductive toxicity in rats (1.2 mg/kg body weight/day; U.S., Environ- Materials and methods mental Protection Agency (EPA 2003)). The acceptable Journal of Endocrinology daily intake (ADI) of V is 600 mg/day per person, Chemicals corresponding to an exposure of 0.01 mg/kg body G with a purity of 99% was synthesized (LCOO, Universite´ weight per day (Eustache et al. 2009). On the day of Bordeaux 1, Talence, France), as previously reported parturition, pups were sexed and weighed; all litters were (El Sheikh Saad et al. 2011). V was extracted from the standardized in order to contain ten offspring (five males commercial Ronilan (BASF, Levallois-Perret, France) by JP and five females). At weaning, rats were dispatched. Female Cravedi’s Laboratory (INRA, Toulouse, France) and the offspring were randomized (one female per litter), ident- final preparation was O95% pure, as described previously ified by implanted chips, and grouped together in (Bursztyka et al. 2008). In addition, the absence of polypropylene cages (five females per cage). Offspring degradation products M1 and M2 was tested by LC–MS (ten pubertal or ten cycling females per treatment group, as described previously (Bursztyka et al. 2008). Molecules or untreated rats) fed the purified diet (phytoestrogen-free) were dissolved in corn oil. until they were killed (PND35 and PND50) (Fig. 1). Food consummation and body weight were measured twice weekly. Animals and treatment All female offspring in these studies were examined Female and male Wistar Han rats (60 females and daily for vaginal opening to determine pubertal stage. 60 males) at 8 weeks of age (Harlan France Sarl, Gannat, Vaginal smears were collected on the morning on which France) were maintained in an animal facility. Procedures the animals were killed (PND35 or PND50). The collected for handling and experimentation followed by ethical vaginal smears were placed on glass slides, air dried, and guidelines have been described elsewhere (El Sheikh stained with Hemacolor. At PND35, all animals but one Saad et al. 2011). At arrival, the rats were allowed to had a vaginal opening. Rats were determined to be in acclimatize to the animal facility for 4 weeks before proestrus, estrus, metestrus, or diestrus. We checked

Pregnant mother Female offspring

Conception Birth Weaning Puberty Virgin cycling females

GD1 PND1 PND21 PND35 PND50

Gesto-lactational exposure to G or/and V 1 mg/kg per day Killing Killing

Figure 1 Protocol for in utero and lactational exposure to low doses (1 mg/kg per (INRA). Details of the diet have been reported elsewhere (El Sheikh Saad day) of G, V and a mixture of these compounds. Animals were force-fed et al. 2011) and are detailed in Materials and methods section. once daily with the appropriate preparation in a phytoestrogen-free diet

whether treated animals were evenly distributed between used for genomic analysis to limit the number of the various phases of the estrous cycle (not shown). microarray experiments required. Two pools (five rat tissues each) for each treatment (G and V) and untreated animals (control (C)) were analyzed at each period of Mammary gland samples observation (PND35 and PND50). We thus analyzed Female offspring were anaesthetized under isoflurane 300 ng pooled RNA (each set to 400 ng/ml) coming from (2.5% in air). The right fourth mammary glands were five different mammary glands from the same group. harvested to perform whole mount analysis (El Sheikh Two pools for GV-treated samples were also prepared at Saad et al. 2011). The left fourth mammary glands were PND50, but a specific pool was prepared for GV-treated harvested for mRNA studies (ten animals/group). The rats presenting abnormal hyperplastic alveolar structures lymph nodes were not removed from the mammary (GV2, as described in our previous study; El Sheikh gland before RNA isolation. An aliquot of the left fourth Saad et al. 2011). Sixteen RNA samples were thus

Journal of Endocrinology mammary gland was also fixed in buffered 10% formalin, analyzed for hybridization. dehydrated through graded alcohols, and paraffin After validation of the RNA quality, 300 ng total RNA embedded. Animals were killed by opening the thorax. was reverse transcribed following the Genechip Whole transcript Sense Target labelling assay kit (Affymetrix, RNA extraction Santa Clara, CA, USA). Briefly, the resulting double- stranded cDNA was used for in vitro transcription with T7 Total RNA was extracted from the same individual RNA polymerase. After purification, 10 mg cRNA was used mammary glands obtained from PND35 and PND50 rats for RT with random primers. The cDNA obtained was then as were used for histological studies (El Sheikh Saad et al. purified and fragmented. After control of fragmentation 2011). Tissues were homogenized in 1 ml TRIzol Reagent using Bioanalyzer 2100, cDNA was end labeled with biotin (Invitrogen) per 50 mg tissue, as described previously using Terminal Transferase (using the whole transcript (El Sheikh Saad et al. 2011). The quantity and purity of terminal labelling kit of Affymetrix). cDNA was then each RNA sample were determined from absorbance values hybridized to GeneChip rat gene (Affymetrix) at 45 8C for at 260 and 280 nm absorbance, and RNA integrity was 17 h. These microarrays have 27342 probe sets represent- determined (Bioanalyzer 2100, using Agilent RNA6000 ing known rat genes. After overnight hybridization, chips nano chip kit, Agilent Technologies, Massy, France). Only were washed on the fluidic station FS450 following specific high-quality total RNA samples, with an RNA integrity protocols (Affymetrix) and scanned using the GCS3000 number (RIN) O8, were considered for further analysis. 7G. The image was then analyzed with Expression Console software (Affymetrix) to obtain raw data (cell files) and Characterization of the genomic profile metrics for quality controls. The observations of some of RNA samples were used for both genomic analysis and these metrics and the study of the distribution of raw data validation experiments. A pooling strategy was initially showed no outlier experiment.

Bioinformatics the samples) and high (Ct values between 18 and 20) respectively. Finally, Tbp was identified as stably expressed Quality assessments, normalization, and statis- across the estrous cycle (Hvid et al. 2011). The nucleotide tics All quality controls, normalization, and statistics sequences of the oligonucleotide primers of the tested were performed using Partek Genomic Suite. Raw data genes are shown in Supplementary Table S1, see section on were preprocessing using the Robust Multichip Algorithm supplementary data given at the end of this article. To (RMA; Irizarry et al. 2003, Nikolsky & Bryant 2009, Mi avoid amplification of contaminating genomic DNA, one et al. 2010). We first made a hierarchical clustering for of the two primers bound to the junction between two unsupervised analysis (Pearson’s dissimilarity and average exons. The specificity of PCR amplicons was checked by linkage) with all pooled samples. To find differentially agarose gel electrophoresis. PCR was performed with the expressed genes, we applied a classical unpaired Student’s SYBR Green PCR Core Reagents kit (Perkin–Elmer Applied t-test between compared groups and computed the fold- Biosystems, Foster City, CA, USA). The experiments were change for each gene. Then, we used these two statistics to performed in duplicate for each data point. filter and select differentially expressed genes. We selected Validation was carried out for each individual RNA genes with P value !0.05 and 1.5-fold-changes. All data (nZ10) for each treatment group at PND35 and PND50 obtained by microarray analysis have been submitted and for each selected gene. When change in the on GEO Omnibus site with the accession number expression level for a specific gene in the transcriptomic GSE32432 (private link for reviewers: http://www.ncbi. analysis was correlated with a significant difference in the nlm.nih.gov/geo/query/acc.cgi?tokenZxdafzascymwswr- Q-PCR, the gene was considered as validated. q&accZGSE32432).

Immunocytochemistry Functional analysis Function and pathway enrich- ments of differentially expressed genes lists were carried The samples were cut into 4 mm sections and processed for out through the use of Ingenuity Pathway Analysis hematoxylin and eosin staining. TEBs, terminal ducts, (Ingenuity Systems, USA, www.ingenuity.com). ABs, and lobular epithelial structures were histologically classified, as previously reported (El Sheikh Saad et al. 2011) according to Russo & Russo (1978). Highlights of Gene expression analysis by real-time RT-PCR the previous work are presented in Table 1. After paraffin Journal of Endocrinology RT of the RNA (1 mg) from each mammary sample (ten removal, the sections were subjected to 15-min microwave animals/group), prepared as described earlier, was per- antigen retrieval in pH 6 citrate buffer and pre-incubated formed with 100 units of Superscript II RNase H-reverse with a blocking solution (5% BSA and 0.5% casein in PBS transcriptase (Invitrogen) and 3 mM random hexamers buffer) to minimize nonspecific binding. Sections were (Pharmacia), as described previously (El Sheikh Saad et al. then incubated overnight at 4 8C in a humid chamber 2011). The fold-change in gene expression was quantified with the following antibodies, as described previously by real-time quantitative RT-PCR as described previously (El Sheikh Saad et al. 2011): rabbit polyclonal antibody (Bieche et al. 2004), using the P0 ribosomal protein as an against b-casein (Csn2) (1:2000 dilution; courtesy of endogenous RNA control. Results are expressed as N-fold M Glukhova and D Medina, Baylor College of Medicine, differences in target gene expression relative to the Houston, TX, USA) (Durban et al. 1985), against Stat5A expression of the housekeeping gene Po and to untreated (Clone L-20, 1:2000 dilution, sc-1081, Santa Cruz Bio- samples (set to 1). We selected Po (Rplp0 also known as technology), cytokeratin 17 (Krt17) (Clone V-17; 1:25 36B4) because this gene is widely used as an endogenous dilution, sc-101931, Santa Cruz Biotechnology), claudin 1 control for northern blot analysis. In some experiments, (1:50 dilution, Sigma–Aldrich), and monoclonal mouse gene expression in mammary gland samples was also antibody against smooth muscle actin (Clone 1A4, 1:400, normalized with respect to the housekeeping gene Tbp. Dako, Carp Carpenteria, CA, USA). Control slides were Tbp was selected as an endogenous control because there incubated with normal rabbit or mouse IgG. After

are no known Tbp retropseudogenes that might lead to the endogenous peroxidase quenching with 3% H2O2 in PBS co-ampliﬁcation of contaminating genomic DNA, thereby for 10 min, the bound antibodies were revealed with interfering with RT-PCR. We also selected Tbp and Po as an Impress REAGENT anti-Mouse and anti-Rabbit Ig endogenous controls because the prevalence of their (Vector Laboratories, Burlingame, CA, USA) according transcripts is moderate (Ct values between 24 and 26 in to the manufacturer’s instructions. Aminoethylcarbazole

Table 1 Quantitative modiﬁcations and other observations on the mammary gland structures at PND35 and PND50 after gestational/lactational exposure to G, V, and GV at 1 mg/kg body weight per day. This table summarizes the ﬁndings of our previous work (El Sheikh Saad et al. 2011)

Branching Proliferation Glandular Periductal (mm) (Ki67 mRNA) area (mm2) stroma Other observations

PND35 C 3.8G0.4 1 183.2G21.8 1 G 5.6G0.3* 2.1* 191.3G10.2 1.5* Large sized TEB; hyperplastic ducts V 6.0G0.4* 0.5 215.2G30.3 0.8 Focal branching defects; hyperplastic ducts Occasional loss of epithelial cell polarization GV 5.7G0.4* 2.4* 223.0G14.5 2.0* Large sized TEB; hyperplastic ducts PND50 C 4.5G0.2 1 447.5G17.8 1 G 5.7G0.3 1.8 433.7G15.2 1.3* V 5.1G0.2 2.8* 430.0G25.5 1.3* Hyperplasic lobuloalveolar structures (20% of animals) GV 5.7G0.6 1.5 510.7G13.7* 1.1 Hyperplasic lobuloalveolar structures (40% of animals)

*P!0.05, compared with control.

was used as a chromogen. Some slides were counterstained groups were analyzed at PND35 (puberty) and PND50 with hematoxylin. The distribution and the intensity of (virgin cycling females). A previous study by our group staining were evaluated by conventional light microscopy, showed histological changes in mammary gland develop- as described previously (El Sheikh Saad et al. 2011). ment in these conditions (El Sheikh Saad et al. 2011). Highlights of the previous work are presented in Table 1. Brieﬂy, the morphological changes observed at PND35 Statistical analysis included increases in TEB size, epithelial branching, All data were analyzed using SigmaStat software. Data and duct proliferation in animals exposed to G- and generated for analyzing vaginal opening or by Q-PCR were GV-exposed animals. Focal branching defects, with an expressed as meanGS.E.M. and analyzed using one-way Journal of Endocrinology occasional loss of epithelial cell polarization, were ANOVA. Where appropriate, intergroup comparisons were observed in animals exposed to V. At PND50, animals performed using Tukey’s multiple comparisons test. exposed to G alone displayed no major histological Differences were considered signiﬁcant if the P value abnormalities. Surprisingly, exposure in utero and during ! was 0.05. lactation to the binary mixture of G and V compounds resulted in postpubertal hyperplasic lobuloalveolar structures, associated with numerous secretion Results vacuoles (Table 1).

Rats were exposed in utero and during the neonatal period (from conception to weaning) to the phytoestrogen G, Mammary gland gene expression profiles at puberty the anti-androgen V, or a GV mixture at a dose of 1 mg/kg and in adult cycling rats previously exposed to G, V or per day as described in Fig. 1. As described previously (El a mixture of G and V Sheikh Saad et al. (2011); see Supplementary Figure S1A, We compared the gene expression profiles of the see section on supplementary data given at the end of mammary glands of pubescent (PND35) and cycling this article), female offspring exposed to the mixture of G and V (GV) exhibited significant early vaginal opening (PND50) female rats exposed to G, V, and GV with those compared with G and V exposed animals (P!0.05; of controls. Two pools (five rat tissues each) for each Supplementary Figure S1). However, early vaginal open- treatment (G and V) and untreated animals (C) (see ing was not associated with modifications in body weight Materials and methods section) were analyzed at each (see Supplementary Figure S1B in El Sheikh Saad et al. period of observation on Affymetrix rat microarray. Nonsupervised hierarchical clustering of the 27342 probe (2011)) or changes in E2 and progesterone levels (N Lalou, unpublished data). Mammary glands from the different sets (Fig. 2) highlighted cluster of genes modified by each

treatment, through comparisons with control rats. The transcriptional effects of the various treatments were clearly different at PND35 and PND50. The criteria for the genes to be analyzed were as follows: up- or down- regulation by 1.5-fold (P!0.05). We then removed features that did not correspond to known genes and duplicate probe sets. This process resulted in the identifi- ° cation of genes displaying significant differential expression between the controls and animals treated with the two compounds alone or together at PND35 PND50-V PND35-V PND35-V PND35-GV PND35-GV PND35-G PND35-C PND35-C PND35-G PND50-GV* PND50-V PND50-C PND50-G PND50-GV PND50-C PND50-G or PND50 (Supplementary Figure S2,seesectionon supplementary data given at the end of this article). At PND35, exposure to G, V, or the GV mixture resulted in changes to the expression of 213 genes: 65 (G), 56 (V), and 113 (GV) genes (Supplementary Figure S2A). Only one gene, Krt17 (encoding the cytokeratin 17), displayed changes in expression in response to all three treatments, G, V, and GV. Most of the genes were upregulated (Supplementary Figure S2A). Twenty genes were regulated by both G and GV (15 up- and five downregulated) (Supplementary Figure S2A and see GEO in Materials and methods section). However, no gene displayed changes in expression in response to both G and V or V and GV (Supplementary Figure S2A). At PND50, exposure to G, V, or the GV mixture resulted in the overall deregulation of 213 genes: 25 (G), 112 (V), and 119 (GV) genes (Supplementary Figure S2B).

Journal of Endocrinology Two genes, Pfkb3 and Nr1d1, displayed changes in expression in response to all three treatments, G, V, and GV. Thirty-three genes were common to both V and GV, ten genes displayed changes in expression in response to G and GV, and two genes displayed changes in expression in response to both G and V (Supplementary Figure S2B and see GEO).

Validation of the microarrays by real-time RT-PCR

ERa, progesterone receptor (PR), and AR are present in the –3.6 0 3.7 developing mammary gland at puberty and in cycling animals. Genes encoding these hormone receptors were Figure 2 analyzed as controls by real-time RT-PCR. Data concerning Transcriptome alterations in rat mammary gland as a function of exposure ERa and AR have been already reported (see Table 2 in to G, V, and GV exposure. Nonsupervised clustering analysis of genes et al displaying differential expression after exposure to G, V, and the GV El Sheikh Saad . (2011)). A significantly higher level of mixture relative to C. Two individual chip arrays (nZ5 animals/ PR expression was observed in the mammary glands of pool) for each treatment were analyzed on the basis of log2 ratios, with the rats exposed to GV and G at PND35, a time point at which scale shown on the bottom right. The heat map of the expression values for each gene differentially expressed relative to control is shown. Transcrip- estrogens are driving development. No significant change tomic alterations in mammary gland were different between PND35 in ERa was detected compared with control (Table 2). (puberty) and PND50 (animals with established cycles). At PND50, the For validation of the microarray results, we selected the samples are GV2 (*) and GV1 (8), as described in Materials and methods section and in El Sheikh Saad et al. (2011). Full colour version of this figure 16 genes (13 at PND35 and 12 at PND50, nine genes available via http://dx.doi.org/10.1530/JOE-12-0395. being in common) most strongly overexpressed in the

Table 2 Validation of microarray results by real-time RT-PCR upregulated by V and GV in the transcriptomic analysis on the mammary glands of rats exposed to G,V or a mixture of (Table 3). We also identified five additional genes the two compounds at PND35. Genes were selected as described displaying significant upregulation in the GV-exposed in Results section. Data are represented as mean fold changes in group (Table 3) that were not identified in the expression for each treated group (ten animals/group), as a transcriptomic assay (Actn2, Csn2, Fabp3, Tnni2, and treated/control group ratio, as detailed in Materials and Wap). The detection of these additional upregulated methods section. Relative expression levels for each gene in genes may reflect the higher sensitivity of RT-PCR than the control was set to 1, and the values shown are meanGS.E.M. of transcriptomic analysis, and/or the heterogeneous values obtained for the animals of the GV group. As we Microarray RT-PCR previously observed abnormal development in four GV-

Genes G V GV G V GV treated animals (called GV2) (see El Sheikh Saad et al. (2011)), we compared the expression of these genes in Casq1 NS NS 3.3 1.8G0.8 0.3G0.2 4.1G2.5* Ckm 1.7 NS 2.8 1.8G0.5 0.4G0.1 3.5G1.2 mammary glands from animals of the GV2 and GV1 Cldn1 NS 5.7 NS 1.7G0.3 6.2G3.6* 1.6G0.3 subgroups. As shown in Table 3, Csn2, Fabp3, and Wap Csn2 NS NS K2.0 2.0G0.9 1.5G0.7 0.9G0.4 were most strongly upregulated in GV2 than in GV1. Elf5 1.6 NS NS 4.0G0.5* 1.4G0.3 1.6G0.2 Krt17 2.1 6.5 1.8 1.5G0.3 8.3G3.6* 2.0G0.2* For identification of the biological processes poten- Mylk2 1.5 NS 2.8 2.2G0.9* 0.5G0.2 3.2G1.5* tially relevant to the observed changes in mammary Pvalb NS NS 4.0 1.8G0.8 0.4G0.1 5.2G3.0* gland treatment that were altered by the various Pygm 1.6 NS 2.8 2.3G1.1* 0.4G0.2 3.6G2.1* Sprr1a NS 38.8 NS 2.4G0.6 55.7G37.1* 1.0G0.3 treatments, we used the Ingenuity Pathway Analysis Tfap2c 1.6 NS NS 2.1G0.3* 1.4G0.2 1.3G0.1 database to cluster the 213 genes displaying differential Tnni2 1.5 NS 3.6 2.3G0.9 0.4G0.2 3.6G1.8* expression (on both PND35 and PND50) into common Wap NS NS K2.4 0.8G0.2 1.0G0.3 0.4G0.1* Ar NS NS NS 1.5G0.3 K1.5G0.1* 1.3G0.2 biochemical pathways and physiological processes. Er NS NS NS 1.3G0.1 K1.4G0.1 1.2G0.1 Each of the treatments appeared to alter the expression Pgr NS NS NS 2.1G0.8 K1.2G0.4 2.0G0.6* of genes with diverse functions, in a different way at different time points (Supplementary Figure S3,see *Statistically significant difference (P!0.05). Comparisons with results from microarrays are also shown. section on supplementary data given at the end of this article and Table 4, Supplementary Tables S2 and S3, see section on supplementary data given at the end of Journal of Endocrinology mammary glands of GV- and V-treated rats with respect to this article). untreated rats at PND35 or PND50 (see GEO). These genes were involved in various cellular and molecular functions including muscle development and morphogenesis Differentially expressed genes among treatments at (Actn2, Casq1, Ckm, Fhl1, Mylk2, Pvalb, Pygm, and Tnni2), PND35 (puberty) development (Cldn1, Elf5, Krt17, Sprr1a, and Tfap2c), and The genes upregulated in response to G (45/65) were differentiation (Csn2, Fabp3, and Wap). Validation was generally not strongly upregulated (1.5- to 2-fold, carried out for each individual RNA (nZ10) for each P!0.05) (Supplementary Table S2). The major categories treatment group, according to the criteria described in identified related to morphogenesis and development Materials and methods section. RT-PCR confirmed the (Supplementary Figure S3 and Table 4), and the genes results obtained in the microarray experiments for 97% at assigned to these categories included Krt14 and Krt17, PND35 and 83% at PND50 using P0 as endogenous control Ckm, Pygm,andMylk2 (Supplementary Table S2). (Tables 2 and 3). The normalization of gene expression Another functional group altered by G included few with respect to Tbp and Po gave similar results, by real-time genes encoding enzymes involved in metabolism RT-PCR, in terms of the effects of the treatments (data not (Supplementary Table S2). The expression of Tfap2c shown). At PND35, the upregulation of Cldn1, Krt17, and and Elf5, encoding transcription factors, was also Sprr1a expression by exposure to V was validated by RT-PCR modulated by exposure to G (Supplementary Table S2). (Table 2); the upregulation of Casq1, Krt17, Mylk2, Pygm, The 20 downregulated genes (see GEO) included the Pvalb, and Tnni2 with GV was also validated, as was the metabolism-related genes Fkbp5 and Pla2g2d (Supple- induction of Tfap2c and Elf5, which encode transcription mentary Table S2). The largest functional categories, factors, by exposure to G (Table 2). The results at PND50 associated with robust changes (greater than fivefold), confirmed the upregulation of all genes shown to be following exposure to V concerned development and lipid

Table 3 Validation of microarray results by real-time RT-PCR on mammary gland from rats exposed to G, V, and GV at PND50. Data are represented as mean fold-changes in expression for each treated group vs untreated, as described in Table 2

Microarray RT-PCR

Genes G V GV G V GV GV1 GV2

Actn2 NS 4.3 NS 2.3G1.2 11.9G8.0* 5.9G1.4* 3.9G3.3 7.2G0.6 Casq1 NS 4.6 NS 0.8G0.4 7.2G3.8* 2.6G0.8 1.7G1.4 3.5G1.1 Ckm NS 4.1 2.1 1.0G0.5 7.2G3.0* 5.3G2.8* 1.6G1.4 9.4G5.1 Csn2 NS 3.5 NS 1.3G0.8 7.3G3.7* 31.4G20.1* 0.4G0.2 62.4G35.3 Elf5 NS NS 1.7 2.3G0.6 2.9G0.4* 2.4G0.4* 1.7G0.3 3.1G0.4 Fabp3 NS 2.2 NS 1.0G0.4 3.3G1.3* 5.8G3.2* 0.6G0.2 10.9G5.4 Fhl1 NS 1.6 NS 1.5G0.2* 2.1G0.6* 1.3G0.1 1.3G0.3 1.3G0.1 Mylk2 NS 1.8 NS 1.2G0.5 3.9G2.6* 2.8G0.8 0.6G0.3 5.1G0.2 Pvalb NS 4 NS 0.6G0.3 4.3G2.4* 1.7G0.6 1.0G0.9 2.4G0.7 Pygm NS 5 2.6 1.2G0.5 6.0G3.0* 3.2G1.0* 1.8G1.4 4.6G1.5 Tnni2 NS 3.1 NS 1.0G0.5 8.4G4.5* 3.5G1.3* 2.3G2.1 4.8G1.6 Wap NS 2.6 NS 0.8G0.2 2.7G0.9* 32.2G20.6* 0.8G0.3 63.7G36.5 Ar NS NS NS 2G0.7 1.3G0.2 1.3G0.2 1.7G0.2* 1G0.3 Er NS NS NS 1.6G0.3* 1.5G0.2* 1.5G0.3* 2G0.5* 1G0.4 Pgr NS NS NS 2.6G1.4 1.3G0.4 1.7G0.8 2G1.1 1.4G0.4

*Statistically signiﬁcant difference (P!0.05).

metabolism (Supplementary Figure S3 and Table 4). 12 downregulated) (Supplementary Figure S2), and the Development-related genes displaying changes in magnitude of the change was low (fold-change !2) expression included Cldn1, Sprra1,numerousKrts, (Supplementary Table S3). The V exposed group displayed and Klks (kallikrein-related peptidase) (Supplementary signiﬁcant upregulation of genes (88/112) mostly involved Table S2). V also induced the expression of a few in development and morphology(SupplementaryFigureS3), genes encoding enzymes involved in steroidogenesis including muscle development (greater than threefold including Cyp17a1 and Cyp2b12 (Cyp2b15)(Supple- induction, such as Actn2, Casq1, Ckm, Pvalb, Pygm,and

Journal of Endocrinology mentary Table S2). Tnni2)(Supplementary Table S3, Supplementary Figure S3 The GV mixture elicited more robust changes in and see GEO). V also induced moderate changes in gene gene expression than G and V alone (two- to four-fold expression related to carbohydrate (Pfkfb3 and Ppp1r3a)and induction, P!0.005, Supplementary Table S2), in terms of lipid (Fabp3 and Lipa)metabolism(Supplementary Figure S3 development and morphology (Supplementary Figure S3). and Supplementary Table S3). Twenty-four genes, including The functional categories concerned muscle development Cxcl13 and Lep, were downregulated by exposure to V (e.g. Actn2, Casq1, Ckm, Pvalb, and Tnni2)(Supplementary (Supplementary Table S3). Table S2 and Supplementary Figure S3) and development For the GV exposure group, the largest functional (Krt17 and Myh4)(Supplementary Table S2). GV also categories (up- and downregulated) were development, induced changes in the expression of genes, such as Pvalb, morphology, metabolism, proliferation, and cell death Pfkm,andPrkg3, related to carbohydrate metabolism (Supplementary Figure S3). The number of genes (Supplementary Figure S3 and Supplementary Table S2). involved in muscle development and morphology The genes downregulated (!K2.2-fold change) after identified was smaller and the level of induction was exposure to the GV mixture included Csn2 and Wap, lower than that observed with V alone (Supplementary which are related to cell differentiation (Supplementary Table S3). Cell differentiation-associated genes (Csn2, Table S2 and see GEO). Fabp3, and Wap) were also significantly upregulated by exposure to V and GV (Supplementary Table S3) in 40% of exposed animals, as shown by real-time RT-PCR (not Differentially expressed genes among treatments at shown). Other categories related to carbohydrate (such PND50 (cycling animals) as Pdk4 and Pfkfb3) and lipid metabolism (such as The G exposed group displayed few significant changes Cyp24a1 and Hmgcs2) were also altered after GV in gene expression (13 genes upregulated and exposure (Supplementary Table S3). Interestingly, the

Table 4 Analysis of the functions found to be differentially expressed in the mammary gland of rats after exposure to G, V, and GV at PND35 (A) and PND50 (B)

Min Max Number of genes

(A) PND35: functions G Skeletal and muscular system development and function 6.74!10K5 2.49!10K2 11 Tissue morphology 6.74!10K5 2.84!10K2 10 K K Carbohydrate metabolism 3.12!10 4 3.54!10 2 9 Development 2.07!10K3 4.92!10K2 22 Cell cycle 3.59!10K3 2.49!10K2 3 K K Cell death 3.59!10 3 4.92!10 2 11 Cell growth and proliferation 3.59!10K3 4.57!10K2 8 Lipid metabolism 3.59!10K3 4.92!10K2 5 V Development 9.91!10K7 1.79!10K2 11 Lipid metabolism 1.23!10K4 4.16!10K2 4 K K Tissue morphology 2.14!10 3 2.14!10 3 1 Carbohydrate metabolism 4.28!10K3 2.33!10K2 1 Cell growth and proliferation 2.96!10K2 2.96!10K2 1 GV Skeletal and muscular system development and function 9.97!10K32 4.66!10K2 41 Tissue morphology 9.97!10K32 3.68!10K2 39 K K Development 1.95!10 8 4.66!10 2 27 Carbohydrate metabolism 1.36!10K3 4.66!10K2 11 Cell death 5.94!10K3 2.94!10K2 3 K K Lipid metabolism 5.94!10 3 4.66!10 2 5 Cell growth and proliferation 1.03!10K2 4.66!10K2 2 Cell cycle 1.19!10K2 2.94!10K2 3 (B) PND50: functions G Cell death 5.79!10K3 5.79!10K3 1 K K Carbohydrate metabolism 1.92!10 2 1.92!10 2 1 Cell growth and proliferation 2.68!10K2 2.68!10K2 1 Development 2.68!10K2 4.64!10K2 1 Journal of Endocrinology V Skeletal and muscular system development and function 6.18!10K25 4.87!10K2 41 Tissue morphology 6.18!10K25 4.87!10K2 32 K K Development 3.17!10 5 4.87!10 2 22 Carbohydrate metabolism 7.08!10K5 4.87!10K2 10 Lipid metabolism 5.64!10K4 4.87!10K2 8 K K Cell cycle 6.22!10 3 4.27!10 2 3 Cell growth and proliferation 6.22!10K3 4.29!10K2 4 Cell death 1.85!10K2 4.87!10K2 6 GV Skeletal and muscular system development and function 2.03!10K5 4.66!10K2 22 Tissue morphology 2.03!10K5 3.51!10K2 19 K K Carbohydrate metabolism 1.04!10 4 4.09!10 2 11 Cell death 5.16!10K4 4.80!10K2 22 Cell cycle 3.83!10K3 4.51!10K2 8 K K Lipid metabolism 5.94!10 3 4.73!10 2 12 Development 5.94!10K3 4.95!10K2 19 Cell growth and proliferation 1.19!10K2 4.66!10K2 23

The minimum (Min) and the maximum (Max) indicate the most and the least signiﬁcant P value for genes.

expression of Cyp24A1 was increased in 60% of GV- Analysis of proteins by immunocytochemistry exposed animals (not shown), compared with controls, Several proteins for which gene expression had been G-, and V-exposed animals; twenty-six genes were shown to be modulated with P!0.05, and which were downregulated by GV exposure, including genes such considered to be of interest on the basis of the functional as Ccnb1 and Cdc20 involved in cell death and the cell categories to which they belonged, were validated by cycle (Supplementary Table S3). immunocytochemistry. We checked for the presence of

cytokeratin 17, claudin 1, and b-casein encoded by genes Sprr1a expression were clearly higher with V in E or D differentially expressed between treated and normal rat stages (Fig. 4A), and Ckm and Tnni2 expression levels were mammary glands. Immunostaining of smooth muscle higher with GV in P and E stages (Fig. 4A). Wap expression actin, a marker for myoepithelial cells, was observed in at D was lower after G, V, and GV treatment than in basal cells surrounding the TEB and ducts and in vessels untreated rats (Fig. 4A). In contrast to these changes (Fig. 3Aa). Cytokeratin 17 immunostaining was detected observed for a few genes (Cldn1, Krt17, Sprr1a, Ckm, and in the cells lining the TEBs (cap cells), in some cells in the Tnni2) at PND35 according to the cycle, changes were body (Fig. 3Ab, d, e and f), and in the ductal myoepithelial most obvious during the estrus cycle at PND50. Levels of (not shown). Immunostaining for claudin 1 (Fig. 3Ac, g, h Csn2, Wap, and Fabp3 expression in P stage were clearly and i), the major component of epithelial tight junctions, higher with GV treatment (mostly due to GV2 rather was strong at the point of contact between epithelial cells than GV1) (Fig. 4B); a smaller effect was observed for these from the TEBs. Both mRNA (Fig. 3C) and protein levels three genes with V in E stage (Fig. 4B), together with an were increased for both cytokeratin 17 and claudin 1 at increase in Ckm and Tnni2 expression in P or E stages PND35 (Fig. 3A e and h respectively). Consistent with an following treatment with V (Fig. 4B). increase in the expression of genes encoding milk proteins Csn2 Wap ( and ) at PND50 in the mammary glands of Discussion animals exposed to V and GV (Fig. 3D), b-casein immunostaining (Fig. 3B a, b, c, d and e) was detected in Exposure to environmental EDCs with estrogenic acti- the epithelium of the secreting alveolar structures from vities during fetal development is thought to contribute GV2 group (Fig. 3Bd), but not in the GV1 group (Fig. 3Be). to abnormal development of the mammary gland. Our Activated STAT5A has been detected in the differentiated data provide the first evidence that exposure to the anti- lactating gland. We therefore also carried out immuno- androgen V and a mixture of G and V in utero and during staining for STAT5A on nuclear sections from mammary lactation, modulates mammary gland morphogenesis at glands of GV-treated animals at PND50. Nuclear STAT5A the molecular level during puberty and in cycling animals. immunostaining (corresponding to activated STAT5A) was G administered alone and at low doses had the smallest also observed in the lobuloalveolar units (Fig. 3Bh), effect of the treatments tested. Changes in gene expression consistent with both the increase in the expression of were observed with V and GV, the precise nature of the genes (Csn2 and Wap) encoding milk proteins (Fig. 3D) changes depending on the period of postnatal develop- Journal of Endocrinology and the secreting alveolar structures observed in the ment (puberty or postpuberty, i.e. PND35 and PND50) and mammary glands of GV-treated animals (El Sheikh Saad the compound/or mixture considered. The data obtained et al. 2011). are consistent with our previous histological observations (El Sheikh Saad et al. 2011). The signaling pathways deregulated and the underlying mechanisms are discussed. Changes in gene expression may be dependent on We investigated the molecular events regulated the estrous cycle by exposure to G, V, and GV, by examining the gene Subtle changes in the development of the mammary expression profiles of whole mammary glands, with gland, such as proliferation and differentiation, have been validation of the results obtained for the expression of described during the 4–5 days of transition during the selected genes by RT-PCR and immunocytochemistry. rat estrous cycle (Schedin et al. 2000, Hvid et al. 2012). Crucial methodological considerations were taken into We reanalyzed the major changes in gene expression account in the interpretation of our data, including the previously reported by quantitative RT-PCR data for use of a phytoestrogen-free diet, and an analysis of vaginal individual RNA expression (as described in Tables 1 opening and of vaginal smears on the day on which the and 2), taking into account the stage of the estrous cycle rats were killed, for all rats. The RNA used for molecular (i.e. proestrus (P), estrus (E), and metestrus-diestrus studies was prepared from the same individual mammary (M-D)). Gene expression in mammary gland samples was glands obtained from PND35 and PND50 rats for normalized against two housekeeping genes: P0 (Fig. 4) histological studies (El Sheikh Saad et al. 2011), allowing and Tbp (not shown). At PND35 in control rats, several a comparison with previous histological observations. The changes were observed in the expression of genes transcriptomic study was performed with a cut off of 1.5 examined (Ki67, Wap, Tcfap2c, and Tnni2 and Ckm) during and a significant threshold of P value !0.05. The largest the estrous cycle (Fig. 4A). Levels of Krt17, Cldn1, and changes in the expression of the selected genes in our

study were normalized to gene expression in mammary results, taking into account the phase of the estrous samples by RT-PCR on ten samples per treatment, with cycle for each animal on the day it was killed. Finally, several housekeeping genes (Po and Tbp), as previously we also validated by immunocytochemistry the detection reported (Hvid et al. 2011). We also carefully analyzed our of several proteins for which commercial antibodies

A (a) (b) (c)

(d) (e) (f)

(g) (h) (i) Journal of Endocrinology

B (a) (b) (c) (d)

(e) (f) (g) (h)

C D 12 * * 100 9 60 * * C 8 20 G 15 3 * V 10 * 2 GV 5 * * GV2 mRNA expression 1 mRNA expression 0 0 Cldn1 Krt17 Csn2 Csn2 Fabp3Wap

were available. Further studies, including tumorigenic encoding cytokeratin 17 was increased by G and GV in studies, are required to determine the full relevance of myoepithelial cells, as shown by RT-PCR and immuno- these ﬁndings. cytochemistry. Cytokeratin 17 was observed in basal cells in TEB, i.e. in the cells that expressed cytokeratin 14 (a marker of basal cells in TEBs). These results suggest that G and GV increase the expression of genes encoding major changes to the actomyosin cytoskeletal architecture cytoskeletal elements, involved in ductal branching, of cells, including myoepithelial cells (Moumen et al. during puberty (PND35) 2011), occur during mammary gland morphogenesis at

During puberty, E2 directs ductal elongation and branch- puberty. Moreover, the significant increase in expression ing (Sternlicht et al. 2006). We have previously reported of Tcfap2c, which encodes the transcription factor AP2g that both an increase in ductal branching and a (Table 2) in response to G treatment, is also consistent proliferation within the TEBs at PND35 following with the increases in epithelial cell proliferation in TEBs exposure to G and GV in utero and during lactation and ductal branching observed at PND35 following (El Sheikh Saad et al. 2011). In this study, we provide the exposure to G (El Sheikh Saad et al. 2011). AP2g is strongly first evidence that in utero exposure to GV and G strongly expressed in the cap cells lining the TEBs (smooth muscle alters gene transcription at puberty and has effects on actin C) and in ductal contractile myoepithelial (cyto- developmental genes, consistent with the changes to keratin 14C) cells in the developing rodent mammary mammary gland morphogenesis observed in our previous gland (Friedrichs et al. 2007, Jager et al. 2010). The loss of study (El Sheikh Saad et al. 2011). The largest category AP2g/Tfap2c has been shown to reduce cellular prolifer- of genes concerned encodes cytoskeletal elements, such ation within TEBs and to impair ductal branching and as actin, actin-associated proteins, and cytokeratins. The elongation (Jager et al. 2010). Interestingly, V, when expression of several genes encoding actin-associated administered alone, did not appear to modulate these proteins including Pvalb, Pygm,andTnni2,andthe genes, except for cytokeratin 17, suggesting that G may expression of Ckm and Mylk2 was increased by the be the driving force behind regulation of these genes by combination of G with V (GV), and G alone to a lesser GV at PND35. extent, as shown in our transcriptomic study and validated by RT-PCR (see Table 2). The modulation of V alters gene expression related to the development these genes has been observed following treatment with Journal of Endocrinology of the mammary gland at PND35 BPA, benzyl butyl phthalate (BBP), and transforming growth factor b (TGFb) in the rodent mammary gland V alone had a different effect on gene expression profile (Xie et al. 2003, Moral et al. 2007, 2008). Creatine kinase from G and GV in the pubertal mammary gland. V increases (Ckm) has also been detected in the mammary gland the expression of Krt17, Cldn1, and Sprr1a, whereas G did (Raymond et al. 2011). Recently, methylated CKM has not, as shown by RT-PCR (Table 2). Claudin 1 was detected been correlated with clinical prognosis in breast cancer in epithelial cells by immunocytochemistry (see Fig. 3). The patients (Jeschke et al. 2012). The expression of the gene increase in Cldn1 expression induced by V in mammary

Figure 3 Immunostaining for specific proteins in the mammary gland. V (c), or GV-treated animals (d and e); b-casein immunostaining was (A) Immunocytochemistry at PND35 (A) and PND50 (B). Mammary sections increased in the epithelium of the secreting alveolar structures of GV2 representing ten animals of each treatment or control group were (d) and increased compared with control and other treatments; GV1 (e). analyzed. Representative micrographs are presented. (A a, b and c) Paraffin (f) H&E of lobuloalveolar structures from the GV2 group. (g and h) Stat5A sections of untreated animals stained with antibodies against alpha immunostaining. In h, strong STAT5A immunostaining was observed in smooth muscle actin (a), cytokeratin 17 (b), and claudin 1 (c). Positive a lobuloalveolar unit from the GV2 group. Arrow, nuclear STAT5A staining of alpha smooth muscle actin (a) and cytokeratin 17 (b) was shown immunostaining. By contrast, nuclear STAT5A was not observed in the in the outer myoepithelial cell layer of TEBs. Staining with an antibody terminal structures from an untreated animal (g). Original magnifications against claudin 1 (c), showing positive staining between epithelial cells !40 (a, b, c, d and e), !20 (f), or !60 (g and h). (C) Histograms showing the from a TEB. (d, e, f, g, h and i) Immunostaining cytokeratin 17 (d, e and f) higher levels of mRNA for both claudin 1 and cytokeratin 17 at PND35 in and claudin 1 (g, h and i) in treated animals. Immunostaining for both animals exposed to V; (D) significant higher levels of mRNA for Csn2, proteins was increased in V (e and h respectively), compared with G Fabp3,andWap in GV (GV2)-exposed animals at PND50. *, Statistically (d and g) or untreated (b and c); GV (f and i). (B) Immunocytochemistry at significant difference (P!0.05). Full colour version of this figure available PND50. (a, b, c, d and e) b-Casein immunostaining of control (a), G (b), via http://dx.doi.org/10.1530/JOE-12-0395.

A mRNA expression at PND35

0.20 Ki67 0.025 Sprr1a 0.05 Cldn1 0.008 Krt17 0.15 0.020 0.04 0.006 0.015 0.03 0.10 0.004 0.010 0.02 0.05 0.005 0.01 0.002 0 0 0 0 P E D P E D PDE D P E

0.6 Ckm 0.5 Tnni2 0.025 Tcfap2c 0.08 Wap 0.5 0.4 0.020 0.06 0.4 0.3 0.015 0.3 0.04 0.2 0.010 0.2 0.02 0.1 0.1 0.005 0 0 0 0 P E D PDE D PDE P E

B mRNA expression at PND50

30 Csn2 2.0 Wap 0.10 Fabp3 25 1.5 0.08 20 0.06 15 1.0 C 0.04 10 0.5 G 5 0.02 0 0 0 V PEE PEP GV

0.4 Ckm 0.20 Tnni2 0.20 Ki67 GV1 Journal of Endocrinology GV2 0.3 0.15 0.15

0.2 0.10 0.10

0.1 0.05 0.05

0 0 0 PEE PEP

Figure 4 Distinctive G-, V-, and GV-associated changes in the mammary gland Tables 2 and 3. Gene expression (y-axis) was shown for each treatment and of treated rats, as a function of position in the estrous cycle. Untreated, each phase of the estrous cycle (x-axis). Full colour version of this ﬁgure G, V, and GV mixture gene proﬁles (determined by RT-PCR) were shown available via http://dx.doi.org/10.1530/JOE-12-0395. for PND35 (A) and PND50 (B). The selected genes are derived from

glands is consistent with the modulation of expression of see Table 2). Together, these findings suggest a role of these this gene by androgens previously reported in mouse two V-regulated proteins, KRT17 and CLDN1, in three- immature Sertoli cells (Gye 2003). Changes in the dimensional organization and cell–cell contacts. The expression of the gene encoding claudin 1 (Cldn1), the significant increase in Sprr1a expression observed after major component of the epithelial tight junctions, have V treatment may be consistent with the increase in also been reported to be associated with tumorigenesis in ductal branching that we have previously reported with V human breast (Myal et al. 2010). We observed, for the first (El Sheikh Saad et al. 2011). SPRR1A is mostly found in the time, the presence of KRT17 in myoepithelial cells (Fig. 3) TEBs and ducts (epithelial cells) of the mouse mammary and its significant increase by V (also upregulated by GV, gland (Morris et al. 2006). This protein has also been

described as an axonal growth and guidance protein, dependent (Liu et al. 1997, Siegel & Muller 2010), also suggesting a possible role in ductal growth and morpho- agrees with the increases in milk protein levels. Finally, at genesis in the developing mammary gland (Morris et al. PND50, we found a significant increase in Elf5 expression 2006). Interestingly, at PND35, the modulation of claudin with V and GV treatment (Table 3) in 80–90% animals 1 (epithelial cells) and cytokeratin 17 (myoepithelial cells) (not shown). ELF5 is a transcription factor that mediates wasmostlyobservedinestrus(Fig.4),suggestinga prolactin signaling (Brisken & Rajaram 2006). It has been coordinated regulation; in addition, the modulation of reported that overexpression of Elf5 in an inducible Sprr1a was observed mostly in diestrus. Further studies are transgenic model causes alveolar differentiation and required to clarify these findings. The expression of milk secretion in virgin mice and disrupt ductal morpho- Cyp17A encoding 17a-hydroxylase/17,20-lyase has been genesis (Oakes et al. 2008). Altogether, differences in gene shown to be modulated in fetal testis by V and prochoraz expression in the mammary gland observed at PND50 (another fungicide with anti-androgen activity), resulting following in utero exposure to V or GV are consistent with in changes to androgen metabolism (Laier et al. 2006). the major histological modifications observed in post- The modulation of genes involved in steroidogenesis pubertal mammary glands (increased glandular area and by V at PND35 (see Supplementary Table S2) may reflect lobuloalveolar development, and possibly focal branching changes in estrogen and androgen balance during fetal defects). In line of our findings, Wang et al. (2006) have development. observed marked feminization of the adult male mammary gland following in utero and onward dietary exposure to the mixture of G and methoxychlor (a pesticide having V and GV increase the expression of genes involved in estrogenic and anti-androgenic activities); prominent ductal side branching and lobuloalveolar development proliferation of both ducts and alveoli; secretory material in virgin cycling animals (PND50) was seen in alveolar lumen, which is normally absent in In the developing mammary gland of cycling animals, untreated male mammary glands (Wang et al. 2006). short ‘tertiary’ side branches of the ductal epithelium form Further studies are required to analyze the complex in response to progesterone, and lobuloalveolar structures mechanism of action of endocrine-active chemicals that develop at the end of the tertiary branches (Atwood et al. may act as agonists or antagonists through one or more 2000, Haslam et al. 2008, El Sheikh Saad et al. 2011). We hormone receptors. found striking differences in gene expression in the There are few studies on the effect of the estrous cycle Journal of Endocrinology mammary gland between PND35 and PDN50 following on mammary gland development. These studies have exposure to V or GV, consistent with the major histo- concerned the effects of the estrous cycle on proliferation logical modifications observed in postpubertal mammary in rats, and conflicting results have been observed glands exposed to V and GV (El Sheikh Saad et al. 2011). (Schedin et al. 2000, Hvid et al. 2011, 2012). This study Indeed, increased glandular area and lobuloalveolar also examined the influence of the estrous cycle on a development were observed after exposure to GV and V number of important markers involved in mammary at PND50 (El Sheikh Saad et al. 2011). First, a subset of gland morphogenesis in young virgin female rats (5 and genes (Actn2, Casq1, Mylk2, Tnni2, Pvalb, and Pygm) for 7 weeks respectively). Changes were observed for a few which expression levels were strongly upregulated by V genes (Cldn1, Krt17, and Sprr1a) at PND35 according to the and GV (as shown using RT-PCR) in cycling animals cycle (as mentioned earlier); most of the genes showed was related to muscle development (Tables 3 and changes during the estrus cycle at PND50, such as those Supplementary Table S3). Our results again suggest that concerning Csn2, Wap, and Fabp3, demonstrating the changes in the actomyosin cytoskeletal architecture of importance of evaluating within a stage for other studies. myoepithelial cells may occur at PND50, depending on b-Casein (Csn2) and whey acidic protein (Wap) are the treatment. The postpubertal significant increase in classical markers of mammary gland differentiation and Wap, Csn2, and Fabp3 by V, and GV is consistent with the widely used as indicators of reduced breast cancer risk. The presence of secretion vacuoles in the alveolar structures higher mRNA expression of these two genes at PND50 in observed at PND50 from animals exposed to V and GV the mammary gland of GV group relative to G or V alone (El Sheikh Saad et al. 2011), suggesting an early terminal may suggest a protective role of maternal exposure to G in differentiation of alveolar cells after V and GV exposure. the presence of V. This is also supported by lower Ki67 The increase in STAT5A activation (as shown by nuclear expression during estrous in GV-exposed animals. Experi- immunostaining in Fig. 3), which is mainly lactation ments using DMBA-induced carcinogenesis are ongoing.

Mechanisms underlying the effects of V has been recently reported throughout the mammary gland of postpubertal mice (12 weeks) (Peters et al. 2011), The mechanisms underlying the patterns of gene whereas a decrease in mammary epithelial cell prolifer- regulation observed at PND35 and PND50 after exposure ation and gland development has been reported in to G, V, and GV during gestation and lactation remain to prepubertal AR knockout mice (Yeh et al. 2003). These be explored. V is an anti-androgenic compound that can findings indicate differential effects of an anti-AR antagon- be metabolized to generate two ER agonists, M1 and M2 ist in the peri- and postpubertal murine mammary gland; (Molina-Molina et al. 2006). Several fungicides have been in mature glands, flutamide promotes the proliferation of shown to have anti-androgenic activity through breast epithelial cells, by allowing estrogen signaling to interaction with ARs. Some xenoestrogens found in food predominate, through interactions between AR and ER contaminants present in the human diet may also have (i.e. between estrogen and androgen signaling) (Peters anti-androgenic actions, especially at low doses (Sohoni & et al. 2009). In our study, exposure during gestation/ Sumpter 1998, Andersen et al. 2002, Lemaire et al. 2004), lactation to a mixture of GV results in both significant which might modify normal mammary gland develop- alterations of mammary gland development in female ment and growth. The action of androgens and their offspring at peri-puberty and lobuloalveolar hyperplasia in antagonists in normal mammary gland morphogenesis cycling rats, consistent with differences in gene profiles remains poorly understood. It has been shown that at PND35 and PND50. These findings also suggest that growth retardation (reduced duct branching at prepuber- the molecular mechanisms underlying the effects of V in tal stages and lower levels of lobuloalveolar development the mammary gland are highly dependent on the stage of in adults) is observed in female mice lacking AR (Yeh et al. mammary gland development (peri- or postpuberty), 2003). Retained nipples in male rats treated with V have possibly with an anti-androgen effect of V on selected suggested that androgens may play an important role in genes at PND35 and a mixed anti-androgen/estrogenic early sex differentiation of the mammary gland (Gray et al. effect of GV in cycling animals (i.e. in the presence of 1994, 1999). Cooperation of signaling pathways between significant levels of E2). epithelium and mesenchyme in embryonic or in the postnatal mammary gland development have been addressed in several reviews (Robinson 2007, Su et al. Conclusion 2011). The AR is abundantly expressed in normal Early hormonal disturbances (in utero and/or during Journal of Endocrinology mammary epithelium, but not detected in mammary lactation) with effects on the mammary gland of the stroma or myoepithelium (Somboonporn & Davis (2004) offspring have been described for the combination and data not shown). The gene expression profile was of G and the fungicide (anti-androgen) V. Detailed gene determined in the whole mammary gland. Thus, the profiling analysis identified new pathways involved in the changes in gene expression observed in our study can be development of the mammary gland after exposure to related to modifications in the biology of the epithelial or this mixture of G and V in utero and during lactation, stromal component (such as adipocytes, fibroblasts, and extending our previous results (El Sheikh Saad et al. 2011). immune cells). Changes in lipid metabolism observed The effects of the anti-androgen V on the mammary gland from the V-exposed group (Supplementary Table S2) may depend on the period considered (peri- or postpubertal). occur in the adipose tissue and changes in the immune- These studies, based on exposure during gestation and related genes (see GEO) may be a reflection of modifi- lactation, demonstrate for the first time that V and GV cations in the number or function of infiltrated immune may be active on the mammary fetal anlage in females and cells. All these cell types have an essential role in the that these compounds may modify mammary gland development of the mammary gland. These results suggest development under hormonal action. Our results suggest that V exposure may alter epithelium–stroma interaction a continuum of development from fetal to pubertal stages. during mammary gland development. Androgens also Finally, this study highlights the health risks associated inhibit estrogen-induced mammary epithelial prolifer- with the presence of products in the environment, even at ation and breast growth (Somboonporn & Davis 2004, low doses, more particularly the potentially adverse effects Dimitrakakis & Bondy 2009), which can be reversed by of these compounds on the mammary gland during anti-androgens (flutamide), suggesting that AR-dependent embryonic development. Further studies are required to mechanisms are involved. An increase in cellular prolifer- assess the potential effect of these products on human ation and branching in response to flutamide treatment breast cancer.

Christiansen S, Scholze M, Dalgaard M, Vinggaard AM, Axelstad M, Supplementary data Kortenkamp A & Hass U 2009 Synergistic disruption of external This is linked to the online version of the paper at http://dx.doi.org/10.1530/ male sex organ development by a mixture of four antiandrogens. JOE-12-0395. Environmental Health Perspectives 117 1839–1846. Daniel CW, Silberstein GB & Strickland P 1987 Direct action of 17b-estradiol on mouse mammary ducts analyzed by sustained release Declaration of interest implants and steroid autoradiography. Cancer Research 47 6052–6057. The authors declare that there is no conflict of interest that could be Dimitrakakis C & Bondy C 2009 Androgens and the breast. Breast Cancer perceived as prejudicing the impartiality of the research reported. Research 11 212. (doi:10.1186/bcr2413) Durando M, Kass L, Piva J, Sonnenschein C, Soto AM, Luque EH & Munoz-de-Toro M 2007 Prenatal bisphenol A exposure induces Funding preneoplastic lesions in the mammary gland in Wistar rats. Environmental This study was supported by INSERM, CNRS, and Region Ile-de-France. Health Perspectives 115 80–86. (doi:10.1289/ehp.9282) H El S S and A T received a fellowship from Region Ile-de-France (France). Durban EM, Medina D & Butel JS 1985 Comparative analysis of casein The funding source had no involvement in any step in the conduct of the synthesis during mammary cell differentiation in collagen and research, preparation of the article, in analysis and interpretation of data, mammary gland development in vivo. Developmental Biology 109 and in the decision to submit the paper. The authors declare they have no 288–298. (doi:10.1016/0012-1606(85)90456-7) competing financial interest. El Sheikh Saad H, Meduri G, Phrakonkham P, Berges R, Vacher S, Djallali M, Auger J, Canivenc-Lavier MC & Perrot-Applanat M 2011 Abnormal peripubertal development of the rat mammary gland following exposure in utero and during lactation to a mixture of genistein and the Author contribution statement food contaminant vinclozolin. Reproductive Toxicology 32 15–25. M P-A conceived and designed the experiments. H El S S performed the (doi:10.1016/j.reprotox.2011.03.001) experiments. H El S S, A T, S V, and I B analyzed the data. M P-A wrote Enoch RR, Stanko JP, Greiner SN, Youngblood GL, Rayner JL & Fenton SE the paper. H El S S and I B involved in critical reading of the manuscript. 2007 Mammary gland development as a sensitive end point after acute prenatal exposure to an atrazine metabolite mixture in female Long–Evans rats. Environmental Health Perspectives 115 541–547. (doi:10.1289/ehp.9612) Acknowledgements Eustache F, Mondon F, Canivenc-Lavier MC, Lesaffre C, Fulla Y, Berges R, The authors thank F Dumont and F Letourneur for the microarray Cravedi JP, Vaiman D & Auger J 2009 Chronic dietary exposure to a hybridization (platform analysis, Hoˆ pital Cochin, Paris), R Berge` sfor low-dose mixture of genistein and vinclozolin modifies the reproduc- technical teams at INRA Dijon, especially. They also thank MC Canivenc- tive axis, testis transcriptome, and fertility. Environmental Health Lavier (INRA, Dijon, France), D Vaiman, J Auger, and F Mondon (INSERM U Perspectives 117 1272–1279. 567, Hoˆ pital Cochin, Paris) for their advice (PNRPE 2006), and N Peyri Fenton SE 2006 Endocrine-disrupting compounds and mammary gland (INSERM U965, Paris). They thank Dr D Medina, Baylor College of Medecine, development: early exposure and later life consequences. Endocrinology Houston TX, USA, for the gift of rabbit polyclonal antibody against b-casein 147 S18–S24. (doi:10.1210/en.2005-1131) (Durban et al. 1985). Foster WG, Younglai EV, Boutross-Tadross O, Hughes CL & Wade MG 2004 Journal of Endocrinology Mammary gland morphology in Sprague–Dawley rats following treatment with an organochlorine mixture in utero and neonatal genistein. Toxicological Sciences 77 91–100. (doi:10.1093/toxsci/kfg247) References Friedrichs N, Steiner S, Buettner R & Knoepfle G 2007 Immunohisto- chemical expression patterns of AP2alpha and AP2gamma in the Andersen HR, Vinggaard AM, Rasmussen TH, Gjermandsen IM & developing fetal human breast. Histopathology 51 814–823. Bonefeld-Jorgensen EC 2002 Effects of currently used pesticides in assays (doi:10.1111/j.1365-2559.2007.02887.x) for estrogenicity, androgenicity, and aromatase activity in vitro. Toxicology Gray LE Jr, Ostby JS & Kelce WR 1994 Developmental effects of an and Applied Pharmacology 179 1–12. (doi:10.1006/taap.2001.9347) environmental antiandrogen: the fungicide vinclozolin alters sex Atwood CS, Hovey RC, Glover JP, Chepko G, Ginsburg E, Robison WG & differentiation of the male rat. Toxicology and Applied Pharmacology 129 Vonderhaar BK 2000 Progesterone induces side-branching of the ductal 46–52. (doi:10.1006/taap.1994.1227) epithelium in the mammary glands of peripubertal mice. Journal of Gray LE Jr, Ostby J, Monosson E & Kelce WR 1999 Environmental Endocrinology 167 39–52. (doi:10.1677/joe.0.1670039) antiandrogens: low doses of the fungicide vinclozolin alter sexual Badger TM, Ronis MJ, Simmen RC & Simmen FA 2005 Soy protein isolate differentiation of the male rat. Toxicology and Industrial Health 15 48–64. and protection against cancer. Journal of the American College of Nutrition (doi:10.1191/074823399678846646) 24 146S–149S. Gye MC 2003 Expression of claudin-1 in mouse testis. Archives of Andrology Bieche I, Tozlu S, Girault I & Lidereau R 2004 Identification of a three-gene 49 271–279. (doi:10.1080/713828170) expression signature of poor-prognosis breast carcinoma. Molecular Haslam SZ, Drolet A, Smith K, Tan M & Aupperlee M 2008 Progestin- Cancer 3 37. (doi:10.1186/1476-4598-3-37) regulated luminal cell and myoepithelial cell-specific responses in Brisken C & Rajaram RD 2006 Alveolar and lactogenic differentiation. mammary organoid culture. Endocrinology 149 2098–2107. Journal of Mammary Gland Biology & Neoplasia 11 239–248. (doi:10. (doi:10.1210/en.2007-1398) 1007/s10911-006-9026-0) Hilakivi-Clarke L, Cho E, Onojafe I, Raygada M & Clarke R 1999a Maternal Bursztyka J, Debrauwer L, Perdu E, Jouanin I, Jaeg JP & Cravedi JP 2008 exposure to genistein during pregnancy increases carcinogen-induced Biotransformation of vinclozolin in rat precision-cut liver slices: mammary tumorigenesis in female rat offspring. Oncology Reports 6 comparison with in vivo metabolic pattern. Journal of Agricultural and 1089–1095. Food Chemistry 56 4832–4839. (doi:10.1021/jf0728045) Hilakivi-Clarke L, Onojafe I, Raygada M, Cho E, Skaar T, Russo I & Clarke R Cameron HL & Foster WG 2009 Developmental and lactational exposure to 1999b Prepubertal exposure to zearalenone or genistein reduces dieldrin alters mammary tumorigenesis in Her2/neu transgenic mice. mammary tumorigenesis. British Journal of Cancer 80 1682–1688. PLoS ONE 4 e4303. (doi:10.1371/journal.pone.0004303) (doi:10.1038/sj.bjc.6690584)

Hoover RN, Hyer M, Pfeiffer RM, Adam E, Bond B, Cheville AL, Colton T, organization of the mouse mammary gland. Biology of Reproduction 65 Hartge P, Hatch EE, Herbst AL et al. 2011 Adverse health outcomes in 1215–1223. women exposed in utero to diethylstilbestrol. New England Journal of Mi H, Dong Q, Muruganujan A, Gaudet P, Lewis S & Thomas PD 2010 Medicine 365 1304–1314. (doi:10.1056/NEJMoa1013961) PANTHER version 7: improved phylogenetic trees, orthologs and Hvid H, Ekstrom CT, Vienberg S, Oleksiewicz MB & Klopfleisch R 2011 collaboration with the Gene Ontology Consortium. Nucleic Acids Identification of stable and oestrus cycle-independent housekeeping Research 38 D204–D210. (doi:10.1093/nar/gkp1019) genes in the rat mammary gland and other tissues. Veterinary Journal Molina-Molina JM, Hillenweck A, Jouanin I, Zalko D, Cravedi JP, 190 103–108. (doi:10.1016/j.tvjl.2010.09.002) Fernandez MF, Pillon A, Nicolas JC, Olea N & Balaguer P 2006 Hvid H, Thorup I, Sjogren I, Oleksiewicz MB & Jensen HE 2012 Mammary Steroid receptor profiling of vinclozolin and its primary metabolites. gland proliferation in female rats: effects of the estrous cycle, pseudo- Toxicology and Applied Pharmacology 216 44–54. (doi:10.1016/j.taap. pregnancy and age. Experimental and Toxicologic Pathology 64 321–332. 2006.04.005) (doi:10.1016/j.etp.2010.09.005) Monosson E, Kelce WR, Lambright C, Ostby J & Gray LE Jr 1999 Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U & Peripubertal exposure to the antiandrogenic fungicide, vinclozolin, Speed TP 2003 Exploration, normalization, and summaries of high delays puberty, inhibits the development of androgen-dependent density oligonucleotide array probe level data. Biostatistics 4 249–264. tissues, and alters androgen receptor function in the male rat. (doi:10.1093/biostatistics/4.2.249) Toxicology and Industrial Health 15 65–79. Jager R, Schafer S, Hau-Liersch M & Schorle H 2010 Loss of transcription Moral R, Wang R, Russo IH, Mailo DA, Lamartiniere CA & Russo J 2007 factor AP-2gamma/TFAP2C impairs branching morphogenesis of the The plasticizer butyl benzyl phthalate induces genomic changes in rat murine mammary gland. Developmental Dynamics 239 1027–1033. mammary gland after neonatal/prepubertal exposure. BMC Genomics 8 (doi:10.1002/dvdy.22239) 453. (doi:10.1186/1471-2164-8-453) Jeschke J, Van Neste L, Glockner SC, Dhir M, Calmon MF, Deregowski V, Moral R, Wang R, Russo IH, Lamartiniere CA, Pereira J & Russo J 2008 Effect Van Criekinge W, Vlassenbroeck I, Koch A, Chan TA et al. 2012 of prenatal exposure to the endocrine disruptor bisphenol A on Biomarkers for detection and prognosis of breast cancer identified by a mammary gland morphology and gene expression signature. Journal of functional hypermethylome screen. Epigenetics 7 701–709. Endocrinology (doi:10.4161/epi.20445) 196 101–112. (doi:10.1677/JOE-07-0056) Kelce WR, Lambright CR, Gray LE Jr & Roberts KP 1997 Vinclozolin and Morris JS, Stein T, Pringle MA, Davies CR, Weber-Hall S, Ferrier RK, Bell AK, p,p0-DDE alter androgen-dependent gene expression: in vivo confir- Heath VJ & Gusterson BA 2006 Involvement of axonal guidance mation of an androgen receptor-mediated mechanism. Toxicology and proteins and their signaling partners in the developing mouse Applied Pharmacology 142 192–200. (doi:10.1006/taap.1996.7966) mammary gland. Journal of Cellular Physiology 206 16–24. (doi:10.1002/ Korach KS, Couse JF, Curtis SW, Washburn TF, Lindzey J, Kimbro KS, jcp.20427) Eddy EM, Migliaccio S, Snedeker SM, Lubahn DB et al. 1996 Estrogen Moumen M, Chiche A, Cagnet S, Petit V, Raymond K, Faraldo MM, receptor gene disruption: molecular characterization and experi- Deugnier MA & Glukhova MA 2011 The mammary myoepithelial cell. mental and clinical phenotypes. Recent Progress in Hormone Research International Journal of Developmental Biology 55 763–771. (doi:10.1387/ 51 159–186 (discussion 186–158). ijdb.113385mm) Korde LA, Wu AH, Fears T, Nomura AM, West DW, Kolonel LN, Pike MC, Munoz-de-Toro M, Markey CM, Wadia PR, Luque EH, Rubin BS, Hoover RN & Ziegler RG 2009 Childhood soy intake and breast cancer Sonnenschein C & Soto AM 2005 Perinatal exposure to bisphenol-A risk in Asian American women. Cancer Epidemiology, Biomarkers & alters peripubertal mammary gland development in mice. Endocrinology Journal of Endocrinology Prevention 18 1050–1059. (doi:10.1158/1055-9965.EPI-08-0405) 146 4138–4147. (doi:10.1210/en.2005-0340) Laier P, Metzdorff SB, Borch J, Hagen ML, Hass U, Christiansen S, Myal Y, Leygue E & Blanchard AA 2010 Claudin 1 in breast tumorigenesis: Axelstad M, Kledal T, Dalgaard M, McKinnell C et al. 2006 revelation of a possible novel "claudin high" subset of breast cancers. Mechanisms of action underlying the antiandrogenic effects of Journal of Biomedicine & Biotechnology 2010 956897. (doi:10.1155/2010/ the fungicide prochloraz. Toxicology and Applied Pharmacology 213 956897) 160–171. (doi:10.1016/j.taap.2005.10.013) Nikolsky Y & Bryant J 2009 Protein networks and pathway analysis. Preface. Lamartiniere CA, Cotroneo MS, Fritz WA, Wang J, Mentor-Marcel R & Methods in Molecular Biology 563 v–vii. Elgavish A 2002 Genistein chemoprevention: timing and mechanisms Oakes SR, Naylor MJ, Asselin-Labat ML, Blazek KD, Gardiner-Garden M, of action in murine mammary and prostate. Journal of Nutrition 132 Hilton HN, Kazlauskas M, Pritchard MA, Chodosh LA, Pfeffer OL et al. 552S–558S. 2008 The Ets transcription factor Elf5 specifies mammary alveolar cell Lehraiki A, Messiaen S, Berges R, Canivenc-Lavier MC, Auger J, Habert R & fate. Genes and Development 22 581–586. (doi:10.1101/gad.1614608) Levacher C 2011 Antagonistic effects of gestational dietary Pei RJ, Sato M, Yuri T, Danbara N, Nikaido Y & Tsubura A 2003 Effect of exposure to low-dose vinclozolin and genistein on rat fetal germ cell prenatal and prepubertal genistein exposure on N-methyl-N-nitro- development. Reproductive Toxicology 31 424–430. (doi:10.1016/j. sourea-induced mammary tumorigenesis in female Sprague–Dawley reprotox.2010.12.005) rats. In Vivo 17 349–357. Lemaire G, Terouanne B, Mauvais P, Michel S & Rahmani R 2004 Effect of Peters AA, Buchanan G, Ricciardelli C, Bianco-Miotto T, Centenera MM, organochlorine pesticides on human androgen receptor activation in Harris JM, Jindal S, Segara D, Jia L, Moore NL et al. 2009 Androgen vitro. Toxicology and Applied Pharmacology 196 235–246. (doi:10.1016/j. receptor inhibits estrogen receptor-alpha activity and is prognostic in taap.2003.12.011) breast cancer. Cancer Research 69 6131–6140. (doi:10.1158/0008-5472. Liu X, Robinson GW, Wagner KU, Garrett L, Wynshaw-Boris A & CAN-09-0452) Hennighausen L 1997 Stat5a is mandatory for adult mammary gland Peters AA, Ingman WV, Tilley WD & Butler LM 2011 Differential effects development and lactogenesis. Genes and Development 11 179–186. of exogenous androgen and an androgen receptor antagonist in the (doi:10.1101/gad.11.2.179) Endocrinology Lydon JP, DeMayo FJ, Funk CR, Mani SK, Hughes AR, Montgomery CA Jr, peri- and postpubertal murine mammary gland. 152 Shyamala G, Conneely OM & O’Malley BW 1995 Mice lacking 3728–3737. (doi:10.1210/en.2011-1133) progesterone receptor exhibit pleiotropic reproductive abnormalities. Raymond K, Cagnet S, Kreft M, Janssen H, Sonnenberg A & Glukhova MA Genes and Development 9 2266–2278. (doi:10.1101/gad.9.18.2266) 2011 Control of mammary myoepithelial cell contractile function Markey CM, Luque EH, Munoz De Toro M, Sonnenschein C & Soto AM by alpha3beta1 integrin signalling. EMBO Journal 30 1896–1906. 2001 In utero exposure to bisphenol A alters the development and tissue (doi:10.1038/emboj.2011.113)

Robinson GW 2007 Cooperation of signalling pathways in embryonic preventive actions of dietary factors. Journal of Nutritional Biochemistry mammary gland development. Nature Reviews. Genetics 8 963–972. 22 605–611. (doi:10.1016/j.jnutbio.2010.09.008) (doi:10.1038/nrg2227) Vandenberg LN, Mafﬁni MV, Wadia PR, Sonnenschein C, Rubin BS & Russo IH & Russo J 1978 Developmental stage of the rat mammary gland as Soto AM 2007 Exposure to environmentally relevant doses of determinant of its susceptibility to 7,12-dimethylbenz[a]anthracene. the xenoestrogen bisphenol-A alters development of the fetal Journal of the National Cancer Institute 61 1439–1449. mouse mammary gland. Endocrinology 148 116–127. (doi:10.1210/ Schedin P, Mitrenga T & Kaeck M 2000 Estrous cycle regulation of en.2006-0561) mammary epithelial cell proliferation, differentiation, and death in the Wang XJ, Bartolucci-Page E, Fenton SE & You L 2006 Altered mammary Sprague–Dawley rat: a model for investigating the role of estrous gland development in male rats exposed to genistein and methoxy- cycling in mammary carcinogenesis. Journal of Mammary Gland Biology chlor. Toxicological Sciences 91 93–103. (doi:10.1093/toxsci/kfj120) & Neoplasia 5 211–225. (doi:10.1023/A:1026447506666) Warri A, Saarinen NM, Makela S & Hilakivi-Clarke L 2008 The role of early Siegel PM & Muller WJ 2010 Transcription factor regulatory networks in life genistein exposures in modifying breast cancer risk. British Journal of mammary epithelial development and tumorigenesis. Oncogene 29 Cancer 98 1485–1493. (doi:10.1038/sj.bjc.6604321) 2753–2759. (doi:10.1038/onc.2010.43) Wu AH, Yu MC, Tseng CC & Pike MC 2008 Epidemiology of soy exposures Sohoni P & Sumpter JP 1998 Several environmental oestrogens are also and breast cancer risk. British Journal of Cancer 98 9–14. (doi:10.1038/sj. anti-androgens. Journal of Endocrinology 158 327–339. (doi:10.1677/joe. bjc.6604145) 0.1580327) Xie L, Law BK, Aakre ME, Edgerton M, Shyr Y, Bhowmick NA & Moses HL Somboonporn W & Davis SR 2004 Testosterone effects on the breast: 2003 Transforming growth factor beta-regulated gene expression in a implications for testosterone therapy for women. Endocrine Reviews mouse mammary gland epithelial cell line. Breast Cancer Research 5 25 374–388. (doi:10.1210/er.2003-0016) R187–R198. (doi:10.1186/bcr640) Sternlicht MD, Kouros-Mehr H, Lu P & Werb Z 2006 Hormonal and local Yeh S, Hu YC, Wang PH, Xie C, Xu Q, Tsai MY, Dong Z, Wang RS, Lee TH & control of mammary branching morphogenesis. Differentiation 74 Chang C 2003 Abnormal mammary gland development and growth 365–381. (doi:10.1111/j.1432-0436.2006.00105.x) retardation in female mice and MCF7 breast cancer cells lacking Su Y, Shankar K, Rahal O & Simmen RC 2011 Bidirectional signaling of androgen receptor. Journal of Experimental Medicine 198 1899–1908. mammary epithelium and stroma: implications for breast cancer – (doi:10.1084/jem.20031233)

Received in ﬁnal form 1 November 2012 Accepted 16 November 2012 Accepted Preprint published online 16 November 2012 Journal of Endocrinology