Abstract
Although soy foods have been recognized as an excellent source of protein, there have been recent concerns regarding potential adverse effects of isoflavone phytochemicals found in soy products, which are known to bind and activate estrogen receptors. Here, we used global hepatic gene expression profiles in ovariectomized female Sprague–Dawley rats treated with 17β-estradiol (E2) or fed with soy protein isolate (SPI) as a means of estimating potential estrogenicity of SPI. Female Sprague–Dawley rats were fed AIN-93G diets containing casein (CAS) or SPI starting at postnatal day (PND) 30. Rats were ovariectomized on PND 50 and infused with E2 or vehicle in osmotic pumps for 14 d. Microarray analysis was performed on liver using Affymetrix GeneChip Rat 230 2.0. Serum E2 levels were within normal ranges for the rat and SPI feeding did not increase uterine wet weight in the absence or presence of E2. SPI feeding altered (P<0.05, ≥±1.5-fold) the expression of 82 genes, while E2 treatment altered 892 genes. Moreover, only 4% of E2-affected genes were also modulated by SPI, including some whose expression was reversed by SPI feeding. The interaction between E2 and SPI uniquely modulated the expression profile of 225 genes including the reduction of those involved in fatty acid biosynthesis or glucocorticoid signaling and an induction of those involved in cholesterol metabolism. The different hepatic gene signatures produced by SPI feeding compared with E2 and the lack of increase in uterine wet weight in rats fed with SPI suggest that SPI is not estrogenic in these tissues.
Introduction
Soy is a popular dietary constituent in Asian countries such as Japan, Korea, and China. Lately, there has been an increasing use of soy products in western diets as the result of purported health benefits. There is epidemiological evidence suggesting that soy consumption reduces the risk of hormonal cancers, such as breast, ovarian, and prostate cancers, in addition to chronic diseases such as obesity, atherosclerosis, diabetes, and renal disease (Clair & Anthony 2005, Goetzl et al. 2007, Velasquez & Bhathena 2007). Mechanistic laboratory studies suggest that soy-associated phytochemicals (>137; Fang et al. 2004), and soy protein, independently or synergistically, may be responsible for the plethora of beneficial effects associated with soy consumption. Some of the phytochemicals associated with soy such as isoflavones, saponins, and phytosterols have endocrine actions (Munro et al. 2003). The isoflavones – genistein and daidzein, as well as the daidzein metabolite equol – have molecular structures similar to endogenous estrogens and bind to estrogen receptors (ER) α and β (Setchell 2001). They have been shown to be estrogenic, anti-estrogenic, or partial agonists depending upon the tissue, cell type, isoflavone concentration, and other conditions, such as age and hormonal status (Barkhem et al. 1998, Patisaul et al. 2001, Hwang et al. 2006). Due to this property, they are often referred to as ‘selective ER modulators (SERMs)’ (Setchell 2001). There has been considerable interest in the estrogenic activity of diets that contain soy isoflavones with regard to potential for alleviation of post-menopausal symptoms. However, the same property has raised legitimate concerns regarding estrogenization in infants fed soy formula, males, and pre-pubertal females (Munro et al. 2003, McLachlan et al. 2006).
Isoflavones occur in soy protein isolate (SPI) and other soy products as polar β-glycoside conjugates, such as genistin and daidzin. Bacterial hydrolysis of these isoflavones, in the GI tract, releases the principal bioactive aglycones, genistein, and daidzein. These aglycones and bacterial metabolites such as the daidzein metabolite equol are absorbed from the intestinal tract and conjugated in both the intestinal mucosa and the liver mainly to glucuronides and sulfates (Ronis et al. 2006); only 1–5% of the aglycones remain unconjugated. Fermented soy products such as miso and tempeh, may contain a higher concentration of aglycones (Barrett 2006, Rozman et al. 2006). The route and concentration of exposure to isoflavones, the animal model, soy processing, gut microflora, and endogenous estrogenic status are among many factors that play crucial roles in determining health outcomes of soy consumption. Soy foods have a long history in Asia of being safe, and evidence continues to mount regarding the health benefits of soy consumption. By contrast, data generated from purified isoflavones, used at pharmaceutical concentrations and/or injected into rodents, suggest a risk of reproductive dysfunctions including endometrial and uterine hyperplasia in females and reduced sperm count in males (Delclos et al. 2001, Nagao et al. 2001, Unfer et al. 2004).
Previous observations from our laboratory suggested reduced carcinogenicity and atherosclerosis in animals fed soy diets (Badger et al. 2005, Singhal et al. 2008a). Moreover, we have reported no compromise of cognitive function in infants fed soy formula (Jing et al. 2008). Nonetheless, the safety concerns regarding potential estrogenicity of soy foods cannot be ignored. The present study examines the question – is SPI estrogenic in the liver? The objective was to compare the effect of SPI diets with actions of the endogenous estrogen, 17β-estradiol (E2). Mature ovariectomized female Sprague–Dawley rats were fed with a diet containing SPI and followed by E2 infusion in half of the rats. The effects of feeding SPI in the presence or absence of E2 were compared with E2 using uterine wet weight and hepatic gene expression profile.
Materials and Methods
Animal care and experiment design
The experiment received prior approval from the Institutional Animal Care and Use Committee at UAMS. Adult female Sprague–Dawley rats were purchased from Charles River Laboratories and were fed diets made with casein (CAS) or SPI. Semi-purified diets were made according to the AIN-93G formula (Reeves et al. 1993) except that corn oil replaced soybean oil and the protein source was either CAS or SPI (Singhal et al. 2007). The SPI diet had 430 mg total isoflavones/kg containing 276 mg/kg genistein and 132 mg/kg daidzein. Average daily consumption of genistein and daidzein were 19.3 mg/kg per day and 9.2 mg/kg per day respectively. Total isoflavone concentrations in 24-h urine pools was ∼40 μmol/l (Ronis et al. 2001).
Rats were fed SPI or CAS diets (N=12 in each group) from postnatal day (PND) 30 until PND 64. On PND 50, rats were ovariectomized. Half of each group was subcutaneously infused with E2 and half with polyethylene glycol vehicle (Sigma) using Alzet 2002 mini-osmotic pumps (Alza Corp., Mountain View, CA, USA) that were calibrated to release 0.5 μl/h for 14 d to produce an E2 dose of 5 μg/kg per day. On PND 64, all the rats were anesthetized with Nembutal (100 mg/kg, i.p.) followed by decapitation. Blood, liver, and uterus were collected and frozen at −80 °C until analysis. Treatment groups were designated as: 1) Control, CAS diet; 2) E2, E2 treatment on CAS background; 3) SPI, SPI diet; and 4) SPI+E2, E2 treatment on SPI background.
Isoflavone and E2 analyses
Serum isoflavones (total serum isoflavones) were measured by LC–MS or electrochemical detection after conjugate hydrolysis, as described in detail by Gu et al. (2006). Serum E2 levels were determined by ultra-sensitive RIA, using DSL-4800 kit (Diagnostic Systems Laboratories, Webster, TX, USA) following the manufacturer's protocol. The kit has a lower detection limit of 2.2 pg/ml of E2.
Microarray preparation, normalization, and data analysis
Microarray preparation and data analysis were performed following minimum information about microarray experiments – supportive relational database (Brazma et al. 2001). Total RNA was isolated and cleaned using ∼100 mg hepatic tissue using TRI reagent (Molecular Research Center Inc, Cincinnati, OH, USA) from three rats in each treatment group, as described previously (Singhal et al. 2008b). First- and second-strand cDNA synthesis, biotin-labeled cRNA synthesis, fragmentation of cRNA, and hybridization reactions were performed using one-cycle cDNA synthesis kit (Affymetrix Inc, Santa Clara, CA, USA). Briefly, 8 μg purified RNA was used to synthesize cDNA. Labeled cRNA was synthesized from cDNA using a GeneChip IVT labeling kit (Affymetrix) according to the manufacturer's instructions. Twenty micrograms cRNA was then fragmented in a solution of 5× fragmentation buffer and RNase for 35 min. Complementary RNAs from each rat liver (n=3) were hybridized to an individual Affymetrix GeneChip Rat genome 230 2.0. for 16 h at 45 °C in the hybridization oven set at 60 r.p.m. The probe array was washed and stained using Affymetrix kit in GeneChip fluidics station 450 and scanned using GeneChip Scanner 3000. For each of the 31 099 genes on the Affymetrix Rat genome 230 2.0 array, the data on induction or repression values were analyzed using GeneChip Operating Software obtained from Affymetrix.
The data files (.CEL files) containing the probe-level intensities were processed using the robust multiarray analysis algorithm (GeneSpring 7.3×, Agilent Technologies Inc., Wilmington, MA, USA) for background correction, normalization, and log2 transformation (Irizarry et al. 2003). Subsequently, the data were subjected to per-chip and per-gene normalization using GeneSpring normalization algorithms. A list of differentially expressed genes was generated by performing one-way ANOVA (P<0.05; Welch) analysis followed by Benjamini and Hochberg false discovery rate and Student–Newman–Keuls multiple testing correction. This list was used to make comparisons between various treatments: SPI versus control; E2 versus control; and SPI+E2 versus control, criteria: >+1.5 or <−1.5 (i.e. ±1.5), and P<0.05. This list of differentially expressed genes was used to evaluate the pattern of gene expression profile by hierarchical clustering, using GeneSpring software or Cluster 2.1.1 and ‘Tree View’ version 1.60 software supplied by Eisen Lab, Stanford University (http://rana.lbl.gov/EisenSoftware.htm). The top networking pathways, canonical functions, and top molecular functions were determined by Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems, Inc, Redwood City, CA, USA).
Microarray validation by real-time PCR
Total RNA (1 μg) was reverse-transcribed using iSCRIPT cDNA synthesis kit (Bio-Rad) following the manufacturer's instructions. cDNA samples were amplified using previously described conditions (Singhal et al. 2008a). RT-PCR was performed on nine randomly chosen genes, some of them were of interest and discussed in the results and discussion section. Expression levels of genes were normalized to Gapdh gene levels.
Statistical analysis
Statistical analysis was performed using Sigma Stat software package (Systat Software Inc, San Jose, CA, USA). Data were analyzed by two-way ANOVA followed by Student–Newman post hoc test and were considered significant if P<0.05. Differences between treatment groups in microarray data were analyzed by ‘fold changes in volcano plot’ (GeneSpring Software) and changes were considered significant at ±1.5-fold and P<0.05 followed by Benjamini and Hochberg multiple testing correction for false discovery rate.
Results
Organ weight and serum E2 levels
E2 infusion in ovariectomized rats for 14 d resulted in a reduction (P<0.05) in body weight (BW) while the relative liver and wet uterus weights (organ weight normalized to the BW) were increased (P<0.05), as compared with the non-E2 infused groups fed with either CAS or SPI diets, n=6. SPI feeding, in the presence or absence of E2 infusion, did not have any effect on total BW or relative liver and uterus weights. Serum E2 levels were higher (P<0.05) in the group infused with E2 compared with non-E2 groups; no effect of SPI feeding was observed on the serum E2 levels. Moreover, no interactions, on either parameter mentioned here, were observed between SPI and E2, statistically determined by two-way ANOVA (Table 1).
Relative organ weights and serum estradiol (E2) levels of rats with different treatments
| Body weight | % Liver weight | % Uterus weight | E2 (pg/ml) | |
|---|---|---|---|---|
| Groups | ||||
| CAS | 291±9.1a | 4.3±0.14a | 0.07±0.009a | 5.0±1.01a |
| CAS+E2 | 244±6.6b | 5.0±0.23b | 0.26±0.017b | 27.9±4.9b |
| SPI | 289±6.4a | 4.0±0.11a | 0.07±0.003a | 3.2±0.26a |
| SPI+E2 | 227±3.6b | 4.6±0.05b | 0.25±0.016b | 30.4±6.2b |
% Liver and uterus weight are relative to body weight. Serum estradiol (E2) levels were determined by RIA as described in Materials and Methods. Means (n=6) with different letters differ significantly, P<0.05.
Serum isoflavones
When fed with SPI the total isoflavone concentrations (aglycone+conjugates) of genistein, daidzein, equol, and glycetin were 0.4±0.081, 0.26±0.053, 0.24±0.7, 0.04±0.005 μg/ml respectively. SPI feeding in the presence of E2 increased (P<0.05, n=6) the levels of equol to 0.57±0.07 μg/ml while genistein, daidzein, and glycetin were 0.26±0.063, 0.16±0.025, and 0.02±0.002 μg/ml respectively. Two-way ANOVA suggested an interaction between SPI and E2 resulting in an increase in equol levels. No effect of E2 on the summed total serum isoflavone levels, 1.27±0.19 μg/ml by SPI feeding in the absence of E2 and 1.35±0.24 μg/ml by SPI feeding in the presence of E2, was observed (Table 2).
Serum isoflavone levels. Total isoflavone and metabolites in the sera of OVX-female rats (n=6) fed with diets containing soy protein isolate (SPI) or casein (CAS) as sole protein source with or without estradiol (E2) supplementation. Data are mean±s.e.m.
| Genistein | Daidzein | Equol | Glycetin | Total isoflavone | |
|---|---|---|---|---|---|
| Groups | |||||
| CAS | 0 | 0 | 0 | 0 | 0 |
| E2 | 0 | 0 | 0 | 0 | 0 |
| SPI | 0.40±0.08 | 0.26±0.05 | 0.24±0.07 | 0.04±0.01 | 1.27±0.19 |
| SPI+E2 | 0.26±0.06 | 0.16±0.03 | 0.57±0.07*,†,‡ | 0.03±0.0 | 1.35±0.24 |
Since no isoflavones were detected in treatment groups CAS and E2, comparisons were made only between SPI and SPI+E2 treatment groups. *,†,‡represent significant effect of – SPI diet, E2 supplementation, and interaction between SPI and E2 respectively, P<0.05.
Hierarchical clustering
The list of ‘differentially expressed genes’ contained 1160 genes with known biological functions. Correlation-based unsupervised hierarchical clustering analysis was performed on the treatment (x-axis) and gene expression type (y-axis). The E2 treated groups – E2 (CAS+E2) and SPI+E2 – clustered together, while the control (CAS) and SPI clustered together, suggesting greater common effects of E2 treatment on hepatic genes irrespective of the diets. Using pseudogene lines, the heat-map was divided into four sub-clusters: I – genes induced by E2; II – genes induced by SPI; III – genes repressed by E2; and IV – genes repressed by SPI (Fig. 1A).


Differential signatures of SPI, E2, and SPI+E2 on hepatic gene expression. (A) Hierarchical cluster analysis of SPI altered (±1.5-fold, P<0.05) hepatic genes in the presence or absence of E2. The heat map was generated from the differentially expressed gene list in GeneSpring software. Using pseudolines, the heat map was divided into four clusters – 1) genes induced by E2; 2) genes induced by SPI; 3) genes repressed by E2; and 4) genes repressed by SPI. Colors – orange, yellow, and blue represent upregulation, no relative effect, and downregulation of hepatic genes respectively. (B) Venn diagram on the differentially expressed genes (described in Materials and Methods).
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059

Differential signatures of SPI, E2, and SPI+E2 on hepatic gene expression. (A) Hierarchical cluster analysis of SPI altered (±1.5-fold, P<0.05) hepatic genes in the presence or absence of E2. The heat map was generated from the differentially expressed gene list in GeneSpring software. Using pseudolines, the heat map was divided into four clusters – 1) genes induced by E2; 2) genes induced by SPI; 3) genes repressed by E2; and 4) genes repressed by SPI. Colors – orange, yellow, and blue represent upregulation, no relative effect, and downregulation of hepatic genes respectively. (B) Venn diagram on the differentially expressed genes (described in Materials and Methods).
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059
Differential signatures of SPI, E2, and SPI+E2 on hepatic gene expression. (A) Hierarchical cluster analysis of SPI altered (±1.5-fold, P<0.05) hepatic genes in the presence or absence of E2. The heat map was generated from the differentially expressed gene list in GeneSpring software. Using pseudolines, the heat map was divided into four clusters – 1) genes induced by E2; 2) genes induced by SPI; 3) genes repressed by E2; and 4) genes repressed by SPI. Colors – orange, yellow, and blue represent upregulation, no relative effect, and downregulation of hepatic genes respectively. (B) Venn diagram on the differentially expressed genes (described in Materials and Methods).
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059
Global hepatic gene signature of SPI versus E2
E2 treatment resulted in significant modulation (fold change – induction or repression by more than 1.5-fold; and P<0.05) of the expression profile of 892 genes; 188 upregulated and 704 downregulated. The top three networks with maximum number of genes affected, as identified by GeneSpring and IPA, were: metabolism and transport; cellular growth and proliferation; and cell cycle. Early genes responsive to stress – leukemia inhibitory factor receptor (Lifr, 43-fold), nephroblastoma overexpressed (Nov, 10.6-fold), and early growth response-1 (Egr1, fivefold) genes expression – were highly induced. The E2 treatment has been associated with altered hepatic glucose homeostasis, improved insulin sensitivity, reduction in obesity, and lipid metabolism (Boverhof et al. 2004, Gao et al. 2006). Similar to other investigators, we observed a modulation in genes involved in glucose homeostasis and insulin sensitivity, including an upregulation in insulin-like growth factor-binding protein 1 (Igfbp-1 and -2) and insulin receptor-related receptor (Insrr) by 4- and 1.6-fold, while insulin receptor substrate 3 (Irs3) and insulin induced gene 2 (Insig2) were downregulated by 3.7- and 1.8-fold respectively. Genes involved in fatty acid metabolism and transport such as carnitine acetyltransferase (Crat), fatty acid desaturase3 (Fad3), fatty acid-binding protein 2 (Fabp2), Cd36, carnitine palmitoyltransferase 1 (Cpt1), and hydroxysteroid 17β dehydrogenase 3 (Hsd7β3), were upregulated. A list of genes altered by E2 treatment is included in online supplementary material (OSM Table 1, see Supplementary data in the online version of the Journal of Endocrinology at http://joe.endocrinology-journals.org/content/vol202/issue1/).
Functional role of estrogens in protection of pre-menopausal women against inflammatory and fibrogenic hepatic diseases has been established (Shimizu & Ito 2007). The mechanisms, however, are not well known. Transforming growth factor-β (Tgfβ) plays a crucial role in hepatic fibrogenesis (Parsons et al. 2007) and estrogens have been known to reduce CCl4-induced liver fibrosis in rats (Xu et al. 2002). Here, we found that E2 supplementation resulted in reduced Tgfβ1 gene expression. IPA identified an enrichment in downregulation of genes encoding the entire TGF1 network – TGF-receptor, TGF-inducible homeobox-1, matrix metallopeptidase 2 (Mmp2), tissue inhibitor of metalloproteinase-3 (Timp3), and Smad-2/3 were down-regulated (OSM Figure 1A, see Supplementary data in the online version of the Journal of Endocrinology at http://joe.endocrinology-journals.org/content/vol202/issue1/).
SPI feeding to the mature female rats in the absence of endogenous estrogens resulted in the modification in the expression profile of 82 genes (Venn diagram Fig. 1B). Among the 46 upregulated genes, many belong to metabolic processes. For example, glucose metabolism included phosphatidyl inositol 3-kinase (Pi3ka) and early hepatic insulin-responsive gene -Eiih genes; xenobiotic metabolism included glutaredoxin1 (Glrx1) and glutathione-s-transferase (Gst) genes; and fatty acid metabolism including the gene coding for CD36. The downregulated genes included genes coding for estrogen metabolism, sulfotransferase (Estsul/Ste) and transcriptional regulator, Cbp/p300-interacting transactivator (Cited2) and hairy and enhancer of split 1 (Hes1). A complete list of genes altered by SPI treatment is provided in OSM Table 2, see Supplementary data in the online version of the Journal of Endocrinology at http://joe.endocrinology-journals.org/content/vol202/issue1/. Networking of genes altered by SPI feeding is represented in OSM Figure 1B.
Analysis of genes altered by both SPI and E2 revealed 39 genes in common (Venn diagram Fig. 1B). Correlation-based hierarchical cluster analysis (OSM Figure 2, see Supplementary data in the online version of the Journal of Endocrinology at http://joe.endocrinology-journals.org/content/vol202/issue1/) of these 39 genes suggested that not all of the genes altered by SPI were in the same direction as E2. Expression of four genes downregulated by ≥1.5-fold by E2 treatment viz. cysteine- and glycine-rich protein 2 (Csrp2), cathepsin C (Ctsc), inositol (myo)-1 monophosphatase 2 (Impa2), and carnitine O-octanoyltransferase (Crot), were upregulated by SPI feeding by ≥1.5-fold, while E2 upregulated arrestin domain-containing 2 (Arrdc2) gene was downregulated by SPI, suggesting whether SPI has estrogen-like actions or anti-estrogenic actions is dependent upon the individual gene and cannot be generalized. Some genes highly upregulated by E2 treatment – Lifr (43-fold) and Cd36 (9-fold) – were increased following SPI feeding only by 2.5- and 2-fold respectively; while, HLA-B-associated transcript 1A (Bat1) and Cyp2C13, genes highly downregulated by E2 treatment by 8.1- and 5.5-fold respectively, were repressed by SPI feeding by only 5.3- and 3.8-fold respectively (Table 3).
Functional characterization and fold changes of hepatic genes common to estradiol (E2) and soy protein isolate (SPI)
| Gene symbol | Gene title | GO biological process term | E2-fold change | SPI-fold change | |
|---|---|---|---|---|---|
| RefSeq | |||||
| NM_031048 | Lifr | Leukemia inhibitory factor receptor | Stress response | 43.27 | 2.41 |
| NM_031561 | Cd36 | cd36 antigen | Fatty acid transport | 8.90 | 1.99 |
| NM_031649 | Klrg1 | Killer cell lectin-like receptor subfamily G, member 1 | Cell surface receptor linked signal transduction | 3.71 | 1.50 |
| NM_022258 | C44 | α-1-B glycoprotein | – | 2.36 | 1.54 |
| NM_024129 | Dcn | Decorin | Extracellular matrix organization | 2.26 | 2.03 |
| NM_001013083 | Cpa2 | Carboxypeptidase A2 | Proteolysis | 2.05 | 1.76 |
| NM_017094 | Ghr | GH receptor | Growth | 1.98 | 1.60 |
| NM_053781 | Akr1b7 | Aldo-keto reductase family 1, member B7 | Cellular lipid metabolic process | 1.80 | 1.82 |
| XM_224720 | Arrdc2 | Arrestin domain-containing 2 | – | 1.74 | 0.63 |
| NM_053549 | Vegfb | Vascular endothelial growth factor B | Angiogenesis | 1.73 | 1.63 |
| NM_130408 | Cyp26a1 | Cytochrome P450, family 26, subfamily A, polypeptide 1 | Electron transport | 1.59 | 1.88 |
| XM_227117 | Pcdh18 | Protocadherin 18 | Homophilic cell adhesion | 1.53 | 1.55 |
| NM_053019 | Avpr1a | Arginine vasopressin receptor 1A | Regulation of blood pressure by vasopressin | 1.51 | 1.52 |
| XM_343169 | Cfd | Complement factor D | Innate immune response | 1.58 | 1.55 |
| NM_053322 | Pom210 | Nuclear pore membrane glycoprotein 210 | Protein targeting | 1.62 | 1.56 |
| NM_001025137 | Ier5 | Immediate early response 5 | – | 1.62 | 1.56 |
| NM_024360 | Hes1 | Hairy and enhancer of split 1 | Transcription | 1.69 | 1.88 |
| NM_177425 | Csrp2 | Cysteine- and glycine-rich protein 2 | Multicellular organismal development | 1.70 | 1.71 |
| NM_017097 | Ctsc | Cathepsin C | – | 1.72 | 1.50 |
| NM_016991 | Adra1b | Adrenergic receptor, α-1b | – | 1.81 | 1.59 |
| NM_133560 | Trak2 | Trafficking protein, kinesin-binding 2 | Protein targeting | 1.86 | 1.79 |
| NM_012883 | Estsul | Sulfotransferase, estrogen preferring | Estrogen metabolic process | 1.87 | 1.70 |
| NM_031154 | Gstm3 | Glutathione S-transferase, μ-type 3 | Metabolic process | 1.87 | 1.51 |
| XM_214217 | Tdh | l-threonine dehydrogenase | Cellular metabolic process | 1.88 | 2.23 |
| NM_001009536 | Coil | Tripartite motif protein 25 | – | 1.88 | 1.68 |
| NM_145089 | Asrgl1 | Asparaginase-like 1 | Glycoprotein catabolic process | 2.01 | 1.57 |
| NM_053698 | Cited2 | Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain, 2 | Transcription | 2.06 | 1.99 |
| NM_012533 | Cpb1 | Carboxypeptidase B1 | Proteolysis | 2.07 | 1.69 |
| NM_031005 | Actn1 | Actinin, α-1 | Regulation of apoptosis | 2.10 | 1.56 |
| NM_172224 | Impa2 | Inositol (myo)-1(or 4)-monophosphatase 2 | Signal transduction | 2.22 | 1.55 |
| NM_212507 | Ltb | Lymphotoxin B | Immune response | 2.28 | 2.34 |
| NM_023965 | Gp91-phox | Cytochrome b-245, β-polypeptide | Electron transport | 2.35 | 1.52 |
| XM_341700 | Cotl1 | Coactosin-like 1 | – | 2.35 | 1.50 |
| NM_031987 | Crot | Carnitine O-octanoyltransferase | Fatty acid transport | 2.36 | 1.73 |
| NM_138510 | Akr1c18 | Aldo-keto reductase family 1, member C18 | Progesterone metabolic process | 2.75 | 2.30 |
| NM_001003711 | Jph4 | Junctophilin-4 | – | 2.91 | 2.04 |
| NM_001008831 | Rt1-Ba | RT1 class II, locus Ba | Immune response | 3.57 | 3.08 |
| NM_138514 | Cyp2c13 | Cytochrome P450 2c13 | Electron transport | 5.49 | 3.72 |
| NM_133300 | Bat1 | HLA-B-associated transcript 1A | – | 8.20 | 5.32 |
Genes changed by the interaction between E2 and SPI
Expression profile of 225 genes was modulated (P<0.05, 1.5-fold) by the interaction between SPI and E2, not by either treatment alone (Venn diagram Fig. 1B). IPA performed on these genes identified lipid metabolism, carbohydrate metabolism, and cellular proliferation and development as the highly influenced biological functions. The common key regulators of fatty acid metabolism, carbohydrate metabolism, and glucocorticoid signaling were down-regulated. For example, sterol regulatory-binding protein factor 1 gene, Srepf1/Srep1, a transcription factor involved in sterol biosynthesis and fatty acid metabolism and glucocorticoid receptor (Nr3c1/GR) were downregulated by 1.9- and 1.5-fold. Consequently, the SREBP-1-responsive genes, fatty acid synthase (Fasn, 2.5-fold), steroyl CoA desaturase (Scd, 2.5-fold), acetyl CoA carboxylase-α, (Acaca, 1.8-fold), ATP citrate lyase (Acly, 1.9-fold), and fatty acid-binding protein 5 (Fabp5, 2.1-fold) involved in fatty acid metabolism; and glucokinase (Gck, 1.94-fold), glucose-6-phosphatase (G6pc, 1.6-fold), pyruvate kinase (Pklr, 1.6-fold) involved in carbohydrate metabolism were downregulated (values in bracket indicate fold reduction). The GCK-regulatory protein, Gckr, was reduced by 1.54-fold. Transcription factors – CCAAT/enhancer-binding protein (C/EBPα and -β) – co-regulated by Srebp and GR were downregulated by 1.5-fold. Gene coding for farensyl-X-receptor, Fxr, a transcription factor involved in the cholesterol metabolism, and regulated by CEBPα was downregulated by 1.6-fold, resulting in the increased expression of the FXR-repressed gene – Cyp7a1, involved in the catabolism of cholesterol to bile acids, by 2.8-fold (Fig. 2). A more detailed classification of gene expression changes as based on known biological functions, derived from NetAffx and GeneSpring programs, is represented in Table 4.


Ingenuity Pathway Analysis (IPA) gene network. The highest significant gene network identified in the Ingenuity Pathway Analysis of the gene changed by the interaction between SPI feeding and E2 treatments. Genes are colored according to the log ratio gene expression values. Colors green and red represent downregulation and upregulation respectively. The solid and dotted arrows indicate direct and indirect interaction respectively, between the two genes as known in the literature and detected by IPA.
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059

Ingenuity Pathway Analysis (IPA) gene network. The highest significant gene network identified in the Ingenuity Pathway Analysis of the gene changed by the interaction between SPI feeding and E2 treatments. Genes are colored according to the log ratio gene expression values. Colors green and red represent downregulation and upregulation respectively. The solid and dotted arrows indicate direct and indirect interaction respectively, between the two genes as known in the literature and detected by IPA.
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059
Ingenuity Pathway Analysis (IPA) gene network. The highest significant gene network identified in the Ingenuity Pathway Analysis of the gene changed by the interaction between SPI feeding and E2 treatments. Genes are colored according to the log ratio gene expression values. Colors green and red represent downregulation and upregulation respectively. The solid and dotted arrows indicate direct and indirect interaction respectively, between the two genes as known in the literature and detected by IPA.
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059
Functional characterization and fold changes of genes altered only in the presence of both soy protein isolate (SPI) and estradiol (E2)a
| Gene symbol | Gene title | GO biological process term | Fold change | |
|---|---|---|---|---|
| RefSeq | ||||
| XM_343823 | Serpina7 | Serine (or cysteine) peptidase inhibitor, clade A, member 7 | Response to drug | 7.20 |
| XM_215985 | Rapgef4 | Rap guanine nucleotide exchange factor (GEF) 4 | Insulin secretion | 3.55 |
| XM_221074 | Abca8b | ATP-binding cassette, subfamily A (ABC1), member 8b | Transport | 3.24 |
| NM_012942 | Cyp7a1 | Cytochrome P450, family 7, subfamily a, polypeptide 1 | Cholesterol metabolic process | 2.78 |
| NM_138826 | Mt1a | Metallothionein 1a | Cellular metal ion homeostasis | 2.76 |
| NM_017136 | Sqle | Squalene epoxidase | Cholesterol metabolic process | 2.61 |
| NM_134329 | Adh7 | Alcohol dehydrogenase 7 (class IV), μ- or σ-polypeptide | Retinoid metabolic process | 2.08 |
| NM_031594 | P2rx4 | Purinergic receptor P2X, ligand-gated ion channel 4 | Transport | 2.03 |
| XM_342340 | Pla2g12a | Phospholipase A2, group XIIA | Lipid catabolic process | 1.99 |
| NM_080886 | Sc4mol | Sterol-C4-methyl oxidase-like | Fatty acid metabolic process | 1.96 |
| NM_013058 | Id3 | Inhibitor of DNA-binding 3, dominant negative helix-loop-helix protein | Negative regulation of transcription from RNA polymerase II promoter | 1.94 |
| NM_012847 | Fnta | Farnesyltransferase, CAAX box, α | Transforming growth factor β-receptor signaling pathway | 1.91 |
| NM_019256 | P2rx7 | Purinergic receptor P2X, ligand-gated ion channel, 7 | Transport | 1.90 |
| NM_172033 | Plekhb1 | Pleckstrin homology domain-containing, family B, member 1 | Signal transduction | 1.87 |
| NM_001008773 | Eif1a | Eukaryotic translation initiation factor 1A | Translation | 1.85 |
| NM_012503 | Asgr1 | Asialoglycoprotein receptor 1 | Endocytosis | 1.84 |
| NM_017235 | Hsd17b7 | Hydroxysteroid (17-β) dehydrogenase 7 | Steroid biosynthetic process | 1.80 |
| NM_001013212 | Snapc3 | Small nuclear RNA activating complex, polypeptide 3 | Transcription | 1.79 |
| NM_001013179 | Hes6 | Hairy and enhancer of split 6 | Regulation of transcription, DNA-dependent | 1.61 |
| NM_012576 | Gr/Nr3c1 | Glucocorticoid receptor | Glucocorticoid receptor signaling pathway | −1.50 |
| NM_017084 | Gnmt | Glycine N-methyltransferase | S-adenosylhomocysteine metabolic process | −1.53 |
| NM_012524 | Cebpa | CCAAT/enhancer-binding protein (C/EBP), α | Signal transduction | −1.53 |
| NM_013120 | Gckr | Glucokinase regulatory protein | Carbohydrate metabolic process | −1.55 |
| NM_024125 | Cebpb | CCAAT/enhancer-binding protein (C/EBP), β | Signal transduction | −1.57 |
| NM_021745 | Fxr/Nr1h4 | Nuclear receptor subfamily 1, group H, member 4 | Cholesterol metabolic process | −1.59 |
| NM_013098 | G6pc | Glucose-6-phosphatase, catalytic | Glycogen biosynthetic process | −1.61 |
| NM_012624 | Pklr | Pyruvate kinase, liver and red blood cell | Carbohydrate metabolic process | −1.61 |
| NM_012753 | Cyp17a1 | Cytochrome P450, family 17, subfamily a, polypeptide 1 | Steroid biosynthetic process | −1.85 |
| XM_213329 | Srebf1/Srebp | Sterol regulatory element-binding factor 1 | Signal transduction | −1.88 |
| NM_016987 | Acly | ATP citrate lyase | Lipid metabolic process | −1.90 |
| NM_012565 | Gck | Glucokinase | Response to glucose stimulus | −1.95 |
| NM_145878 | Fabp5 | Fatty acid-binding protein 5, epidermal | Lipid metabolic process | −2.16 |
| NM_033234 | Hbb | Hemoglobin-β chain complex | Transport | −2.17 |
| NM_012651 | Slc4a1 | Solute carrier family 4, member 1 | Transport | −2.26 |
| NM_017206 | Slc6a6 | Solute carrier family 6, member 6 | Transport | −2.35 |
| NM_017272 | Aldh1a7 | Aldehyde dehydrogenase family 1, subfamily A7 | Metabolic process | −2.40 |
| NM_017332 | Fasn | Fatty acid synthase | Fatty acid biosynthetic process | −2.45 |
| NM_001007722; NM_001013853 | Hba-a1 | Hemoglobin-α 2 chain | Transport | −2.47 |
| NM_139192 | Scd1 | Stearoyl-Coenzyme A desaturase 1 | Fatty acid biosynthetic process | −2.51 |
| XM_214935 | Sult2b1 | Sulfotransferase family, cytosolic, 2B, member 1 | Lipid metabolic process | −2.58 |
| NM_012886 | Timp3 | Tissue inhibitor of metalloproteinase 3 | Response to estrogen stimulus | −2.59 |
| XM_216194 | Dguok | Deoxyguanosine kinase | Nucleic acid metabolic process | −2.64 |
| XM_213590 | Hrasls | HRAS-like suppressor | Regulation of cell growth | −2.70 |
| NM_198776 | Hbb | β-Glo | Transport | −3.30 |
| NM_013197 | Alas2 | Aminolevulinic acid synthase 2 | Response to hypoxia | −3.31 |
| NM_012531 | Comt | Catechol-O-methyltransferase | Catecholamine metabolic process | −4.33 |
For complete listing of genes refer to Online Supporting Material.
Corroboration of the microarray results by QRTPCR for a selected group of genes is shown in (Fig. 3). Statistics performed on nine selected genes expression values from the real-time PCR suggest similar interactions to those observed in our microarray analysis.


Pattern of change in mRNA expression level, N=6; in nine selected genes, normalized with the expression level of GAPDH (internal control), determined by real-time PCR. Both real-time PCR and microarray data are represented as percent of control. Real-time PCR were analyzed by two-way ANOVA followed by Student–Newman post hoc test (P<0.05). #, $ and @ represent effects of E2, SPI, and interaction between SPI and E2 respectively.
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059

Pattern of change in mRNA expression level, N=6; in nine selected genes, normalized with the expression level of GAPDH (internal control), determined by real-time PCR. Both real-time PCR and microarray data are represented as percent of control. Real-time PCR were analyzed by two-way ANOVA followed by Student–Newman post hoc test (P<0.05). #, $ and @ represent effects of E2, SPI, and interaction between SPI and E2 respectively.
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059
Pattern of change in mRNA expression level, N=6; in nine selected genes, normalized with the expression level of GAPDH (internal control), determined by real-time PCR. Both real-time PCR and microarray data are represented as percent of control. Real-time PCR were analyzed by two-way ANOVA followed by Student–Newman post hoc test (P<0.05). #, $ and @ represent effects of E2, SPI, and interaction between SPI and E2 respectively.
Citation: Journal of Endocrinology 202, 1; 10.1677/JOE-09-0059
Discussion
Several in vivo and in vitro studies suggest that consumption of soy foods, such as soy infant formula, may have adverse effects on development, breast cancer, and reproduction, as the result of estrogenic actions of soy isoflavones (Duffy et al. 2007, Bhatia & Greer 2008, Boucher et al. 2008). These purported estrogenic effects have been attributed to soy isoflavones binding to and activating ERα and/or ERβ (Casanova et al. 1999, Beck et al. 2003). If consumption of soy foods activates estrogen signaling pathways, one would expect that animals treated with estrogens or fed diets with SPI would have similar gene expression profiles. We used two ‘bioassays’ of estrogen actions: a) uterine wet weight is a traditional and widely employed bioassay of classical estrogenicity; and b) hepatic genomic profiling in ovariectomized rats fed SPI- or CAS-containing diets, with or without restoration of E2 to physiologically relevant plasma concentrations. Our results clearly demonstrate that SPI does not elicit responses characteristic of E2, the major estrogen in women, in these tissues. On the other hand, and in agreement with findings made by others (Adlercreutz et al. 1993, Welshons et al. 2006), we observed that estrogenic status is a key factor in determining the SPI-mediated response.
Our rat studies were designed to closely model soy consumption by pre- and post-menopausal women, conditions of ‘normal’ and low E2 levels respectively. Our results clearly demonstrate that irrespective of the estrogenic status, feeding SPI-containing diets to rats does not restore OVX-mediated loss in uterine wet weight whereas E2 supplementation does. Although normally performed in immature rodents, the uterotrophic response is used as a classic bioassay for determining ERα agonists (OECD 2001, Padilla-Banks et al. 2001). Thus, current findings and our previous data (Badger et al. 2001) demonstrate that SPI feeding does not elicit classical estrogenic responses on ERα-responsive tissues such as the uterus, even when utilized as the sole protein source. By contrast, mature OVX rats fed purified equol (400 mg/kg) or injected with purified genistein (54 mg/kg) displayed mild estrogen-like uterine stimulation (Rachon et al. 2007a, Rimoldi et al. 2007). Studies utilizing purified soy isoflavones were shown to regulate gene expression similar to estrogens in female (Naciff et al. 2002) and male reproductive tissues (Naciff et al. 2004, 2005). However, other studies comparing SPI with genistein at the gene expression level in female reproductive tissues revealed a completely different set of genes expressed following SPI feeding compared with feeding purified genistein at the same levels present in SPI (Eason et al. 2005, Su et al. 2007). Thus, when the results of these published studies are considered in combination with our results, it is clear that the effects of purified soy isoflavones differ from that of SPI feeding, even when the dose of isoflavones are approximately the same.
Analysis of total serum isoflavone concentrations showed that while feeding diets in which SPI represented the sole protein source resulted in substantial concentrations of genistein, daidzein, and equol (levels of genistein equaling those of infants fed soy formula), these concentrations were not sufficient to exhibit classical estrogenic responses mediated through ERα. One interesting finding was that serum equol concentrations in SPI-fed rats treated with E2 were greater than those of vehicle-treated SPI-fed rats. Equol is the most potent isoflavone metabolite and is produced in all or most rodents and monkeys, but in only about 25% of human subjects, and is not produced in human infants (Gu et al. 2006). We have no explanation of the elevated serum equol concentrations, but since equol is formed in the gut by bacterial metabolism of daidzein, E2 may promote daidzein metabolism. This may have functional significance for women who are equol producers and who also take birth control pills or estrogen replacement therapy.
Our current weight gain data support those of others showing an attenuation of BW following E2 supplementation in OVX rats (Cooke & Naaz 2004, Rachon et al. 2007a,b). However, no such effects were observed with SPI feeding. Additionally, SPI feeding did not alter serum E2 levels. Treatment with purified genistein, however, reduces the activity of aromatase, the enzyme that metabolizes androgen to estrogens, resulting in reduced serum E2 concentrations (Adlercreutz et al. 1993). This is another example of differing actions of SPI and purified isoflavones.
It has been noted that making conclusions based on the estrogen-dependent changes in the whole genomic profile is more sensitive than relying on single bioassays such as uterine hyperplasia (Naciff et al. 2004). Previously, it was demonstrated that the gene expression profile in uterus and ovary of immature Sprague–Dawley rats fed with Purina 5001 diet (total genistein+daidzein=0.49 mg/g was completely different from low doses of ethinyl E2 (0.1 μg/kg per day for 4 d). Additionally, similar to our current data, no uterine stimulation was observed, suggesting no estrogenic effect of laboratory animal feed on the reproductive system in the immature female rat (Naciff et al. 2004). In the present study, liver was selected to compare the gene expression profile of SPI with E2. Although not involved in reproduction, liver is a highly estrogen-responsive tissue as evident from the role of estrogens in the regulation of liver regeneration, inflammation and injury, and hepatic insulin sensitivity, as well as lipid- and sterol-mediated signaling (Hall et al. 2001, Murphy & Korach 2006). The liver also predominately expresses ERα (Kuiper et al. 1997), although low levels of ERβ have been reported (Petersen et al. 1998) and serves as a relatively clean tissue to study the effect of exogenous and endogenous estrogens on classical ERα signaling. We have identified a significant alteration in the gene expression profile of 892 genes, which itself suggests that estrogens are robustly involved in hepatic signaling. Genes involved in physiological processes including lipid and cholesterol metabolism, fibrogenesis, xenobiotic metabolism, transport, inflammation, cellular proliferation, and differentiation were altered by E2 supplementation, as previously reported by other laboratories (Gao et al. 2006, 2008).
The relevance of the current study is in the comparison of gene expression signatures of SPI fed with E2-treated ovariectomized rats. SPI feeding resulted in significant alterations in the expression of only 82 genes. IPA identified SPI-altered genes that are also regulated by estrogens and progesterone, such as aquaporin1 (Aqp1), Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain, 2 (Cited2), and phosphatidylinositol 3-kinase, regulatory subunit, polypeptide 1 (Pik3r1), revealing that some E2-regulated genes are also regulated by SPI (OSM Table 2). Naciff et al. (2004) reported a downregulation of Aqp1, a gene encoding a protein that is a member of a family of membrane channel proteins which facilitate bulk water transport and aquaporins, by soy/alfalfa-based diet in immature rat uterus, which was opposite to the effect of estrogenic stimulation by a high E2 dose (Li et al. 1997). Interestingly, we also found a suppression of this gene by SPI feeding; however, no estrogenic stimulation was observed.
Further comparative analysis determined that genes common to SPI feeding and E2 treatment, 39 genes, were only 4% of the genes changed by E2, suggesting an involvement of different biologic pathways in SPI-mediated signaling. Among these 39 genes, very few were identified by PROMOSER (Halees et al. 2003), a large-scale mammalian promoter and transcription start site identification service available online. The promoter analysis performed on the selected genes identified by PROMOSER revealed that only a few of these genes possessed estrogen-responsive element (ERE), binding sites for ER, as identified by DRAGON ERE version 2.0 program (Bajic et al. 2003). Immediate early genes – Lifr (E2: 43-fold; and SPI: 2.5-fold induction) and Egr1 (E2: fivefold; and SPI: no induction) – were identified with ERE-like consensus sequences (GGTCAnnnTGACC) and half EREs (GGTCA) in the promoter element upstream -5000 of transcription start site, suggesting a direct E2-mediated transcriptional increase in these genes, while Cd36 gene (E2: ninefold; and SPI: twofold induction) lacked any ERE-like sequence. This suggests that E2- or SPI-mediated hepatic signaling does not necessarily involve classical ER/ERE pathways. Genome-wide identification of ERα-binding sites in mouse liver, using ChIP on Chip molecular technique, suggested that besides ERE a number of motifs including forkhead sites, AP-1, Sp1, bHLH, and ETS sites are important for E2-mediated hepatic gene expression (Gao et al. 2008). Our data suggest that SPI-feeding results in a unique signature in hepatic gene expression. Phytoestrogens present in SPI may change expression of a few genes by ERα recruitment to EREs or by modulating other transcription factors that mediate estrogenic responses. Hierarchical clustering revealed that SPI feeding altered the expression of five genes in the direction opposite to E2 treatment, demonstrating a SERM-like action of SPI on selective genes. Whether the effect of SPI would be the same or opposite to E2 depends solely upon the gene type and cannot be generalized. Interestingly, 225 genes were uniquely modulated by the combination of E2 plus SPI and not by either treatment alone. The biochemical and physiological effects of SPI are known to be affected by the estrogenic status (Adlercreutz et al. 1993, Welshons et al. 2006). For example, both SPI and E2 enhanced DMBA-mediated Cyp1a1 induction in ovariectomized females but this was repressed by SPI feeding in the presence of E2 (Singhal et al. 2008b). Interaction between SPI-associated phytochemicals and peptides, and E2 may alter the expression profile and binding pattern of transcription factors, co-activators, and co-repressors resulting in the modulation of gene expression. Here, we observed changes in the expression profile of many transcription factors, including RNA polymerase II (Pol II), responsible for transcriptional elongation of genes. Repression in the expression of Pol II was linked with reduced expression of many transcription factors such as Cebp-a and -b, Dbp, glucocorticoid receptor, Fxr, and Srebp1, suggesting altered metabolic profile by the combination of SPI+E2, as compared with SPI or E2 alone. This also strengthens the point that the SPI gene expression signature in: children prior to puberty, men and women, and post-menopausal women would all be predicted to differ.
Conclusion
The present data demonstrate that SPI, the sole protein source of infant formula, is not uterotrophic and it has a unique pattern of hepatic gene expression that differs substantially from that of E2. These data call into question concerns about the potential estrogenic effects of soy formulas in infants (5, 6, 12–15, 59, 60). It should be noted that the rats with access to SPI ad libitum consumed 19.3 mg/kg per day of genistein (Ronis et al. 2001), which is much higher than the average intake of genistein in US (0.014–0.14 mg/kg per day) or in Japan (0.21–0.43 mg/kg per day; Rozman et al. 2006) or infants on soy formula 4–6 mg/kg per day (Setchell et al. 1997). It is unlikely that the general population would consume sufficient amounts of phytoestrogens in the diet to cause any adverse effects. Our conclusion in liver is also supported by a series of studies performed by Naciff et al. in reproductive tissues in immature rats where soy-containing diets were not observed to be estrogenic. However, further validation of the current data by examining the SPI gene signature in estrogen-responsive reproductive tissues containing both ERα and ERβ in neonatal animals such as a piglet, which have a more similar pattern of isoflavone metabolites to humans than rodents (Gu et al. 2006) is required to completely rule out the possibility of estrogenic effects of soy infant formula.
Declaration of interest
R Singhal, K Shankar, T M Badger, and M J Ronis declare that there are no conflicts of interest.
Funding
The work presented here is supported in part by USDA (CRIS# 6251-51000-005-03S), a CAGSRF award from UAMS (R S) and a Technology Transfer Award from SCC-SOT (R S).
Acknowledgements
The authors are thankful to Drs Shanmugam Nagarajan and John C Marecki for their critical suggestions in compiling the manuscript. Online Supporting Material (OSM) is available with the online posting of this paper.
References
Adlercreutz H, Bannwart C, Wahala K, Makela T, Brunow G, Hase T, Arosemena PJ, Kellis JT Jr & Vickery LE 1993 Inhibition of human aromatase by mammalian lignans and isoflavonoid phytoestrogens. Journal of Steroid Biochemistry and Molecular Biology 44 147–153.
Badger TM, Ronis MJ & Hakkak R 2001 Developmental effects and health aspects of soy protein isolate, casein, and whey in male and female rats. International Journal of Toxicology 20 165–174.
Badger TM, Ronis MJ, Simmen RC & Simmen FA 2005 Soy protein isolate and protection against cancer. Journal of the American College of Nutrition 24 146S–149S.
Bajic VB, Tan SL, Chong A, Tang S, Strom A, Gustafsson JA, Lin CY & Liu ET 2003 Dragon ERE Finder version 2: a tool for accurate detection and analysis of estrogen response elements in vertebrate genomes. Nucleic Acids Research 31 3605–3607.
Barkhem T, Carlsson B, Nilsson Y, Enmark E, Gustafsson J & Nilsson S 1998 Differential response of estrogen receptor alpha and estrogen receptor beta to partial estrogen agonists/antagonists. Molecular Pharmacology 54 105–112.
Barrett JR 2006 The science of soy: what do we really know? Environmental Health Perspectives 114 A352–A358.
Beck V, Unterrieder E, Krenn L, Kubelka W & Jungbauer A 2003 Comparison of hormonal activity (estrogen, androgen and progestin) of standardized plant extracts for large scale use in hormone replacement therapy. Journal of Steroid Biochemistry and Molecular Biology 84 259–268.
Bhatia J & Greer F 2008 Use of soy protein-based formulas in infant feeding. Pediatrics 121 1062–1068.
Boucher BA, Cotterchio M, Kreiger N & Thompson LU 2008 Soy formula and breast cancer risk. Epidemiology 19 165–166.
Boverhof DR, Fertuck KC, Burgoon LD, Eckel JE, Gennings C & Zacharewski TR 2004 Temporal- and dose-dependent hepatic gene expression changes in immature ovariectomized mice following exposure to ethynyl estradiol. Carcinogenesis 25 1277–1291.
Brazma A, Hingamp P, Quackenbush J, Sherlock G, Spellman P, Stoeckert C, Aach J, Ansorge W, Ball CA & Causton HC et al. 2001 Minimum information about a microarray experiment (MIAME)-toward standards for microarray data. Nature Genetics 29 365–371.
Casanova M, You L, Gaido KW, Archibeque-Engle S, Janszen DB & Heck HA 1999 Developmental effects of dietary phytoestrogens in Sprague–Dawley rats and interactions of genistein and daidzein with rat estrogen receptors alpha and beta in vitro. Toxicological Sciences 51 236–244.
Clair RS & Anthony M 2005 Soy, isoflavones and atherosclerosis. Handbook of Experimental Pharmacology 170 301–323.
Cooke PS & Naaz A 2004 Role of estrogens in adipocyte development and function. Experimental Biology and Medicine 229 1127–1135.
Delclos KB, Bucci TJ, Lomax LG, Latendresse JR, Warbritton A, Weis CC & Newbold RR 2001 Effects of dietary genistein exposure during development on male and female CD (Sprague–Dawley) rats. Reproductive Toxicology 15 647–663.
Duffy C, Perez K & Partridge A 2007 Implications of phytoestrogen intake for breast cancer. CA: A Cancer Journal for Clinicians 57 260–277.
Eason RR, Till SR, Velarde MC, Geng Y, Chatman L Jr, Gu L, Badger TM, Simmen FA & Simmen RC 2005 Uterine phenotype of young adult rats exposed to dietary soy or genistein during development. Journal of Nutritional Biochemistry 16 625–632.
Fang N, Yu S & Badger TM 2004 Comprehensive phytochemical profile of soy protein isolate. Journal of Agricultural and Food Chemistry 52 4012–4020.
Gao H, Bryzgalova G, Hedman E, Khan A, Efendic S, Gustafsson JA & Dahlman-Wright K 2006 Long-term administration of estradiol decreases expression of hepatic lipogenic genes and improves insulin sensitivity in ob/ob mice: a possible mechanism is through direct regulation of signal transducer and activator of transcription 3. Molecular Endocrinology 20 1287–1299.
Gao H, Falt S, Sandelin A, Gustafsson JA & Dahlman-Wright K 2008 Genome-wide identification of estrogen receptor α-binding sites in mouse liver. Molecular Endocrinology 22 10–22.
Goetzl MA, Van Veldhuizen PJ & Thrasher JB 2007 Effects of soy phytoestrogens on the prostate. Prostate Cancer and Prostatic Disease 10 216–223.
Gu L, House SE, Prior RL, Fang N, Ronis MJ, Clarkson TB, Wilson ME & Badger TM 2006 Metabolic phenotype of isoflavones differ among female rats, pigs, monkeys, and women. Journal of Nutrition 136 1215–1221.
Halees AS, Leyfer D & Weng Z 2003 PromoSer: a large-scale mammalian promoter and transcription start site identification service. Nucleic Acids Research 31 3554–3559.
Hall JM, Couse JF & Korach KS 2001 The multifaceted mechanisms of estradiol and estrogen receptor signaling. Journal of Biological Chemistry 276 36869–36872.
Hwang CS, Kwak HS, Lim HJ, Lee SH, Kang YS, Choe TB, Hur HG & Han KO 2006 Isoflavone metabolites and their in vitro dual functions: they can act as an estrogenic agonist or antagonist depending on the estrogen concentration. Journal of Steroid Biochemistry and Molecular Biology 101 246–253.
Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U & Speed TP 2003 Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4 249–264.
Jing H, Pivik RT, Gilchrist JM & Badger TM 2008 No difference indicated in electroencephalographic power spectral analysis in 3- and 6-month-old infants fed soy- or milk-based formula. Maternal and Child Nutrition 4 136–145.
Kuiper GG, Carlsson B, Grandien K, Enmark E, Haggblad J, Nilsson S & Gustafsson JA 1997 Comparison of the ligand binding specificity and transcript tissue distribution of estrogen receptors alpha and beta. Endocrinology 138 863–870.
Li XJ, Yu HM & Koide SS 1997 Regulation of water channel gene (AQP-CHIP) expression by estradiol and anordiol in rat uterus. Yao Xue Xue Bao 32 586–592.
McLachlan JA, Simpson E & Martin M 2006 Endocrine disrupters and female reproductive health. Best Practice and Research. Clinical Endocrinology and Metabolism 20 63–75.
Munro IC, Harwood M, Hlywka JJ, Stephen AM, Doull J, Flamm WG & Adlercreutz H 2003 Soy isoflavones: a safety review. Nutrition Reviews 61 1–33.
Murphy E & Korach KS 2006 Actions of estrogen and estrogen receptors in nonclassical target tissues. Ernst Schering Foundation Symposium Proceedings 1 13–24.
Naciff JM, Jump ML, Torontali SM, Carr GJ, Tiesman JP, Overmann GJ & Daston GP 2002 Gene expression profile induced by 17alpha-ethynyl estradiol, bisphenol A, and genistein in the developing female reproductive system of the rat. Toxicological Sciences 68 184–199.
Naciff JM, Overmann GJ, Torontali SM, Carr GJ, Tiesman JP & Daston GP 2004 Impact of the phytoestrogen content of laboratory animal feed on the gene expression profile of the reproductive system in the immature female rat. Environmental Health Perspectives 112 1519–1526.
Naciff JM, Hess KA, Overmann GJ, Torontali SM, Carr GJ, Tiesman JP, Foertsch LM, Richardson BD, Martinez JE & Daston GP 2005 Gene expression changes induced in the testis by transplacental exposure to high and low doses of 17α-ethynyl estradiol, genistein, or bisphenol A. Toxicological Sciences 86 396–416.
Nagao T, Yoshimura S, Saito Y, Nakagomi M, Usumi K & Ono H 2001 Reproductive effects in male and female rats of neonatal exposure to genistein. Reproductive Toxicology 15 399–411.
OECD 2001 Third meeting of the validation management group for the screening and testing of endocrine disruptors (mammalian effects). Joint meeting of the chemicals Committee and the working party on chemicals, pesticides and biotechnology. Organization of Economic Cooperation and Development. http://www.oecd.org. Acessed October 31, 2002. Ref Type: Internet Communication..
Padilla-Banks E, Jefferson WN & Newbold RR 2001 The immature mouse is a suitable model for detection of estrogenicity in the uterotropic bioassay. Environmental Health Perspectives 109 821–826.
Parsons CJ, Takashima M & Rippe RA 2007 Molecular mechanisms of hepatic fibrogenesis. Journal of Gastroenterology and Hepatology 22 S79–S84.
Patisaul HB, Dindo M, Whitten PL & Young LJ 2001 Soy isoflavone supplements antagonize reproductive behavior and estrogen receptor alpha- and beta-dependent gene expression in the brain. Endocrinology 142 2946–2952.
Petersen DN, Tkalcevic GT, Koza-Taylor PH, Turi TG & Brown TA 1998 Identification of estrogen receptor beta2, a functional variant of estrogen receptor beta expressed in normal rat tissues. Endocrinology 139 1082–1092.
Rachon D, Vortherms T, Seidlova-Wuttke D, Menche A & Wuttke W 2007a Uterotropic effects of dietary equol administration in ovariectomized Sprague–Dawley rats. Climacteric 10 416–426.
Rachon D, Vortherms T, Seidlova-Wuttke D & Wuttke W 2007b Effects of dietary equol on body weight gain, intra-abdominal fat accumulation, plasma lipids, and glucose tolerance in ovariectomized Sprague–Dawley rats. Menopause 14 925–932.
Reeves PG, Nielsen FH & Fahey GC Jr 1993 AIN-93 purified diets for laboratory rodents: final report of the American Institute of Nutrition ad hoc writing committee on the reformulation of the AIN-76A rodent diet. Journal of Nutrition 123 1939–1951.
Rimoldi G, Christoffel J, Seidlova-Wuttke D, Jarry H & Wuttke W 2007 Effects of chronic genistein treatment in mammary gland, uterus, and vagina. Environmental Health Perspectives 115 62–68.
Ronis MJ, Rowlands JC, Hakkak R & Badger TM 2001 Inducibility of hepatic CYP1A enzymes by 3-methylcholanthrene and isosafrole differs in male rats fed diets containing casein, soy protein isolate or whey from conception to adulthood. Journal of Nutrition 131 1180–1188.
Ronis MJ, Little JM, Barone GW, Chen G, Radominska-Pandya A & Badger TM 2006 Sulfation of the isoflavones genistein and daidzein in human and rat liver and gastrointestinal tract. Journal of Medicinal Food 9 348–355.
Rozman KK, Bhatia J, Calafat AM, Chambers C, Culty M, Etzel RA, Flaws JA, Hansen DK, Hoyer PB & Jeffery EH et al. 2006 NTP-CERHR expert panel report on the reproductive and developmental toxicity of genistein. Birth Defects Research. Part B, Developmental and Reproductive Toxicology 77 485–638.
Setchell KD 2001 Soy isoflavones – benefits and risks from nature's selective estrogen receptor modulators (SERMs). Journal of the American College of Nutrition 20 354S–362S.
Setchell KD, Zimmer-Nechemias L, Cai J & Heubi JE 1997 Exposure of infants to phyto-oestrogens from soy-based infant formula. Lancet 350 23–27.
Shimizu I & Ito S 2007 Protection of estrogens against the progression of chronic liver disease. Hepatology Research 37 239–247.
Singhal R, Badger TM & Ronis MJ 2007 Reduction in 7,12-dimethylbenz[a]anthracene-induced hepatic cytochrome-P450 1A1 expression following soy consumption in female rats is mediated by degradation of the aryl hydrocarbon receptor. Journal of Nutrition 137 19–24.
Singhal R, Badger TM & Ronis MJ 2008a Rats fed soy protein isolate (SPI) have impaired hepatic CYP1A1 induction by polycyclic aromatic hydrocarbons as a result of interference with aryl hydrocarbon receptor signaling. Toxicology and Applied Pharmacology 227 275–283.
Singhal R, Shankar K, Badger TM & Ronis MJ 2008b Estrogenic status modulates aryl hydrocarbon receptor – mediated hepatic gene expression and carcinogenicity. Carcinogenesis 29 227–236.
Su Y, Simmen FA, Xiao R & Simmen RC 2007 Expression profiling of rat mammary epithelial cells reveals candidate signaling pathways in dietary protection from mammary tumors. Physiological Genomics 30 8–16.
Unfer V, Casini ML, Costabile L, Mignosa M, Gerli S & Di Renzo GC 2004 Endometrial effects of long-term treatment with phytoestrogens: a randomized, double-blind, placebo-controlled study. Fertility and Sterility 82 145–148 (quiz).
Velasquez MT & Bhathena SJ 2007 Role of dietary soy protein in obesity. International Journal of Medical Sciences 4 72–82.
Welshons WV, Nagel SC & vom Saal FS 2006 Large effects from small exposures. III. Endocrine mechanisms mediating effects of bisphenol A at levels of human exposure. Endocrinology 147 S56–S69.
Xu JW, Gong J, Chang XM, Luo JY, Dong L, Hao ZM, Jia A & Xu GP 2002 Estrogen reduces CCL4 induced liver fibrosis in rats. World Journal of Gastroenterology 8 883–887.
(R Singhal is now at Michigan State University, 222 Food Safety and Toxicology Building, East Lansing, Michigan 48824, USA)
