Online Issues

ILAR Journal Vol 45(4)

Selecting the Appropriate Rodent Diet for Endocrine Disruptor Research and Testing Studies

Julius E. Thigpen, Kenneth D. R. Setchell, H. E. Saunders, J. K. Haseman, M. G. Grant, and D. B. Forsythe

Julius E. Thigpen, Ph.D., is Head of the Quality Assurance Laboratory (QAL), Comparative Medicine Branch, National Institute of Environmental Health Sciences (NIEHS), National Institutes of Health, Research Triangle Park, North Carolina. Kenneth D. R. Setchell, Ph.D., is Director of Clinical Mass Spectrometry, Department of Pediatrics, Cincinnati Children's Hospital Medical Center, Cincinnati, Ohio. H. E. Saunders, B.S., is a Microbiologist at QAL, NIEHS. J. K. Haseman, Ph.D., is a retired Biostatistician, the Biostatistics Branch, NIEHS. M. G. Grant, V.M.D., and D. B. Forsythe, D.V.M., are Deputy Chief and Chief, respectively, of the Comparative Medicine Branch, NIEHS.

Abstract

Selecting the optimum diet for endocrine disruptor (ED) research and testing studies in rodents is critical because the diet may determine the sensitivity to detect or properly evaluate an ED compound. Dietary estrogens can profoundly influence many molecular and cellular event actions on estrogen receptors and estrogen-sensitive genes. The source, concentration, relative potency, and significance of dietary estrogens in rodent diets are reviewed, including dietary factors that focus specifically on total metabolizable energy and phytoestrogen content, which potentially affect ED studies in rodents. Research efforts to determine dietary factors in commercially available rodent diets that affect uterotrophic assays and the time of vaginal opening in immature CD-1 mice are summarized. A checklist is provided of important factors to consider when selecting diets for ED research and testing studies in rodents. Specific metabolizable energy levels are recommended for particular bioassays. Discussions include the between-batch variation in content of the phytoestrogens daidzein and genistein, the effects of total metabolizable energy and phytoestrogens on the timing (i.e., acceleration) of vaginal opening, and increased uterine weight in immature CD-1 mice. It is concluded that rodent diets differ significantly in estrogenic activity primarily due to the large variations in phytoestrogen content; therefore animal diets used in all ED studies should ideally be free of endocrine-modulating compounds.

Key Words: daidzein; endocrine disruptors; genistein; phytoestrogen; rodent diets; uterotrophic assay

Introduction

Selection of the most appropriate diet for endocrine disruptor (ED¹) research involving the use of rodents is not only a timely subject, given the current proposals to mass screen a wide range of endocrine-disrupting compounds (EDCs¹), but is also a decision that should be given serious consideration before standardizing or initiating such studies. All selected diets should be based on the principle of minimizing variables that have the potential of altering endocrine-sensitive endpoints in reproductive, comparative estrogenicity, toxicity, or carcinogenicity studies. Although many different formulations of commercially available rodent diets are in use, diets basically comprise two types: natural-ingredient diets and purified or chemically defined diets.

Natural-ingredient diets are formulated with agricultural products (e.g., ground wheat, corn, oats, and soybeans), mill by-products (e.g., wheat bran, wheat middling, and corn gluten meal), or high-protein meals (e.g., fish meal and alfalfa meal). Natural-ingredient diets are relatively inexpensive to manufacture, are palatable for most laboratory animals, and can be in the form of open- or closed-formula diets. In open-formula diets, information is provided about the ingredients and their concentrations, whereas in closed-formula diets, only the ingredients are stated and their concentrations are proprietary. There are also fixed-formula, natural-ingredient, and purified diets in which ingredients and respective concentrations do not vary from batch to batch; and these diets are also classified as either open- or closed-formula.

Purified or chemically defined diets are formulated with a set of ingredients that is more refined and restricted than those of the natural-ingredient category. Casein and soybean protein isolate are used as sources of protein, sugar and starch provide the sources of carbohydrates, vegetable oil is used as a source of fat and essential fatty acids, and cellulose is the source of fiber. These diets are open formula, and the nutrient concentrations are less variable and more easily controlled via formulation than in a natural-ingredient diet. Purified diets are more expensive to produce than the natural-ingredient diets; however, the potential for chemical contamination with pesticide or herbicide residues is much less than with natural-ingredient diets (Subcommittee on Laboratory Animal Nutrition 1995).

Diets may contain multiple sources of protein, fat, carbohydrates, minerals, and vitamins; and the bioavailability of nutrients may be lower in natural-ingredient diets than in purified diets. The bioavailability of nutrients may be influenced by their chemical form, by the presence of constituents such as phytate, tannins, and lignin that bind nutrients, by nutrient-matrix interactions, or by the effects of processing of the diet and its constituents. Many diets are formulated with a specific objective in mind, for example, for breeding and reproduction, for maximizing growth, for maintenance of adult animals, for carcinogenicity studies, or for specialized research projects. Most diets can be sterilized by autoclaving or irradiation to prevent transfer of pathogenic microorganisms to research animals. The method of sterilization may affect pellet hardness and the texture of the diet. These factors, along with diet palatability, may affect food consumption and body weights. These dietary factors are critical because they may affect total metabolizable energy (ME¹) intake, thereby potentially altering measured endpoints such as uterine weights and timing of vaginal opening (VO¹).

It is evident from the scie.jpgic literature that often the specific diet used in a study is either not evident to the investigators or is seemingly considered unimportant, as evidenced by the frequent statement that "a standard rodent diet was used." Investigators appear to be more concerned about environmental estrogenic contaminants such as pesticide residues in feed and water than the possible presence of naturally occurring estrogenic compounds in rodent diets. Many investigators seem to be unaware that most rodent diets contain the phytoestrogens daidzein and genistein (D&G¹) in variable levels. These isoflavones are found in any natural-ingredient diet that is formulated with either soybean meal or soy protein and alfalfa, which is a source of the phytoestrogen coumestrol (Price and Fenwick 1985). Furthermore, the concentration of phytoestrogens in any diet is directly correlated with the soybean content (Thigpen et al. 1992, 1999, 2001b). D&G, and the important intestinal bacterial metabolite equol (Setchell et al. 2002a), show affinity for estrogen receptors and, in particular, selectivity as a ligand for estrogen receptor (ER¹) β. When circulating levels of phytoestrogens are elevated, as is typical in rodents consuming the most commonly used commercial diets (Brown and Setchell 2001; Thigpen et al. 2003), significant effects at the genetic, molecular, and cellular levels are observed (Camper-Kirby et al. 2001; Diel et al. 2000). The effects of D&G can be more potent than the effects of environmental estrogens such as dichlorodiphenyltrichloroethane (DDT¹) and bisphenol A (BPA¹) (Shelby et al. 1996), as shown in Table 1.

Source, Variation in Concentration, Significance, and Relative Potency of Dietary Estrogens

Major sources of exogenous estrogenic substances in laboratory rodent diets include the phytoestrogens (isoflavones and coumestans), estrogenic mycotoxins (e.g., zearalenone), and contaminating estrogenic pesticide residues such as DDT (Figure 1). Phytoestrogens are nonsteroidal estrogens of plant origin. Their metabolites exert an estrogenic effect on the central nervous system, which induces estrus and stimulates cell division and growth of the genital tract of female animals (Lieberman 1996). The relative potency of phytoestrogens depends on the dose, the time, duration, and route of exposure, the animal species, strain, age, and sex, and the monitored responses (Figure 2). The isoflavones D&G, formononetin, and biochanin A, and the coumestan coumestrol, all are structurally similar to diethylstilbestrol (DES¹) and estradiol (Figure 3), and consequently show affinity for rat or mouse ERα and ERβ. Genistein shows particular specificity for ERβ (Kuiper et al. 1997; Yamasaki et al. 2002). Equol, an isoflavone metabolite formed in the intestinal tract by the action of bacteria, is especially important in the case of rodents because when rodents are fed soy, equol becomes the predominant isoflavone in adult rat and mouse plasma whereas D&G are present in relatively low proportions (Brown and Setchell 2001). Equol not only has significant affinity for ERβ (Morito et al. 2003; Muthyala et al. 2004; Setchell et al. 2002a) but also was recently shown to be a potent antiandrogen (Lund et al. 2004).

Figure 1 Schematic diagram showing the potential sources of dietary and environmental exogenous estrogens affecting animals used in endocrine disruptor studies. DDT, dichlorodiphenyltrichloroethane; PCB, polychlorinated biphenyl.

Figure 2 Schematic diagram showing factors that affect the relative potency of dietary estrogens and their impact on endocrine disruptor research and testing studies.

Figure 3 Comparative structures of the exogenous estrogens genistein, daidzein, coumestrol, and zearalenone in rodent diets to estradiol and diethylstilbesterol.

Phytoestrogens can be active at low or high concentrations, can affect different target tissues, and can act as agonists or antagonists (Figure 4). The phytoestrogens have nonhormonal properties that can influence basic cell biology due to their effects on cellular enzymes, including cytochrome P-450 (CYP) enzymes, as well as their ability to influence growth factors such as cytokines and to regulate genes and exert antioxidant actions even at low levels (Brown and Setchell 2001). Thus, even though gross macroscopic or physiological effects on reproduction may be difficult to observe, investigators using diets containing phytoestrogens may unwittingly observe effects due to their presence, especially in the case of estrogen-sensitive signaling pathways (Camper-Kirby et al. 2001). Based on published data, the recent proposal (Owens et al. 2003) that diets could be used for rat uterotrophic assays for evaluating EDCs if they contain certain threshold levels of phytoestrogens (< 325-350 µg/g of total genistein equivalent [TGE¹]) is absurd given the potency of these compounds and their ability to significantly alter molecular endpoints.

Figure 4 Schematic diagram showing the potential effects of dietary phytoestrogens on the endocrine, nervous, skeletal, and cardiovascular systems of laboratory animals, and their potential effects on endocrine disruptor studies.

It is very difficult to control the isoflavone content of a diet that contains soybean meal. Although it was reported that manufacturers processing soybeans can adequately control for protein content, there is in fact no way to control for the natural variation in the isoflavone content of soybeans (Setchell and Cole 2003), which can vary up to 5-fold (Eldridge and Kwolek 1983; Njiti et al. 1999). Differences in geographical location of the soybean environment, processing procedures, and plant genotype (Hoeck et al. 2000) all contribute to large variations in phytoestrogen levels from season to season, and year to year. Such variations are unacceptable when data clearly show that the variation in the concentration of D&G between different mill dates (Figure 5) of the same Purina Mills, Inc. (PMI¹) #5002 negative control diet produces significant (p < 0.01) differences in the time of VO in immature CD-1 mice. For example, we have shown that a diet containing 431 µg of D&G /g diet resulted in different effects on the timing of VO when compared with the same manufacturer's diet containing 159 µg of D&G/g diet, and the latter diet differed from one that had less than 10 µg of D&G/g diet (Thigpen et al. 2003). Such differences in effects were also independent of the amount of total ME among the different batches of diets, all of which were coincidentally similar (Thigpen et al. 2003) and yielded no differences in body weights among mice fed these different diets.

Figure 5 Batch-to-batch variation in total daidzein (D) and genistein (G) content versus vaginal opening (VO) data in CD-1 mice fed different mill dates of Purina Mills, Inc. (PMI) # 5002 diet. The total D & G content can vary 3-fold in different mill dates producing significant (p <0.05) differences in the time of VO between different mill dates at postnatal day 24. The low phytoestrogen diet in the chart is PMI #5K96 but similar results were seen with Harlan Sprague Dawley (HSD) 2014S, HSD 2016S, Zeigler 5412-00, and Zeigler 5412-01. Data from Thigpen JE, Haseman JK, Saunders HE, Setchell KDR, Grant MF, Forsythe DB. 2003. Dietary phytoestrogens accelerate the time of vaginal opening in immature CD-1 mice. Comp Med 53:607-615.

In recent studies, we (Thigpen et al. 2003) have shown that the PMI #5002 diet containing 431 µg/g of D&G has the same effect on the time of VO as that induced by 4 or 6 ppb of DES added to a different mill date of this diet, which contained 159 µg/g of D&G (Figure 5). The PMI #5002 diet containing 431 µg/g of D&G is in effect a positive, not a negative, control diet, as investigators may be misled to believe (Thigpen et al. 2003). Ashby (2003) monitored different batches of the PMI #5002 diet during a 3-yr period and observed that the PMI #5002 diet showed a 6-fold variation in total phytoestrogen content consistent with what can be expected based on the natural variation in isoflavone content of soybeans and soy protein isolates (Setchell and Cole 2003). A 3- to 6-fold variation in phytoestrogen content in different batches of the PMI #5002 reinforces our contention that for EDC studies, it is important to use diets that are relatively free or contain virtually undetectable levels of phytoestrogens. For isoflavones, this lower limit would equate to a diet having less than 20 ppm because only then can one be certain that there will be negligible endocrine effects of the background diet.

The mission of the National Institute of Environmental Health Sciences (NIEHS¹) is to determine the potential adverse effects of environmental chemicals on human health and wildlife. Since the mid-1970s, the NIEHS has conducted many comparative studies on estrogens. In the early 1980s, NIEHS investigators visited the Quality Assurance Laboratory (QAL¹) in the NIEHS Comparative Medicine Branch in an effort to determine why scientists were unable to duplicate data from their own uterotrophic assays, or to reproduce data from uterotrophic assays conducted in outside laboratories. As a result, the QAL began evaluating environmental, dietary, and animal factors that affect the results of estrogenic studies utilizing uterotrophic assays. Initial investigations and the review of literature focused on differences in animal factors such as animal strain, age, date of weaning, length of assay, route of exposure, and methodology of removing, trimming, blotting, and weighing the uteri and other organs. Initial results indicated that in some cases, investigators were using animals that were not only of different strains but also of different ages. Moreover, the most consistent and significant finding was that in most cases, the animals were being fed vastly different diets. Therefore, it was hypothesized that diet could influence the results of uterotrophic assays, and studies were undertaken to address this issue.

Uterotrophic Assays

Effects of Rodent Diets

A series of studies was conducted to determine the potential effects of rodent diets on the results of uterotrophic bioassays. First, a standard procedure for conducting the mouse bioassay was published (Thigpen et al. 1987a). As early as 1987, it was reported that commercially available rodent diets differed significantly in estrogenic activity (Thigpen et al. 1987b), as determined by their ability to increase uterine weights and uterine:body weight (U:BW¹) ratios of immature CD-1 mice used in the 7-day bioassay postnatal day (PND¹) 15 to PND 22. It was also shown that sucrose, glucose, and corn starch could stimulate uterine weight in immature CD-1 mice (Thigpen et al. 1987c). It was concluded that a diet low in estrogenic activity should be used for comparative estrogenic studies or for studies that could be influenced by the presence of exogenous estrogens. At that time, it was unclear which dietary ingredients(s) or compounds were responsible for the observed estrogenic activity of these diets. Ashby and colleagues (2000, 2001) and Odum and coworkers (2001) subsequently reported that rodent diets could advance the timing of VO and alter endocrine-related endpoints, which may not be surprising given that Brown and Setchell (2001) and we (Thigpen et al. 2003) showed that high isoflavone levels are found in the plasma and urine of animals fed the most commonly used commercial rodent diets.

Effects on Clinical Endpoints

It had been reported that a rodent diet containing large amounts of D&G (140 and 210 mg/kg, respectively) induced a near-maximal uterotrophic response in control and ovariectomized, 30-day-old Sprague Dawley rats, and altered their normal response to administered estradiol (Boettger-Tong et al. 1998). It was later reported that up to 310 µg/L of bisphenol A (BPA¹), a weak estrogen, could leach from used polycarbonate animal cages or polycarbonate water bottles into animals' drinking water (Howdeshell et al. 2003). No BPA was detected in water incubated in glass water bottles or used polypropylene cages. A study was designed to determine whether immature CD-1 mice potentially exposed to 310 µg/L (ppb) of BPA in drinking water would experience a significant increase in uterine weights compared with mice not exposed to BPA. The uterine weights of immature CD-1 mice housed in used polycarbonate cages with used polycarbonate water bottles were compared with the uterine weights of negative control mice housed in new polypropylene cages and glass water bottles from PND 19 to PND 26 (Howdeshell et al. 2003). In our opinion, this study was compromised once the mice were fed a PMI #5001 diet containing 350 to 500 µg/g (ppm) of total D&G, because D&G are more potent estrogens than BPA (Table 1). Additionally, the mice were exposed to doses of D&G from the PMI #5001 diet that exceeded the potential exposure dose of BPA (310 µg/L; ppb) from the used polycarbonate cages and used water bottles. These examples highlight why it is critical to use a phytoestrogen-free diet when conducting uterotrophic assays. There have been other examples of animals exposed to higher doses of estrogens from the diet than from the test compound (Brown and Setchell 2001).

The NIEHS QAL has historically used the PMI #5002 diet containing 0, 4, or 6 ppb of added DES as a control diet for conducting a 7-day mouse uterotrophic bioassay (PND 15 to PND 22). To conduct a valid uterotrophic assay, the uterine weights of mice fed the PMI #5002 diet with 4 or 6 ppb added DES must be significantly greater than the uterine weights of the mice fed the PMI #5002 negative control (background) diet without any added DES. This is difficult because of wide variation in the concentration of D&G in these diets (Thigpen et al. 2003). When the PMI #5002 control diet contained a 159 µg/g diet of total D&G plus 4 or 6 ppb of added DES, we were able to conduct a valid assay (which yielded a significant difference in uterine weight between the negative and positive control diets). However, when the PMI #5002 diet contained 431 µg/g diet of total D&G, which is within natural variation, and 4 or 6 ppb of DES were added, it was not possible to produce a reliable uterotrophic assay at PND 22. The problem relates to high levels of the phytoestrogens D&G in rodent diets, which reduce the sensitivity of uterotrophic assays or other endocrine-related endpoints. If the negative control diet is a phytoestrogen-reduced or phytoestrogen-free diet, then adding 4 or 6 ppb of added DES to diets such as PMI #5K96, Harlan Sprague-Dawley (HSD¹) 2014S, HSD 2016S, and Zeigler 5412-01 (Group 1, Table 2) results in reproducible and valid bioassays (Thigpen et al. 2003).

Origin of Phytoestrogen-reduced Diets

The phytoestrogen content of seven rodent diets and the six dietary ingredients (soybean meal, wheat, oats, corn, wheat middlings, and alfalfa meal) used in their formulation was ide.jpgied more than 12 yr ago (Thigpen et al. 1992). Not surprisingly, soybean meal was the primary source of the phytoestrogens D&G, and the content of D&G was directly correlated with the proportion of soybean meal in the diet. Subsequent studies of a range of commercial rodent diets have confirmed this early finding (Brown and Setchell 2001). Based on the relative potency (Table 1) of these dietary phytoestrogens (Bickoff et al. 1962), Thigpen and colleagues (1992) reported that the total concentration of D&G in the PMI #5001 diet was equal to 4.3 ppb of equivalent DES activity, which is significant because 4.0 ppb of added DES markedly increases uterine weight in the mouse uterotrophic bioassays when compared with a negative control diet (Thigpen et al. 1987b, 2002). Recognizing the potential impact of diets containing high levels of phytoestrogens inspired our proposal that diets should be formulated to reduce the concentration of phytoestrogens in rodent diets by omitting the soybean and alfalfa meals (Thigpen et al. 1998, 1999). Alfalfa meal should be omitted because it is the primary source of coumestrol, a phytoestrogen with even greater affinity for the estrogen receptor than genistein or daidzein (Patisaul et al. 1999; Whitten and Naftolin 1992) even though the percentage of alfalfa meal in diets is low compared with soybean meal.

Thigpen and coworkers (1998) described the comparative estrogenic activity of three new closed-formula natural-ingredient test diets (diets A, B, and C) formulated to reduce the concentration of phytoestrogens, and subsequently compared these diets with 12 other rodent diets (Thigpen et al. 1999). In test diets A and B, both the soybean and alfalfa meals were omitted. In test diet C, only the alfalfa meal was excluded. Estrogenic activity was assayed by measuring the ability of the diet to increase mouse U:BW ratios in 15-day-old female CD-1 mice during a 7-day mouse bioassay. The mouse U:BW ratios from test diets A and B were significantly lower than the mouse U:BW ratios from test diet C. Test diet A was lower, but not significantly lower, than test diet B. Test diet A was significantly lower than the NIH-31 (Knapka 1983), HSD Teklad LM-485-7012, and PMI #5058 diets.

The results described above confirmed that mice consuming the phytoestrogen-free test diets A and B, formulated with no soybean or alfalfa meals, had lower uterine weights gains and lower U:BW ratios than mice consuming the NIH-31, HSD Teklad LM485-7012, or PMI #5058 diets. These results, which established that the three new diets differed significantly in estrogenic activity, suggested that these three diets could not be used interchangeably for comparative estrogenic studies. These differences deserve further consideration for two reasons: (1) Test diet C had only the alfalfa removed, yet it still had a high concentration of soybean meal that was responsible for its estrogenic activity. (2) Test diets A and B were formulated with different concentrations of different dietary ingredients resulting in minor differences in estrogenic activity, total ME, and palatability.

It was clear from the results of these studies (Thigpen et al. 1998, 1999) that the future use of phytoestrogen-free diets would have a profound, global effect on improving the reliability of uterotrophic assays aimed at evaluating EDCs. The effects of low, medium, or high phytoestrogen-containing diets on the timing of VO in immature CD-1 mice are well documented and are summarized in Table 2 (Thigpen et al. 2003). As a consequence of these reports, the three main vendors of rodent feed now market phytoestrogen-free diets that are suitable for comparative estrogenic studies and are relatively economically priced. Yet these diets appear not to have been adopted in government programs aimed at testing for EDCs.

D&G as Dietary Factors

A comparison of U:BW ratios of mice fed phytoestrogen-reduced diets confirm that the dietary phytoestrogens D&G play a significant role in the estrogenic activity of rodent diets. Examples of previous reports in the literature that the animals' diet has altered the results of endocrine, as well as nonendocrine, or nonhormonal endpoints include the following: (1) Brown and Setchell (2001) reported that the anti-inflammatory effects of genistein, using the carrageenan-induced paw edema rat model of inflammation, were severely compromised by feeding the PMI #5001 rodent diet containing a high level of the phytoestrogen genistein, which reduced the overall anti-inflammatory effects of administered genistein (Salzman et al. 1999). (2) In a cardiomyopathy mouse model, Camper-Kirby and colleagues (2001) documented profound effects on the expression of Akt (serine/threonine protein kinase, also known as protein kinase B). Feeding a diet high in genistein negated the effects of the experimentally administered genistein that was being tested in one model, and altered the phenotypic expression in a gene knockout model of cardiomyopathy. It is difficult to ignore these effects, which were independent of any direct uterotrophic action.

Uterine Weights and VO

Effects of Metabolizable Energy

Studies have been conducted to determine how dietary factors such as percentage of protein, fat, carbohydrate (total ME), crude fiber, and phytoestrogen content correlate with changes in uterine weights of immature CD-1 mice. Thigpen and colleagues (1987c) evaluated the AIN-76A (Reeves et al. 1993) purified diet as the negative and positive control diets for conducting the 7-day mouse uterotrophic assay. At that time, it was thought logical to use this diet as the negative and positive control diets because the AIN-76A diet tested negative for known phytoestrogens. However, it proved difficult to produce valid bioassays when mice were fed the AIN-76A diet containing 0, 4, or 6 ppb added DES. To achieve valid assays, the uterine weights of mice fed the DES-positive control diets must be significantly increased over the uterine weights of mice fed a negative control diet. Uterine weights from mice fed the purified diet were greater than those of mice fed natural-ingredient diets. Mouse uterine weights were greater in all groups, and the wide variation in weights occurring within the groups led to nonsignificant differences. It was concluded that the high total ME (3.83 kcal/g diet) of the purified diet caused this effect. A high level of ME reduces the sensitivity of the uterotrophic mouse bioassay. The ME in purified diets is approximately 3.8 kcal/g diet. Therefore, a diet with a lower level of ME (approximately 3.1 kcal/g diet) that does not reduce the sensitivity of the assay is essential for conducting uterotrophic bioassays.

The initial conclusions about total ME described above were confirmed in a recent publication (Thigpen et al. 2002) documenting the evaluation of 20 different diets fed to 1160 mice to determine dietary factors that more highly correlate with the increase in uterine weights of immature CD-1 mice in the uterotrophic assay. Total ME in diets was found to correlate significantly with, and be more predictive of, uterine weights in immature mice (PND 15-22). Furthermore, total ME appeared to be more highly correlated with the increase in uterine weights in immature CD-1 mice than was the phytoestrogen content of the diets (Thigpen et al. 2002). Thus, ME introduces a second variable in considering the optimal diet to use for EDC testing.

Effects of Phytoestrogens

The relative importance of differences in dietary ME and phytoestrogen content in the uterotrophic assays has been examined recently by studying the effect on advancement of VO (Thigpen et al. 2002, 2003). Dietary phytoestrogens significantly influence the results of ED studies when either uterine weight or VO is used as the primary endpoint. We (Thigpen et al. 2003) recently described the effects of D&G and total ME on the time of VO based on our study of 1059 mice exposed to 21 different diets. This study established that both the phytoestrogen content and total ME were highly significantly (p < 0.0001) correlated and predictive of the time of VO. However, the phytoestrogen content was more highly predictive of the time of VO than was the total ME.

Nevertheless, it was concluded that both the phytoestrogen content and total ME should be taken into account when examining uterotrophic responses and timing of VO. Interestingly, Markey and coworkers (2001) reported that BPA advances the time of VO at doses that have no effect on uterine weights, suggesting that the time of VO is a more sensitive endpoint than the increase in uterine weights for evaluating estrogenicity. In addition, Jefferson and colleagues (2002) reported that genistein, coumesterol, and zearalenone all increased uterine weights in immature CD-1 mice. They also noted that other phytoestrogens, daidzein and biochanin A, affect hormonal endpoints such as increasing uterine gland number and cell height but without also increasing uterine weight. These results highlight the complexity of the problem facing investigators and illustrate that subtle effects on reproductive physiology can occur in the absence of overt changes in uterine weight.

Effects of Dietary Phytoestrogens on Plasma and Urine Phytoestrogen Levels

Previous reports by Brown and Setchell (2001) and ourselves (Thigpen et al. 2003) clearly demonstrate that the level of phytoestrogen in plasma and urine of rats and mice is directly correlated with the isoflavone content of the diet (Table 3). The plasma and urinary concentration of D&G also correlates with advancement in the timing of VO in immature CD-1 mice. These reports and others indicate that dietary phytoestrogens alter endocrine-related endpoints (Figure 6).

Figure 6 Schematic diagram showing the potential effects of dietary phytoestrogens on the endocrine, nervous, skeletal, and cardiovascular systems of laboratory animals, and their potential effects on endocrine disruptor studies.

Setchell and Cole (2003) examined several batches of soy foods and reported that the protein concentration varied less than 3%, whereas the phytoestrogen content varied 200 to 300%. Thus, while it is possible to control the protein content in rodent diets, it is very difficult to control the phytoestrogen content, especially when soybean meal is used as the source of protein in the diet. We therefore believe it is incumbent upon manufacturers of rodent diets to make available to investigators pertinent information regarding the level of phytoestrogens in the diet, and it is essential that scientists conduct studies to target the assessment of an EDC in the absence of phytoestrogens. If these criteria are not fulfilled, the reliability of the findings will undoubtedly be questionable.

Effects on Molecular and Morphological Endpoints

Dietary phytoestrogens, even at levels that have no observable effect on uterine weights in rats (Whitten and Patisaul 2001), can have significant effects on gene expression and regulation of ERα mRNA and ERβ mRNA in the rat brain, thus altering or modifying estrogenic and antiestrogenic effects (Patisaul et al. 1999), sexual behavior, and neurobehavioral effects (Lephart et al. 2002, 2003). The effects of dietary phytoestrogens on the reproductive system, bone, central nervous system (brain), and cardiovascular system (heart, blood pressure, and cholesterol levels) are shown in Figures 4 and 6.

Effects on Tumor Endpoints

Given the important chemopreventive (Barnes et al. 1997; Lamartiniere et al. 1995) or chemopromotive (Muthyala et al. 2004) effects of isoflavones in rodent models of breast cancer or prostate cancer, and especially the significance of early exposure to phytoestrogens, which causes desensitization to chemical carcinogens in later life (Lamartiniere et al. 1995), the selection of diet in carcinogenicity studies is of critical importance. We (Thigpen et al. 2001) conducted a study to determine the effects of dietary D&G on the incidence of spontaneous vulvar carcinomas in 129/J mice using three natural-ingredient diets and two purified diets containing predetermined levels of D&G. Within 1 mo, the incidence of vulvar carcinomas in mice fed the AIN-76A modified soy protein diet was significantly (p < 0.05) increased over those of mice fed the AIN-76A modified casein, the PMI #5K96, or the PMI #5058 diet. At 3 mo, the incidence of vulvar carcinomas in mice fed the soy protein diet was significantly (p < 0.05) increased over those of mice fed the NIH-31 or the PMI #5K96 diet.

These results confirmed that dietary D&G were associated with the increased incidence of vulvar carcinomas in 129/J mice. This study indicates that the rodent diet can readily alter the results of carcinogenic studies, especially given that early exposure to soy isoflavones such as genistein has been shown to reduce the susceptibility to mammary carcinoma induced by dimethylbenz[a]anthracene in rats (Lamartiniere et al. 1995). Generally, an investigator has no knowledge of the diet that was fed to young animals when purchased from commercial suppliers. However, reduced sensitivity to carcinogenic induction by chemicals clearly would be expected in some models if animals had prior exposure to diets with high soy levels. This is yet another example of the need for awareness of the diet used, not only for feeding of adult animals on maintenance diets but also for young animals being bred for such studies.

Implanted Tumors in Athymic Nude Mice

Ju and colleagues (2001) determined the effects of added dietary genistein (0, 125, 250, 500, or 1000 µg/g diet) on cellular proliferation of MCF-7 tumor cells implanted in athymic nude mice at a level similar to that of the estradiol-positive control group. Dietary genistein increased tumor size in a dose-dependent manner similar to that of the estradiol control group. The percentage of proliferating cells and total plasma genistein concentrations were significantly increased with increasing dietary genistein levels (250 µg/g diet), highlighting the effectiveness of phytoestrogens in influencing the results of endocrine-related endpoints and of carcinogenicity studies.

Detection of Estrogens in Diets, Plasma, and Urine

Methods for the detection and qua.jpgication of estrogenic compounds in rodent diets, plasma, and urine are summarized in Table 4 together with the relative merits of each approach. The current method of choice for the qua.jpgication of phytoestrogens in rodent diets is high-performance liquid chromatography coupled with electrospray ionization mass spectrometry (Barnes et al. 1997; Brown and Setchell 2001; Setchell et al.1987). This method is accurate, provides specific concentrations, and is considerably more economical than conducting uterotrophic assays. It does, however, preselect for the measurement and is therefore "biased" in the choice of compounds being measured. Uterotrophic assays, which are bioassays, use animals and are time consuming and expensive. Although uterotrophic assays do not provide specific "estrogen" concentrations, they do detect any estrogenic substance, whether known or unknown (e.g., phytoestrogens, mycotoxins, pesticides), and in this regard are nonselective in their assessment. More importantly, a bioassay will account for metabolic transformation of a particular compound, which is not the case for in vitro assays using estrogen receptors. With the latter assay, it is possible to miss the importance of a particular environmental agent that undergoes metabolism from an inactive to an active estrogen. For example, the major isoflavone of soybeans is the β-glycoside genistin, but this has no estrogenic activity in vitro and is not bioavailable (Setchell et al. 2002a,b). In the intestine, it is metabolized to genistein, which is absorbed and shows significant binding to the ER. Daidzin, the other major isoflavone in soybeans, is also nonestrogenic, but undergoes hydrolysis to daidzein and is then biotransformed to equol (Axelson et al. 1984; Setchell et al. 1984), a metabolite that has much greater estrogenicity (Morito 2003; Muthyala et al. 2004; Setchell et al. 2002a). These examples illustrate the drawbacks of current testing for EDCs, which are only compounded by the use of diets that contain estrogenic substances.

Thigpen and colleagues (2001a) compared the Lumi-Cell ER^TM bioassay with the uterotrophic mouse bioassay for the detection of DES, coumestrol-, and zearalenone-spiked rodent diets and found the former to be the more sensitive assay. The Lumi-Cell ER^TM bioassay is a sensitive, fast, reliable, relatively inexpensive assay for the detection of multiple estrogenic contaminants in rodent diets (Table 4); however, it still lacks the ability of accounting for in vivo metabolism of an inactive estrogenic substance to an active metabolite.

Diet Recommendations

The Organisation for Economic Cooperative Development (OECD¹) and the Environmental Protection Agency (EPA¹) have placed a high priority on developing standardized in vivo and in vitro protocols to determine the estrogenic activity or antiestrogenic activity of approximately 87,000 environmental EDCs (EDSTAC 1999, 2001; Kanno et al. 2001, 2003a,b; Owens et al. 2003). At the time this article went to press, the EPA was planning to use uterotrophic assays in intact immature and ovariectomized rats fed the PMI #5002 ce.jpgied rodent diet. The decision to select the #5002 diet was based in part on an OECD report (Kanno et al. 2001, 2003a,b; Owens et al. 2003) in which the phytoestrogen content in rodent diets from 20 laboratories was evaluated in an effort to determine the effects of dietary phytoestrogens on uterotrophic bioassays evaluating the estrogenic activity of EDCs. The objective of the combined reports was to determine the effects of the dietary phytoestrogens on the results of rat uterotrophic assays evaluating EDCs. The authors concluded that laboratories conducting the rat uterotrophic bioassay for research or regulatory purposes may routinely use diets containing levels of phytoestrogens (< 325-350 µg/g of TGE) without impairing the responsiveness of the bioassay (Owens et al. 2003).

Most investigators use mice rather than rats in estrogenic studies, and they may draw from the OECD conclusion above that it is acceptable to use high phytoestrogen diets (325-350 µg/g of TGE) for evaluation of molecular and cellular hormonal endpoints. Such statements prolong the inappropriate use of high phytoestrogen diets in ED studies. Testing of an unknown compound for potential estrogenic activity utilizing a diet that already naturally contains a high concentration of estrogenic substances appears illogical and makes no scie.jpgic sense, especially when there is no possible way to control for the level of these phytoestrogens,. There is compelling evidence from our recent studies that the only acceptable level of phytoestrogens in diets for such studies should be minimal. Trace, or undetectable, levels would be set at a content of < 20 ppm, which is close to the detection limit of most methods of analytical measurement.

Other investigators (Casanova et al. 1999a,b; Kanno et al. 2002; You et al.2002) have evaluated the use of phytoestrogen-reduced diets for estrogenic studies. Casanova and coworkers (1999a,b) evaluated the effects of the NIH-07 diet and a soy-/alfalfa-free diet (SAFD¹) on sexual development in the Sprague Dawley rat. Only one statistically significant difference was detected between groups fed the SAFD and the NIH-07: The anogenital distance (AGD¹) of female neonates on PND 1 whose dams were fed NIH-07 was 12% larger than that of neonates whose dams were fed SAFD. Their results suggest that normal amounts of phytoestrogens in natural-ingredient rodent diets may affect female AGD, and that higher doses can affect several parameters in both males and females. Despite their findings, Casanova and colleagues did not suggest replacing soy- and alfalfa-based rodent diets with phytoestrogen-free diets in most developmental toxicology studies. However, phytoestrogen-free diets were recommended for low-dose endocrine toxicology studies to determine whether interactive effects may occur between dietary phytoestrogens and synthetic chemicals (Cassanova et al. 1999b). In addition, Kanno and coworkers (2002) compared the hormonal effects of the NIH-07 diet with a modified NIH-07 open-formula phytoestrogen-low diet (PLD¹) on uterine weight and uterine luminol epithelial labeling index in ovariectomized CD rats. The uterine weight was significantly lower in rats given the PLD than in those given the NIH-07 diet. In addition, the luminol epithelial labeling index was significantly higher in rats fed the NIH-07 diet. The authors stated that phytoestrogens in the diet is a concern for sensitivity as well as specificity in highly sensitive ED experiments. They concluded that the "phytoestrogen-low" diet enhances the sensitivity and reduces the interlaboratory/interexperimental variations. Furthermore, Kanno (2003) states that for more precise endpoints such as gene expression, which could be part of the newer ED screening and testing program, a PLD diet is very important.

Recent reports (Thigpen et al. 2002, 2003) have confirmed that dietary phytoestrogens and total ME significantly alter the results of VO endpoints and uterotrophic assays in immature CD-1 mice. Yet because most of our studies (Thigpen et al. 2002, 2003) have utilized mice, comparisons with results from the aforementioned rat studies (Kanno et al. 2001, 2003a,b; Owens et al. 2003; Yamasaki et al. 2002) should be made with caution. Factors to consider when comparing results from different studies are shown in Figures 2 and 4. Owens and coworkers (2003) concluded that administration of < 325 to 350 µg/g of TGE does not impair the responsiveness of the rat uterotrophic bioassay. Their conclusions were based on uterine weights in immature and ovariectomized rats administered different doses of BPA, genistein, or methoxychlor, and fed various diets containing approximately 100 to 541 µg of phytoestrogen/g diet. Yamasaki and colleagues (2002) compared uterine weight changes in an immature Sprague Dawley rat uterotrophic assay of multiple doses of ethinylestradiol, BPA, nonylphenol, or genistein in rats given diets having different phytoestrogen content. The diets and their phytoestrogen content µg/g diet are shown in parentheses: MF diet (299.8), NIH-07 (139), and the NIH-07 low-phytoestrogen diet (31.9). They reported observing no essential differences in uterine weights among the various phytoestrogen content diets, and they concluded that low levels of phytoestrogens do not affect the sensitivity of the rat uterotrophic assay. Differences in their results and ours may be due to animal species and endpoints evaluated. For example, we have not performed any rat uterotrophic assays.

Nevertheless, recent studies (Thigpen et al. 2004) show that dietary D&G (223 µg/g diet) can significantly advance the time of VO in F344 rats. In addition, these reports (Thigpen et al. 2003, 2004) show that the batch-to-batch variation in the D&G content between different mill dates of the same PMI #5002 diet fed to F344 rats (Figure 7) or mice (Figure 5) produced significant differences in the time of VO. We have shown that D&G (200 µg/g diet) can significantly advance the time of VO in CD-1 mice (Thigpen et al. 2003). Ju and coworkers (2001) reported that dietary genistein (125 and 250 µg/g) significantly accelerated cell division and increased the size of tumors implanted into nude mice, respectively. These results indicate that diets containing less than 325 to 350 µg/g TGE have the potential to alter the results of VO endpoints and uterotrophic assays evaluating the estrogenic activity of EDCs.

Figure 7 Batch-to-batch variation in total daidzein (D) and genistein (G) content versus vaginal opening (VO) data in F344 rats fed different mill dates of Purina Mills, Inc. (PMI) # 5002 diet. The total D & G content can vary 3-fold in different mill dates producing significant (p < 0.05) differences in the time of VO between different mill dates at postnatal days 34 and 36. The low phytoestrogen diet in the chart is PMI #5K96, but similar results were seen with Harlan Sprague Dawley (HSD) 2014S and HSD 2016S.

More attention should be given to the choice of diet to be used in standardizing uterotrophic assays and in assays aimed at assessing the estrogenicity of a particular substance. For the accuracy and reproducibility of results involving the assessment of estrogenicity, it is critical to eliminate as many confounding variables as possible, and in this regard, it is clearly evident that the phytoestrogen content and the total ME of the diet are perhaps the most important confounding variables requiring consideration in present and future ED studies. It is also critical to use a "standardized diet" from which the potential effects of factors such as pellet hardness, diet palatability, and food consumption have been eliminated because these factors may alter total ME intake and consequently alter uterine weights and time of VO.

Study Factors to Consider When Selecting a Diet

We recommend careful and advance consideration of the quantitative and qualitative details of each study. The resolution of each aspect of the study will benefit researchers, institutional animal care and use members, the animals, and ultimately our understanding of endocrine disruptors. We suggest asking and answering the following questions:

What are the specific objectives of the study?
What animals will be used: species, stock/strain, sex, and age at weaning?
Does the study involve reproduction, lactation, growth, endocrine toxicity, carcinogenicity, or uterotrophic bioassays?
What are the specific endpoints (e.g., increase or decrease in tissue weights; time of vaginal opening)?
Are these endpoints influenced by exogenous estrogens including the phytoestrogens? If the endpoints are influenced by exogenous estrogens, or phytoestrogens, then the selection of a phytoestrogen-free diet becomes an important consideration. It is important to realize that dietary phytoestrogens have profound effects at the molecular and cellular level (Brown and Setchell 2001; Diel et al. 2000), which may not be obvious when considering only gross development and morphology of the genital tract.

Diet Factors to Consider for a Specific Study

Because the diet may determine the success or failure of a study, we also recommend providing accurate answers to all of the following questions before initiating a study:

Will dietary estrogens or total ME alter the endpoint(s)?
Will pellet hardness, diet palatability, and food consumption alter results?
Does the diet remove variables or add more variables to the study?
What diet(s) were the dams fed during gestation/lactation, and what are the potential effects of this diet on the fetus and on fetal genetic imprinting?
What diet did the weanlings or test animals consume before the study, and what are the potential effects on subsequent exposure to estrogens?
Will environmental exogenous estrogens such as BPA alter the endpoint(s)?
Quality control of the diet: Which estrogens should be tested for and which assay should be used for specific estrogens (Table 4)?

Conclusions

Based on the findings presented in this paper and on a review of the literature, our conclusions may be listed as follows:

Rodent diets differ significantly in estrogenic activity, due primarily to the phytoestrogen content that is directly correlated with the proportion of soybean meal in the diet.
The phytoestrogen content can vary 3- to 6-fold between different batches of the same diet producing significant differences in the time of VO in CD-1 mice and in F344 rats.
Diets high in phytoestrogens can have the same effects on the time of VO and uterine weights in mice as diets containing 4 or 6 ppb of added DES, and this factor may mask the ability to detect weak but significantly estrogenic substances.
Dietary estrogens affect the timing of sexual development in rodents and can influence outcomes of reproductive, carcinogenicity, and endocrine disruptor studies.
Dietary estrogens, including the phytoestrogens D&G, and total ME significantly accelerate the time of VO and increase uterine weights in CD-1 mice, thus altering the results of ED studies.

If dietary estrogens alter endocrine-related endpoints, then one should select a phytoestrogen-free diet defined as < 20 µg/g diet that contains the same low level of ME and consequently will not have one or more of the following effects: to reduce the sensitivity of the assay for conducting estrogen bioassays comparing the time of VO; to affect uterotrophic assays, where increases in uterine weight are measured; or to influence assays using molecular endpoints for evaluating the estrogenic or antiestrogenic activity of an EDC.

Acknowledgments

We thank Patricia Deese for excellent help in preparing the manuscript, and Frank Kari and John Roberts for their time and helpful comments during the review of the manuscript. We thank Jacqueline Locklear, Tanya Whiteside, Gordon Caviness, and Leslie Echerd for excellent technical support. The technical support of Linda Zimmer-Nechemias from the Cincinnati Children's Hospital Medical Center, in conducting the isoflavone assays of the diets and plasma, is also gratefully acknowledged.

¹Abbreviations used in this article: AGD, anogenital distance; BPA, bisphenol A; DDT, dichlorodiphenyltrichloroethane; DES, diethylstilbestrol; D&G, daidzein and genistein; ED, endocrine disruptor; EDC, endocrine-disruptor compounds; EPA, Environmental Protection Agency; ER, estrogen receptor; HSD, Harlan Sprague-Dawley; ME, metabolizable energy; NIEHS, National Institute of Environmental Health Sciences; OECD, Organisation for Economic Cooperative Development; PLD, phytoestrogen-low diet; PMI, Purina Mills, Inc.; PND, postnatal day; QAL, Quality Assurance Laboratory; SAFD, soy-/alfalfa-free diet; TGE, total genistein equivalent; U:BW, uterine:body weight; VO, vaginal opening.

References

Asby J. 2003. Comparative estrogenic effects of phytoestrogens. In: Lawton G, ed. Endocrine/Estrogen Letter. Brisbane CA: Online Enterprises. Vol 9 No 4 (176):13.

Ashby J, Odum J, Tinwell H. 2001. Predictive value of the uterotrophic assay for genistein carcinogenicity in the neonatal mouse: Relevance to infants consuming soy-based formula. Environ Health Perspect 109:568A.

Ashby J, Odum J, Tinwell H, Kimber I, Brooks AN, Pate I, Boyle CC. 2000. Diet and the aetiology of temporal advances in human and rodent sexual development. J Appl Toxicol 20:343-347.

Axelson M, Sjovall J, Gustafsson B, Setchell KDR. 1984. Soy--A dietary source of the non-steroidal estrogen equol in humans and animals. J Endocr 102:49-56.

Barnes S, Coward L, Kirk M, Sfakianos J. 1997. HPLC-mass spectrometry analysis of isoflavones. Proc Soc Exp Biol Med 217:254-262.

Bickoff EM, Livingston AL, Henderson AP, Booth AN. 1962. Forage estrogens, relative potencies of several estrogen-like compounds found in forages. J Agric Food Chem 10:410-412.

Boettger-Tong H, Murthy L, Chiappetta C, Kirkland JL, Goodwin B, Adlercreutz B, Stancel GM, Makela S. 1998. A case of laboratory animal feed with high estrogenic activity and its impact on in vivo responses to exogenously administered estrogens. Environ Health Perspect 106:369-373.

Brown NM, Setchell KDR. 2001. Animal models impacted by phytoestrogens in commercial chow: Implications for pathways influenced by hormones. Lab Invest 81:735-747.

Camper-Kirby DC, Welch S, Walker A, Shiraishi I, Setchell KDR, Schaefer E, Kajstura J, Anversa P, Sussman M. 2001. Myocardial Akt activation and gender: Increased nuclear activity in female versus males. Circ Res 88:1020-1027.

Casanova M, You L, Archibeque-Engle S, Heck H d'A. 1999a. Effects of dietary genistein on reproductive development of Sprague-Dawley rats. Toxicologist Suppl Toxicolog Sci 48:375.

Casanova M, You L, Gaido KW, Archibeque-Engle S, Janszen DB, Heck H d'A. 1999b. Developmental effects of dietary phytoestrogens in Sprague-Dawley rats and interactions of genistein and daidzein with rat estrogen receptors α and β in vitro. Toxicol Sci 51:236-244.

Diel P, Schulz T, Strunck E, Vollmer G, Micha H. 2000. Ability of xeno- and phytoestrogens to modulate expression of estrogen-sensitive genes in rat uterus; estrogenicity profiles and uterotropic activity. J Steroid Biochem Mol Biol 73:1-10.

Eldridge AC, Kwolek WF. 1983. Soybean isoflavones: Effect of environment and variety on composition. J Agric Food Chem 31:394-396.

EDSTAC [Endocrine Disrupter Screening and Testing Advisory Committee]. Final Report. 1999. Washington DC: US EPA Office of Prevention, Pesticides, and Toxic Substances.

EDSPW [Endocrine Disruptor Screening Program Website]. 2001. Office of Science Coordination and Policy (http://www.epa.gov/scipoly/oscpendo/index.htm).

Hoeck JA, Fehr WR, Murphy PA, Welke GA. 2000. Influence of genotype and environment on isoflavone contents of soybean. Crop Sci 40:48-51.

Howdeshell KL, Peterman PH, Judy BM, Taylor JA, Orazio CE, Ruhlen RL, vom Saal FS, Welshons W. 2003. Bisphenol A is released from used polycarbonate animal cages into water at room temperature. Environ Health Perspect 111:1180-1187.

Jefferson WN, Padilla-Banks E, Clark G, Newbold RR. 2002. Assessing estrogenic activity of phytochemicals using transcriptional activation and immature mouse uterotrophic responses. J Chromatogr B 777:179-189.

Ju YH, Allred CD, Allred KF, Karko KL, Doerge DR, Helferich WG. 2001. Physiological concentrations of dietary genistein dose-dependently stimulate growth of estrogen-dependent human breast cancer (MCF-7) tumors implanted in athymic nude mice. J Nutr 131:2957-2962.

Kanno J. 2003. Q&A with Jun Kanno. In: In: Lawton G, ed. Endocrine/Estrogen Letter. Brisbane CA: Online Enterprises. Vol 9 No 4 (176):11.

Kanno J, Kato H, Iwata T, Inoue T. 2002. Phytoestrogen-low diet for endocrine disruptor studies. J Agric Food Chem 50:3883-3885.

Kanno J, Onyon L, Peddada S, Ashby J, Jacob E, Owens W. 2003a. The OECD Program to validate the rat uterotrophic bioassay. Phase 2: Dose-response studies. Environ Health Perspect 111:1530-1549.

Kanno J, Onyon L, Peddada S, Ashby J, Jacob E, Owens W. 2003b. The OECD Program to validate the rat uterotrophic bioassay. Phase 2: Coded single-dose studies. Environ Health Perspect 111:1550-1558.

Kanno JL, Onyon L, Haseman J, Fenner-Crisp P, Ashby J, Owens W. 2001. The OECD program to validate the rat uterotrophic bioassay to screen compounds for in vivo estrogenic responses. Phase 1. Environ Health Perspect 109:785-794.

Knapka J. 1983. Nutrition. In: Foster HL, Small JD, Fox JG, eds. The Mouse in Biomedical Research. New York: Academic Press. p 51-67.

Kuiper GG, Lemmen JM, Carlsson B, Corton JC, Safe SH, van der Saag PT, van der Burg B, Gustafsson JA. 1997. Interaction of estrogenic chemicals and phytoestrogens with estrogen receptor beta. Endocrinology 139:4252-4263.

Lamartiniere CA, Moore J, Holland M, Barnes S. 1995. Neonatal genistein chemoprevents mammary cancer. Proc Soc Exp Biol Med 208:120-123.

Lephart ED, Rhees RW, Setchell KDR, Bu LH, Lund TD. 2003. Estrogens and phytoestrogens: Brain plasticity of sexually dimorphic brain volumes. J Steroid Biochem Mol Biol 85:299-309.

Lephart ED, West TW, Weber KS, Rhees RW, Setchell KDR, Adlercreutz H, Lund TD. 2002. Neurobehavioral effects of dietary soy phytoestrogens. Neurotoxicol Teratol 24:5-16.

Lieberman S. 1996. Are the differences between estradiol and other estrogens, naturally occurring or synthetic, merely semantical? J Clin Endocrinol Metab 81:850.

Lund TD, Munson DJ, Haldy MD, Setchell KDR, Lephart ED, Handa RJ. 2004. Equol is a novel anti-androgen that inhibits prostate growth and hormone feedback. Biol Repro 70:1188-1195.

Markey CM, Michaelson CL, Veson EC, Sonnenschein C, Soto AM. 2001. The mouse uterotrophic assay: A re-evaluation of its validity of assessing the estrogenicity of bisphenol A. Environ Health Perspec 109:55-60.

Morito K, Hirose T, Kinjo J, Hirakawa T, Okawa M, Nohara T, Ogawa S, Inoue S, Muramatsu M, Masamune Y. 2003. Interaction of phytoestrogens with estrogen receptors α and β. Biol Pharm Bull 24:351-356.

Muthyala RS, Ju YH, Sheng S, Williams LD, Doerge DR, Katzenellenbogen BS, Helfrich WG, Katzenellenbopgen JA. 2004. Equol, a natural estrogenic metabolite from soy isoflavones: Convenient preparation and resolution of R- and S-equols and their differing binding and biological activity through estrogen receptors alpha and beta. Biorganic Medicinal Chem 12:1559-1567.

Njiti VN, Meksem K, Yuan J, Lightfoot DA, Banz WJ, Winters TA. 1999. DNA markers associated with loci underlying seed phytoestrogen content in soybeans. J Med Food 2:185-187.

Odum J, Tinwell H, Jones K, Van Miller JP, Joiner RL, Tobin G, Kawasaki H, Deghenghi R, Ashby J. 2001. Effect of rodent diets on the sexual development of the rat. Toxicol Sci 61:115-127.

Owens W, Ashby J, Odum J, Onyon L. 2003. The OECD program to validate the rat uterotrophic bioassay. Phase 2: Dietary phytoestrogen analyses. Environ Health Perspect 111:1559-1567.

Patisaul HB, Whitten PL, Young LJ. 1999. Regulation of estrogen receptor beta mRNA in the brain: Opposite effects of 17β-estradiol and the phytoestrogen, coumestrol. Mol Brain Res 67:165-171.

Price KR, Fenwick GR. 1985. Naturally occurring oestrogens in foods--A review. Food Add Contam 2:73-106.

Reeves PG, Nielsen FH, Fahey GC Jr. 1993. AIN-93 purified diets for laboratory rodents: Final report of the American Institute of Nutrition and Ad Hoc Committee on the Reformulation of the AIN-76A Rodent Diet. J Nutr 123:1939-1951.

Salzman A, Preiser J-C, Setchell KDR, Szabo C. 1999. Isoflavone-mediated inhibition of tyrosine kinase: A novel anti-inflammatory approach. J Med Food 2:179-181.

Setchell KDR, Borriello SP, Kirk DN, Axelson M. 1984. Non-steroidal estrogens of dietary origin: Possible roles in hormone-dependent disease. Am J Clin Nutr 40:569-578.

Setchell KDR, Brown NM, Olsen-Lydeking E. 2002a. The clinical importance of the metabolite equol--A clue to the effectiveness of soy and its isoflavones. J Nutr 132:3577-3584.

Setchell KDR, Brown NM, Zimmer-Nechemias L, Brashear WT, Wolfe B, Kirschner AS, Heubi JE. 2002b. Evidence for lack of absorption of soy isoflavone glycosides in humans, supporting the crucial role of intestinal metabolism for bioavailability. Am J Clin Nutri 76:447-453.

Setchell KDR, Cole SJ. 2003. Variation in isoflavone levels in soy foods and soy protein isolates and issues related to isoflavone databases and food labeling. J Agric Food Chem 51:4146-4155.

Setchell KDR, Welsh MB, Lin CK. 1987. High-performance liquid chromatographic analysis of phytoestrogens in soy protein preparations with ultraviolet, electrochemical, and thermospray mass spectrometric detection. J Chrom 386:315-323.

Shelby MD, Newbold RR, Tully DB, Chae K, Davis VL. 1996. Assessing environmental chemicals for estrogenicity using a combination of in vitro and in vivo assays. Environ Health Perspect 104:1296-1300.

Subcommittee on Laboratory Animal Nutrition. 1995. Nutrient requirements of laboratory animals. Washington DC: National Academy Press.

Thigpen JE, Clark GC, Saunders H, Byman C, Locklear J, Caviness G, Eatmon T, Grant MG, Forsythe DB. 2001a. Evaluation of a novel estrogen responsive BG1Luc4E2 cell line for the detection of estrogenic activity in laboratory animal diets (Abstract). Contemp Top Lab Anim Sci 40:75.

Thigpen JE, Haseman J, Locklear J, Ahlmark K, Caviness GF, Williamson RL, Goelz MF, Forsythe D. 1998. Comparative estrogenic activity of three new closed formula natural ingredient diets formulated to reduce the concentration of phytoestrogens (Abstract). Contemp Top Lab Anim Sci 37:103.

Thigpen JE, Haseman JK, Saunders HE, Setchell KDR, Grant MF, Forsythe DB. 2003. Dietary phytoestrogens accelerate the time of vaginal opening in immature CD-1 Mice. Comp Med 53:607-615.

Thigpen JE, Lebetkin EH, Dawes ML, Richter CB, Crawford D. 1987a. The mouse bioassay for the detection of estrogenic activity in rodent diets: III. Stimulation of uterine weight by dextrose, sucrose and cornstarch. Lab Anim Sci 37:606-609.

Thigpen JE, Li LA, Richter CB, Lebetkin EH, Jameson CW. 1987b. The mouse bioassay for the detection of estrogenic activity in rodent diets: I. A standardized method for conducting the mouse bioassay. Lab Anim Sci 37:596-601.

Thigpen JE, Li LA, Richter CB, Lebetkin EH, Jameson CW. 1987c. The mouse bioassay for the detection of estrogenic activity in rodent diets: II. Comparative estrogenic activity of purified ce.jpgied and standard open and closed formula rodent diets. Lab Anim Sci 37:602-605.

Thigpen JE, Locklear J, Caviness GF, Stokes WS, Stechell K. 1992. Concentration, source, and role of plant hormones (phytoestrogens) in laboratory animal diets (Abstract). Contemp Top Lab Anim Sci 31:31.

Thigpen JE, Locklear J, Haseman JK, Saunders HE, Caviness G, Grant MF, Forsythe DB. 2002. Dietary factors affecting uterine weights of immature CD-1 mice used in uterotrophic bioassays. Cancer Detect Prev 26:381-393.

Thigpen JE, Locklear J, Haseman JK, Saunders HE, Grant MF, Forsythe DB. 2001b. Effects of the dietary phytoestrogens daidzein and genistein on the incidence of vulvar carcinomas in 129/J mice. Cancer Detect Prev 25:527-532.

Thigpen JE, Saunders H, Haseman JK, Setchell KDR, Caviness G, Echerd L, Grant MG, Forsythe DB. 2004. Variation in phytoestrogen content between different mill dates of the same diet produces significant differences in the time of vaginal opening in F344 rats (Abstract no. PS47). Contemp Topics 43(4):55.

Thigpen JE, Setchell KDR, Ahlmark KB, Locklear J, Spahr T, Caviness GF, Goelz MF, Haseman JK, Newbold RR, Forsythe DB. 1999. Phytoestrogen content of purified open and closed formula laboratory animal diets. Lab Anim Sci 49:530-536.

Underhill KL, Rotter BA, Thompson BK, Prelusky DB, Trenhom HL. 1995. Effectiveness of cholestyramine in the detoxification of zearalenone as determined in mice. Environ Contam Tox 54:128-134.

Whitten PL, Naftolin F. 1992. Effects of a phytoestrogen diet on estrogen-dependent reproductive processes in immature female rats. Steroids 57:56-61.

Whitten PL, Patisaul HB. 2001. Cross-species and interassay comparisons of phytoestrogen action. Environ Health Perspect 109:5-20.

Yamasaki K, Sawaki M, Noda S, Wada T, Hara T, Takatsuki M. 2002. Immature uterotrophic assay of estrogenic compounds in rats given different phytoestrogen content diets and the ovarian changes in the immature rat uterotrophic of estrogenic compounds with ICI 182,780 or antide. Arch Toxicol 76:613-620.

You L, Casanova M, Bartolucci EJ, Fryczynski MW, Dorman DC, Everitt JI, Gaido KW, Ross SM, Heck H d'A. 2002. Combined effects of dietary phytoestrogen and synthetic endocrine-active compound on reproductive development in Sprague-Dawley rats: Genistein and methoxychlor. Toxicol Sci 66:91-104.