Effects of Postnatal Exposure to a Mixture of Polychlorinated Biphenyls,p,p ′-dichlorodiphenyltrichloroethane,and p - p ′-dichlorodiphenyldichloroethene in Prepubertal and Adult Female Sprague-Dawley Rats

Abstract

The postnatal period is a critical phase of development and a time during which humans are exposed to higher levels of persistent organic pollutants (POPs), than during subsequent periods of life. There is a paucity of information describing effects of postnatal exposure to environmentally relevant mixtures of POPs, such as polychlorinated biphenyls (PCBs), p,p′-dichlorodiphenyltrichloroethane (DDT), and p,p′-dichlorodiphenyldichloroethene (DDE). To provide data useful for the risk assessment of postnatal exposure to POPs, mixtures containing 19 PCBs, DDT, and DDE were prepared according to their concentrations previously measured in the milk of Canadian women, and dose-response effects were tested on the proliferation of MCF7-E3 cells in vitro, and in vivo experiments. Female neonates were exposed by gavage at postnatal days (PNDs) 1, 5, 10, 15, and 20 with dosages equivalent to 10, 100, and 1000 times the estimated human exposure level over the first 24 days of life. The MCF7-E3 cells showed a 227% increase in the AlamarBlue proliferation index, suggesting estrogen-like properties of the mixture, but this was not confirmed in vivo, given the absence of uterotrophic effects at PND21. An increase (511%) in hepatic ethoxyresorufin-o-deethylase activity at the dose 100× was the most sensitive endpoint among those measured at PND21 (organ weight, mammary gland and ovarian morphometry, hepatic enzyme inductions, serum thyroxine and pituitary hormones). In liver samples from older female rats (previously involved in a mammary tumor study [Desaulniers et al., Toxicol. Sci. 75:468–480, 2001]), hepatic metabolism of ¹⁴C-estradiol-17β (E2) at PND55 to PND62 was significantly higher in the 1000× compared to the control group, but hepatic detoxification enzyme activities had already returned to control values. The production of hepatic 2-hydroxy-E2 decreased, whereas that of estrone increased with age. In conclusion, the smallest dose of the mixture to induce significant effects was 100×, and mixture-induced changes in the hepatic metabolism of estrogens might be a sensitive indicator of persistent effects.

Polychlorinated biphenyls (PCBs), p,p′-dichlorodiphenyltrichloroethane (DDT), and its major metabolite p,p′-dichlorodiphenyldichloroethene (DDE) are lipid-soluble persistent organochlorine chemicals banned from most industrialized countries since the 1970s; however, their presence in human tissues still raises health concerns (Robertson and Hansen 2001). Due to regulatory actions, levels of these contaminants are declining (World Health Organization 1996; Craan and Haines 1998; Solomon and Weiss 2002; Dallaire et al. 2003), but PCBs, DDT, and DDE are still the most abundant breast milk contaminants (Jaga and Dharmani 2003; Guvenius et al. 2003). In vitro and in vivo studies have shown that PCB exposure can cause alterations in growth, reproduction, thyroid function, porphyria, immunotoxicity, hepatotoxicity, neurotoxicity, dermal toxicity, carcinogenicity, and endocrine and biochemical changes (Safe 1994; Ulbrich and Stahlmann 2004; Robertson and Hansen 2001; Hansen 1999). Human health effects in adults resulting from accidental exposure to PCBs include mostly chloracne, hyperpigmentation, and peripheral neuropathy (Guo et al. 2004). Prenatal exposure from these accidents elicits similar symptoms but can also lead to stillbirth and infant death, retarded growth, delayed cognitive development, increased otitis media, behavioral problems, nail and teeth developmental defects, and decreased penile length and abnormal semen parameters in young men (Guo et al. 2004). Effects from occupational exposure are less evident and include increases in various cancers, porphyria, serum triglyceride, and dermatologic problems (Persky 2001). In segments of the general population having an elevated body burden of PCBs, a reduction in cognitive function during infancy and childhood is frequently reported (Schantz, Widholm, and Rice 2003). DDT is an insecticide that was banned from most countries, mostly because of its negative impact on wildlife, but it is still being used to control malaria in approximately 23 countries (Turusov, Rakitsky, and Tomatis 2002). In addition to possible carcinogenic effects, DDT and its metabolites have been associated with central nervous system toxicity, premature birth, shorter duration of lactation, immunosuppression, hepatotoxicity, and endocrine disruption with both estrogenic and antiandrogenic properties (Gray et al. 2001; Bilrha et al. 2003; Dewailly et al. 2000; Chen and Rogan 2003; Turusov, Rakitsky, and Tomatis 2002).

In rats, acute or short-term exposure to high doses of xenobiotics—2,3,7,8-tetrachlorodibenzodioxin (TCDD) (Graham et al. 1988); phenobarbital, dexamethasone, 3-methylcholanthrene (Suchar et al. 1996); PCB 126 and 153 (Painter et al. 2001); TCDD, 2,4-dichlorophenoxyacetic acid, dieldrin (Badawi, Cavalieri, and Rogan 2000)—increased the expression of cytochrome P450 (CYP) CYP1A, CYP1B1, CYP2B, and CYP3A enzymes and hepatic levels of estrogen metabolites. Changes in estrogen metabolism and signaling affect hypertension, lipid metabolism, brain development, numerous organs (ovaries, uterus, placenta, testicles, breast, bones), systems (vascular, immune), and cancers (breast, endometrium, ovary, prostate, liver, kidney, brain) (Zhu and Conney 1998a; McEwen and Alves 1999; Cavalieri et al. 2002; Rogan et al. 2003). At least 31 estrogen metabolites have been identified (Rogan et al. 2003) and hydroxylation in three positions (2OH, 4OH, and 16αOH) are associated with important biological consequences. 2-Hydroxylation is the major metabolic oxidation pathway. In the female rat liver, it is predominantly achieved by CYP2B1/2 and CYP3A1/2, with a minimal role for CYP1A (Suchar et al. 1996) or other CYPs (Zhu and Conney 1998a). 2OH-estradiol-17β(E2), like the other catecholestrogens (CEs), is rapidly methylated by the enzyme catechol-o-methyltransferase (COMT) into 2-methoxyestra-diol (Zacharia et al. 2004). 2-Methoxyestradiol has beneficial effects on the cardiovascular system, renal function (Dubey, Tofovic, and Jackson 2004), obesity (Tofovic, Dubey, and Jackson 2001), and cancers (Seeger et al. 2004), acting as an antioxidant (more potent than vitamin E [Seeger, Mueck, and Lippert 1997]), antiangiogenic, and apoptotic agent (Shimada et al. 2003; Dubey, Tofovic, and Jackson 2004; Zhu and Conney 1998a, 1998b). 16α-Hydroxylated estrogens are highly estrogenic, might stimulate the proliferation of damaged cells, and act as promoters of carcinogenesis (Stresser and Kupfer 1997; Osborne et al. 1993). Although in humans CYP3A4 has strong 16α-hydroxylation activities, the identity of the CYP enzyme responsible for this activity in the female rat liver is controversial, but a member of the CYP2D family is suspected (Zhu and Conney 1998a). Given the different biological effects of 2OH- and 16αOH-estrogens, their ratios (2/16αOH or 16α/2OH) in urine, serum, tissue extracts, and culture medium have been used as biomarkers of breast cancer risk, or as an indicator of altered estrogen metabolism (Jernstrom et al. 2003; Bradlow et al. 1995). In liver tissue, 4-hydroxylation is a minor pathway and, in female rat, is attributed to CYP2B1/2 and CYP3A1/2 activity (Suchar et al. 1996). In extrahepatic tissues 4-hydroxylation occurs through the TCDD-inducible CYP1B1 enzyme, and constitutes a mutagenic pathway (Jefcoate et al. 2000; Li et al. 2004; Badawi, Cavalieri, and Rogan 2000; Hayes et al. 1996). Evidence of gene mutations induced by estrogens have been reviewed (Cavalieri et al. 2000).

Given the high exposure of infants to persistent environmental contaminants through breast milk, and the sometimes greater sensitivity of infants to toxic effects compared to adults (Ginsberg, Hattis, and Sonawane 2004), there are concerns that exposure to environmental contaminants during the perinatal period could have adverse developmental and/or long-term effects. A balanced perspective on this issue has recently been published, highlighting research needs (LaKind, Amina, and Berlin 2004). There is little information available describing the effects of exposure to reconstituted mixtures of environmental contaminants present in breast milk. Woodhouse and Cooke (2004) reported that a reconstituted mixture of ortho-PCBs present in breast milk had no effect on human aromatase activity in vitro, but individually, some components (2, 4, 4′-trichlorobiphenyl, 2, 3, 3′, 4, 4′-pentachlorobiphenyl) inhibited the activity. Parkinson, Robertson, and Safe (1980) observed that intraperitoneal injections of a reconstituted breast milk mixture of 13 ortho-PCBs was seven times more potent for stimulating hepatic aryl hydroxylase activity, compared to the more complex commercial PCB mixture Kanechlor 500. A German laboratory exposed dams to a reconstituted mixture of 14 PCBs present in breast milk and reported endocrine, metabolic, and long-lasting changes in sex-dependent behaviors (Lilienthal et al. 2000; Hany et al. 1999; Kaya et al. 2002). We prepared mixtures of aryl -hydrocarbon receptor (AhR) agonists (3 non-ortho-PCBs, 6 polychlorinated dibenzodioxins [PCDDs], 7 polychlorinated dibenzofurans [PCDFs]) based on concentrations found in breast milk and administered them to baby rats by gavage during the postnatal period (Desaulniers et al. 2003, 2004). Nonadditive/antagonistic interactions among these AhR agonists were observed that could have caused an overestimation of the toxicity of the complete mixture (Desaulniers et al. 2003). Using the same exposure protocol, a different mixture of the most abundant organochlorines in breast milk, including 19 PCBs, DDT, and DDE, initially delayed the development of methylnitrosourea-induced mammary tumors, but at necropsy the final number of mammary lesions was increased (Desaulniers et al. 2001). The current investigation is now presenting the dose-response effects of this mixture in three experiments. First, estrogen-like effects were assessed by measuring the proliferation of MCF7-E3 cells. Second, effects on organ weight, mammary gland and ovarian morphometry, hepatic enzyme induction, thyroxin, and pituitary hormones are presented in female rats at postnatal day 21 (PND21). Finally, using liver samples from rats of a previous experiment testing the effects of this mixture on mammary tumor development (Desaulniers et al. 2001), new data on hepatic transformation of radioactive estradiol-17β in 55- to 308 days-old female rats were obtained, and associations were tested between mixture exposure and mammary tumorigenic outcomes.

METHODS

Mixture Compositions for In Vitro and In Vivo Experiments

The compositions of the test mixtures are described in Table 1. DDT, DDE, and PCB congeners detected in more than 75% of breast milk samples obtained from Canadian Caucasian women were included in the mixture (Dewailly et al. 1992; Mes et al. 1993a, 1993b). Non-ortho-substituted PCBs (PCB-77, -126, and -169) are also ubiquitous environmental contaminants, which are present in small amounts relative to other PCBs. The concentrations of non-ortho PCBs added to the mixture were derived from the concentrations measured in pooled breast milk samples (Dewailly et al. 1992). It was not determined if they were present in more than 75% of women, but they were included because of their high toxicity. For the in vivo experiments, mixtures were prepared in corn oil to reach exposure levels equivalent to 10, 100, and 1000 times the typical intake of PCBs, DDT, and DDE by a human baby that would be exclusively breastfed over its first 24 days of life. The daily intake of each organochlorine was calculated using its concentration in breast milk fat, an average milk consumption of 120 ml of per kg body weight per day (Ayotte, Carrier, and Dewailly 1996), with a milk fat composition of 3.7% (Frank et al. 1988). A blind analysis of the PCB/DDT/DDE mixture used for in vivo studies confirmed individual congener concentrations (Wellington Laboratories, Guelph, ON). This analysis, and a complete protocol for the preparation of the mixture, have been described previously (Desaulniers et al. 2001). Dose levels added to MCF-7 cell culture medium for in vitro experiments are presented in Table 1. These doses represented 1, 10, 100, 500, 1000, and 5000 times the concentration present in breast milk. Chemicals were dissolved in DMSO before they were added to the culture medium. Some required heating at 55°C and sonication for 2 h prior to dilution. All chemicals present in the mixtures were >99% pure (AccuStandard, New Haven, CT).

MCF7-E3 Cell Culture

MCF-7 cell proliferation assays were conducted as previously described (Desaulniers et al. 1998). Briefly, MCF-7 cells from the estrogen responsive clone E3 (Butler et al. 1986) were grown for 8 days in 2 ml of medium Dulbecco’s modified Eagle’s medium [DMEM] without phenol red, supplemented with 15% charcoal-stripped human serum, with the lowest concentration (16.44 ng/ml) of mixture in the culture medium reflecting the quantity present in human milk containing 3.7% fat (Table 1). Subsequent tested concentrations were increased by 10, 100, 500, 1000, and 5000 times (82,199 ng/ml). The medium was removed on day 8, and replaced with a protein-free culture medium containing 10% AlamarBlue (AccuMed International, Westlake, OH, USA). This proliferation assay is based on the intracellular reduction of AlamarBlue reagent by oxidoreductases and the mitochondrial electron transport chain, with a corresponding shift in fluorescence (Goegan, Johnson, and Vincent 1995). After a 4 h incubation, fluorescence of the medium samples (250 μl) was measured on a Cytofluor 2300 microplate reader (excitation 560 nm, emission 590 nm; Millipore, Bedford, MA). Treatment effects were tested in four wells and the experiments were repeated three times.

Animal Treatment

The experimental protocol was approved by the Health Canada Animal Care Committee in accordance with the Canadian Council on Animal Care guidelines (Canadian Council on Animal Care 1993).

Litters (usually 4 to 5 pups/dam) of only female Sprague Dawley rats were prepared in the afternoon by the supplier (Charles River, St-Constant, QC), using females born that day (day 0 is the day of birth), and shipped to our laboratory the next morning at PND1. As we could only manipulate a limited number of rats per week, six to seven litters (approximately 30 pups), were received on a weekly basis until a minimum of 12 female neonates per treatment group was reached. In total, 27 litters were received. To avoid litter effects, all pups were removed from their mothers before being randomly assigned to a treatment group. They were weighed individually, marked for identification, exposed by gavage to their first treatment, then returned to a dam. Given that the mothers are cleaning their pups, to avoid possible cross-contamination of the pups through the mothers milk, each dam only received pups that were dosed with the same treatment. Consequently, six to seven of the treatment groups were represented each week, although the entire experiment included nine treatment groups: water, corn oil–DMSO, 10×, 100×, and 1000× of the PCB/DDT/DDE mixture, 1000× non-ortho PCBs, 1000× mono-ortho, 1000× di-ortho, and 1000× DDT/DDE (Table 1). Female neonates were exposed to the mixtures by gavage at PND1 (a cumulative dose for days 0, 1, 2, 3, and 4), PND5 (for days 5, 6, 7, 8, and 9), PND10 (for days 10, 11, 12, 13, 14), PND15 (for days 15, 16, 17, 18, 19), and PND20 (for days 20, 21, 22, 23, and 24). Doses were based on body weights measured prior to each treatment (5.18 ml of the mixture per kg body weight [bw]). The rats were sacrificed at PND21 under isoflurane anesthesia, through blood collection via the abdominal aorta. Organs were dissected, and the weights of the liver, ovaries, uterus, and pituitary gland were recorded. Skin pelts were fixed in 10% neutral-buffered formalin 24 to 48 h before dissection of the mammary glands.

In older rats, long-term effects of the complete PCB/DDT/DDE mixture on hepatic CYP detoxification enzyme induction, estrogen metabolism, and ovarian follicle populations were tested using samples collected from another experiment, which investigated the mixture effects on the development of methylnitrosourea (MNU)-induced mammary tumors (Desaulniers et al. 2001). Neonates in that experiment were treated using the same exposure protocol as described in the previous paragraph. Briefly, neonates were divided into seven groups of 28 to 41 females. One group received corn oil as control (dose 0), and another received only the mixture at the highest dose, 1000×. Some rats from the control (n = 7) and 1000× (n = 9) groups were sacrificed between PND 55 and PND62, and the remaining control (n = 20) and 1000× (n = 20) rats were sacrificed at PND224. Four groups received increasing doses of the mixture prior to a single intraperitoneal (ip) injection of MNU (30 mg/kg bw) on PND21 to initiate the mammary carcinogenic process (MNU-0, MNU-10×, MNU-100×, and MNU-1000× ). In addition, a final group was gavaged at PND18 with 2.5 μg TCDD per kg bw (dissolved in corn oil) before MNU administration on PND21. MNU-treated rats were sacrificed when their palpable mammary tumors reached 1 cm in size, or by PND308 if no palpable tumor was detected. Rats were sacrificed as previously described.

Analysis of Mammary Gland Structures at PND21

Whole mounts of abdominal inguinal mammary glands were prepared as previously described (Desaulniers et al. 2001), and carefully examined using a stereoscope and a microscope at 4× magnification. Various mammary structures were identified according to Russo and Russo (1987), and terminal end buds (TEBs), terminal ducts (TDs), and alveolar buds (ABs) were counted. Mammary gland surface areas (excluding the inter-brachial areas) were measured using image analysis software (Image-Pro Plus; Media Cybernetics, Silver Spring, MD).

Assessment of Ovarian Structures

A method was adapted from Plowchalk, Smith, and Mattison (1993) to compare immediate- and long-term effects on ovarian follicle populations at PND21 (control, n = 6; 1000×, n = 6) and PND224 (control, n = 5; 1000×, n = 5). One ovary per rat was fixed in 10% neutral-buffered formalin, paraffin embedded, and serially sectioned (6 μm/section). Every fifth section was mounted and stained with hematoxylin and eosin, then follicles were counted with a computer imaging system (Image-Pro Plus). Nuclear membrane breakdown occurs in atretic follicles, therefore, only follicles with a visible nucleus were counted (except for primordial follicles, which were all counted). Five morphologically distinct categories of follicles were counted: (i) primordial follicles, identified by the presence of an oocyte surrounded by a single layer of flattened granulosa cells; (ii) primary follicles, characterized by an oocyte surrounded by a single layer of cuboidal granulosa cells; (iii) growing follicles, having an oocyte surrounded by multiple cuboidal granulosa cell layers; (iv) antral follicles, with an oocyte surrounded by multiple layers of cuboidal granulosa cells, a visible zona pellucida and a developing antrum; (v) Graafian follicles, identified by a large antrum, an acentric oocyte, and zona pellucida surrounded by a corona radiata supported by a cumulus oophorus. Given the large number of primordial follicles, every 10th section was counted, whereas primary and growing follicles were counted on every 5th section. To ensure that antral and Graafian follicles were counted only once, they were viewed in three sections (covering a thickness of ≈60 μm) simultaneously on the computer screen, so each follicle was individually identified in relation to the next section.

Radioimmunoassays

Serum thyroxin levels were determined using a commercial I¹²⁵ radioimmunoassay (RIA) kit (ICN Biomedicals, Costa Mesa, CA), while pituitary follicle-stimulating hormone (FSH), luteinizing hormone (LH), prolactin (PRL), growth hormone (GH), and thyroid-stimulating hormone (TSH) contents were determined by RIA as previously described (Desaulniers et al. 1999). The amount of serum available from females at PND21 was limited, so initially only serum thyroxin was analyzed. Serum PRL was later measured because of statistically significant changes in pituitary PRL (see Table 3) were observed. Briefly, pituitary glands were sonicated in 200 μl of 0.05 M phosphate-buffered saline with 1% bovine serum albumin (BSA) and protease inhibitor cocktail (Sigma Chemicals, St. Louis, MO), and the homogenates were centrifuged (1300 × g, 30 min, 4°C) to recover the supernatant. The sonication and centrifugation steps were repeated two more times, resulting in a final supernatant volume of 0.6 ml, from which hormones were measured. Pituitary hormone preparations for iodination and standard curves, as well as primary antisera for radioimmunoassays, were rFSH-I-8 (AFP-11454B), rFSH-RP-2 (AFP-4621B), and rFSH-11 (AFP-CO972881) for FSH; rLH-I-9 (AFP-10250C), rLH-RP-3 (AFP-7187B), and rLH-S-11 for LH; rTSH-I-9 (AFP-11542B), rTSH-RP3 (AFP-5512B), and rTSH-RIA-6 (AFP 329691Rb) for TSH; rPRL-I-6 (AFP-10505B), rPRL-RP-3 (AFP-4459B), and rPRL-S-9 (AFP-13581570) for PRL; rGH-I-6 (AFP-5676B), rGH-RP-2 (AFP-3190B), and rGH-S-5 for GH (all provided by the National Hormone and Pituitary Program, National Institute of Diabetes and Digestive and Kidney Diseases). The secondary precipitating antibodies were goat anti-rabbit IgG (Sigma Immunochemicals, St. Louis, MO) for FSH, LH, PRL, and TSH assays, and goat anti-monkey IgG (Antibodies Incorporated, Davis, CA) for the GH assay. The sensitivity of the FSH, LH, TSH, GH, and thyroxin assays were, respectively, 0.03 ng/tube, 0.02 ng/tube, 0.04 ng/tube, 0.3 ng/tube, and 0.63 μg/dl. The intra-assay coefficients of variation for these assays were 9.7%, 6.9%, 6.2%, 4.9%, and 7.4%, respectively.

Hepatic Detoxification Enzyme Activities

The activity of hepatic CYP1A1 and CYP2B1/2 in 10,000 g supernatant was determined through examination of ethoxyresorufin- (EROD), pentoxyresorufin- (PROD), and benzyloxyresorufin-o-deethylase (BROD) activities in accordance with the method of Burke et al. (1985).

Hepatic Metabolism of ¹⁴C-Estradiol-17β at PND55 to PND62 and in Older Rats

The methodology was adapted from Segura-Aguilar, Castro, and Bergman (1997). At the time of necropsy, liver was dissected, snap-frozen, and stored at − 80°C. Liver microsomal fractions were prepared by differential centrifugation as follows: ≈1 g of liver from female rats was homogenized (Polytron homogenizer, Brinkmann, Westbury, NY) in 4 parts (w/w) 50 mM Tris buffer, pH 7.4, containing 20% glycerol, 10 mM ascorbic acid, and 0.1 mM EDTA, and the homogenate centrifuged at 10,000 × g (Sorvall RC 5C+, Dupont, Wilmington, Delaware) for 10 min at 4°C. The supernatant was transferred to clean tubes and recentrifuged under the same conditions. The supernatant was then centrifuged at 4°C for 1 h at 176,000 ×. g _max (Beckman L7 Ultracentrifuge; Beckman Instruments, Montreal, QC), after which the cytosol was discarded and the microsomal pellet resuspended in the same buffer. Resuspended microsomes were used immediately for enzyme assays. Protein concentrations were determined by the method of Lowry et al. (1951). Aliquots of microsomes were incubated with [4-¹⁴C]estradiol (10 μM, 100,000 dpm) (in the same buffer as was used for microsomal preparation) containing NADPH (0.25 mM) and incubated in a shaking water bath at 37°C until terminated after 60 min by the addition of ethyl acetate (5 ml) containing carrier steroids (E2, estrone [E1], 2OH-E2, 4OH-E2, and estriol [16αOH – E2]), each at 10 μg. Tubes were vortexed to extract the steroids and phase separation was achieved by centrifugation at 800 × g for 10 min. The organic phase was transferred to conical tubes and evaporated (Savant, Speedvac Plus; Fisher Scientific, Ottawa, ON). Extracts were separated on plastic coated thin layer plates (Whatman PE SIL G) in one dimension in chloroform–ethyl acetate (7:3 ν/ν run once). Carrier steroids were detected by ultraviolet (UV) illumination (4-ene-3-oxosteroids) and iodine vapour. Radioactive metabolites were detected by autoradiography for 24 h (Biomax MR; Eastman-Kodak, Rochester, NY). Areas corresponding to carrier steroids were excised and counted in a scintillation counter (TriCarb 2100TR; Packard, Meriden, CT). Metabolic conversions were corrected for microsomal protein content.

Data Analysis and Statistical Procedures

All analyses were performed using the software JMP (SAS Institute Inc. 1998). Homogeneity of variance (O’Brien and Brown-Forsythe tests) and normality (Shapiro-Wilk test) of the data were initially tested. If one of these tests failed, the data were log transformed prior to conducting an analysis of variance (ANOVA). Following a significant ANOVA, means that were significantly different were identified using the Student-Newman-Keuls test. In cases where normality or homogeneity of variance of the transformed data could not be achieved, and for the counts of mammary gland structures and ovarian follicles, the nonparametric Kruskal-Wallis test and Mann-Whitney U test were applied. To test for an effect on the proliferation of MCF-7 cells, the statistical model included the effects of experiment, dose, and the interaction Experiment × Dose. To account for differences in baseline among MCF-7 cell experiments, the effect of dose was tested using the interaction Experiment × Dose as residual error (Littell, Freund, and Spector 1991). Using this procedure, the effect of dose becomes significant (p < .05) only if it is larger than the variation among experiments. In this experiment, differences between means were identified with Duncan’s multiple range test, also using the interaction Experiment × Dose as residual error (Littell, Freund, and Spector 1991). Changes in body weight were analyzed by analysis of variance on repeated measures through time (Allen, Burton, and Holt 1983). The ratio of two normally distributed variables automatically leads to non-normal distribution, therefore changes in organ/body weight ratios were analyzed by the nonparametric Kruskal-Wallis test and Mann-Whitney U test. Statistical significance was considered at p ≤ .05.

RESULTS

Test of In Vitro Estrogenicity Using the MCF7-E3 Cell Proliferation Assay

The proliferation indices obtained as arbitrary fluorescence readings from the AlamarBlue assay were 66 ± 5 (control), 73 ± 5 (mixture: 16.44 ng/ml), 73 ± 5 (164 ng/ml), 85 ± 6 (1.6 μg/ml), 66 ± 5 (8.2 μg/ml), 72 ±4 (16.4 μg/ml), and 150 ± 6 (82.2 μg/ml). These results show a small but statistically significant proliferative effect at 1.6 μg/ml, and a large increase (227%) in the highest dose group (p < .05). These effects occurred at doses 100 and 5000 times more concentrated than human milk, and did not follow a dose-response pattern.

Effects at PND21

No statistically significant effects of treatment on body weight were determined (data not shown). There were significant correlations between body weight and either liver weight ( r ² = .70, p < .0001), or uterine weight (r ² = .06, p = .007). As a result, the effects of treatment were also assessed using organ/body weight ratios. In comparison with the oil-DMSO group, liver weight was increased by the 1000× treatment, whereas both the non-ortho and 1000× treatments had increased liver/body weight ratios (Table 2). The DDT-DDE treatment significantly reduced the uterus/body weight ratio. Weights of the pituitary gland and ovaries were not affected by any treatment (data not shown). Indicators of mammary gland development, summarized here by overall means for surface area (18.2 ± 0.7 mm²), number of TD/mm²(1.4 ± 0.2), number of TEB plus LB/mm² (29.6 ± 0.9), and finally number of AB/mm² (3.7 ± 0.3), were also not significantly affected by any treatment.

Results describing the inductions of detoxification CYP enzymes are summarized in Table 2. Among the groups treated with the PCB/DDT/DDE mixture, statistically significant increases in EROD activity were detected in the 100× group (511% = 100 × 0.046/0.09), and in the 1000× group (2022%). Although EROD activity was not statistically increased by the di-ortho PCB and DDT-DDE components of the complete mixture, it was significantly increased by the mono-ortho (344% increase), and particularly by the non-ortho, PCB component (1511%). Therefore, most EROD activity measured in the 1000× PCB/DDT/DDE treated group could be attributed to the non-ortho PCB component of the mix, which accounted for only 0.03% of the mass of organochlorines present in the PCB/DDT/DDE mixture (Table 1). PROD activity was significantly increased by the 1000× treatment, likely due to the di-ortho PCB and DDT-DDE components. BROD activity was significantly increased by the 100× treatment but, surprisingly, not by 1000× because of the large standard error in that group, which prevented the detection of a significant effect. BROD activity was also increased by non-ortho and di-ortho PCBs, as well as DDT-DDE.

Analyses of thyroxin and pituitary hormone levels are described in Table 3. The pituitary content of PRL was significantly increased by DDT-DDE. Although the pituitary PRL was statistically higher in the 100× than in the oil-DMSO group, it was not significantly different from the water group. The serum PRL level, which was not statistically affected by any treatments, was the highest in the DDT-DDE group, as similarly observed for pituitary PRL. The pituitary gland FSH content in the oil-DMSO group was significantly higher than those measured in the mono-ortho, di-ortho, and the 1000× mixture groups; however, these were not different from the water group. Other hormones were not significantly altered by any treatments.

Ovarian Structures at PND21 and in Adults

The 1000× mixture had no effect on ovarian follicular population or the number of corpora lutea (CL) at PND21 and PND224 (data not shown). Given that the mixture also had no effects on estrous cycle length and time to puberty (Desaulniers et al. 2001), which are both dependent on ovarian function, there were no indications that an analysis of the ovarian structures in lower dose groups would provide new information. However, there were effects of age, and the overall mean number of follicles at PND21 and PND224, respectively, were primordial, 1011 ± 91 versus 335 ± 40; primary, 107 ± 10 versus 34 ± 5; growing, 128 ± 7 versus 68 ± 6; antral, 167 ± 13 versus 24 ± 4; Graafian follicles, 11 ± 2 versus 12 ± 2; and CL, 0 ± 0 versus 17 ± 4. As expected, primordial follicles were the most abundant and Graafian follicles were the least abundant in both age groups. The number of follicles in most classes was lower in the older group (p < .05), with the exception of the Graafian follicles, which were similar in number for both age groups. As expected from prepubertal rats, there were no corpora lutea at PND21.

Hepatic Metabolism of ¹⁴C-Estradiol-17β at PND55 to PND62 and in Older Rats

Rats treated with the 1000× mixture and sacrificed at PND55 to PND62 metabolized more ¹⁴C-E2 than control rats; however, this difference was no longer detectable at PND224 (Treatment × Age: p = .0006; Figure 1A ). Surprisingly, this difference at 55 to 62 days of age was not associated with significant differences between control and 1000× rats, respectively, in hepatic BROD (0.041 ± 0.002 versus 0.044 ± 0.004 nm/min/mg), PROD (0.013 ± 0.001 versus 0.015 ± 0.001 nm/min/mg), or EROD (0.053 ± 0.004 versus 0.060 ± 0.007 nm/min/mg protein) activities. The rate of transformation into E1 was significantly higher at PND224 compared to PND55 to PND62 (p < .0001), but there were no significant effects of treatment, or interactions (Figure 1B ). The transformation rates into 2OH-E2 (Figure 1C ) and into 16αOH-E2 (Figure 1D ) were not significantly affected by the mixture, but were both lower at PND224 (p < .0001 and p < .007, respectively). Similarly, 2OH-E2/16αOH-E2 ratios had a tendency (p = .056) to decrease with age (means from the pooled data from control and 1000× at PND55 to PND62 and at PND224 were 4.04 ± 0.55, and 2.8 ± 0.15, respectively). 16αOH-E2 was the least important transformation pathway.

Figure 2 summarizes the regressions between age and hepatic microsomal transformation of ¹⁴C-E2 into 2OH-E2 (Figure 2 A ), 16αOH-E2 (Figure 2B ), and their ratio, 2OH-E2/16αOH-E2 (Figure 2C ), in rats treated by gavage with increasing dose of the mixture (10×, 100×, 1000× ) from PND1 until PND20, or with TCDD (2.5 μg/kg) at PND18, followed by an intraperitoneal injection of the mammary cancer initiator MNU (30 mg/kg) at PND21. This protocol and data on mammary tumor development have been published (Desaulniers et al. 2001). Here, no effects of the mixture or TCDD treatments on the hepatic transformation of ¹⁴C-E2 were observed. The pooled data from all groups showed statistically significant correlations between age and the transformation rate of the substrate into 2OH-E2 (r ² = .33, p < .0001; Figure 2A ), and with the 2OH-/16αOH-E2 ratio (r ² = .40, p < .0001; Figure 2C ), but not with 16αOH-E2 (Figure 2B ). There was also a slight but significant correlation between age and the transformation rate of the substrate into E1 (r ² = .08, p = .001; data not shown). Figure 2 supports the effects of age on enzyme activities also observed from different rats in Figure 1. Consequently, hepatocytes were exposed to a different estrogenic environment with age, with a decreasing proportion of 2OH-E2 relative to 16αOH-E2, particularly apparent after 200 days of age. No significant correlations were found between this new data on hepatic 16αOH-E2, 2OH-E2, and 2OH-/16αOH-E2 ratio, and previously published data on the same rats on the development of malignant, benign, or total number of mammary lesions (Desaulniers et al. 2001).

DISCUSSION

The current investigation revealed that exposure to a reconstituted mixture of breast milk contaminants (19 PCBs, p,p′-DDT, p,p′-DDE), at a dosage equivalent to 100 times the estimated amount that a human baby would consume over the first 24 days of life, significantly induced hepatic EROD activity in the female rat at PND21. This dose represented a daily intake of 197 μg/kg bw. Hepatic EROD activity was the most sensitive endpoint examined at PND21. Other important findings relate to the metabolism of estrogens. First, the 1000× treatment increased hepatic transformation of ¹⁴C-estradiol-17β over a longer period than induction of detoxification enzyme activities (EROD, BROD, PROD). Second, hepatocytes are exposed to a different estrogenic environment with aging, whereas the hepatic production of 16αOH-E2 remained constant, that of 2OH-E2 decreased in older females.

Some hepatic endpoints suggest additive effects. The level of EROD activity induced by the complete mixture at 1000× appears to be mostly attributed to the additive effects of the non-ortho and mono-ortho PCB components, even if they represented only 0.03% and 10.8%, respectively, of the mass of chemicals in the mixture (Table 2). The increase in liver weight in the 1000× group, and in liver/bw ratio in non-ortho PCB-treated rats, are also in accordance with additive hepatotoxic effects. These results support the concept of the toxic equivalency (TEQ) system (Safe 1998) for assessing effects related to AhR activation, even in the presence of the complex mixture tested here. EROD induction is related to the activity of the CYP1 enzyme family, which are induced by the activation of the AhR. Among the chemicals tested in the current mixture, the non-ortho PCBs have the highest affinity for the AhR, followed by the mono-ortho PCBs (Van den Berg et al. 1998). These different affinities for the AhR explain their relative contributions to EROD activity (Table 2). These results further demonstrate the specificity of this mechanism, given the abundance of other chemicals (di-ortho PCBs, DDT, DDE) that did not significantly change EROD activity. Interactions among chemical classes for induction of PROD and BROD are less evident. Di-ortho PCBs and DDT are known inducers of PROD and BROD, and as expected, they independently increased PROD and BROD activities above control values. The induction of BROD by non-ortho PCBs, although not frequently reported, is similar to the effect of TCDD which sustains induction of BROD activity by enzymes other than CYP1A or CYP2B (Iba et al. 2000). However, effects of the various treatments on BROD and PROD activities were not statistically different from those of the 1000× treatment, and this does not support additive actions of the mixture. Large interindividual differences in BROD and PROD induction were noticed, as previously observed by others (Desaulniers et al. 2003; Chang, Bandiera, and Chen 2003). Interindividual differences possibly involve complex regulatory mechanisms. In contrast to CYP1A1, which is regulated through the AhR, the expressions of CYP2B and CYP3A (responsible for most PROD and BROD activities) are regulated through heterodimerization of the retinoid X receptor with either the nuclear orphan constitutively active receptor (CAR), or the pregnane X receptor, respectively (Cai et al. 2002; Masahiko and Honkakoski 2000; Sueyoshi and Negishi 2001; Ueda et al. 2002; Zelko and Negishi 2000).

Little effects were detected on hormone levels (Table 3), likely because of the adaptability of the endocrine system. All treatments slightly increased pituitary and serum PRL, but a statistical difference was reached only for pituitary PRL in the DDT-DDE and 100× treatment groups (Table 3). Others showed that PRL levels could be increased by estrogenic effects, stress, changes in the dopaminergic system (dopamine is the major repressor of PRL synthesis and secretion), and increases in prolactin-releasing factor such as thyrotropin-releasing hormone (TRH) (O’Connor et al. 1996; Severino et al. 2004; Fujikawa et al. 2004; Steinmetz et al. 1997; Cooke et al. 2004). The current data do not permit an identification of the mechanisms affected, but the effects of DDT-DDE are reminiscent of those induced by the dopamine receptor antagonist haloperidol, which increases serum PRL and decreases uterine cytosolic estrogen receptor concentration in the rat (O’Connor et al. 1996). This would explain a possible decrease in uterine weight in the DDT-DDE group (Table 2), given that estrogen signaling is required for uterine growth. Alternative effects mediated through TRH are also not obvious in the current study. Hypothalamic TRH stimulates the release of pituitary TSH, which then stimulates the production of thyroid hormones, but both TSH and thyroxin levels (although slightly reduced) were not significantly affected. The thyroid system is a classical target of PCB exposure (Meerts et al. 2002; Hood et al. 2003; Khan et al. 2002), and the absence of effects on this system suggest a low toxicity of the mixture even at the highest dose. Pituitary FSH levels were the highest in the non-ortho PCB group, and the lowest in di-ortho PCB and high-dose groups. FSH is modulated through complex interactions involving estradiol–17β, peptides of the inhibin family, follistatin, as well as the hypothalamic neuropeptide, gonadotropin-releasing hormone (GnRH) (Fink 1988). The chemical family components of the 1000× mixture could differently affect these numerous regulatory systems, and further interpretation of these results is not attempted. Collectively, the absence of important effects on serum and pituitary hormones suggests that these end points are not robust indicators of effects, as previously indicated (O’Connor et al. 1996), and that hepatic enzyme inductions are more sensitive indicators of exposure to low doses of PCBs, or AhR-agonists (Desaulniers et al.1999, 2003).

The proliferation of MCF7-E3 cells, suggesting an estrogen-like effect, was increased by the mixture of PCBs, DDT, and DDE, particularly at a concentration of 82.2 μg/ml, which is 5000 times higher than the levels found in human milk. This occurred despite that the mixture included chemicals reported to have antiestrogenic (non-ortho PCBs [Krishnan and Safe 1993; Jansen et al. 1993]) and estrogenic activities (mono- and di-ortho PCBs [Li, Zhao, and Hansen 1994; Soto et al. 1995; Jansen et al. 1993; Desaulniers et al. 1999], non-ortho PCBs [Ohtake et al. 2003], p,p′-DDT, and p,p′-DDE [Soto et al. 1995; Andersen et al. 1999; Payne, Scholze, and Kortenkamp 2001]). Mechanisms different from estrogen receptor (ER) activation can modulate the behavior of mammary epithelial cells exposed to environmental contaminants (Kang et al. 1996; Frigo et al. 2004), and might explain the proliferative effect observed at the lower dose of the mixture. Payne,Scholze, and Kortenkamp (2001) indicated that the MCF-7 cell proliferation induced by p,p′-DDE remains, even with coincubation with tamoxifen, an ER antagonist. Nevertheless, the MCF-7 cell proliferation assay, as it is based on a human cell response, is a useful in vitro screening tool, but effects must be confirmed by additional methods (Gray et al. 1997; Andersen et al. 1999). The “gold standard” in vivo test for the detection of estrogenic xenobiotics is the measurement of the uterotrophic response (Gray et al. 1997; Kanno et al. 2003). The lack of concordance on estrogen-like effects observed here between the MCF-7 cell data, the absence of a uterine response, and absence of clear effects on PRL levels, suggest that the mixture had no important estrogen-like effects in vivo in the prepubertal rat. Interestingly, a very similar mixture (but not including DDT and DDE) administered to dams had uterotrophic effects at PND21 (Hany et al. 1999). Treatment of the dams by Hany et al. might have led to larger exposures to hydroxylated PCB metabolites, some of which have much more potent estrogenic effects than their parent PCB congeners (Layton et al. 2002).

The transient increase in the total hepatic metabolism of estrogens occurred at least until PND55 to PND62 (Figure 1A ), but at that time EROD, BROD, and PROD activities had already returned to baseline values. This suggests that hepatic transformation of estrogens may be a good indicator of persistent changes. Steroids are transformed by a number of hepatic phase I and II enzymes (Zhu and Conney 1998a), and measurement of their total transformation rate might be a more sensitive indicator of long-term hepatic effects than using a substrate that is specific to a particular, or to a limited number of enzymes. However, the use of probe substrates for specific CYP activities (e.g., BROD for CYP2B1/2, EROD for CYP1A1/2, etc.) allows the detection of enzyme induction several-fold higher than nonspecific estrogen hydroxylation end points (Suchar et al. 1996). In the current study, the specific transformation that could explain the transient increase in the total hepatic metabolism of ¹⁴C-E2 was not identified. The mixture did not induce detectable changes in the production of hepatic 16αOH-E2, 2OH-E2, and E1. The 16αOH-E2 results are consistent with the absence of an increase in hepatic 16αOH-E2 production following exposure to many prototype CYP inducers (phenobarbital [CYP2B1/2], dexamethasone [CYP3A1/2], 3-methylcholanthrene [CYP1A1], clofibrate [CYP4A family]) (Suchar et al. 1996). It has been reported that all inducers of CYP2B and CYP3A enzymes also increased the production of hepatic 2OH-E2 in the female rat, but the CYP1A1 inducer was the least potent (Suchar et al. 1996). We could not detect changes in 2OH-E2 either, despite EROD, BROD, and PROD all being elevated at PND21 (Table 2). The 17β-hydroxysteroid dehydrogenase (17β-HSD) is a family of eight non-CYP enzymes responsible for bidirectional E1-E2 in-terconversion (Labrie et al. 2003; Corton et al. 1997). Type 2 17β-HSD is associated with the membranes of the endoplasmic reticulum in the liver, whereas type 4 17β-HSD is localized in peroxisomes of rat livers, and both convert E2 to E1 (Corton et al. 1997). Despite that peroxisome proliferators increase the expression of type 4 17β-HSD (Corton et al. 1997), and PCBs are peroxisome proliferators (Gustafsson 1995), the current investigation did not lead to detectable changes in E1. The limited effects on estrogen metabolism (Figure 1A) is likely due to the low dose of exposure used in our study, and the delay between the end of the treatment at PND20 and necropsy at PND55 to PND62, and later in older rats. Nevertheless, the inducible enzymes responsible for the persistent transformation of ¹⁴C-E2 at PND55 to PND62 remain to be identified, along with the estrogen metabolites that are produced, and further investigation of their implications on adverse effects is required.

Although the production of hepatic 16αOH-E2 did not significantly change with age, that of 2OH-E2 decreased, whereas that of E1 increased. This was confirmed by two different sets of data (Figures 1 and 2). These changes are consistent with reported aging effects on fertility (vom Saal, Finch, and Nelson 1994) and enzymatic activities. The level of CYP3A2 transcripts increase during ontogeny, reaching a maximum at PND60, then decreasing as early as 8 months of age (Rosati et al. 2003; Agrawal and Shapiro 2003). Age effects on preinduction levels of CYP2B are not as evident (Agrawal and Shapiro 2003). CYP1A1 activity decreases with age, with a 60% to 70% reduction at 25 to 26 months, a time when NADPH reductase, a required CYP accessory protein, is also reduced (Warrington, Greenblatt, and von Moltke 2004). The increased production of hepatic E1 with age (Figure 1) is consistent with reports of increased 17β-HSD hepatic activity from prepuberty to 6 (Akinola et al. 1997) or 15 weeks of age (Murray and Horsfield 1990), and slight increases in circulating E1 observed between 3 and 4 months and 10- to 11-month-old female rats (Steger and Peluso 1982). Given the increased production of E1 with age (Figure 1), that E1 is more prone to be metabolized to 4-hydroxyestrogen than E2 (Lee et al. 2002), and the decrease in “protective” (Zhu and Conney 1998b; Zacharia et al. 2004) 2-hydroxylation (Figures 1 and 2), perhaps an increase in the relative exposure to the mutagenic 4OH-estrogens is occurring, which might increase the risk of compromising liver function and precipitate diseases of aging and tumorigenicity.

The 1000× treatment was previously shown to delay the initial development of palpable mammary tumors, but increased the final number of mammary lesions counted at necropsy (Desaulniers et al. 2001). It is difficult to explain these dual effects on mammary tumors. The 1000× treatment had no effects on the development of ovarian follicles (current study), estrous cycle length and time to puberty (Desaulniers et al. 2001), suggesting that ovarian production of E2 was normal. Perhaps a transient increase in the metabolism of estrogens (which in the liver occurred at least until 55 to 62 days of age but not at PND224 [Figure 1A ]) might have reduced the estrogenic signal to the mammary gland and thus the early development of palpable tumors (being mostly estrogen dependent). Alternatively, the production of 2OH-E2 by CYP1A1, which is rapidly converted to 2-methoxyestradiol, might be associated with the delay in tumor development. Indeed, treatment with low doses (1 mg/kg/day for 20 days) of 2-methoxyestradiol inhibited, whereas high doses (5 mg/kg) stimulated MNU-induced mammary tumors in the rat (Lippert et al. 2003). The 1000× treatment might also have interfered with the ER signaling pathway in the mammary tissue, as others have observed following exposure to AhR agonists (Wormke et al. 2003; Safe, Wormke, and Samudio 2000). The increased tumorigenic response observed at necropsy might have been caused by an increase in the metabolism of estrogens, augmenting the production of the mutagenic 4OH-E2 by CYP1B1 (Yue et al. 2003), and perhaps a reduction in COMT. Note that hepatic COMT mRNA is reduced by AhR agonist treatment (Desaulniers et al. 2005). The measurement of 4OH-E2 metabolites was not possible with the current TLC method, consistent with others reporting undetectable hepatic 4OH-E2 from female rats (Suchar et al. 1996), and the fact that 4-hydroxylation is a minor pathway in the female rat liver (Zhu and Conney 1998a). Given the interest in testing associations between breast cancer and the 2OH-/16αOH-E2 ratio (Bradlow et al. 1995; Jernstrom et al. 2003), Figure 2 summarized hepatic 16αOH-E2, 2OH-E2, and the 2OH-/16αOH-E2 ratio. However, no significant correlations were found between these new data, and previously published data on the development of malignant, benign, or total number of mammary lesions from the same rats (Desaulniers et al. 2001). An investigation of the metabolism of estrogens in the mammary tissue was not possible in the current investigation because that tissue was used to count all mammary lesions from whole mount preparations. Note that effects of organochlorine exposure on MNU-induced mammary tumor development might have been influenced by changes in a number of other mechanisms related to cancer initiation and/or promotion.

There are limitations to the extent at which the current findings could be extrapolated to the human. First, despite similarities, the enzymes involved in metabolizing xenobiotics (CYP1–3 family) differ between species. CYP1A1, CYP1A2, and CYP1B1 are orthologous enzymes in human and rats (Nebert et al. 1991; Choudhary et al. 2004); however, the orthologue for the rat CYP2B1 is the human CYP2B6 (Lewis et al. 1999), and for the rodent CYP3A1/2 the human counterpart are CYP3A7 prenatally, but CYP3A4/5 postnatally (Lu and Waxman 2005; Lacroix et al. 1997). Second, there are species differences in the age at which these enzymes are expressed, which may lead to age differences in the accumulation and production of metabolites (Johnson 2003; Rich and Boobis 1997; Rosati et al. 2003). In humans, for example, hepatic CYP1A1 and CYP1B1 are declining with increasing age in utero, and although they become no longer detectable in neonates, CYP1A2 is increasing and becomes readily measurable in infants at 1 month of age (Hines and McCarver 2002; Sonnier and Cresteil 1998). In contrast, CYP1A1 is the predominant CYP1 family member in the rat (Hines and McCarver 2002; Choudhary et al. 2004). These differences lead to difficulties in the extrapolation of data between species and age groups, and therefore for risk assessment calculations, the reference doses or acceptable daily intake values are corrected by various uncertainty factors, including some for interspecies differences (Dorne 2004), and a recently proposed factor to ensure protection of infants and children (Faustman and Omenn 2001). Finally, the levels of organochlorines in breast milk have been declining (World Health Organization 1996; Craan and Haines 1998; Norén and Meironyté 2000; Solomon and Weiss 2002), possibly by half every 9.6 years for AhR agonists (World Health Organization 1996), 9% per year for DDE, and 8% per year for PCBs and DDT (Dallaire et al. 2003).

In conclusion, (1) 100× was the smallest dose of the mixture to induce statistically significant effects; (2) changes in the metabolism of estrogens may provide a good indication of persistent effects given that they remain for a longer duration than other end points; and (3) regardless of treatment, hepatic estrogen metabolism changes with age, with a reduction in 2OH-E2.