Environ. Sci. Technol. 2009, 43, 2144–2150
Endocrine Disrupting Effects of Herbicides and Pentachlorophenol: In Vitro and in Vivo Evidence F R A N C E S O R T O N , * ,† I L K A L U T Z , ‡ WERNER KLOAS,‡ AND EDWIN J ROUTLEDGE† Institute for the Environment, Brunel University, Kingston Lane, Uxbridge UB8 3PH, U.K., and Leibniz-Institute of Freshwater Ecology and Inland Fisheries, Department of Aquaculture and Ecophysiology, Mu ¨ ggelseedamm 310, 12 587 Berlin, Germany
Received October 15, 2008. Revised manuscript received January 20, 2009. Accepted January 23, 2009.
The potential for agricultural chemicals to cause endocrine disruption (ED) in humans and wildlife is an increasing concern; however, the effects of commonly used pesticides at environmentally relevant concentrations are largely unknown. Therefore, 12 environmentally relevant pesticides (11 herbicides and pentachlorophenol (PCP)) were tested for their endocrine disrupting potential in two in vitro assays. A recombinant yeast screen was used to detect receptor mediated (anti-) estrogenic and (anti-) androgenic activity (concentration range: 0.01-1000 µM), and cultured Xenopus oocytes were used to measure effects on the ovulatory response and ovarian steroidogenesis (concentration range: 0.00625-62.5 µM). Eleven pesticides were active in at least one assay (isoproturon, diuron, linuron, 4-chloro-2-methylphenoxy acetic acid (MCPA), mecoprop, atrazine, simazine, PCP, trifluralin, chlorpropham, bentazone), and one had no effect (2,4-dichlorophenoxy acetic acid (2,4,-D)). The most common effects were antiestrogenic/ antiandrogenic activity in the yeast screen, and inhibition of ovulation in vitro, accompanied by decreased testosterone production. Estrogenic activity was never observed. In addition, the most potent compound identified in vitro (PCP) was tested for ED activity in vivo. A short-term exposure (6 days) of adult female Xenopus to low concentrations (0.1 or 1 µg/L; 0.375 or 3.75 nM) resulted in minor alterations in plasma hormone levels and toxic effects on the ovary. Changes in in vitro human chorionic gonadotropin (hCG) stimulated hormone production in ovarian follicles from exposed individuals was also observed. In conclusion, novel effects of herbicides and PCP at environmentally relevant concentrations were found, and the effects of these compounds on humans and/or wildlife warrant further investigation.
Introduction The potential for agricultural chemicals to cause endocrine disruption (ED) is an increasing concern, both in humans (1) and in wildlife (2, 3). The majority of published research on pesticides and ED has focused on organochlorine * Corresponding author phone 07531773517;
[email protected]. † Brunel University. ‡ Department of Aquaculture and Ecophysiology. 2144
9
e-mail:
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 6, 2009
insecticides due to their persistence and ability to bioaccumulate in the environment; however, their use has been largely superseded by less persistent pesticides. Indeed, the most commonly used pesticides worldwide are herbicides (4), which generally have shorter half-lives in soil and water (5), but are found at much higher concentrations in the environment. Indeed, herbicides and the general use biocide PCP were the most frequently detected pesticides, and were detected at the highest concentrations, in UK freshwaters in 2004/2005 (Supporting Information (SI) Table S1). The agricultural environment has been shown to cause ED in humans (6, 7) and wildlife (8, 9), and pesticides have been detected in drinking water (10); however, relatively few environmentally relevant pesticides have been tested for specific effects on endocrine end points. Furthermore, those compounds that have been tested more extensively have been reported to be endocrinologically active, suggesting that structurally related, but untested, compounds may also cause ED. For example, the phenylurea herbicide linuron binds to the androgen receptor (AR), and is an antiandrogen in vitro (11), and has also been shown to inhibit prostatic 5R-reductase activity in human tissue homogenates (50% at 86 µM (12)). Studies have also demonstrated antiandrogenic effects of linuron on reproductive parameters in vivo in mice (100 mg/kg/day (13)), and in sticklebacks (150 µg/L (14)). However, the effects of the phenylurea herbicides isoproturon and diuron are poorly understood. Similarly, the triazine herbicide atrazine upregulated aromatase in human adrenocortical and placental carcinoma cells in vitro (0.3-30 µM (15)), inhibited 5R-reductase in fish testicular homogenate (100 µM (16)), inhibited 3R-hydroxysteroid dehydrogenase (HSD)/17β-HSD activity in cultured rat pituitary cells (0.92 µM (17)), and caused gonadal abnormalities in amphibians (9, 18), However, the effects of the related triazine herbicide simazine have not been extensively tested. The phenoxy acid herbicide 2,4-D has been described as a “hormonal” herbicide, and was reported to inhibit ovulation of Xenopus oocytes (19), however, the effects of other common phenoxy acids have barely been investigated (e.g., MCPA and mecoprop). In addition, the general use biocide PCP was selected for testing as it is common in UK freshwaters and has been shown to inhibit Zebrafish ovulation in vitro (0.6 µM (20)) and to reduce the number of eggs laid, their subsequent hatching rates and to induce formation of testis-ova in Japanese medaka (50-100 µg/L (21)), although it is mechanism of action is unknown. Therefore, in this study, eleven environmentally relevant herbicides (phenoxy acids -4chloro-2-methylphenoxy acetic acid (MCPA), mecoprop, 2,4dichlorophenoxy acetic acid (2,4,-D); phenylurea: isoproturon, diuron, linuron; triazines: simazine, atrazine; others: bentazone, trifluralin, chlorpropham); and PCP were tested for receptor-mediated and steroidogenic activity in two in vitro assays. Estrogenic/androgenic and antiestrogenic/ antiandrogenic receptor-mediated activity of the selected pesticides was tested in a recombinant yeast assay (22) and disruption of the steroidogenic pathway was tested using cultured Xenopus oocytes. Xenopus oocytes have long been used as a model for investigating cellular pathways, as they are large, easy to maintain in the laboratory and enzymatic pathways, including steroidogenesis, are highly conserved in vertebrates (23). Incubation of oocytes with hCG stimulates steroidogenesis, especially testosterone production, and results in ovulation of the oocyte (SI Figures S1 and S2). Although steroidogenesis is a less well researched area than receptor-mediated effects, it is increasingly gaining attention as an important target for endocrine disrupting compounds 10.1021/es8028928 CCC: $40.75
2009 American Chemical Society
Published on Web 02/17/2009
(24) and ovulation has previously been shown to be sensitive to pesticide exposure including methoxychlor (25) and PCP (20). In vitro tests are a useful first step for prioritising chemicals that require further testing and therefore, the predictive ability of these screens was tested with the most active pesticide identified in vitro (PCP), by means of a short in vivo exposure using adult female Xenopus. The overall aim of this study was to investigate the effects of common herbicides and PCP on endocrine end points at environmentally relevant concentrations.
Materials and Methods Chemicals. 17β-estradiol (>98% pure), testosterone (>98% pure), progesterone (>99% pure), flutamide (98% pure), 4-hydroxytamoxifen (>98% pure), and all pesticides (>97% pure) were obtained from Sigma Chemical Company Ltd. (Dorset, UK). Cell culture media components, radioimmunoassay buffer components, and hCG were purchased from Sigma (Dorset, UK). Epostane was a gift from V. Luu-The´ (Oncology and Molecular Endocrinology Research Center, Canada). Hormones and epostane were dissolved in ethanol (HPLC grade) to make a stock solution of 10 mM. Pesticides were dissolved in ethanol (HPLC grade) to make a stock solution of 20 mM, except simazine, which was dissolved in methanol (HPLC grade) at a concentration of 2 mM. Yeast Screen. The methods for (anti-) yeast estrogen screen (YES) and (anti-) yeast androgen screen (YAS) assays were described previously (YES (22), YAS (26)). Briefly, stimulation of the transfected human estrogen or androgen receptor causes a color change in the media, which is measured by absorbance at 540 nm (Spectramax 340pc, Molecular Devices, CA). Plates were also measured at 620 nm to measure the cell growth (turbidity) and therefore check for any cytotoxic effects that may have occurred. The antiscreens function by inhibition of receptor binding, via coincubation with the agonist (estradiol (0.25 nM) or testosterone (2.5 nM)) and the test compound or the positive control agent (OHT: 4-hydroxytamoxifen (0.01-25 µM) or flutamide (0.02-50 µM)). Initially, pesticides were added to wells over the range of 1000-0.49 µM, but turbidity readings were unacceptably low for some pesticides (statistically significant decreases), indicating a toxic effect on the yeast cells. Therefore, isoproturon and trifluralin were subsequently tested over the range of 125-0.06 µM, diuron, linuron, and chlorpropham over the range of 15.6-0.008 µM, PCP over the range of 7.8-0.004 µM. Where cell turbidity was significantly reduced compared to the ethanol or media only controls, data was omitted from statistical analysis. Pesticides were tested in triplicate over three plates, and over two separate experiments. Absorbance was measured after 3-6 days incubation, depending on the assay. Ethanol and media only controls were also run in each assay. Ovulation Assay. Sexually mature female Xenopus laevis (a gift from Jane Kirk, Cancer Research Institute, UK) were anaesthetised by submersion in 3-aminobenzoic acid ethyl ester (MS222) until reflexes ceased and were then sacrificed by pithing the spinal cord. The ovaries were removed and placed in a glass Petri dish containing Modified Barth’s Media (MBS: 88 mM NaCl, 1 mM KCl, 0.41 mM CaCl2, 2.5 mM NaHCO3, 0.82 mM MgSO4, 0.33 mM Ca (NO3)2, buffered 10 mM HEPES, pH 7.6, Penicillin/Streptomycin (5000 U/50 mg per mL), sterilized by filtration). They were cut into 10 oocyte fragments (7 or 8 stage V/VI and 2 or 3 stage I-IV), and two fragments were cultured per well in 24-well plates. After 20 h incubation, media was sampled, frozen on dry ice, and stored at -80 °C until hormone analysis (by radioimmunoassay (RIA)). In addition, the oocytes were tested for viability with trypan blue solution (0.2%), fixed with trichloroacetic acid (TCA: 5%) and the number of ovulated oocytes were counted. Ovulated oocytes were identified by dissociation from
follicular tissue and verified by presence of the “white spot”. Oocytes were initially incubated with 50, 25, 12.5, 6.25 IU hCG or control media, for 20 h. On day two a submaximal concentration of hCG (∼ 60% of maximum ovulatory response) was chosen for coincubation with pesticides. Oocytes were cultured in control media, hCG alone, or coincubated with a pesticide (sextuplicate wells). The ovulation assay was repeated three times with modifications to pesticide concentrations. Initially, all pesticides were tested at 62.5 and 6.25 µM, and pesticides that had no effect or only an effect at 62.5 µM, were retested at the same concentrations. Those that had an effect at 62.5 and 6.25 µM were additionally tested at 0.625, 0.0625, and 0.00625 µM (PCP, chlorpropham, atrazine). The complete range was then repeated for each pesticide, and therefore, each concentration of each pesticide was tested two or three times. In addition, the 3β-HSD inhibitor epostane was tested at 1, 0.1, 0.01, and 0.001 µM, to verify inhibition of ovulation via inhibition of steroid hormone synthesis (SI Figure S2, ref 27). In Vivo Exposure. Fifty-four adult female Xenopus laevis (4 years old) were taken from the breeding stock of the Leibniz-Institute of Freshwater Ecology and Inland Fisheries (IGB, Berlin). The frogs were fed twice per week prior to exposure and the light:dark cycle was 12:12 h. They were placed in 9 × 12 L glass aquaria (three per treatment), containing 10 L reconstituted tap water (distilled water supplemented with 2.5 g marine salt, Tropic Marin Meersalz, Tagis, Dreieich, Germany), at a loading density of six animals/ tank, and were not fed during the exposure period (6 days). PCP has a log Kow of 5.18 and has been shown to bioaccumulate (28), thus it was tested for effects in vivo at environmentally relevant concentrations (0.1 and 1 µg/L ) 0.375 and 3.75 nM), with the assumption that it would be bioaccumulated to higher levels within the animal. A PCP stock solution of 1 mg/L was prepared in distilled water at the beginning of exposure (day 0), and was stored at 4 °C during the exposure period and tanks were redosed every 48 h during a complete water change (days 2 and 4). Temperature and pH were also measured prior to and following each water change, and water samples were taken from each tank at the same time for chemical analysis of PCP. Analysis of PCP was undertaken using HPLC (for method details see SI text S1). At the end of the exposure period, animals were bled, total weight, liver weight, and ovary weight were recorded. Blood was centrifuged at 6000g for 1 min, and plasma was removed and snap frozen in liquid nitrogen. The liver and ovary were removed and weighed, and ovarian follicles were cultured in MBS supplemented with 20 IU hCG in sextuplicate wells. Media was removed after 20 h, frozen, and stored at -20 °C. RIA. Unextracted media samples and plasma were tested for progesterone, testosterone, and estradiol concentration. Prior to plasma hormone RIA, hormones were twice extracted with ethyl acetate (3:1 ratio addition, shaken vigorously (10 min), centrifuged (2000g, 2 min)). The supernatant (ethyl acetate) was removed and dried with purified nitrogen gas. 650 µL MBS was then added to each dry tube, and they were vortexed vigorously. Media assay tubes contained 300 µL test media (for testosterone, 200 µL of test media was diluted with 100 µL of culture media), 100 µL of antibody, and 100 µL of radiolabeled steroid (total ) 500 µL). Plasma hormone tubes contained 100 µL of extracted sample, 100 µL of antibody, and 100 µL radiolabeled steroid (total ) 300 µL). Antibodies (AbD Serotec, Morphosys AG, Germany) were added at final concentrations of 1:500 for progesterone (cross reactivity: 0.02% cortisone only), 1:4000 for testosterone (cross reactivity: 11β-hydroxy testosterone 3.3%, 5R dihydroxy testosterone 2%), and 1:2000 for estradiol (cross reactivity: estrone 14%, estriol 5%). Radiolabeled steroids, [1,2,6,7-3H] progesterone, [1,2,6,7-3H] testosterone, and [2,4,6,7-3H] esVOL. 43, NO. 6, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2145
FIGURE 1. Activity of standards (gray line) and pesticides in the yeast antiestrogen (A) and antiandrogen screen (B-D). Values are mean ( SE, n ) 3 or 6 for the pesticides, and 4 for the standards. Cont ) ethanol control, OHT ) hydroxy-tamoxifen, Flut ) flutamide, Diu ) diuron, Lin ) linuron, Iso ) isoproturon, Atr ) atrazine, Sim ) simazine, PCP ) pentachlorophenol, Cprop ) chlorpropham, Tri ) trifluralin. tradiol (Perkin-Elmer, MA), were added at 120 000 counts/ value to allow for variation between experiments prior to mL. Antibodies and radiolabeled steroid were diluted in 0.2 statistical analysis. Data was pooled from replicate experiM phosphate buffered saline, supplemented with 0.02% ments, resulting in n values of 3 or 6 for yeast data, 6, 12, or sodium azide and 0.5% bovine serum albumin (sPBS) (Sigma, 18 for ovulation data. Media hormone data was analyzed Dorset, UK). After addition of all components, tubes were without conversion. For the in vivo exposure, liver and ovary vortexed and incubated overnight at 4 °C. To separate boundweight were converted to hepatic somatic index (HSI) and free hormone, 500 µL of activated charcoal slurry (0.5% gonadal somatic index (GSI), by dividing organ weights by charcoal and 0.05% dextran in sPBS), was added to each the total weight of the individual and these values were tube. Tubes were incubated on ice (10 min), and centrifuged analyzed statistically. Plasma hormone data, ovulation data, (2000g, 15 min). The supernatant and scintillation cocktail and oocyte abnormality data were log transformed for (5 mL) were added to scintillation vials., and radioactivity normalization, prior to statistical analysis; however, media was measured (Tri-Carb, Packard Instrument Company, CT). hormone data could not be normalized in this way and were Media samples were tested singly over two assays, and plasma analyzed nonparametrically. samples tested in duplicate in one assay. The detection range Results for estradiol and progesterone was 109-3500 pM, and for Yeast Screen. All yeast screen data was analyzed paratestosterone was 875-7000 pM. For media samples, intermetrically. None of the pesticides tested were agonistic in assay coefficients of variance were 22% for progesterone, 6% the estrogen or androgen screen (data not shown), however, for testosterone, and 12% for estradiol. For plasma samples, antiestrogenic/antiandrogenic activity was often observed intrassay coefficients of variance were 22% for progesterone, (Figure 1). Cytotoxicity was also observed, therefore, the top 15% for testosterone, and 12% for estradiol, and extraction concentrations reported refer to the highest concentration efficiencies were 68% for progesterone, 86% for testosterone, eliciting an effect on the receptor, and without affecting and 73% for estradiol. turbidity readings (pesticides were cytotoxic above these Statistics. Data was tested for normality using the Shapiroconcentrations). PCP was the most potent compound tested Wilk W test. For normally distributed data, ANOVA was used and was antiestrogenic from 0.015 to 7.8 µM (p < 0.004), and to find differences between groups. For nonparametric data, antiandrogenic from 0.015 to 3.9 µM (p < 0.02). Other Kruskal-Wallis test was used to find differences between antiestrogenic pesticides were diuron (0.98-31.3 µM, p < groups. Dunnett’s test was used as a posthoc test in both 0.004), linuron (1.9-31.3 µM, p < 0.02), isoproturon (15.6-250 cases, to test differences from the control. Yeast data and µM, p < 0.02), atrazine (125-1000 µM, p < 0.04), and ovulation data were converted into percentages of the control 2146
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 6, 2009
TABLE 1. Ovulatory Response and Hormone Concentrations in Oocyte Incubation Media after Incubation with Pesticides and/or hCG (20 hrs)a treatment hCG ISO DIU LIN MCPA MPROP ATR
SIM PCP
CPROP
conc. µM (µg/L)
prog. (pM)
test. (pM)
E2 (pM)
ovulation (%)
N/A 62.5 (12893) 6.25 (1289) 62.5 (14569) 6.25 (1457) 62.5 (15569) 6.25 (1557) 62.5 (12539) 6.25 (1254) 62.5 (13413) 6.25 (1341) 62.5 (13481) 6.25 (1348) 0.625 (135) 0.0625 (14) 62.5 (12606) 6.25 (1261) 62.5 (16646) 6.25 (1665) 0.625 (167) 0.0625 (17) 0.00625 (2) 62.5 (13354) 6.25 (1335)
374 ( 28 394 ( 37 515 ( 93 372 ( 30 416 ( 64 634 ( 58 396 ( 43 493 ( 48 399 ( 38 363 ( 41 430 ( 53 762 ( 68 609 ( 58 419 ( 50 389 ( 13 398 ( 22 512 ( 87
