Potential endocrine disruptor activity of drinking water samples

Abstract

Conventional water treatment plants (WTP) do not completely remove contaminants with endocrine activity which may then be present in drinking water (DW). The potential for endocrine disruption of 2 DW samples collected in 2010 and 2012 from a conventional WTP in São Paulo, Brazil was investigated. In vivo assays were conducted with 21-day old female rats exposed to DW extracts for 3- (uterotrophic assay) or 20-days (pubertal assay). The exposure represented a daily ingestion of 2 L, 10 L and 20 L of DW per 60 kg-body weight. Caffeine (5.8 – 21 ug/L), estrone (1 ng/L), atrazine (2.2 – 11.2 ng/L), carbendazim (0.22 ng/L), azoxystrobin (0.23 ng/L), tebuconazole (0.19 ng/L) and imidacloprid (0.88 ng/L) were detected in DW extracts by LC-MS/MS. No increase in uterus wet weight in the uterotrophic assay, and no alteration of vaginal opening in the pubertal assay were observed. However, there were increased absolute blotted uterus weights in animals treated for 3-days with the 3 doses of both DW samples. LH and FSH levels showed significant dose-response increases in the uterotrophic assay using the 2010 DW sample, in association with a significantly increased incidence of vaginal keratinization after the 3-day exposure. The pubertal animals exposed to the 2010 DW had a significant body weight gain and decreased LH at the highest dose. Results suggest that DW samples tested exerted estrogenic and hypothalamic-hypophysis activity alterations in vivo.

Keywords:

• emerging contaminants
• endocrine disruption
• pesticides
• pubertal female rat development
• uterotrophic assay

abbreviations

DW	=	drinking water
ED	=	endocrine disruptors
E1	=	Estrone
E2	=	17-β estradiol
FSH	=	Follicle-Stimulating Hormone
LC-MS/MS	=	Liquid Chromatography Tandem Mass Spectrometry
PND	=	Post Natal Day
LH	=	Luteinizing Hormone
P4	=	Progesterone
PRL	=	Prolactin
T	=	testosterone
VO	=	vaginal opening
WTP	=	water treatment plants
ER-α/β	=	estrogen receptor α or β
PR	=	progesterone receptor
Hoxa-10	=	Hox multigene family homeobox A
CaBP-9k	=	cytosolic calcium binding protein.

Introduction

The increasing release of raw or treated sewage into surface waters has drawn attention to an important current issue: the water quality. Surface waters are treated for drinking purposes usually through conventional processes which are not able to completely remove several groups of compounds including endocrine disruptors.^1,2 In Brazil, many studies confirmed that sewage discharges, with or without previous treatment, are responsible for the presence of estrogens and xenoestrogens in waters.^3-5 São Paulo State has only 50% of its sewage production treated, and high levels of different classes of chemicals (pharmaceuticals, personal care products, hormones, drugs of abuse, pesticides, industrial chemicals) have been found in surface and drinking water (DW).^6-9 As a consequence, humans are continuously exposed to several estrogens, anti-estrogens, androgens, anti-androgens and steroidogenic substances, and this has stimulated interest in assessing the effects associated with simultaneous exposure to these chemicals^10,11 because low concentrations of some environmental contaminants may cause adverse effects both in humans and biota.^12,13

A significant increase in reproductive problems in some regions of the world over the last few decades points to unidentified environmental factors in disease etiology, and ED exposures have recently been linked with obesity, cardiovascular disease, diabetes and metabolic syndrome in human adults.¹⁴ Given the adverse effects that these substances can cause in humans and wildlife, it is extremely important to elucidate the possible consequences of the occurrence of these endocrine disruptors in the aquatic environment. Although the scientific literature deals with the characterization of local industrial/domestic effluents and river quality, endocrine disruptors occur globally,^15-17 and these observations indicate that environmental ED compounds have already been found in DW, leading to direct human exposure.

In Brasil, water bodies have been chemically evaluated to determine ED contaminant concentrations of chemicals such as triazoles, triazines, strobilurins, organophosphates, phenyl pyrazole, benzimidazole, caffeine, bisphenol A, natural and synthetic hormones, phenols, bactericides, inseticides and herbicides.^4,5,7,9,18 The potential of the concentrations of these contaminants present in 2 DW samples collected at a Water Treatment Plant (WTP) of São Paulo, Brazil to induce endocrine effects in juvenile female rats was measured using the uterotrophic and pubertal female development in vivo rat assays. Additional sensitive end-points were added to identify the mode of action of the effects seen.

Results

Chemical analyses

Atrazine, caffeine, estrone, carbendazim, imidacloprid, azoxistrobin and tebuconazole were found, in one or both DW extracts (Table 1). For both DW samples the concentrations of 24 of the 31 chemicals analyzed were below the limit of quantification. We did not analyze the 2010 DW for the other pesticides because the methods were not validated at the time. Estrone was detected at 1 ng/L in the 2010 DW, but not in the 2012 DW.

Table 1. Compounds identified by Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS)in the drinking water (DW) extracts

	DW extracts
	2010		2012
Target compounds	Concentration ng/L	LOQ ng/L	Concentration ng/L	LOQ ng/L
17α-ethynylestradiol	<LOQ	13.9	<LOQ	13.9
17β-estradiol	<LOQ	5.9	< LOQ	5.9
4-n-nonylphenol	<LOQ	10.6	<LOQ	10.6
4-n-octylphenol	<LOQ	5.1	<LOQ	5.1
Atrazine	11.2	0.7	0.8	0.2
Azoxystrobin	Not analyzed		0.2	0.2
Bisphenol A	<LOQ	1.5	<LOQ	1.5
Bromacil	Not analyzed		<LOQ	0.5
Caffeine	5,800	15.4	56.0	0.6
Carbendazim	Not analyzed		0.2	0.2
Chlorpyrifos			<LOQ	0.9
Dietilstilbestrol	<LOQ	3.4	<LOQ	3.4
Difenoconazole	Not analyzed		<LOQ	0.2
Epoxiconazole			<LOQ	0.7
Estriol	<LOQ	0.8	<LOQ	0.8
Estrone	1	0.8	<LOQ	1.0
Fipronil	Not analyzed		<LOQ	18.0
Hexythiazox			<LOQ	1.5
Imidacloprid			0.9	0.7
Levonorgestrel	<LOQ	0.9	<LOQ	0.9
Mestranol	<LOQ	8.8	<LOQ	8.8
Phenolphthalein	<LOQ	3.7	<LOQ	3.7
Picoxystrobin	Not analyzed		<LOQ	0.7
Profenofos			<LOQ	1.5
Progesterone	<LOQ	2.7	<LOQ	2.7
Pyraclostrobin	Not analyzed		<LOQ	0.3
Tebuconazole			0.2	0.2
Testosterone	<LOQ	1.7	<LOQ	1.7
Triclosan	<LOQ	0.7	<LOQ	0.7
Trifloxystrobin	Not analyzed		<LOQ	0.2

LOQ – limit of quantification.

In vivo studies

Uterotrophic bioassay

In-life parameters and organ weights

The 2010 DW sample caused a significant decrease in food intake and final body weight, especially in the highest DW extract treatment group, which reflected in the body weight gain at the end of the experiment. We also found a significant reduction of the absolute liver weights of these animals (Table S1). There was no effect on food or water consumption, or body or organ weights in animals exposed to the 2012 DW sample (data not shown).

There were no differences in uterine wet weight among the treated groups for both DW extracts. The absolute blotted uterus weights increased, but the increase was statistically significant only in the immature rats exposed to the middle and highest 2010 DW doses (Table 2). The overall data on this parameter show that there was a uterine weight alteration in the 2010 and 2012 DW -treated groups when compared to the ultrapure water extract group (extract process negative control). We also observed an increase in the blotted and wet uterine weights for the estrogenic positive control group (17-α-ethynylestradiol), as expected.

Table 2. Absolute blotted uterine weight and endometrial thickness observed in the uterotrophic assay of the drinking water (DW) extracts and atrazine

	Uterine weight (mg) Median (Q1-Q3), n = 12		Endometrial thickness (m) Median (Q1-Q3), n = 6
Groups	2010	2012	2010	2012
Deionized water (vehicle control)	30 (24–37)		94 (61–140)
Extract of ultrapure water (extract process control)	36 (34–39)	26 (23–36)	95 (65–130)	92 (63–123)
Extract of water sample: 33.3 ml/kg bw	38 (35–41)	30 (23–36)	99 (69–144)	103 (57–135)
Extract of water sample: 166.5 ml/kg bw	48* (46–54)	43 (33–63)	146* (115–199)	114* (78–170)
Extract of water sample: 333 ml/kg bw	49* (36–52)	32 (23–37)	91 (70–114)	114* (80–159)
Estrogenic positive control 17α-ethynylestradiol (1ug/kg bw)	117* (113–132)	128* (123–172)	216* (173–248)	177* (99–269)
Atrazine (3.73 ng/l/kg bw)	30 (27–36)		96 (73–119)

(n) number of animals per group; (*) statistically different (Dunn's test) from the Extract of ultrapure water group at p < 0.05.

Histology and morphometry of reproductive organs

The histology of ovaries, adrenal glands, pituitary, thyroid glands and liver was unaltered after the DW exposure, but there was a significant increase of the endometrial thickness in animals exposed to the 2010 DW middle dose (Table 2). The middle and highest doses of the 2012 DW also induced increases in the thickness of the uterine tissue.

There was no VO before the day of necropsy. However, there was a significant increase in the incidence of animals with keratinization of the vaginal epithelium after exposure to the highest 2010 DW (Fig. 1), but not to the 2012 DW. There were no vaginal epithelium alterations in the deionized water or atrazine groups.

Figure 1. Percentage of animals with the keratinization of the vaginal epithelium after 3 days of exposure to the 2010 drinking water extract. Groups: 1: Extract of ultrapure water (extract process control); 2: Extract of water sample: 33.3 ml/kg bw; 3: Extract of water sample: 166.5 ml/kg bw; 4: Extract of water sample: 333 ml/kg bw; 5: Estrogenic Positive control 17-α-ethynylestradiol (1 μg/kg bw). (*) statistically different (Fischer test) from the Extract of ultrapure water group at P < 0.05.

The numbers of total follicles per ovary area were reduced at the middle and highest 2010 DW doses (Figure S1), especially the primordial and (pooled) primary follicles (p = 0.048, middle; p = 0.011 highest dose group).

Hormone measurements

The 2010 and 2012 DW hormonal measurements are presented in Figure 2. LH and FSH serum levels were increased by treatment with the highest dose of 2010 DW, but no changes in the estradiol serum levels were observed in either DW sample. The serum level of estrone was just measured in the 2010 DW sample, and the DW treated groups did not show a change when compared to the ultrapure water extract group. There were also no significant differences between groups in the progesterone and prolactin serum concentrations (hormones only quantified in the 2010 DW). Serum levels of testosterone did not decrease significantly after treatment with either DW sample.

Figure 2. Hormone levels of immature rats exposed to 2010 and 2012 water samples. Data is represented as mean ± standard error. FSH: follicle-stimulating hormone; LH: luteinizing hormone; P4: progesterone; E₂: estradiol; E₁: estrone; T: testosterone; PRL: prolactin. (*) statistically different (Dunn's test) from the Extract of ultrapure water group at P < 0.05. The number of animals sample assayed is presented below each column.

Gene transcripts

The uterine transcript levels from the 2010 and 2012 DW treatments were within the normal limits of relative quantitation (qR) expression (0.5–2.0) calculated by 2^(-ΔΔCt) (Fig. 3), except the 2010 DW 333 mL eq CaBP-9k gene expression (median of 2.74 among 5 animals). There were differences in gene expression for the estrogenic-positive EE2 control group: the ER- α/β decreased and the PR, Hoxa-10 and CaBP-9k increased. The DW results were very different from the EE2 response, but we observed a trend in the 2010 DW and the ultrapure water extract groups: the Hoxa-10 was reduced and the CaBP-9k was increased.

Figure 3. Relative quantitation of target genes from immature female uterus rats after 3-days exposure to the 2010 and 2012 water sample. Data is represented as mean ± standard deviation. Number of animals per group: 05. (*) statistically different (Dunn's test) from the Extract of ultrapure water group at P < 0.05.

Additional groups (Figures 2, and 3, Table 2, and Table S1)

The group treated for 3 days only with atrazine showed similar results as the negative control with ultrapure water extract for all the evaluated parameters, except for a significant increase in the CaBP-9k gene expression. The group treated with pure deionized water (vehicle control group) was not different from the one that was exposed to the ultrapure water extract.

Pubertal development and thyroid function in intact juvenile/peripubertal female rats

In-life parameters and organ weights

The 2010 DW sample had no effect on food or water consumption, but body weight gain was significantly reduced in the middle and highest dose groups. The liver absolute weight was also reduced in the same groups (Table 3).

Table 3. Body weight gain and absolute blotted uterine and liver weights in the female pubertal assay for the drinking water (DW) extracts

	Body weight (g) Median (Q1-Q3) (n = 15)			Uterus (g) (n = see below)	Liver (g) (n = 15)
Groups	Initial	Final	Gain^1	Absolute weight Median (Q1-Q3)
Extract of ultrapure water (extract process control)	40 (34.75–46)	155 (149.25–161)	115 (112.25–116.75)	0.26 (0.18–0.3)^2	7.46 (7.11–7.28)
Extract of water sample: 33.3 mL/kg bw	46 (43–52.75)	151 (139.25–162.75)	105 (100.25–113)	0.15 (0.13–0.22)^3	6.52 (6.17–8.01)
Extract of water sample: 166.5 mL/kg bw	44 (42.25–47)	148 (139.5–156.5)	105* (98.25–110.75)	0.18 (0.13–0.29)^3	6.59* (5.79–6.96)
Extract of water sample: 333 mL/kg bw	44 (39.25–47.5)	148 (139.25–160.5)	105* (99–111.5)	0.22 (0.18–0.27)^2	6.78 (6.33–7.7)

(n) number of animals per group; ¹ Body weight gain: final minus initial body weight;² n = 9; ³ n = 12; (*) statistically different (Dunn's test) from the Extract of ultrapure water group at P < 0.05.

Histology, vaginal opening and estrus cyclicity

The histology of the thyroid, uterus, ovaries, vagina, liver and adrenals were not altered after the 2010 DW exposure. In contrast to the uterotrophic assay, there were no changes in any class of ovarian follicles of these animals treated with the same DW sample.

Female pubertal parameters were not affected by the DW extracts; there were no changes in the age of VO, body weight at VO, mean age at first vaginal estrous, and mean cycle length (not shown). The percentage of animals cycling at the end of experiment was 100% in all groups, except for one female at the highest dose that stopped cycling. The percentage of treated females that were regularly cycling (having recurred 4- to 5-day cycles) was 53.3%, 66.7% and 80% for the lowest, middle and highest DW doses, respectively. In the lowest DW dose group (53.3%), the percentage of females cycling regularly was significantly reduced when compared to the extract of ultrapure water group (93.3%).

Hormone measurements

The LH levels of the lower and highest doses of the 2010 DW, but not the middle dose, decreased significantly. However, no significant alterations in FSH, P4, E2 or E1 levels were observed at any tested dose (Fig. 4).

Figure 4. Hormone levels of female pubertal development assay exposed to 2010 water sample. Only animals in estrous cycle were analyzed. Data is represented as mean ± standard error. FSH: follicle-stimulating hormone; LH: luteinizing hormone; P4: progesterone; E₂: estradiol; E₁: estrone. (*) statistically different (Dunn's test) from the Extract of ultrapure water group at P < 0.05. The number of animals sample assayed is presented below each column.

Gene transcripts

The transcript levels of the genes investigated in the uteri of female rats examined during estrus were within the normal limits of relative quantitation, but the Hoxa-10 and CaBP-9 gene expressions were modified, although not statistically significant, and similar to what was seen in the uterotrophic assay (2010 DW sample) (Fig. 5).

Figure 5. Relative quantitation of target genes from female pubertal development assay exposed to 2010 water sample. Data is represented as mean ± standard deviation. Number of animals per group: 5. (*) statistically different (Dunn's test) from the Extract of ultrapure water group at P < 0.05.

Discussion

Exposure to drugs, natural and synthetic hormones, and chemical pollutants can affect the endocrine system, and some of these endocrine disrupting chemicals may also interfere with the developmental processes of humans and wildlife species.^14,19,20 Studies carried out around the world^15-17 including in Brazil^3,18 confirmed that many estrogens and xenoestrogens, may end up in water bodies through sewage discharges, with or without previous treatment. Contaminated rivers are conventionally treated by Water Treatment Plants (WTP) to produce drinking-quality water, but many contaminants may remain in the water served to the public. DW distributed to a specific city in São Paulo State (Campinas) was used in this study, but the same scenario is relevant to many places around the world.

In the present study, we selected a sampling site that has been extensively evaluated in earlier studies^4,5,7,9,18 to verify if the concentrations of contaminants found in the water could induce endocrine effects in juvenile female rats. As a part of this comprehensive study, raw (untreated) water from Atibaia River showed estrogenic activity when evaluated using the Bioluminescent Yeast Estrogen Assay (BLYES).²¹ Bisphenol A and estrone were the most frequently detected compounds (60%), followed by estriol (40%), 4-n-octylphenol (40%), and 4-n-nonylphenol (20%).⁶ In another study carried out in this same region, Sodré et al.²² detected 6 contaminants in drinking water samples; caffeine, bisphenol A, cholesterol and stigmasterol. Estrone and estradiol were detected only in samples collected during the dry season. These authors also showed that the concentrations of caffeine and bisphenol A were higher than the median levels found in similar matrices around the world. Antibiotics also were detected in drinking water samples from the Atibaia River^7,9 in similar levels as in the present study.

We used Tier 1 in vivo assays to explore effects of low doses of these compounds after in vivo metabolism: the uterotrophic assay, a very sensitive bioassay for estrogenic activity in immature female rats that have an intact hypothalamic-pituitary-gonadal (HPG) axis, and the female pubertal assay, which is capable of detecting EDs that operate at a variety of modes of action, such as strong and weak (anti)estrogenic and (anti)androgenic compounds, steroid biosynthesis, aromatase or 5 α-reductase inhibitors, thyroid-active agents, and compounds affecting the HPG or HPT (hypothalamic-pituitary-thyroid) axis.²³

The uterotrophic bioassay responds to substances that interact with the HPG axis rather than with the estrogen receptor, alone.²⁴ After 3 days of exposure to the 2010 and 2012 DW samples, we observed a significant increase in uterus weight and endometrial thickness that indicates an estrogenic response (Table 3). That occurs because the uterotrophic actions of estrogenic substances involve multiple events like hypertrophy/hyperplasia, secretory protein production, fluid imbibition, and cellular proliferation.²⁵ We observed that the DW was not able to induce changes in biological responses at a high level of organization (such as uterus wet weight - with fluid imbibition), but did increase the absolute blotted uterus weights and endometrial thickness. It is well known that activation of most or all the different gene cascades and multiple molecular pathways responsible for the increase uterus weight is dependent not only on the potency, affinity and efficacy of an endocrine disruptor, but also on the number of receptors and ligand bioavailability.^23,26 In the 3-day DW study, some of the multiple estrogenic events were possibly activated and the sensitive uterine tissues started to respond to the exposure.

The uterotrophic assay is a screening test used to detect estrogenic activity, but the assessment of adverse effects resulting from the exposure to individual chemicals or mixtures is better visualized with the results of combined toxicological datasets. We therefore conducted additional assays and evaluated more end-points to provide a complete coverage of possible effects induced by these water samples.

The concentration of the natural hormone 17-β-estradiol is consistently low throughout the prepubertal development and increases after PND 28th.²⁷ These levels decrease between PND 20th and puberty by an inhibitory action of prolactin and/or progesterone on pituitary gonadotropin release.²⁸ Some of this hormonal fluctuation was observed in our studies, especially the serum LH levels. It is also recognized that disruption of GnRH/LH pulsatility leads to a reduction in gonadotrophin secretion and to a loss of reproductive function.^29,30 Alteration of the HPG hormonal release during prepupertal/immature development could be reflected in an estrous cycling deregulation through adulthood. In our results, we found hormonal alterations in both immature and pubertal development, especially in the FSH and LH levels. This hormonal evidence could indicate a central nervous system interaction that could influence reproductive functionality. However, our data could also represent a pulse of these hormone release cycles at the moment of euthanasia. Therefore a serial blood collection would be more appropriate for the evaluation of hormone levels, and would be better able to confirm a LH release disruption and other hormonal alterations.

The high throughput evaluation using endocrine gene transcripts is a very sensitive baseline for endocrine activity and we used qPCR for in vivo estrogen and progesterone receptors, and 2 endocrine target genes, Hoxa-10 and CaBP-9k, responses. Hox genes are evolutionarily conserved and necessary for body axis patterning during embryogenesis and adult functional differentiation. The Hox multigene family homeobox A, Hoxa-10, is affected by bisphenol A (BPA), methoxychlor and diethylstilbestrol (DES) after in vivo and in vitro exposures.^31,32 It has been reported that CaBP-9k (from a cytosolic calcium binding protein group) is rapidly and strongly induced in the uterus of mammals by estrogenic compounds like DES.³³ Although not statistically significant, the gene transcript evaluations of Hoxa-10 and CaBP-9k showed treatment alterations after the 2010 and 2012 DW exposures: the expression levels of Hoxa-10 were reduced and those of CaBP-9k were increased. This result was also seen in the pubertal development assay with the 2010 DW. We could observe a DW exposure interaction in the ED target genes of the animals at a molecular level, but additional experiments are needed to elucidate these effects.

When comparing the results of the uterotrophic with the pubertal female assay, the immature HPG axis was more sensitive than the pubertal to the chemicals present in the DW. We also observed that in the uterotrophic assay, the vagina epithelium of these animals clearly responded faster than the uterus, and we found a significant increase in the incidence of vaginal keratinization at the highest 2010 DW. This indicates precocious puberty of the animals at this stage of life. These results should not just be taken as flags of concern, because this time-phase interaction influences other stages of maturation.³⁴ We do not consider the evidence found in this study as adverse effects or end-points per se, but they are important preliminary findings of a potential DW endocrine disrupting activity which can be linked to some human diseases, such as female infertility, endometriosis, etc.¹⁴

Also important is that the chemical composition of environmental samples can be underestimated because of the technical procedures involved, and that the actual human exposure could be higher based on consumption over the lifetime. To also support this conclusion, we observed a reduction in body weight following 20 days exposure to the 2010 DW sample, which suggests that additional effects may be seen with a longer treatment and/or observation time. In addition, there is also the possibility of effects being manifested after exposures during other critical developmental periods, like pregnancy and lactation.

The chemical compositions of the DW samples studied could help to link the in vivo results to a specific substance. Caffeine was present in both water extracts, which demonstrates human contamination^9,35 although it was 100 times higher in 2010 DW. Studies with caffeine showed that this substance (5 – 100 mg/kg/day) is able to cause adverse effects in endocrine tissues, and changes in developmental parameters and hormone levels in rodent assays.^36-38 Atrazine, the herbicide present in both samples, although 13 times higher in the 2010 DW sample, has also been shown to exert indirect effects on several reproductive and endocrine parameters in female rats through the central nervous system, including blunting of the hormone-induced LH surge and inhibition of pulsatile LH release.³⁹ Because atrazine could be related to our findings in the present study, it was also tested it in an extra group separately, but there were no results related to our DW findings. Carbendazim, imidacloprid, azoxystrobin and tebuconazole are also pesticides with some evidence of ED in in vivo studies, but their modes of action are not completely elucidated or their individual results are the not the same as what was found in our previous studies.^40-42

The other uterotrophic group, treated with pure deionized water, did not differ from the group receiving the ultrapure water extract eluted in 1% DMSO. These results excluded the possibility of DMSO and also extraction procedure interference.

Estrone is a hormone considered a weak form of estrogen. It is produced by the ovaries and adipose tissue, and is the predominant estrogen in postmenopausal women. It is not directly active in as many tissues as estradiol, but can be readily converted to estradiol in vivo.⁴³ Although the conversion can go both ways, estrone can also be considered a breakdown or storage form of estrogen. We found estrone in the 2010 DW, but at relatively low levels (near the chemical limit of quantification). Nevertheless, those levels were not sufficient to exert a clear in vivo activity in our study, when considering that the estrone or estradiol radioimunoassay levels were not affected during the animals’ exposure. However, synergistic/antagonistic effects may become an important issue when dealing with a complex mixture, as drinking water, especially when a myriad of compounds are present at low concentrations, or below the levels of detection by chemical analytical methods.

In conclusion, the potential effects of the detected chemicals of this study could, in association with each other, and also with unidentified chemicals present in the DW samples, interact in different ways and produce biological changes, including estrogenic responses. In the present study, the increased blotted uterine weight and endometrial epithelium cell height, and the keratinization of vaginal epithelium observed in immature rats after exposure to 2010 DW samples provide strong evidence of estrogenic activity of the water. The 2012 DW also showed a significant increase in endometrial thickness and an increased trend in the blotted uterine weights. Additionally, we conclude that these results, together with the alterations in hormonal levels and ED target gene expression, represent a hypothalamic-hypophysis-gonadal axis effect on the prepubertal and pubertal animals exposed to the 2 DW samples analyzed in this study.

Material and Methods

DW collection and pre-concentration

We collected 2 DW samples, one in 2010 and another in 2012, 26 months apart, from the same source in Campinas, São Paulo, Brazil. The water samples were collected at the tap, and came from the same Water Treatment Plant (WTP). The source water from this WTP is the Atibaia River, which has been shown to contain high levels of estrogenic substances.^5,21 This WTP provides DW for around 1,000,000 people and uses a conventional treatment, very common around the world, called chloramination (NH₂Cl), which is produced by the reaction of free chlorine and ammonia.⁴⁴ The sampling site was established based on the available Public Water Supply Quality Index (IAP), as published by the São Paulo State Environmental Agency (CETESB).^1,2 The IAP index is focused on the evaluation of toxic substances and organoleptical properties, combined with classic water quality parameters. This index is similar to the WQI developed by the National Sanitation Foundation.^3,4 Despite the IAP classification as Good, waters from this sampling site are known to receive raw sewage drained from adjacent urban areas.

The DW was collected in cleaned amber glass bottles, filtered through glass fiber filters (Schleicher & Schuell; Dassel, Germany) and 0.45 μm pore size cellulose acetate membranes (Sartorius; Gottingen, Germany). Drinking water (4 L) was solid-phase extracted using 500 mg HLB Oasis cartridges (Waters; Milford, USA) fitted to a lab-made extraction system⁴⁵ using cartridge conditioning and elution with methanol as described elsewhere.⁵ Before being used in the in vivo assays, the methanol in the resulting extracts was slowly dried in an N₂ flux.

Chemical analysis

We determined the contaminants by liquid chromatography tandem mass spectrometry (LC-MS/MS) (Agilent 1200 LC system coupled to an Agilent 6410 TripleQuad mass spectrometer).^5,6 Most of these substances do not have established DW standards or health advisories at this time. We selected the pesticides based on a São Paulo State, Brazil, priority pesticides list⁴⁶ and method⁴⁷ availability at the time. The ED activities of these substances are not completely elucidated and are currently under investigation. The contaminants belonged to different pesticide types (acaricides, insecticides, fungicides, herbicides) from 6 different chemical groups: triazoles (difenoconazole, epoxiconazole, tebuconazole), triazines (atrazine), strobilurins (azoxystrobin, picoxystrobin, pyraclostrobin, trifloxystrobin), organophosphates (chlorpyrifos, profenofos), phenyl pyrazole (fipronil) and benzimidazole (carbendazim). Other compounds were also evaluated and had also been detected elsewhere^3-7,18,22,48: caffeine, bisphenol A, natural and synthetic hormones (estrone, 17β-estradiol, estriol, progesterone, testosterone, mestranol, levonorgestrel, dietilstilbestrol, 17α-ethynylestradiol), phenols (phenolphthalein, 4-n-nonylphenol, 4-n-octylphenol), bactericide (triclosan), insecticides (imidacloprid, hexythiazox), and herbicide (bromacil). Table 1 shows the quantification and the Limits of Quantification (LOQ) for the targeted chemicals.

In vivo studies

Animals and housing

The study was approved by the Ethical Committee on Animal Experimentation (Protocol CEEA 823–2010, Botucatu Medical School, UNESP). We used newborn female Wistar rats weaned, weighed and allocated to treatment groups at postnatal day (PND) 21. They were maintained in an environmentally controlled facility: room temperature at 22 ± 2°C, humidity at 55 ± 10%, and 12-h light/dark cycle (lights on at 7:00 am). We allocated 3 animals per propylene cage with autoclaved pinewood bedding changed 3 times a week and fed them a balanced laboratory diet (pellet chow; Purina-Labina Evialis, Paulínia, Brazil) with genistein-equivalent content (described in Table S2) under the limits of the Pubertal Development Assay guideline.⁴⁹ Ultrapure water was provided ad libitum from glass bottles. Our group has previous work^50,51 and experience with pubertal Wistar rats and complex environmental mixtures that assure a good and reliable assay performance.

Dose selection

The experimental design is presented in Figure 6. After the DW extraction process, the residue was eluted in methanol, dried down, and reconstituted in 100% DMSO (dimethyl sulfoxide) and then diluted to a final concentration for animal treatment of 1% DMSO in 99% ultrapure water (stock solution). A daily working concentration was then prepared with this stock solution diluted in the vehicle (1 ml ultrapure water/kg body weight of each animal). Female rats were exposed by daily gavage to DW extracts at 33.3, 166.5, and 333.0 mL equivalent of water/kg body weight, proportionally modeling a daily ingestion of 2 L, 10 L and 20 L of DW by a 60 kg human being.⁵² As the extract was 1,000 times concentrated, the exact administered rat doses were of 33.3, 166.5 or 333.0 μL/kg bw per each animal/day. The morning exposure occurred during 3 days (uterotrophic bioassay),²⁴ with 12 animals per group, or 20 days (pubertal development female rat assay),⁴⁹ with 15 animals per group. Both assays also included negative control groups (extraction process control), which consisted of ultrapure water extract, within the same DW extract conditions, administered at the highest dose (333.0 μL/kg bw).

Figure 6. In vivo experimental design of the drinking water extracts.

Additional groups - uterotrophic bioassay

Most estrogen ligands or their metabolic precursors tend to be hydrophobic and DMSO was used to assure that all organic compounds present in the water extract tested were diluted and equally suspended. In order to guarantee that the DMSO would not affect the treatment, we included another experimental group of animals that received only deionized water (not an extract), as a vehicle control group. Then, 12 animals in the same conditions of the uterotrophic assay, received only deionized water, daily by gavage, for 3 days. This group represents another extraction process control and also a DMSO solvent control.

Because 11.2 ng/L of atrazine was chemically detected in the 2010 DW, we also included a control group with this pesticide administered alone, in a dose of 3.73 ng/L/kg bw, which is equivalent to the highest dose water extract amount (333 mL eq). We evaluated the same parameters as in the uterotrophic assay.

Uterotrophic bioassay

We weighed, dosed and checked the female rats for vaginal opening (VO) daily from PND 22 to 24. The groups exposed to the water extracts consisted of 12 animals each; and an additional group was used for a 3-day administration of the positive estrogen control (17-α-ethynylestradiol, 1 μg/kg bw, by gavage) dissolved in corn oil. The animals were sacrificed in the morning by exsanguination and under deep anesthesia in a CO₂ chamber, 24 hours after the last dose. Blood was collected from the left ventricle and stored at -20°C until analyzed for sex hormone concentrations. At necropsy, selected organs (uterus, ovaries, adrenals, pituitary, thyroid, and liver) were excised, trimmed free of fat and connective tissues, and weighed. The uterus was weighed wet (with luminal fluid content) and blotted (after the luminal content had been removed). We separated the uterus and cervix from the vagina after the wet weighing, so that the blotted weight represented only the uterine horns and cervix. The blotted uteri of 6 animals from each group was flash-frozen in liquid nitrogen and stored at −80°C until transcriptomic analyses.

Pubertal development and thyroid function in intact juvenile/peripubertal female rats

This assay was conducted only with the 2010 DW sample. We used 15 animals per dose group and they were weighed, dosed and checked for VO daily from PND 22 to 42. This duration of treatment is not necessary for detection of estrogenic chemicals, but is required for the detection of pubertal delay and anti-thyroid effects.⁴⁹ Beginning on the day of VO, to and including the day of necropsy, we obtained daily vaginal smears which were evaluated under a low-power light microscope for the presence of leucocytes, nucleated epithelial cells, or cornified epithelial cells.⁵³ The same systemic serum, organ collections, and procedures were used as for the uterotrophic assay.

Hormone measurements

Serum was assayed by radioimmunoassay (RIA); serum follicle-stimulating hormone (FSH), luteinizing hormone (LH) and Prolactin assays were conducted using the double-antibody method with specific kits provided by the National Hormone and Peptide Program (Harbor-University of California at Los Angeles). The FSH primary antibody was anti-rat FSH-S11, and the standard was FSH-RP2. The antiserum for LH was LH-S10 using RP3 as reference. The Prolactin antiserum and reference preparations were anti-rat PRL-S9 and PRL-RP3, respectively. The lower limits of detection for FSH, LH and Prolactin were 0.2, 0.04 and 0.19 ng/mL and the intra-assay coefficients of variation were 3%, 3.4% and 4%, respectively. Serum concentrations of E2, T and P4 were determined using specific kits provided by MP Biomedicals (Orangeburgh, NY, USA). The intra-assay coefficients of variation were 4.7%, 4% and 3.6% and the lower limits of detection were 8.6 pg/mL, 5 pg/mL and 0.0 2 ng/mL, respectively for E2, T and P4. We measured Estrone using the DSL-8700 kit (Webstern, Texas, USA), with the lower limit of detection of 1.2 pg/mL and intra-assay error of 6.4%. We assayed all serum in the same RIA to avoid inter-assay variation.

In the pubertal development assay, only rats in estrus cycle were analyzed at the moment of euthanasia for steroid hormone levels. The n for each group is provided in each hormone figure column group (X axis). Quantitation of P4/E1/PRL in the 2012 uterotrophic and T/PRL in the 2010 pubertal samples was not performed because of insufficient blood samples.

Analysis of reproductive organs

At necropsy, we excised ovaries, adrenal glands, uterus, vagina, pituitary, thyroid glands and liver. They were fixed in 10% buffered formalin and processed for histological analysis using hematoxylin and eosin (H&E) staining following OECD guidelines.⁵⁴ Every excised organ was histologicly analyzed in every assay and for both 2010 and 2012 DW samples.

Three 4 μm thick sections were prepared from each ovary (40 μm apart), mounted on glass slides and stained with H&E. The ovarian follicles and corpora lutea from the right ovaries in 3 sections per animal were counted and classified as primordial, primary, secondary and tertiary per ovary area.⁵⁴ Only follicles with an oocyte were counted. A total follicle count per area was a sum of all follicle classes and corpora lutea together, in the same animal and in the same group. The uterine endometrial thickness was also measured in 3 sections per animal, and 5 different regions of each section were analyzed.⁵⁵ Both the ovary and uterus tissue measurements used an optical microscope (40×) coupled to a digital camera and a PC with software Image-Pro Plus (Media Cybernetics, MD, USA). Image analyses and tissue area measurements were performed in the rats exposed to the 2010 DW with the Image-J software (available at http://rsbweb.nih.gov/ij/). The vaginal tissue was also processed in formalin and the 5 μm H&E stained slides were analyzed using an optical microscope (40×), following the OECD histologic guidance to distinguish every different layer of the anatomical regions of this organ.⁵²

Gene transcript measurements

Total RNA isolation

Total RNA was isolated from the uteri of 5 animals per group using RNeasy Mini kit -Qiazol (Qiagen, 74804) according to the manufacturer's instructions. The resulting RNA was quantified using Nanodrop ND-1000 (NanoDrop Technologies, Inc.). RNA integrity (28 S/18 S ratio) and purity were assessed using RNA 6000 nano assay LabChips® (Agilent Technologies, 5067–1513) and analyzed on a 2100 Bioanalyzer (Agilent Technologies) using an RNA integrity number (RIN) > 7.0 (Agilent software).

RT-qPCR analysis

Real-time, quantitative reverse-transcriptase-polymerase chain reaction (RT-qPCR) with selected transcripts used Applied Biosystems reagents and equipment. cDNA was synthesized with 2 μg of total uterine RNA from control and treated animals using reverse transcription with random hexamer primers using the High Capacity cDNA Reverse Transcription kits (4374966) according to the manufacturer's protocol. RT-qPCR reactions were performed using Taqman probes (Assay on demand - 4331182, 4331372; RefSeq) and Taqman Universal PCR Master Mix (4364340) on a Step One Plus machine. The genes studied from LSG (Life Science Germany)- Taqman assay were: Er-α (NM_012689.1, 87 pb, Rn00664737_m1), Er-β (NM_012754.1, 89 pb, Rn00562610_m1), Pgr (NM_022847.1, 105 pb, Rn00674394_m1), CaBP-9k (NM_012521.1, 99 pb, Rn00560940_m1) and Hoxa-10 (NM_001129878.1, 90 pb, Rn01410200_m1). Values of cDNA expression were normalized in relation to the expression of β-actin (NM_031144.2, 91 pb, Rn00667869_m1).

Statistical analyses

Statistical analyses were performed using Jandel Sigma Stat software (Jandel Corporation, San Rafael, CA). The final body weights and body weight gains, relative and absolute organ weights, food consumption, hormone, and gene expressions were analyzed using relative quantitation data by analysis of variance (ANOVA), and followed by the Kruskal–Wallis test to determine significance. The percentage of animals with vaginal keratinization in the uterotrophic assay and the pubertal parameters of the peripubertal development bioassay were analyzed by the Chi-square test, followed by the Fischer test when necessary. Significance was set at P < 0.05, compared to the group treated with the ultrapure water extract.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Supplemental Material

Supplemental data for this article can be accessed on the publisher's website.

Acknowledgments

The authors also thank Mary Rosa Rodrigues de Marchi (Institute of Chemistry, UNESP Araraquara/SP), Igor Cardoso Pescara (Institute of Chemistry, UNICAMP), Flávio de Oliveira Lima, Paulo Roberto Cardoso, Paulo César Georgette and Maria Luísa Ardanaz (UNESP, Botucatu/SP Brazil) for technical support.

Funding

This study was supported by the Center for the Evaluation of the Environmental Impact of Human Health (TOXICAM), at UNESP, the School of Dentistry, at USP, the State of São Paulo Agency for Support of Research (FAPESP) and by the National Council for Technological and Scientific Development (CNPq), Brazil.