Environ. Sci. Technol. 2005, 39, 2461-2471
Mixtures of Estrogenic Contaminants in Bile of Fish Exposed to Wastewater Treatment Works Effluents R. GIBSON,† M. D. SMITH,† C. J. SPARY,‡ C . R . T Y L E R , ‡ A N D E . M . H I L L * ,† Centre for Environmental Research, Chichester Building, University of Sussex, Falmer, Brighton BN1 9QJ, U.K., and Environmental and Molecular Fish Biology Group, Department of Biological Sciences, Hatherley Laboratory, University of Exeter, Devon EX4 4PS, U.K.
Most effluents from wastewater treatment works (WwTWs) contain estrogenic chemicals that include steroidal estrogens and xenoestrogens. We investigated the nature of mixtures of estrogenic contaminants taken up by two species of fish exposed to two WwTWs effluents. Sexually immature rainbow trout, Oncorhynchus mykiss, and sexually mature roach, Rutilus rutilus, were exposed to tap water, river water, or one of two estrogenic WwTWs effluents for up to 10 days, when the fish were sacrificed and tissues removed for chemical analysis. Estrogenic contaminants in the bile and gonads were hydrolyzed, concentrated by solid-phase extraction, and fractionated by RP-HPLC. Active fractions were detected and quantified using a yeast estrogen receptor transcription screen (YES assay) and the identities of estrogenic components in the fractions determined by GC-MS. Bile from rainbow trout exposed to either tap water or river water contained low amounts of 17β-estradiol (E2) and estrone (E1) with a total estrogenic activity (mean ( standard error) of 10 ( 5 and 31 ( 9 ng of E2 equivalents/mL (ng of E2eq/mL) for male and female fish, respectively. In effluent-exposed trout the total estrogen content of bile was considerably higher with the following composition and concentrations (ng of E2eq/mL) of individual estrogens: E2 (/, 591 ( 125; ?, 710 ( 207), E1 (/, 338 ( 75; ?, 469 ( 164), ethinylestradiol, EE2 (/, 32 ( 2; ?, 40 ( 6), nonylphenol (NP) and shortchain NP polyethoxylates (/, 21 ( 4; ?, 22 ( 3). An additional estrogenic compound, 17β-dihydroequilenin (DHQ), was identified for the first time in effluent-exposed fish, and was present in trout bile at concentrations of (/) 40 ( 9 and (?) 30 ( 5 ng of E2 eq/mL. DHQ, E2, E1, and EE2, but not NP or NP polyethoxylates, were also detected in bile of effluentexposed roach, and the concentrations of all these steroidal estrogens in ng of E2eq/mL were lower in male (E2, 62 ( 2; E1, 35 ( 11; EE2, 10 ( 2; DHQ, 1 ( 1) compared with female (E2, 740 ( 197; E1, 197 ( 37; EE2, 40 ( 6; DHQ, 8 ( 2) roach. The synthetic estrogen EE2 was also detected in the testes and ovaries of effluent-exposed roach. This study shows that a mixture of estrogenic contaminants present in WwTWs effluents bioconcentrate in fish * Corresponding author phone: 44 1273 678382; fax: 44 1273 677196; e-mail:
[email protected]. † University of Sussex. ‡ University of Exeter. 10.1021/es048892g CCC: $30.25 Published on Web 03/09/2005
2005 American Chemical Society
tissues, resulting in the induction of vitellogenin, and are likely to contribute to feminizing effects in wild fish living in U.K. rivers. The composition of the mixture of estrogenic contaminants in the bile is species dependent and may determine the susceptibility of fish to the effects of exposure to estrogenic effluents.
Introduction The detection of populations of intersex fish (individuals with gonads containing both female and male reproductive tissue) in freshwater and estuarine environments of a number of countries has been associated with exposure to estrogenic chemicals in wastewater effluents. A high incidence of intersex fish containing primary or secondary oocytes within the testes has been observed downstream of wastewater treatment works (WwTWs) effluents in many U.K. rivers (1). Male fish captured or held below WwTWs discharges show high levels of plasma vitellogenin (a precursor for yolk in eggs, which is not normally detected in male or immature female fish), indicating exposure to estrogenic contaminants (1-3). Furthermore, controlled exposures of roach to WwTWs effluents has resulted in some of the feminized responses observed in wild fish, namely, vitellogenin induction and disruption of gonadal duct development (4). Some of the estrogenic chemicals present in WwTWs effluents have been identified using bioassay-directed fractionation combined with chemical identification (5). Using this approach, three steroidal estrogens, E2, E1, and the synthetic chemical EE2, were detected in WwTWs effluents. These are excreted in human waste as inactive glucuronide or sulfate conjugates, but are hydrolyzed back to the active native molecule in the sewers and the treatment plant (6). In addition, other less potent xenoestrogens arising from industrial uses such as alkylphenols, short-chain ethoxylated alkylphenols, and bisphenol A have also been detected in effluents (7, 8). Laboratory studies have demonstrated that these chemicals can induce vitellogenin synthesis and, in some instances, gonadal intersex (9, 10). The levels of estrogenic contaminants in WwTWs effluents depend on a number of factors including the nature and dilution of the influent, the proportion of industrial inputs, the type and efficiency of any secondary and tertiary treatments, and the residence time within the plant (6). Concentrations of contaminants have been measured by mass spectrometry and in final WwTWs effluents can range from 1 to 100 ng/L for E2, from 1 to 200 ng/L for E1, from 0.1 to 10 ng/L for EE2, from 0.2 to 100 µg/L for nonylphenol and nonylphenol ethoxylates (NP and NPEOs), and from 0.01 to 1 µg/L for bisphenol A (5-8, 11, 12). In addition, the estrogenic strength of WwTWs effluents has been measured using in vitro screens, such as estrogen receptor-linked reporter gene assays where the activity of the sample is measured in E2 equivalents (E2eq). Using these assays, the total estrogenic activity of final WwTWs effluents has ranged between 0.03 and 100 ng of E2eq/L (13-15). Although the composition and concentration of estrogenic contaminants in WwTWs effluents has received considerable study, it is not clear which contaminants, or mixtures thereof, are responsible for causing intersex in fish. In addition, we know very little about the uptake, bioconcentration, and composition of estrogenic contaminants in fish tissues. A number of studies have shown that estrogenic chemicals can be detected at high concentrations in fish bile. Certain xenoestrogens such as waterborne alkylphenols concentrate VOL. 39, NO. 8, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2461
between 20000- 70000-fold in fish bile, depending on the species (16, 17). In rainbow trout the alkylphenols are mainly in the form of biologically inactive glucuronide conjugates, but these can be readily hydrolyzed by glucuronidase enzymes back to the parent molecule (17). High concentrations of E2, E1, EE2, NP, and bisphenol A were detected in bile from trout caged below WwTWs effluents in Sweden (18). The estrogenic activity of bile sampled from bream captured at locations in The Netherlands showed a good correlation with the estrogenic activity in surface waters and induction of plasma vitellogenin in these fish (19). These studies suggest that analysis of bile fluid could provide useful information about the nature and levels of estrogenic contaminants in fish. The aim of this study was to determine the nature of estrogenic contaminants in fish exposed to estrogenic effluents and in addition to ascertain whether there were other additional estrogenic compounds in exposed fish not previously detected in estrogenic effluents. We exposed rainbow trout (Oncorhynchus mykiss) and roach (Rutilus rutilus) to effluents from two WwTWs in continuous-flowthrough tanks and analyzed the uptake of estrogenic contaminants into the fish. Previous work (20) from this study has shown that elevated concentrations of plasma vitellogenin in effluent-exposed rainbow trout were positively correlated with the estrogenic activity of bile, showing that differences in the vitellogenic responses of trout were related to differences in the amount of estrogenic chemical taken up from the water. In the study reported here we investigated the nature of estrogenic contaminants within the effluents and fish tissues by RP-HPLC fractionation of bile and gonad samples, followed by detection and analysis of active fractions using the YES assay and GC-MS. This analytical approach has been successfully used before to determine the nature of estrogenic contaminants in effluents and sediments (5, 21).
Experimental Section Materials. 17β-Dihydroequilenin and 17R-dihydroequilenin were obtained from Steraloids Ltd. (Rhode Island). E1, E2, EE2, NP, bis(trimethylsilyl)trifluoroacetamide (BSTFA) containing 1% trimethylchlorosilane (TMCS), β-glucuronidase (type VII-A extracted from Escherichia coli), β-glucosidase (type H-1 extracted from almonds), sulfatase (type VI from Aerobacter aerogenes), and all other chemicals were obtained from Sigma-Aldrich (Poole, U.K.). [2,4,16,16-D4]E1, -E2 and -EE2 (isotope purity 96%, chemical purity >98%) were obtained from Cambridge Isotope Laboratories (Andover, MA). All solvents were of HPLC-grade purchased from Rathburn Chemicals (Walkerburn, U.K.). WwTWs Sites. Fish were held in tap water, river water (abstracted upstream of the effluent), or two undiluted treated WwTWs effluents. Two effluents were chosen for study: WwTWs A receives an influent load which is primarily domestic (population equivalent of 138000) with industrial inputs contributing 6% of the load. The influent had been subjected to primary as well as fine bubble-diffusion activated sludge and percolating filter treatment. WwTWs A has been shown to be estrogenic to fish (12). WwTWs B has a population equivalent of 312000, and 24% of the influent was derived from industrial sources. The influent was treated by primary treatment followed by bubble-diffused activated sludge and biological phosphorus removal, and the effluent from WwTWs B has also been shown to be estrogenic (20). Fish Exposure. Juvenile rainbow trout (mixed sex) of approximately 250-320 g of body mass were obtained from two commercial fish farms, Houghton Springs Farm, Dorset, U.K., and Hawkshead Trout Farm, Cumbria, U.K. Adult roach 3+ years old, 95-100 g, were obtained from a wild fishery, Framlingham Fisheries, Suffolk, U.K. A total of 25 fish of 2462
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 8, 2005
each species were held in aerated continuous-flow-through 1 m3 tanks containing 100% effluent, river water (no effluent), or dechlorinated tap water fed with a flow rate of approximately 9-10 L/min. Each tank was divided into two with a mesh partition to separate the roach and trout. Prior to exposure fish were tagged and blood samples taken under anaesthetic for plasma vitellogenin analysis. Fish were fed commercial trout pellets once daily throughout the exposure period until 48 h prior to sampling when food was withheld to maximize bile production for collection. Mortalities for trout in undiluted WwTWs effluent were 4% and 2% at sites A and B, respectively. For roach, mortalities were similar at WwTWs A (3%), but somewhat higher at WwTWs B (12%). At the end of the studies there were a minimum of 20 fish (n ) between 20 and 25) per treatment group for both species at both study sites. At the end of the exposure period (10 days; WwTWs A, 8 days; WwTWs B) fish were sampled for blood and then sacrificed and bile sacs and gonads removed. The tissues were stored immediately in dry ice and then at -70 °C on return to the laboratory. The estrogenic responses in these fish, as defined by induction of vitellogenin, are reported in an accompanying paper (20). Samples (2.5 L) of effluent, river water, and tap water were taken from the pipe entering the holding tanks on the first and last days of fish exposure. Methanol (25 mL) was added as a preservative before processing of the effluent samples the same day. Extraction of Effluent, River Water, and Tap Water Samples. Effluent or river or tap water (2 L grab samples) was acidified with glacial acetic acid (25 mL) and then passed through a 500 mg OASIS HLB solid-phase extraction (SPE) cartridge (Waters Ltd., Herts, U.K.) that had been conditioned with methanol (10 mL) followed by water acidified with 1% acetic acid (10 mL). The cartridge was washed with water (5 mL) and then dried under vacuum for 10 min before being eluted with methanol (8 mL), ethyl acetate (6 mL), and hexane (6 mL). The combined eluents were evaporated to dryness and then reconstituted in ethanol (1 mL) for YES assay or methanol (100 µL) and water acidified with 1% acetic acid (100 µL) for HPLC separation. Preparation of Bile Samples. Bile was deconjugated with β-glucuronidase (1000 units/mL), sulfatase (2 units/mL), and β-glucosidase (20 units/mL). The enzymes (200 µL of each) were added to a mixture of bile (100 µL), 0.1 M phosphate buffer at pH 6.0 (1500 µL), and water (800 µL). The solution was incubated for 16 h at 37 °C. Glacial acetic acid (300 µL) and water (2 mL) were added and the samples either assayed for estrogenic activity directly in the YES assay or concentrated on 200 mg OASIS HLB solid-phase extraction (SPE) cartridges prior to HPLC separations. SPE cartridges were conditioned with methanol (5 mL) followed by water acidified with 1% acetic acid (5 mL). The cartridge was washed with water (2 mL) and then dried under vacuum. The cartridge was eluted with methanol (5 mL), ethyl acetate (3 mL), and hexane (3 mL). The combined eluents were evaporated to near dryness and then reconstituted as described above. Extraction of Tissue Samples. Gonadal tissue (1-2 g) was sonicated (2 × 15 s) in 2 volumes of methanol. The extract was centrifuged (3000g, 10 min), the supernatant removed, and the pellet reextracted two more times in fresh methanol. The methanolic supernatants were combined, reduced to approximately 1-2 mL, and diluted to 5% methanol by the addition of aqueous acetic acid (1%) prior to purification by OASIS SPE as described above. High-Performance Liquid Chromatography. Samples were fractionated on a Waters Ltd. HPLC system comprising a model 600 pump and controller, model 717 autosampler, and model 996 photodiode array detector. The system was characterized with standards of 6-R-hydroxyestradiol, β-estriol, 16-R-hydroxyestrone, bisphenol A, E2, EE2, E1, and NP. An aliquot of the sample (100 µL) was injected onto a
Novapak C18 column (5 µm particle size, 250 × 4.6 mm, Waters Ltd.). Mobile-phase solvents were water acidified with 0.2% acetic acid (A) and acetonitrile (B) in an initial ratio (A:B) of 69:31. Separation was achieved at room temperature using a flow rate of 1.0 mL/min with the following gradient program: 0 min (69:31), 35 min (65:35), 50 min (0:100), 60 min (0:100). Under these conditions E2 eluted at 27-29 min, EE2 at 37-39 min, E1 at 42-43 min, and NP and NPEOs at 56 min. Fractions were collected at 1 min intervals and analyzed for estrogenic activity using the YES assay and by GC-MS to identify estrogenic chemicals present in the fraction. Some fractions, which did not contain known effluent-associated estrogenic chemicals, were further purified by reversed-phase liquid chromatography using a C18 column as described above but with a methanol/water solvent system using water acidified with 0.2% acetic acid (A) and methanol (B) solvents in an initial ratio (A:B) of 55: 45. A flow rate of 1.0 mL/min was used with the following gradient program: 0 min (55:45), 35 min (50:50), 50 min (0:100), 60 min (0:100). Fractions were collected at 1 min intervals and analyzed using the YES assay, and any fractions containing estrogenic activity were further analyzed by GCMS. YES Bioassay. The estrogenic activity of extracts and HPLC fractions was determined using the YES bioassay (as described in ref 22). This bioassay has been validated in the detection of a wide range of estrogen receptor agonists including 17βestradiol, estrone, ethinylestradiol, and xenoestrogens such as alkylphenolics and bisphenol A (22, 23). Briefly, extracts of samples and blanks were added in a series of dilutions to a test multiwell plate, and the ethanol was allowed to evaporate at room temperature. Concentrations of E2 were analyzed in parallel as a positive control. Yeast and assay medium containing the chromogenic (revelation) substrate were added to the wells, the plate was incubated for 5 days, and the absorbance of each sample was determined at 540 nm to measure the estrogenic response to the sample, and at 620 nm to measure yeast turbidity. The absorbance of each sample at 540 nm was corrected for untreated controls and yeast growth, and compared with that of the E2 standard. The estrogenicity measured in the samples was expressed as E2 equivalent values which were determined from the linear range of concentration-response curves (recorded between 10 pM and 1 nM). In agreement with published values (22, 23), the median effect concentration was typically around 100 pM. GC-MS. The identities of E1, E2, EE2, NP, and NPEOs in active HPLC fractions were determined by GC-MS after derivitization to their trimethylsilyl ethers. Target HPLC fractions were evaporated to dryness, 20 µL of pyridine and 20 µL of BSTFA were added, and the sample was incubated for 15 min at 65 °C. Samples were analyzed on an HP 5890 gas chromatograph, fitted with a 30 m HP5-MS fused silica capillary column (30 m × 0.25 mm × 0.25 µm film thickness), and connected to a Kratos MS80 mass-selective detector. The carrier gas was helium, at a constant pressure of 7.5 psi, and the sample was introduced using a 1 µL splitless injection. The injection port temperature was 250 °C, and the GC interface temperature was 280 °C. The MS detector was used in selected ion mode (SIM) for analysis of E1, E2, and EE2. It was used in full scan mode for the determination of NPs and NPEOs. The source temperature was 280 °C with an electron energy of 50 eV. The oven program was hold at 100 °C for 2 min, increase by 10 °C/min to 280 °C, and then hold at 280 °C for 11 min. The ions monitored were 342, 257 (E1), 416, 285 (E2), and 425, 440 (EE2). Ions in italics were used for quantification and the others for confirmation. Deuterium (D4) labeled standards of E1, E2, and EE2 were used as internal standards, and the ions monitored were 346, 261 (E1), 420, 287 (E2), and 429, 444 (EE2). NPs and NPEOs were quantified
using an external calibration method using NP as a reference standard. Calibration standards for E1, E2, and EE2 covered the concentration range 0.1-10 ng/µL; for NPs and NPEOs the range was 1-20 ng/µL.
Results Fractionation and Analysis of Estrogenic Components in the WwTWs Effluents. The estrogenic activity of WwTWs A was very variable and ranged between 26 and 99 ng of E2eq/ mL at the beginning and end of the exposure period, whereas the activity of WwTWs B was lower but more consistent, averaging 5 ng of E2eq/mL throughout the study period (20). Extracts of the two WwTWs effluents, sampled at the end of the exposure period, were fractionated by reversed-phase HPLC, and each fraction was analyzed for estrogenic activity using the YES assay. A profile (representative of two analyses) of the extract of effluent A, sampled at the end of the exposure period, showed that there were three highly active fractions corresponding to the retention times of E1, E2, and EE2 (Figure 1a). Quantitative GC-MS analysis of the estrogens in the fractions confirmed that all the estrogenic activity could be attributed to the appropriate estrogen (data not shown). The concentration of each component in the effluent at the end of the exposure period was (E1) 195 ng/L, (E2) 38.9 ng/L, and (EE2) 7.9 ng/L. Analysis of effluent B by HPLC/YES assay revealed a pattern of active fractions similar to that of effluent A, with the three estrogens E1, EE2, and E2 predominating (Figure 1b). The concentration of each estrogen at the end of the exposure period was (E1) 10.3 ng/L, (EE2) 1.1 ng/L, and (E2) 0.8 ng/L. Two other fractions with minor estrogenic activity were also detected in effluent B at retention times of 10 and 46 min. GC-MS analyses of these fractions did not detect other known estrogenic steroids or nonylphenolic substances, and the identities of estrogenic compounds in these fractions are currently unknown. The concentrations of individual estrogens in the effluents were within the range reported for these WwTWs sites in other studies (WwTWs A, ref 12; WwTWs B, K. Liney, personal communication). Estrogenic Activity in Fish Bile. There was no apparent difference in the volume of bile collected from tap water and effluent-exposed trout or from tap water and effluent-exposed roach (Table 1). There was little estrogenic activity in any of the rainbow trout bile samples prior to enzymatic hydrolysis (Table 1). After deconjugation, the activity of bile from trout exposed to either tap water or river water was greater, but still low, averaging less than 31 ( 9 ng of E2eq/mL for both sexes. Enzymatic hydrolysis substantially increased the estrogenic activity of the bile of trout held in the two WwTWs effluents. In trout held in WwTWs effluent A, the estrogenic activity of the bile was 100-fold higher after hydrolysis to give a concentration of approximately 1.0 µg of E2eq/mL for both sexes. In trout held in WwTWs effluent B, hydrolysis of bile resulted in a 28-fold increase in estrogenic activity. In contrast to the trout samples, hydrolysis of bile from effluent (WwTWs A)-exposed roach resulted in only a 3-12-fold increase in estrogenic activity for male and female roach respectively, indicating that it contained significant amounts of nonconjugated estrogenic compounds. Fractionation and Qualitative Analysis of Estrogenic Components in Fish Bile. Extracts of hydrolyzed fish bile were fractionated on reversed-phase HPLC, and each fraction was assayed for estrogenic activity using the YES screen. The HPLC profiles of the estrogenic activity of hydrolyzed bile from male trout and roach exposed to tap water are given in part a and b of Figure 2, respectively. GC-MS analysis of active fractions eluting at 27-28 and 40-42 min confirmed that E2 and E1 were the major estrogenic components in bile of both fish species. In some bile samples of reference VOL. 39, NO. 8, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2463
FIGURE 1. Estrogenic profiles of final WwTWs effluents: (a) effluent A, (b) effluent B. Effluent extracts were fractionated by reversed-phase high-performance liquid chromatography, and fractions were analyzed for estrogenic activity by measuring the response (absorbance) in the YES assay. The identity of the estrogenic chemical in the active fractions was determined by GC-MS.
TABLE 1. Estrogenic Activity of Fish Bile before and after Enzymatic Hydrolysis
site
exposure
WwTws A
tap water
WwTWs A
river water
WwTWs A
effluent
WwTWs B
effluent
WwTWs A
tap water
WwTWs A
effluent
fish (volume, µL, of bile collected, mean ( SE)
no. of samples
female trout (259 ( 44) male trout (ndb) female trout (nd) male trout (nd) female trout (280 ( 41) male trout (275 ( 48) female trout (nd) male trout (nd) female roach (148 ( 27) male roach (122 ( 22) female roach (118 ( 14) male roach (83 ( 15)
5 5 7 1 12 3 8 2 3 3 3 3
estrogenic activitya (ng of E2eq/mL bile, mean ( SE) before after hydrolysis hydrolysis 10
103 113 27 28 12 3
a Trout and roach were exposed to tap water, river water (effluent free), or 100% effluent for up to 10 days. An aliquot of the fish bile was hydrolyzed by glucuronidase, sulfatase, and glycosidase enzymes. The estrogenic activity of native and hydrolyzed bile samples was determined using the YES assay in comparison with concentrations of standard E2. b nd ) not determined c Range of two replicates.
male trout an additional unidentified estrogenic component, which eluted at 13 min, was also detected. Similar HPLC profiles of bile were also obtained for reference female trout 2464
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 8, 2005
and roach (data not shown). In contrast, the HPLC profiles of bile from rainbow trout or roach exposed to the effluents were very different from those of bile of reference fish and
FIGURE 2. Estrogenic profiles of bile from (a) a male rainbow trout exposed to tap water and (b) a male roach exposed to tap water. Extracts of hydrolyzed bile were fractionated by reversed-phase high-performance liquid chromatography, and fractions were analyzed for estrogenic activity by measuring the response (absorbance) in the YES assay. The identity of the estrogenic chemical in the active fractions was determined by GC-MS. contained a number of additional estrogenic fractions as well as, in the case of trout, greater concentrations of E2 and E1. Typical profiles for bile from male trout and roach exposed to effluent A and male trout exposed to effluent B are given in parts a-c of Figure 3, respectively. GC-MS analysis of the estrogenic fractions in bile of both fish species confirmed the identity of E2, EE2, and E1 in active fractions in the regions 27-29, 37-39, and 39-43 min, respectively. A comparison of the concentrations of individual estrogens in each fraction by YES assay and by GC-MS confirmed that the majority of the estrogenic activity of each fraction could be accounted for by the estrogen alone (data not shown). GC-MS analysis of the fraction in trout bile eluting at 55-56 min detected both NP and NPnEOs, where n, the number of ethoxy (EO) units in the molecule, was 1-4. In addition another estrogenic fraction (labeled U1 in Figure 3, retention time 19-20 min) was detected in bile profiles of trout exposed to effluent A and to a lesser extent in roach exposed to effluent A and trout exposed to effluent B, but was not detected in bile of reference fish. A number of other minor estrogenic fractions were detected in bile from trout exposed to effluent A, but their concentration was too low to be investigated further. Within each fish species the profiles of estrogenic activity were similar for both sexes (data not shown). The compositions of bile from trout exposed to either effluent A or effluent
B were similar in that they both comprised E2, E1, EE2 NP, NPEOs, and fraction U1. However, the profiles between the trout and roach exposed to effluent A were markedly different in that the fraction containing NP and NPEOs was not detected in roach bile. Identification of the Estrogenic Component in U1. The estrogenic fraction U1 was further purified by HPLC using a methanol/water solvent gradient, and the estrogenic fractions, which eluted at 39-40 min, were concentrated on an OASIS cartridge and then silylated with BSTFA before GC-MS analysis. Many peaks on GC-MS were discounted as potential estrogens as they either occurred in adjacent nonactive HPLC fractions (38 and 41 min) or were identified as long-chain carboxylic acids. The most likely candidate for the estrogenic component of U1 was a peak at a retention time of 24.25 min on GC-MS which had a mass spectrum identified as 17β-dihydroequilenin (Figure 4). Characteristic fragments of the mass spectrum are m/z 412 (M+), 397 (M+ - CH3), 307 (M+ - CH3 - TMSOH), and 281 (M+ - TMSOH - C3H5) corresponding to loss of the D ring from the trimethylsilyl ether of 17β-dihydroequilenin. A commercially available standard of 17β-dihydroequilenin coeluted with U1 on both GC-MS and on acetonitrile/water and methanol/ water HPLC programs, with a UV spectrum similar to that of U1 (λmax 229, 280 nm) and a mass spectrum identical to VOL. 39, NO. 8, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2465
FIGURE 3. Estrogenic profiles of bile from (a) a male rainbow trout exposed to effluent A, (b) a male roach exposed to effluent A, and (c) a male rainbow trout exposed to effluent B. Extracts of hydrolyzed bile were fractionated by reversed-phase high-performance liquid chromatography, and fractions were analyzed for estrogenic activity by measuring the response (absorbance) in the YES assay. The identity of the estrogenic chemical in the active fractions was determined by GC-MS. U1 represented a fraction containing unidentified estrogenic substance(s). that shown in Figure 4a. A standard of 17R-dihydroequilenin eluted 4 min later on acetonitrile/water HPLC and 42 s earlier on GC-MS than U1 and the 17β-dihydroequilenin standard. These data confirmed that dihydroequilenin present in U1 was the 17β-stereoisomer. A comparison of the estrogenic activity of U1, determined by serial dilution in the YES assay, and the concentration of 17β-dihydroequilenin in U1, as measured by GC-MS, revealed that at least 71% of the estrogenic activity of U1 was accounted for by 17β-dihydroequilenin. 2466
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 8, 2005
Fractionation and Qualitative Analysis of Estrogenic Components in Roach Gonads. Gonadal extracts of male and female roach were subjected to HPLC/YES assay analysis of estrogenic activity. GC-MS analysis of active fractions from testes of reference fish revealed the presence of E2 and E1 (Figure 5a), and that of ovaries predominantly E2 (data not shown). Analysis of testes from roach exposed to effluent A revealed the presence of E2, E1, and EE2 (Figure 5b). Ovary extracts showed a profile similar to that of testicular extracts (data not shown). The relative reproductive mass (gonadal
FIGURE 4. Identity of the estrogenic steroid identified in fraction U1 in bile of rainbow trout exposed to effluent A. Extracts of hydrolyzed bile were purified by high-performance liquid chromatography and prior to analysis by GC-MS. (a) Mass spectrum of the trimethylsilyl ether of the estrogenic component in fraction U1. (b) Structure of 17β-dihydroequilenin identified in fraction U1. mass as a percentage of body mass) for reference and effluentexposed roach was 6.1 ( 0.5 and 6.2 ( 0.3, respectively, for males and 6.6 ( 0.4 and 7.3 ( 0.6, respectively, for females. Quantitative Analysis of Estrogenic Contaminants in Fish Bile. The distribution of estrogenic components in the bile of fish exposed to effluent from WwTWs A was quantified by YES assay analysis (Table 2). The mean recovery of estrogenic activity for the bile samples after HPLC fractionation was 87% ( 10%, which was calculated as the sum of the estrogenic activities of the fractionated components compared with the activity of the original bile sample prior to fractionation. The fish were grouped according to sex and species. Female rainbow trout were subdivided into two further groups according to their estrogenic response to the effluent as measured by the level of plasma vitellogenin induction during the exposure period (one group ranged between a 3- and 31-fold increase and the other group between a 120- and 1650-fold increase in vitellogenin concentration). Concentrations of E2 in all the fish bile samples varied from 62 to 710 ng of E2eq/mL, and this estrogen was the dominant component in bile from both sexes of trout and roach. E1 was the next most prevalent component in all samples with concentrations of 35-469 ng of E2eq/mL bile. In bile of trout, EE2, NP/NPEOS, and 17β-dihydroequilenin all showed similar activity ranging between 20 and 48 ng of E2eq/mL. Low levels of EE2 and 17β-dihydroequilenin, but not NP/NPEOs, were also detected in roach bile. There was no significant difference in the concentrations of the individual components in bile between male and female trout or between the two groups of female trout with high and low levels of the vitellogenin induction. However, significantly greater concentrations of E2 and E1 were detected in bile of female roach compared with bile of male roach, reflecting differences in sexual status in these maturing fish.
Discussion Previous work using controlled exposures of fish to effluents has shown that there is a positive correlation between the induction of the estrogenic biomarker vitellogenin and the concentration of estrogenic activity in the bile of fish (20). To investigate the nature of the estrogenic contaminants in fish tissues, we chose to expose fish to two WwTWs effluents, both of which received primarily domestic inputs and proportions of trade effluents but which differed in their total estrogenic activities. The major route of excretion of many of the estrogenic components present in sewage effluents, such as steroidal estrogens and nonylphenolics, is via the bile as well as the urine of fish, and a significant proportion of these phenolic contaminants will be conjugated in the liver prior to excretion in the bile (24). Glucuronidation is a major pathway of conjugation of steroids and alkylphenolics in fish (17, 25); however, both sulfate and glucose conjugates of phenolic xenobiotics have been detected in fish bile (16), so in this study we used a mixture of hydrolytic enzymes to ensure all conjugates of xenobiotics were hydrolyzed. Our work revealed that treatment of bile from effluent-exposed rainbow trout and roach with a mixture of the three hydrolytic enzymes resulted in a substantial increase (3-113-fold) in estrogenic activity as detected by the YES assay, which suggested that most of the estrogenic substances present in the bile of both species were inactive conjugates. In contrast to bile from trout held in clean water, the composition of bile from trout exposed to either effluent contained a number of contaminants that were present in the effluents, namely, EE2, as well as higher concentrations (compared to endogenous levels) of E1 and E2. The finding that NP and short-chain NP(1-4)EOs also accumulated in trout bile was not unexpected, as NP and NP1-2EOs have also been reported in effluent A (12), and in another study bisphenol A, NP, and NPEOs have also been reported in both WwTWs effluents (K. Liney, personal communication). VOL. 39, NO. 8, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2467
FIGURE 5. Estrogenic profiles of testes of (a) a male roach exposed to tap water and (b) a male roach exposed to effluent A. Testes extracts were hydrolyzed and fractionated by reversed-phase high-performance liquid chromatography. Fractions were analyzed for estrogenic activity by measuring the response (absorbance) in the YES assay. The identity of the estrogenic chemical in the active fractions was determined by GC-MS.
TABLE 2. Estrogenic Activity of Individual Components in the Bile of Trout Exposed to Effluent from WwTWs A fish male trout female trout (low vitellogenin) female trout (high vitellogenin) male roach female roach
no. of samples
estrogenic activitya (mean ( SE, ng of E2eq/mL of bile) (% of total bile estrogenic activity) DHQb E2 EE2 E1 NP
fold induction of vitellogenin
3 3
40 ( 9 (3.9) 591 ( 125 (57.8) 32 ( 2 (3.1) 338 ( 75 (33.1) 25 ( 8 (2.5) 591 ( 72 (59.2) 48 ( 19 (4.8) 314 ( 78 (31.5)
21 ( 4 (2.1) 20 ( 11 (2.0)
163 ( 12 19 ( 8
3
30 ( 5 (2.4) 710 ( 207 (55.9) 40 ( 6 (3.1)
469 ( 164 (36.9)
22 ( 3 (1.7)
1136 ( 505
3 3
1 ( 1 (0.9) 62 ( 2 (57.4) 10 ( 2 (9.2) 8 ( 2 (0.8) 740 ( 197 (75.1) 40 ( 6 (4.1)
35 ( 11 (32.4) 197 ( 37 (20.0)
