Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

A considerable body of evidence (Field et al., 1990) indicates that many classes of environmental contaminants, including dioxins, polychlorobiphenyls (PCBs), polycyclic aromatic hydrocarbons, and 4-alkylphenols, have the ability to interfere with normal hormonal activity by mimicking or blocking the action of natural hormones (McLachlan, 1993). Thus, such chemicals may give rise to a range of toxic effects including developmental abnormalities and disruption of endocrine function. The primary consideration for assessing the impact of a given environmental contaminant is the potential to produce a biological effect since this criterion can be used to eliminate inactive compounds from the risk assessment process. Determination of the absorption, tissue distribution, metabolism, and excretion of potentially biologically active contaminants is a prerequisite for risk assessment purposes since these parameters are fundamental in establishing in vivo activity and lend further biological significance to environmental abundance data. After absorption, the lipophilic nature of many persistent environmental contaminants predisposes to accumulation, whereas intrinsic hormonal activity may be appreciably modulated by biotransformation. The potency of the natural oestrogens, oestrone and oestradiol, is significantly reduced by conjugation to glucuronic and sulfuric acids (Klein et al., 1994), whereas biotransformation of certain proestrogens such as PCBs, by the action of cytochrome P₄₅₀ enzymes, may give rise to oestrogenic phenolic metabolites (Korach et al., 1987). A reasonable scenario for the action of proestrogens under physiological circumstances would be, for instance, mobilization from fat depots and biotransformation to oestrogenic metabolites in organs such as the liver, resulting in appreciable uptake into hormonally sensitive tissues.

4-Alkylphenol polyethoxylates are widely used nonionic surfactants (Haupt, 1983) and common contaminants of waste water. They are biodegraded under anaerobic conditions by microbes during sewage sludge treatment to various branch chain 4-alkylphenol isomers (Naylor, 1992). Although 4-nonylphenol has been found in high concentrations (0.45-2.53 g/kg) in anaerobically stabilized sewage sludge (Giger et al., 1984), concentrations of alkylphenols in river water (Blackburn and Waldock, 1995), sediments and sludge amended soils (Marcomini and Giger, 1987) are typically several orders of magnitude lower. Alkylphenols are lipophilic (Nimrod and Benson, 1996) and as such accumulate in a wide range of marine and aquatic life including algae, crustacea, molluscs, and fish (Ahel et al., 1993; Shiraishi et al., 1989; Lewis and Lech, 1996; Eklund et al., 1990). A recent study using [¹⁴C]4-nonylphenol demonstrated uptake from water into trout tissues with apparent bioaccumulation factors between 40-100 and significant deposition in many tissues, including muscle and liver (Lewis and Lech, 1996). However, information is sparse concerning the metabolic fate of alkylphenols in aquatic animals. Microsomes prepared from trout liver have been used to identify three -glucuronidase sensitive metabolites as conjugates of hydroxylated (in the C8 position of the alkyl side chain) 4-nonylphenol (Meldahl et al., 1996); these conjugates are also present in trout bile.

The ability of 4-alkylphenols to displace 17-oestradiol from the oestrogen receptor was first reported by Mueller and Kim (1978). Recently studies using a wide range of in vitro assay systems (White et al., 1994), including induction of vitellogenin secretion by trout hepatocytes (Jobling and Sumpter, 1993), confirmed that 4-alkylphenols are weak oestrogens. Clearly, oestrogenic potency values derived from in vitro assays are dependent on the principles of the assay and, as such, may not take account of species specific parameters such as biotransformation or tissue distribution of the test compound. A recent review (Nimrod and Benson, 1996) summarized potency estimations of 4-alkylphenols and indicated values of 1000-10,000 times less than E2; these values were similar to those of many other environmental contaminants. Association between aquatic oestrogenic environmental contaminants and potential biological activity in animals has been established since fish maintained in effluent from certain sewage treatment works have exhibited oestrogen dependent changes, although the precise identity of the chemical entity responsible remains unclear (Purdom et al., 1994).

The objectives of the present study were to investigate the tissue distribution, metabolism, and excretion of 4-nonylphenol in rainbow trout (Oncorhynchus mykiss) and to determine pharmacokinetic factors that may influence oestrogenic activity in vivo. The utility of in vitro liver preparations for predicting the metabolic fate of such environmental contaminants was evaluated by comparing biotransformation of 4-nonylphenol by trout hepatocytes with data derived from these in vivo studies.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Chemicals. Technical grade 4-nonylphenol (Aldrich, Gillingham, Dorset, UK) was selected for use in this study. This commercial preparation has been analyzed (Bhatt et al., 1992) and contains a mixture of 25 branch chain and 6 di-nonylphenol isomers, reflecting those likely to occur in the environment. 4-n-Nonyl[ring-2,6(n)-³H]phenol (53 Ci/mmol; Amersham International, UK) was synthesized from identical technical grade 4-nonylphenol (Aldrich) by catalytic tritiation and supplied at a stated radiochemical purity of 99%. Radiochemical purity was confirmed before use by radio-HPLC¹ using conditions described below for metabolite analysis. Helix pomatia -glucuronidase/aryl sulfatase was obtained from Sigma (Poole, Dorset, UK).

Experimental Design. Immature female rainbow trout (mean weight 122 g) were maintained in 50-liter glass aquaria, four fish per tank, supplied with a constant flow (100 ml/min) of water derived from Windermere (pH 7.1, Na⁺ 216, Ca²⁺ 165, Mg²⁺ 46, Cl 260 (µmol l¹) and organic carbon < 1 mg/ml) at ambient temperature (5^o C). Prior to the study fish were fed with Trout Aquaculture trout feed; this was withdrawn after dosing. Trouw (N = 28) were administered 4-nonylphenol (0.375 mg containing 37.5 µCi ³H label) dissolved in propylene glycol (100 µl) via the caudal vessels. This dosing rate was selected to facilitate subsequent identification of minor metabolites rather than provide a reflection of environmental exposure which will be highly variable depending on the extent of water pollution with 4-nonylphenol (Blackburn and Waldock, 1995). Blood, bile, excreta (contents of hind gut proximal to the vent), and tissues (liver, kidney, brain, pyloric caecae, intestine, skin, muscle, heart, spleen, gonad, and gills) were sampled post-mortem from trout sacrificed in groups (N = 4) 1, 2, 4, 24, 48, 72, and 144 hr after dosing. Similarly, tissues and body fluids were sampled from untreated control trout (N = 4) prior to dosing. For whole body autoradiography studies, an additional three trout were administered 112.5 µCi ³H-labeled and 0.375 mg of 4-nonylphenol via the caudal vein and frozen in a dry ice/acetone bath 24, 48, and 72 hr after dosing. All samples were then stored at 70^oC until analyzed.

	Tissue Distribution of Tritium Labeled Residues

Combustion Analysis. Tissue samples (approx. 100 mg), plasma (50 µl) and bile (20 µl) were analyzed for [³H]residue content following oxidation in a Tri-Carb Model 306 oxidizer (Packard, Pangbourne, Berks). Recovery of ³H was found to be 94 ± 2.4%. The ³H₂O was collected into 15 ml of scintillant (Starscint, Packard) and quantitated by LSC (LKB-Wallac 1214). Total residue concentrations were calculated from the specific activity and radioactivity per mass or volume and expressed as molar equivalents of 4-nonylphenol .

Whole Body Autoradiography (WBA). Trout were embedded in 3% w/v carboxymethyl cellulose and whole body slices cut to a thickness of 20 µm using a cryotome (PMV-450, Stockholm, Sweden). Slices were mounted on adhesive tape and dehydrated in the presence of silica gel at 50^o C. Distribution of tritium was demonstrated by concurrent exposure of Hyperfilm (Amersham, Little Chalfont, Bucks) to mid-line slices and ³H-calibration scales (Microscales, Amersham) for 6 months.

Analysis of Tissue Metabolites. Tissues and faeces (250 mg) and plasma (200 µl) were homogenized in 5 ml of methanol. The homogenates were centrifuged at 2000g and the supernatants taken to dryness under nitrogen. Radiolabeled metabolites in bile, tissues, and plasma were analyzed by reversed phase radio-HPLC using a 4.6 × 100 mm Hypersil C18 Elite column connected in series to an A500 radiodetector (Packard). Residues were chromatographed at a flow rate of 1 ml/min using a linear gradient mobile phase ranging from 20:80 methanol:water to 100% methanol over 15 min and maintained at 100% methanol for a further 5 min.

Solid Phase Extraction of Metabolites from Bile and Hepatocyte Medium. Bile, sampled at post-mortem from trout 144 hr after dosing, was diluted 1 in 20 with 0.1% acetic acid and applied to prewashed (10 ml methanol followed by 10 ml with 0.1% acetic acid) C18 Bond Elut cartridges. Cartridges were washed with 10 ml 0.1% acetic acid, and radioactivity was subsequently eluted with 3 ml methanol containing 0.1% acetic acid. The eluates were taken to dryness under nitrogen at ambient temperature. Culture medium (1 ml) sampled after 20-hr incubation with trout hepatocytes was extracted in a similar manner. Typical recovery of radioactivity by this procedure was 107 ± 22%.

	Tentative Identification of Metabolites

-Glucuronidase/Sulfatase Hydrolysis of Metabolites. Solid phase extracts of bile and trout hepatocyte culture medium were dissolved in 50 mM sodium acetate buffer (pH 4.5, containing 100 mM NaCl) and incubated for 24 hr at 37^o C, with or without Helix pomatia -glucuronidase/aryl sulfatase (1.4 mg/ml), prior to analysis by LC-MSⁿ.

LC-MSⁿ Analysis. Solid phase extracts of bile or trout hepatocyte culture medium (50 µl) were analyzed for the presence of metabolites by reversed phase radio-HPLC, using a 2.1 × 100 mm Hypersil C18 Elite column connected to an A500 radiodetector (Packard) with the MS in series. Metabolites and parent compound were eluted from the column at a flow rate of 0.3 ml/min using a linear mobile phase gradient as described above, with the exception that 100% methanol was maintained from 15 to 25 min. Metabolites were tentatively identified by negative-ion electrospray atmospheric pressure ionization mass spectrometry (LCQ, Finnigan, Hemel Hempstead, Herts). HPLC eluate was scanned from m/z 200-450 to provide selected ion chromatograms of potential metabolites. A collision energy of 25% was used to produce spectra from deprotonated pseudomolecular or daughter ions where appropriate.

Covalent Binding. Covalently bound [³H]residues in homogenates of liver (350 mg) from trout 48 hr after dosing and control liver were evaluated by exhaustive solvent extraction (Sun and Dent, 1980). Briefly, liver homogenate (2 ml) was washed three times with 5 ml 10% (w/v) TCA and the precipitate sequentially extracted with 5 ml of the following solvents: absolute ethanol, 70% (v/v) ethanol, twice with 0.25M Tris-HCl (pH 7.4), 70% (v/v) ethanol, twice with boiling 95% (v/v) ethanol, twice with acetone-chloroform 4:1 (v/v), and finally with 70% (v/v) ethanol. After each extraction the precipitate was centrifuged (2000g for 10 min), a 0.5 ml aliquot of the supernatant sampled and radioactive content determined by LSC; the remaining supernatant was decanted to waste. Pellets were resuspended in the next extraction solvent with the aid of sonication. Representative control samples were prepared by the addition of [³H]4-nonylphenol (400,000 dpm) to homogenized liver (350 mg) from untreated trout and incubated for 1 hr at 4^o C to enable the development of any interactions prior to analysis. The proportion of ³H-labeled residues in solvent extracts and covalently bound to the insoluble residual pellet were calculated as a percentage of the total present in the samples prior to extraction.

Preparation of Trout Hepatocytes. Trout were obtained from a local fish farm, euthanized by concussion and administered 2000 units of heparin into the blood stream by intracardiac injection. Fish were transported to the laboratory in Earles balanced salt solution on ice. Trout hepatocytes were isolated and maintained in suspension culture using a method similar to that described previously for bovine liver (Coldham et al., (1995); liver was perfused in situ at a flow rate of 8 ml/min at room temperature (18^o C) via an 18 gauge cannula needle placed in the vena porta. Perfusion buffers drained to waste via a small cut made in the heart. Biotransformation of [³H]4-nonylphenol was evaluated by incubation of suspension cultures (10 ml) of trout hepatocytes (1 × 10⁶ml¹) with 10µM 4-nonylphenol (1µCiml¹). Culture medium (1 ml) was sampled after 1, 2, 4, and 20 hr and separated from hepatocytes by centrifugation at 2000g and stored at 20^o C. A second aliquot (100 µl) of cell suspension was also sampled at each time point and analyzed for cell viability by the trypan blue exclusion test. The profile of metabolites in all samples of hepatocyte medium was analyzed by radio-HPLC without prior extraction as described for bile. The profiles of metabolites present in 20-hr hepatocyte medium samples were investigated after solid phase extraction by radio-HPLC and tentatively identified by selective -glucuronidase/aryl sulfatase enzyme hydrolysis and LC-MSⁿ as described above.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

³H-Labeled Residues in Tissues. The concentration of 4-nonylphenol equivalents (nmol/g ± 1 SD) found in trout tissues, excreta and body fluids (nmol/0.1 ml) determined by sample combustion and LSC are presented with distribution () and elimination () phase half-lives in table 1. Highest concentrations of residues were found in bile, faeces, and pyloric caecae. Elimination of residues from most tissues was biphasic and fitted a two compartment model (TopFit 2.0 pharmacokinetic and pharmacodynamic data analysis system for the PC; Heinzel et al., 1993) with weighting of x = 1/y². Prolonged -phase half-lives were found in muscle, skin, and liver although the concentrations were appreciably higher in liver (table 1).

Whole Body Autoradiography. An autoradiograph of a mid-line slice from a trout 24 hr after dosing, exposed alongside ³H calibration scales, is shown in fig. 1. Levels of radioactivity were substantially higher in liver, pyloric caecae, and hind gut than in kidney; residues were also present at the injection site. Exposure of film for 6 months provided images of 7 of the 8 ³H calibration scales.

View larger version (46K): [in this window] [in a new window]   Fig. 1.   Whole body autoradiograph of 3H labeled residues distribution in trout 24 hr after administration of 112.5 µCi [3H] and 0.375 mg of 4-nonylphenol.

Radio-HPLC Analysis of Tissue Metabolites. The chemical nature of tritium labeled residues found by sample combustion in tissues and excreta was further investigated by radio-HPLC. Extracts of bile, faeces, liver, pyloric caecae, and plasma sampled 1, 4, 24, and 144 hr after dosing contained a single major metabolite (retention time (r.t.) 10 min) and several poorly resolved minor metabolites (r.t. 5-7 min). A representative profile of these tritium labeled metabolites found in liver 144 hr after dosing is shown in fig. 2. For skeletal muscle, 24 hr after dosing, radio-HPLC provided evidence for a single radiolabeled compound with r.t. of 16 min, consistent with that of the parent compound. Radio-HPLC was less sensitive than sample combustion, and as a consequence ³H-labeled residues in samples of skeletal muscle 144 hr after dosing were below the limit of detection. The quantity of gonad obtained from juvenile fish was insufficient to enable analysis of metabolites by these procedures. Although not present at later sampling points, substantial quantities of the parent compound were evident in radio-HPLC chromatograms (not shown) of pyloric caecae, liver, and plasma up to 1, 4, and 24 hr, respectively, after dosing.

View larger version (7K): [in this window] [in a new window]   Fig. 2.   Radio-HPLC chromatogram of tritium labeled residues present in an extract of liver sampled 144 hr after dosing. [3H]4-Nonylphenol eluted with a retention time of 15.8 min under these conditions.

Enzyme Hydrolysis of Metabolites in Bile and Hepatocyte Medium. Radio-HPLC chromatograms of solid phase extracts of bile and trout hepatocyte medium before and after treatment with -glucuronidase/aryl sulfatase are shown in fig. 3. The major metabolite (III) present in bile and hepatocyte culture medium (VII) was hydrolyzed by incubation with -glucuronidase/aryl sulfatase to products V and IX, respectively, which had similar retention times to 4-nonylphenol. The cluster of minor metabolites in bile (I) and hepatocyte culture medium (VI) were hydrolyzed to products IV and VIII, respectively.

View larger version (17K): [in this window] [in a new window]   Fig. 3.   Radio-HPLC chromatograms of solid phase extracts of bile (144 hr after dosing) and trout hepatocyte medium (20 hr sample) before and after treatment with 1.4 mg/ml Helix pomatia -glucuronidase. Metabolites present in chromatograms have been labeled I-IX and tentatively identified by LC-MSn (table 2). [3H]4-Nonylphenol eluted with a retention time of 18.6 min under these conditions.

Analysis of Metabolites by LC-MSⁿ. Selected ion chromatograms of deprotonated pseudomolecular ions (M^-) of potential metabolites present in solid phase extracts of bile and trout hepatocyte medium before and after treatment with -glucuronidase/aryl sulfatase were prepared and matched for co-elution with peaks in the radio-HPLC chromatograms. Daughter ion (MS²) and granddaughter ion (MS³) spectra of metabolites I-XI are summarized in table 2. Metabolites I and VI, IV and VIII, V and IX present in bile and hepatocyte medium, respectively, had similar retention times and daughter ion spectra (MS²), whereas metabolites II, III present in bile, and VII in hepatocyte medium had dissimilar retention times but similar spectra. Subsequent investigation indicated that small differences in elution times of metabolites II and III in bile and VII in hepatocyte medium were related to the influence of matrix and sample pH.

Tentative Identification of Metabolites

I, II, III, VI, and VII. Daughter ion spectra (MS²) of these -glucuronidase/aryl sulfatase sensitive metabolites contained aglycone ions of the parent (m/z 219, metabolites II, III, and VII) or hydroxylated parent (m/z 235, metabolites I and VI) compound and ions from glucuronic acid (m/z 175) and fragments thereof (m/z 113). This indicated that the metabolites were glucuronide conjugates of 4-nonylphenol (II, III, and VII) or hydroxylated 4-nonylphenol (I and VI). The presence of an ion at m/z 133 corresponding to an aromatic ring fragment in the granddaughter ion spectrum (MS³) of metabolite VI indicates that this metabolite is a glucuronide conjugate of 4-n-hydroxynonylphenol.

IV, VIII. These products were produced by treatment of bile and hepatocyte medium with -glucuronidase/aryl sulfatase. Their spectra contained a deprotonated molecular ion of m/z 235. Similar daughter ion spectra (MS²) of daughters of 4-nonylphenol including the ion at m/z 133 provide further evidence (c.f. granddaughter ion m/z 133 of metabolite VI) that the position of alkylhydroxylation is distal to the phenolic ring and is lost during fragmentation to provide a similar spectrum to that of the parent compound.

V, IX. These products were also produced by treatment of bile and hepatocyte medium with -glucuronidase/aryl sulfatase and had similar HPLC retention times and daughter ion mass spectra to 4-nonylphenol.

Non-Radiolabeled Metabolites. HPLC eluates were analyzed for nonradiolabeled metabolites, since hydroxylation within the aromatic ring by cytochrome P₄₅₀ enzymes may eliminate the tritium label. A nonradiolabeled -glucuronidase/aryl sulfatase sensitive metabolite (X) was found in hepatocyte medium (r.t. 9.9 min) with an M of m/z 315. The loss of 80 atomic mass units from this ion to produce a single daughter of m/z 235 is consistent with the loss of sulfate. The granddaughter ion (MS³) spectrum of metabolite X contained a base ion at m/z 149 which was not evident in the daughter ion spectra of 4-nonylphenol (M m/z 219) or metabolites IV and VIII (M m/z 235). A plausible explanation for this intense ion at m/z 149 is hydroxylation within the aromatic ring and elimination of tritium. These data suggest that the metabolite X is 4-nonyl-n-hydroxyphenol sulfate. -Glu-curonidase/aryl sulfatase treated bile and hepatocyte medium both contained a product (metabolite XI, r. t. 17.2 min) with a deprotonated molecular ion at m/z 235. The longer HPLC retention time and loss of ³H provided preliminary evidence that this putative 4-nonylphenol metabolite was distinct from that of metabolites IV and VIII found in bile and medium, respectively. The daughter ion spectrum (MS²) of this ion (m/z 235) was similar to the granddaughter spectrum of putative 4-nonyl-n-hydroxyphenol sulfate found in hepatocyte medium rather than metabolites IV and VIII, which indicates that metabolite XI is 4-nonyl-n-hydroxyphenol. The unconjugated form of this metabolite in bile has not been identified. The relatively low abundance of M m/z 411 in bile has prevented successful MS³ analysis of the daughter ion 235 in bile.

Covalent Binding. The percentage of radioactivity present in solvent extracts and covalently bound to pellets from liver of treated fish 48 hr after dosing and to control liver pellets is shown in fig. 4. Covalently bound ³H-labeled residues (mean ± 1 SD) were significantly lower (P<0.0001, by Student's t-test) in control 0.19 ± 0.07% compared with treated liver 1.93 ± 0.09%. Although the majority of radioactivity in control and treated liver was extracted with ethanol, TCA extracted a greater percentage of ³H from treated than control liver, perhaps reflecting the presence of hydrophilic metabolites in treated liver. Recovery of radioactivity during solvent extraction was estimated by summation of radioactivity in all of the solvent extracts and was 101 ± 2.8 and 107 ± 3.6% for control and treated liver, respectively.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Exhaustive solvent extraction of control liver (open bars) and liver from treated (hatched bars) trout. Radioactivity in solvent extracts and in the final pellet (covalently bound material) after exhaustive extraction with the solvent shown is expressed as a percentage of the total. Error bars represent ± 1 SD

Metabolism of [³H]4-Nonylphenol by Trout Hepatocytes. Hepatocyte preparations were initially 92% viable as determined by the trypan blue exclusion test. These were used to investigate the potential value of in vitro biotransformation studies of 4-nonylphenol by comparison with metabolite profiles produced in vivo. Although hepatocyte viability was well maintained (79% at 20 hr), the rate of 4-nonylphenol metabolism decreased rapidly during the 20-hr incubation period. After solid phase extraction, metabolites in culture media at 20 hr were tentatively identified by selective enzyme hydrolysis and LC-MSⁿ (fig. 3 and table 2) as described above.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Appreciation of the pharmacokinetics of pollutants in susceptible animals, such as trout, is an important step in evaluating the risk to less conspicuous forms of wildlife and to the consumer.

In the present study highest concentrations of residues were found in trout bile and intestinal contents/faeces, indicating a biliary/fecal route of excretion. Others (Kleinow et al., 1992; Cravedi et al., 1985) have shown that fish are extremely active biliary concentrators of xenobiotics. Although the concentration of residues were relatively low in muscle, only biologically active 4-nonylphenol was detected, and notably this had a prolonged -phase half-life. Chronic exposure in polluted waters may thus facilitate continuous accumulation of parent compound into skeletal muscle to form a substantial depot since this tissue represents the majority of body mass. The presence of parent compound in skeletal muscle but not in liver may reflect the high capacity for xenobiotic metabolism of the latter tissue, combined with other factors such as lipophilicity and protein binding (Monro, 1994). The relatively high concentrations of residues found in liver are of potential biological significance since 4-alkylphenols have been shown to bind to oestrogen receptor prepared from rainbow trout liver (Jobling and Sumpter, 1983) and induce the oestrogen dependent synthesis of vitellogenin (Maitre et al., 1986; Jobling et al., 1995). However, excretion and concentration of metabolites in bile suggest that this organ is unlikely to represent a significant depot of parent compared with skeletal muscle. In contrast, plasma contained relatively low concentrations of total residues despite the presence of sex hormone binding protein (Pottinger and Pickering, 1990) which is associated with the binding of lipophilic substances such as oestrogens and many other steroid hormones. Although a similar distribution of [¹⁴C]4-nonylphenol residues in trout tissues has been reported by Lewis and Lech (1996), their study was shorter than the present one and demonstrated liver and muscle -phase half-life values for [¹⁴C]nonylphenol residues that were 5 and 17 times lower, respectively, than the -phase half-lives reported here. However, the absence of any information regarding the pharmacokinetic modeling methods used by Lewis and Lech (1996) and duration of tissue depletion studies precludes further meaningful discussion to explain these disparities.

WBA revealed a pattern of tissue distribution of ³H-labeled 4-nonylphenol residues similar to that evident from sample combustion and LSC. However, imaging of ³H disposition in tissue slices presents a number of technical difficulties related to the weak energy of emitted  particles which were addressed by the use of unprotected autoradiography emulsion films and long exposure times. The presence of radioactive residues in the posterior intestine/rectum provided further evidence for significant excretion in faeces. The presence of radioactivity in the kidney suggest a role in the excretion of 4-nonylphenol residues. Extensive excretion of xenobiotics may also occur across the gills (Kleinow et al., 1992) although the design of the present study did not allow evaluation of this route. The whole body autoradiograms revealed a relatively small amount of dose leakage into the muscle surrounding the injection site. This dosing route was selected because others, such as via downstream indwelling cannulae, produce chronic stress, particularly in immature fish, which may affect the kinetics of xenobiotic metabolism. Clearly, elimination of ³H through biotransformation in the aromatic ring presents the possibility that a proportion of unlabeled metabolites will evade detection by all analytical procedures used in this study (WBA, radio-HPLC detectors and liquid scintillation counters) with the exception of LC-MS.

Although tritium exchange may provide a contribution towards the -phase half-lives found in trout tissues by sample combustion, two lines of experimental evidence argue against a substantial influence. Tissue and excreta sampled throughout the study contained radio-HPLC profiles of tritium labeled residues similar to those present in bile. These residues in bile were subsequently identified by LC-MSⁿ as metabolites of 4-nonylphenol. Further, exhaustive solvent extraction of liver revealed that only 2% of the residues were covalently bound, and since the majority of tritium labeled residues were extracted from liver by organic rather than aqueous solvent (fig. 2), this indicates that they are lipophilic metabolites. Thus, the -phase half-lives of tritium in trout tissues are attributed to the presence of 4-nonylphenol metabolites with a small contribution associated with covalent binding in the case of liver or solely to parent compound in the case of muscle. The detection of very low levels (0.19%) of apparent covalent binding in samples of control liver may reflect either noncovalent interactions such as protein binding, sequestration by entrapment of radiolabel during precipitation, tritium exchange with macromolecules, or the outcome of minimal metabolism during incubation at 4^o C for 1 hr.

Metabolites in bile and hepatocyte medium were extracted with solid phase media prior to tentative identification by enzyme hydrolysis/radio-HPLC and LC-MSⁿ. Although this extraction strategy did not entirely eliminate matrix interference since the major metabolite in bile (III) had a longer retention time than that in hepatocyte medium (VII), similarity of mass spectra and the sensitivity of these metabolites to enzyme hydrolysis provided corroboration of common identity. Enzyme hydrolysis studies with Helix pomatia does not enable discrimination between glucuronide and sulfate conjugates since this preparation contains -glucuronidase and aryl sulfatase activities. Electrospray LC-MSⁿ analysis provided a level of confirmation to identification of metabolites since both conjugates and hydrolysis products were analyzed. Furthermore, mass spectral analysis provided detection of nonradiolabeled metabolites and enabled evaluation of the approximate site of hydroxylation. Since HPLC does not have the power to resolve the 22 or so para isomers found in commercial preparations of 4-nonylphenol (Bhatt et al., 1992; Wheeler et al., 1997), the present study has not considered alkyl side chain isomerism of metabolites. Neutral loss scanning by triple quadrupole mass spectrometry is a recognized strategy for the detection and discrimination of glucuronide and sulfate xenobiotic conjugates (Brownsill et al., 1992) since fragmentation may be induced by collisionally activated dissociation (CAD) with the loss of 176 and 80 a.m.u., respectively (Draper et al., 1990). Although neutral loss scanning is not possible on the ion trap LCQ mass spectrometer, MSⁿ analysis of aglycone ions present in daughter ion spectra, provide strong structural information. This was exemplified by analysis of granddaughter ion spectra of the putative 4-nonyl-n-hydroxyphenol sulfate metabolite. Aglycone and dehydroglucuronic acid daughter ions have been described in the CAD spectrum of 4-nitrophenyl--D-glucuronide (Draper et al., 1989). Similarly, CAD spectra of 4-nonylphenol and hydroxylated 4-nonylphenol glucuronide conjugates yielded aglycone and dehydroglucuronic acid daughter ions, whereas the sulfate conjugate fragmented by single loss of 80 a.m.u.

The relative abundance of 4-nonylphenol metabolites in bile and hepatocyte medium could not be accurately assigned since a proportion of tritium was eliminated during biotransformation. However, radio and ultra-violet (254 nm) HPLC analysis suggests that direct phase II metabolism, namely glucuronidation of the parent compound, predominates over other routes of metabolism in both trout liver and isolated hepatocytes. Relatively smaller quantities of phase I side chain and aryl ring hydroxylated metabolites were produced, as indicated in the proposed pathway for the biotransformation of 4-nonylphenol in trout and by isolated hepatocytes (fig. 5). The paucity of literature regarding the metabolic fate of alkylphenols in fish, or other species for that matter, limits informed discussion. Gas chromatography/mass spectrometry has been used to identify three -glucuronidase sensitive metabolites of 4-nonylphenol found in trout bile as 2-, 3-, and 4-(4-hydroxyphenyl)-8-hydroxy-nonanes (Meldahl, et al., 1996). Side chain ( or -1) hydroxylation is a common pathway of metabolism for compounds with aliphatic side chains such as phthalates (Kluwe, 1982). Phenols are hydroxylated by fish and readily excreted as glucuronic acid and sulfate conjugates (Kasokat et al., 1987), while 4-nonylphenol is excreted as a glucuronic acid conjugate by the rat (Knaak et al., 1966).

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Proposed pathways for the biotransformation of 4-nonylphenol in trout. Roman numerals correspond to metabolite designations described in fig. 3 and table 2. The bold arrow indicates the major pathway of biotransformation for metabolites found in bile and trout hepatocyte medium and the broken arrow indicates conjugation pathway for metabolite XI by trout hepatocytes. Gluc refers to glucuronic acid.

A similar profile of 4-nonylphenol metabolism was found in trout bile and hepatocyte culture medium with the exception of a sulfate conjugate present in the latter. This may be a result of the lower K_m of sulfo compared with UDP-glucuronosyl transferases (Pang, 1990) which will favor sulfatation at low substrate concentrations. A further possibility is that production of the lower molecular weight sulfate conjugate may favor excretion by the kidney and gills. Given these considerations, the results of the present study indicate that isolated hepatocytes are suitable for investigating the metabolic fate of 4-nonylphenol and may be usefully employed further in the large scale production of metabolites for subsequent purification and assessment of oestrogenic activity in future studies. Such a strategy may be suitable for assessing the environmental impact of other contaminant metabolites.

The unexpected finding that muscle provides a substantial depot of parent compound has implications for the putative hormonal action of 4-nonylphenol and for any risk associated with residues in edible tissue. Further consideration of this finding may be warranted given the likelihood that natural exposure may be of a chronic nature. Although in vitro tests have an important role to play in defining the potency of such residues, the present study serves to emphasize the necessity of pharmacokinetic studies for informed risk assessment of the likely impact of environmental contaminants on wildlife and man.