MARGARET M. KRAHNt, M. S. MYERS,. D. G. BURROWS and ...... We thank Drs William D. MacLeod, Jr and Usha Varanasi for reviewing this paper, Dr Donald ...
1984, VOI.. 14, NO. 8, 633-646
Determination of metabolites of xenobiotics in the bile of fish from polluted waterways MARGARET M . K R A H N t , M. S. MYERS, D . G . BURROWS and D. C. M A L I N S Northwest and Alaska Fisheries Center, Environmental Conservation Division, 2725 Montlake Boulevard East, Seattle, WA 981 12, USA
Received 9 August 1983 1. An h.p.1.c.-fluorescence technique was used to estimate relative concentrations of metabolites of xenobiotics in bile of 103 English sole (Parophryswetulus) from both polluted and minimally polluted (reference) sites in Puget Sound, WA. 2. Fish from polluted sites had concentrations of xenobiotics in bile with naphthalene-, phenanthrene- and benzo[a]pyrene-like fluorescence that averaged 9, 14 and 19 times, respectively, those of fish from reference sites.
3. Within a polluted site, fish with liver lesions had significantly higher bile concentrations of xenobiotics with benzo[a]pyrene-like fluorescence than did fish without liver lesions. 4. Individual metabolites of fluorene, phenanthrene, anthracene, biphenyl and dimethylnaphthalene were determined bv g.1.c.-mass spectrometry in extracts of hydroIysed bile of three English sole from polluted waterways; concentrations ranged from 90 to 19000 ng/g, wet wt. Other xenobiotics were tentatively identified, but not quantified.
Introduction Marine waters and sediments located near urban centres contain hundreds of xenobiotics (Malins 1980, Malins et al. 1980, Gossett, Brown and Young 1982, Malins et al. 1982). Aquatic organisms can accumulate many of these compounds and metabolize some of them to toxic and/or carcinogenic products that may be more harmful than the parent structures (Roubal, Collier and Malins 1977, Yang, Deutsch and Gelboin 1978, Gruger et al. 1981, Malins and Hodgins 1981, Varanasi and Gmur 1981, Varanasi, Stein and Hom 1981). A number of pathological conditions, including neoplasms, have been observed in liver tissue of fish from polluted waters (Falkmer et al. 1977, Pierce, McCain and Wellings 1978, Smith et al. 1979, Malins et al. 1980, Pierce, McCain and Wellings 1980, Baumann, Smith and Ribick 1982, McCain et al. 1982, Malins et al. 1982). For example, recent studies in Puget Sound, WA, USA have linked certain idiopathic liver lesions, including neoplasms, in bottom-dwelling fish to aromatic hydrocarbons (AH) associated with the sediments (Malins et al. 1982). However, routine analyses for parent AH in tissue samples from fish captured in polluted areas often show only traces of AH, even when the sediments contain high concentrations of these compounds (Baumann et al. 1982, Malins et al. 1982). It is well established that AH taken up by the fish are readily converted to metabolites (e.g. dihydrodiols and phenols) (Roubal et al. 1977, Malins 1980, Gruger et al. 1981, Malins and Hodgins 1981, Varanasi and Gmur 1981), but current routine analyses for AH do not assay for metabolites or other polar compounds.
t To whom correspondence should be addressed. llention of trade names is for information only and does not constitute endorsement by the US Department of Commerce.
M. M. Krahn et al.
Metabolites of a single AH, or of a simple mixture of AH, are readily determined in biological systems (Krahn et al. 1980, Krahn, Collier and Malins 1982). For example, naphthalene metabolites in fish bile were quantified by direct-injection, high-performance liquid chromatography with fluorescence detection (h.p.1.c.fluorescence) (Krahn et al. 1980). Alternatively, Solbakken and co-workers (Solbakken et al. 1980, Solbakken and Palmork 1981) fed phenanthrene to several species of marine organisms and identified the trimethylsilyl ethers of the metabolic products by gas chromatography-mass spectrometry. In addition, Gmur and Varanasi (1982) used fluorescence spectrometry and g.1.c.-mass spectrometry to characterize several metabolites of benzo[a]pyrene (BaP) in English sole (Parophrys vetulus). Until recently, however, no techniques had been reported for determining individual AH metabolites in mixtures of several AHs. Although a recent study described the determination by g.1.c.-mass spectrometry of several AH metabolites in fish exposed to fuel oil in the laboratory (Krahn and Malins 1982), more definitive methods are needed to characterize the complex mixtures of metabolites in environmentally exposed organisms. In this paper, we describe an h.p.1.c.-fluorescence technique for estimating relative concentrations of AH metabolites in bile of English sole from four sites in Puget Sound. In addition, liver lesions in these fish were diagnosed histologically and lesion presence was statistically related to metabolite concentrations in bile. We also describe an advance in methodology which permits determination of some individual metabolites in the bile of fish exposed to contaminated sediments.
Experimental Fish capture, necropsy and histopathology Adult English sole (Parophrys vetulus) were captured by otter trawl from the Duwamish Waterway and from the Lake Washington Ship Canal, Seattle, WA, USA (contaminated sites) and near Meadow Point and Port Madison, Puget Sound, WA (reference sites). These sites were selected on the basis of previous data (Malins et al. 1980, Malins et al. 1982, McCain et al. 1982), which documented high concentrations of AH and other xenobiotics in sediments of the Duwamish Waterway and the Lake Washington Ship Canal and substantially lower concentrations in sediments from the reference sites. After capture, fish were placed in holding tanks containing fresh seawater until they could be necropsied aboard the research vessel. Before necropsy, the length of each fish was measured to the nearest millimeter. Bile was removed from the gall bladder and frozen until analysed. Tissue was then excised from each liver, processed by standard histological techniques (Malins et al. 1980), and examined by light microscopy for the presence of lesions. All diagnoses were made using a blind system in which the examiner had no knowledge of site of capture or results of bile-metabolite measurements. Chemicals and reference standard 2-Hydroxybiphenyl, 4-hydroxybiphenyl, 9-phenanthrol and 9-hydroxyfluorene were obtained from Aldrich Chemical Co. (Milwaukee, WI, USA), and 4-methyl-1-naphthol and 2,2’-dihydroxybiphenyl from K & K Labs (Plainview, NY, USA). Phenanthrene-9-carboxaldehyde,9-fluorenylmethanol, 9( 1OH)-anthracenone, 9,10-anthraquinone, anthracene-9-carboxaldehyde,9,lO-phenanthraquinone and 9-anthracenemethanol were purchased from Fluka Chemical (Hauppauge, NY, USA). 1-Naphthyl-P-Dglucuronic acid (naphthyl glucuronide), 1-naphthyl sulphate, 2-naphthyl-a-~-glucoside (naphthyl glucoside), 1-naphthol and P-glucuronidase No. G-8132 were obtained from Sigma Chemical (St Louis, MO, USA). 2,6-Dimethyl-3-naphthol,6-methyl-2-naphthalenemethanol and trans-3,4-dihydroxy-3,4dihydro-2,6-dimethylnaphthalenewere prepared in our laboratories (Gruger et al. 1981). T h e following benzo[a]pyrene metabolite standards were provided by the National Cancer Institute (Bethesda, M D , USA): 9,10-dihydro-9,10-dihydroxybenzo[a]pyrene (BaP 9,lO-dihydrodiol); 7,8-dihydro-7,8dihydroxybenzo[a]pyrene (BaP 7,8-dihydrodiol); 9-hydroxybenzo[a]pyrene (9-hydroxy BaP); 1-hydroxybenzo[a]pyrene (1 -hydroxyBaP); 3-hydroxybenzo[a]pyrene (3-hydroxy BaP); benzo[a]pyrene 3-sulphate, sodium salt (BaP 3-sulphate); 1-benzo[a]pyrenyl-~-~-glucopyranosiduron~c acid (BaP 1glucuronide). Citric acid monohydrate (U.S.P.) was obtained from Mallinckrodt (St Louis, MO, USA) and bis(trimethylsily1)-trifluoroacetamide (BSTFA) from Regis Chemical (Morton Grove, IL, USA).
Metabolites of xenobiotics in jish bile
Disodium hydrogen orthophosphate heptahydrate, Ultrex acetic acid and h.p.1.c. grade solvents, 2propanol and water, were purchased from J. T . Baker Chemical (Phillipsburg, NJ, USA). T h e ‘distilled in glass’ solvents, methanol, acetone and methylene chloride, were obtained from Burdick and Jackson Laboratories (Muskegon, M I , USA). A g.1.c.-mass spectrometry standard of the following reference compounds, dissolved in methanol, was prepared (listed in order of g.1.c. elution, nglpl): hexamethylbenzene (internal standard 4.10), 1naphthol (46.7), 2-hydroxybiphenyl (23.1), 4-methyl-1 -naphthol (26.7), 2,2’-dihydroxybiphenyl (47.1). 6-methyl-2-naphthalenemethanol (7.9, 4-hydroxybiphenyl (65.5), 2,6-dimethyl-3-naphthol (14.2), trans 3,4-dihydroxy-3,4-dihydro-2,6-dimethylnaphthalene (8.2), 9-hydroxyfluorene (46.2), 9-fluorenylmethanol (16.7), 9(10H)-anthracenone (6.7), 9,lO-anthraquinone (18.5), phenanthrene-9-carboxaldehyde (17.3), 9-phenanthrol (44.3), anthracene-9-carboxaldehyde (16.2), 9,lO-phenanthraquinone (1 5.2) and 9-anthracenemethanol (16.4).
Reverse-phase h.p.1.c. Separations were performed with a Spectra-Physics model 8000B high-performance liquid chromatograph (Santa Clara, CA, USA). A Perkin-Elmer model MPF-44A fluorescence spectrometer equipped with a ‘square’ flow cell was used for detection. Samples were injected using a Waters model 710A variable volume automatic sampler (Milford, MA, USA). T h e analytical column was a 0.26 x 25 cm reverse-phase HC-ODS column (Perkin-Elmer Corp.). A 0.39 x 2.3 cm stainless-steel guard column (Waters Assoc.) was dry packed with Vydak 37 pm size reversephase packing (The Separations Group, Hesperia, CA, USA), and the column was replaced regularly as system pressure increased. T h e guard column effectively reduced degradation and contamination of the analytical column. Acetic acid-water (5 pl/l) (solvent A) and methanol (solvent B) were used in a linear gradient as follows: 100% solvent A to 100% solvent B in 15 min; 7 min at 100% B; 3 min to return to 100% A; and 10 rnin reequilibration at 100% A. Note that the concentration of acetic acid in Solvent A is far lower than reported previously (Krahn and Malins 1982, Krahn et al. 1982). This decrease was necessitated by a change in the properties of the column packing (Perkin-Elmer Corp., personal communication). Flow was 1.0 ml/min, except during re-equilibration when it was increased to 1.5 ml/min. The oven temperature was 50°C. Bile was injected directly into the h.p.1.c. for analytical runs and, for most samples, the chromatograms were recorded at representative fluorescence wavelength pairs (Krahn et al. 1981): excitation/emission, respectively, 290/335 nm for naphthalenes (naphthalene wavelengths), 256/380 nm for phenanthrenes (phenanthrene wavelengths) and 380/430 nm for BaP (BaP wavelengths). All samples were recorded at naphthalene wavelengths; those from a first sampling were also recorded at phenanthrene wavelengths and those from a later sampling at BaP wavelengths. Ultraviolet absorption at 254nm was also recorded for each chromatogram. Nonequilibriurn chromatography and fraction collection T h e gradient elution conditions were identical to those described above except that re-equilibration at 100% A was at a flow of l.Oml/rnin. This resulted in an h.p.1.c. column which was not fully equilibrated with solvent for the next injection, as determined by a plot of system pressure. T h e pressure in the unequilibrated system was higher than the nearly constant pressure at equilibrium. Bile from ) a fraction was collected from experimental or from reference fish was injected into the h.p.1.c. ( 4 0 ~ 1and 1.9 to 3.9 min for enzymic hydrolysis. Enzymic hydrolysis and isolation of nonconjugated metabolites Acetate buffer (0.4 M acetic acid, 0.4 M sodium acetate, p H 5, 1 ml) and 8-glucuronidase dissolved in buffer (0.1 ml, 2000units glucuronidase activity and 25 units sulphatase activity) were added to each fraction from h.p.1.c. (above). T h e mixture was shaken at 40°C for three hours to hydrolyse the conjugates (Varanasi et al. 1982). T h e hydrolysis appeared to be complete as determined by comparison of the h.p.1.c.-fluorescence chromatograms of the sample before and after hydrolysis; chromatograms of a sample hydrolysed for 16 h showed no additional products. Citrate buffer (1.0 M citric acid, 0.2 M disodium hydrogen orthophosphate heptahydrate, p H 2.5, 1 ml) was added to each hydrolysis mixture and the metabolites were extracted by the automated extractor/concentrator method described previously (Krahn et al. 1982).Wash solvent was 1.7 ml water and extraction solvents were 0 8 3 ml acetone-methanol (50 : 50, v/v) and 0.83 ml methylene chloride-isopropyl alcohol-water (75 : 25 : 2, by vol.) for each sample. T h e extract was reduced in volume, under a stream of nitrogen, to 20p1 for analysis by g.1.c.-mass spectrometry. T h e extract from an additional hydrolysis was evaporated to dryness under a stream of nitrogen in preparation for silanization. Preparation of trimethylsilyl ethers of AH nonconjugates T h e dried extract from the hydrolysis step was dissolved in 5 0 ~ of 1 BSTFA and the solution was heated for one hour at 60°C (Brooks et al. 1982). Silanization was complete after this treatment, and the samples were stored in the freezer at -20°C until needed for g.1.c.-mass spectrometry. T o prepare a silanized g.1.c.-mass spectrometry standard, solvent was removed (under nitrogen) from a mixture of
M. M. Krahn et al.
reference compounds (see list that follows), 1 0 0 ~ of 1 BSTFA was added, and the solution was treated as above. T h e amount (ng/g bile) of a T M S derivative necessary for g.1.c.-mass spectrometry identification was estimated: 1-naphthol (1000), 2-hydroxybipheny1(3000),2,2’-dihydroxybiphenyl(3000),6-methyl2-naphthalenemethanol (lOOO), 4-hydroxybiphenyl (lOOO), trans-3,4-dihydroxy-3,4-dihydro-2,6dimethylnaphthalene (20000), 9-hydroxyfluorene (lOOO), 9-phenanthrol (1000) and BaP-7.8-dihydrodiol (2000). All silanized standards were identified by g.1.c.-mass spectrometry.
Gas chromatography and mass spectrometry Portions (2 PI) of the concentrated extract containing nonconjugated metabolites and also of the BSTFA solution of silanized nonconjugates were each injected, with the purge valve closed, into a Hewlett-Packard model 5840 gas chromatograph (Palo Alto, CA, USA) equipped with a flame-ionization detector. T h e purge valve was opened after 18 s. G.1.c.-mass spectrometry was performed using an identical g.1.c. system interfaced with a Finnigan 3200 mass spectrometer (Finnegan M A T Corp., Sunnyvale, CA, USA) and an IKCOS 2300 data system (Finnigan M A T Corp.). T h e fused-silica capillary column was coated with DB-5 (30m x 0.25 mm int. diam., J & W Scientific, Rancho Cordova, CA, USA). Helium carrier gas was adjusted to a linear velocity of c. 33 cmjs at 300°C. Column temp. was programmed from 80 to 300°C at 4”C/min. G.1.c.-mass spectrometry identifications and quantifications Metabolites were identified by comparison to reference compounds, silanized reference compounds, or to spectra in the expanded EPA/NIH Mass Spectral Data Base (on computer disk). In general, commercially available reference compounds were not the same isomers present in the experimental samples. For example, only two of the reference compounds were found in our bile sample. Therefore, many compounds were quantified using a g.1.c.-mass spectrometry response factor for an isomer (e.g. the phenanthrol isomers were quantitated using the response factor for 9-phenanthrol). Estimated minimum detectable amounts of reference compounds in g.1.c.-mass spectrometry analysis of hydrolysed bile extracts are (ng/g of bile): 1-naphthol (40), 2-hydroxybiphenyl (40), 4-methyl-lnaphthol (70), 2,2-dihydroxybiphenyI (80), 6-methyl-2-naphthalenemethanol (60), 4-hydroxybiphenyl (loo), (20), 2,6-dimethyl-3-naphthol (30), trans-3,4-dihydroxy-3,4-dihydro-2,6-dimethyl-naphthalene 9-hydroxyfluorene (40), 9-fluorenyImethanol(20),9( 10H)-anthracenone (SO), 9,lO-anthraquinone (140), phenanthrene-9-carboxaldehyde(40), 9-phenanthrol (1lo), anthracene-9-carboxaldehyde (40), 9,lOphenanthraquinone (40) and 9-anthracenemethanol (40). Stop-flow fluorescence spectra T h e h.p.1.c. flow was stopped at c. 17 min after injection of a sample of hydrolysed bile to record the fluorescencespectra of a compound having a large response at BaP wavelength pairs (excitation/emission, 380/430 nm). Similar stop-flow spectra were recorded for BaP metabolite reference standards. T h e excitation spectrum was recorded from 250 to 410nm with emission set at 430 nm; the emission spectrum was recorded from 375 to 470nm with excitation at 360nm.
Results Pathology of English sole livers and examination of fish length variability Histopathological examination of liver tissue from individual fish led to diagnoses of a spectrum of idiopathic (of unknown etiology) and parasitic/infectious lesions. Only the lesions which were of an idiopathic nature (i.e. excluding parasitic/infectious conditions) were included in subsequent treatment of the histopathology data. Specific idiopathic liver lesions detected in this study were placed within one of eight general diagnostic categories as follows: (1) specific degenerative lesions (megalocytic hepatosis); (2) nonspecific degenerative lesions (hepatocellular coagulation necrosis, hepatocellular hyalinization); (3) hepatocellular storage disorders (hemosiderosis, fatty change); (4) non-neoplastic proliferative lesions (cholangiofibrosis, hepatocellular regeneration); (5) ‘preneoplastic’ hepatocellular lesions (nodular eosinophilic hypertrophy, basophilic foci, clear-cell foci); (6) hepatocellular neoplasms (minimum deviation nodule, liver-cell adenoma, hepatocellular carcinoma); (7) nonparasitic chronic inflammatory lesions (parenchymal fibrosis/fibroplasia); (8) vascular disorders (congestion, haemorrhage, intravascular fibrin deposition. Most of these lesions have been described in detail by
Metabolites of xenobiotics in jish bile
McCain et al. (1982). Any fish affected by one or more of these lesion types, regardless of severity, was included in the lesion-containing group(s) of fish in subsequent statistical analyses. T h e prevalences of lesions detected in all sole sampled in the present study are listed by general diagnostic categories and area of capture (table 1 ) . T h e vast majority of idiopathic lesions were found in fish from the Duwamish Waterway and Lake Washington Ship Canal, with only minor nonspecific degenerative, non-neoplastic proliferative and hepatocellular storage disorders detected in fish from the Meadow Point and Port Madison sampling sites. Fish lengths (206458 mm) were not significantly different among the four sampling sites when compared by one-way analysis of variance (ANOVA) (Scheffe 1959), indicating that homogeneous lengths (which approximate age) of sole were sampled. Table 1 .
Frequencies and prevalences (%) of idiopathic liver lesions detected in English sole from four sampling areas in Puget Sound, WA, listed by general diagnostic categories. Sampling areat Duwamish Waterway (N=58)
Lake Washington Ship Canal (N=12)
Specific degenerative lesions
Port Madison (N=6)
Meadow Point (N=27)
Nonspecific degenerative lesions
Son-neoplastic proliferative lesions
(7) 0 (0) 1
‘Preneoplastic’ hepatocellular lesions
Hepatocellular storage disorders
Hepatocellular neoplasms Son-parasitic chronic inflammatory lesions Vascular disorders Without significant lesions
See text for specific lesions comprising each category.
t Totals may be greater than 100%because some fish have multiple
Estimation of concentrations of metabolites in fish bile H.p.1.c. chromatograms from direct injections of 5 p1 of fish bile (under equilibrium conditions) were recorded at fluorescence excitation/emission wavelength pairs of typical petroleum hydrocarbons: the naphthalenes, phenanthrenes and BaP. Some types of compounds which fluoresce at a particular wavelength pair include ( a ) the parent and its metabolites, ( b ) alkylated derivatives of the parent and their metabolites and ( c ) N-, S- or 0-containing compounds with the same aromatic ring structure. T h e chromatograms of bile from fish caught at polluted sites showed complex mixtures of fluorescent xenobiotics (figure 1 A and D). Most of the fluorescence was attributed to AH metabolites or the other fluorescent xenobiotics described above.
M . M . Krahn et al.
ected Retention time (rnin)
: 5 :tion ected 1
Retention time (min)
Figure 1. H.p.l.c./fluorescence chromatograms at BaP (A-C) and naphthalene (D-F) wavelengths from a direct injection of fish bile ( 2 ~ 1 ) . (A, D), Bile of English sole from the Duwamish Waterway (polluted site) chromatographed under ordinary gradient elution conditions (see Experimental). (B, E), Bile of sole from the Duwamish Waterway chromatographed on an h.p.1.c. column which had not been returned to equilibrium before injection (see Experimental). A fraction (indicated), from 1.9 to 3.9 min, was collected for further analysis. (C, F), Bile of English sole from Meadow Point (reference site) chromatographed under ordinary gradient elution conditions.
Metabolites of xenobiotics in fish bile
However, response from naturally occurring compounds could not be ruled out, though their contribution is expected to be small, because chromatograms of bile from reference fish had few peaks (figure 1 C and F). Integrated areas of peaks eluting after seven minutes in the chromatograms were summed (the 'area sum') per bile injection ( 5 pl); compounds eluting before seven minutes were excluded from this study because no known AH metabolite reference standards, including the conjugates, eluted before that time. English sole from the polluted sites had naphthalene, phenanthrene and BaP area sums that averaged 9 , 1 4 and 19 times, respectively, those of fish from the reference sites (tables 2, 3 and 4).
Statistical analysis relating data of metabolites in bile of captured fish to sampling site Because the standard deviations increased with the mean area sums, naphthalene, phenanthrene and BaP area sums in bile from fish captured from polluted and Table 2. Mean h.p.1.c.-fluorescence peak areas (area sums), ranges and one-way analysis of variance of area sums at naphthalene wavelengths in bile of adult English sole captured in Puget Sound, WA. Reference sites
Range of area sumst (x area units) Mean area s u m f S . D . 1 (x area units) Number of samples (N)
009 to 0.18
0.0 1 to 0.57
018 kO.15 27
Polluted sites Lake Wash. Ship Canal
029 to 4.98
0.06 to 8.63
1.5q - 1.25
?All area sums are based on 5pI injections of bile. Naphthalene wavelengths were 290/335nm, excitation/emission. 1Significant difference (PGO.001) among all sampling sites by analysis of variance (Scheffe 1959), F=42.27. Data were log transformed. Q Significantly different (PGO.001) from reference sites by Scheffe's (1959) multiple comparison procedures.
Table 3. Mean h.p.1.c.-fluorescence peak areas (area sums), ranges and one-way analysis of variance of area sums at phenanthrene wavelengths in biie of adult English sole captured in Puget Sound, WA. Reference sites
Range of area sumst (x area units) Mean area sumfS.D.1 (x area units) Number of samples (N)
0.04 to 0 1 4 0.083 f 0.041 6
0.01 to 0.16 0.057 +0052 6
Polluted sites Lake Wash. Ship Canal 0.1 1 to 2.83 0%8Q 072 12
Duwamish Waterway 0.02 to 3.01 1.019 k 0.83 19
"All area sums are based on 5p1 injections of bile. Phenanthrene wavelengths were 256/380nm, excitation/emission. 1Significant difference (PGO.001) between sampling sites by analysis of variance (Scheffe 1959), F= 10.62. Data were log transformed. 9 Significantly different from Port Madison (PGO.05)and from Meadow Point (Pg001)by Scheffe's (1959) multiple comparison procedures.
M . M . Krahn et al.
Table 4. Mean h.p.1.c.-fluorescence peak areas (area sums), ranges and results of one-way analysis of variance for area sums at benzo[a]pyrene wavelengths in bile of adult English sole captured in Puget Sound, WA. Reference site (Meadow Point) Liver lesionst
Range of area sums (x area units)$
Mean area sum+S.D. (x area units)
Number of samples (N)
Degrees of freedom
Source of variation: All samples All Meadow Point sole v. all Duwamish sole Meadow Point sole, lesions v. normal Duwamish sole, lesions v. normal Error
0.003 to 0.16
Polluted site (Duwamish Waterway) Liver lesions
0032 to 5.83
Summary of analysis of variance§: Mean F square
Level of significance