Determination of Xenobiotic Intrinsic Clearance in Freshly Isolated

The in vivo metabolic stability (represented by hepatic clearance) of ... Table 1. Chemical Structures, LogKOW Values, Industrial or Agricultural ... ...
0 downloads 0 Views 165KB Size
Download PDF
Environ. Sci. Technol. 2007, 41, 3269-3276

Determination of Xenobiotic Intrinsic Clearance in Freshly Isolated Hepatocytes from Rainbow Trout (Oncorhynchus mykiss) and Rat and Its Application in Bioaccumulation Assessment XING HAN,* DIANE L. NABB, ROBERT T. MINGOIA, AND CHING-HUI YANG DuPont Haskell Laboratory for Health and Environmental Sciences, Newark, Delaware 19714

Bioaccumulation in fish depends on the dynamics of various processes that involve fish uptake, storage, and elimination of xenobiotics. Elimination via fish biotransformation is a primary process that can be evaluated in an in vitro system to improve the performance of the prediction of xenobiotic bioaccumulation potentials. In this study, values of intrinsic clearance (CLint) of seven reference compounds (atrazine, molinate, 4,4-bis(dimethylamino)benzophenone, 4-nonylphenol, 2,4-di-tert-butylphenol, trifluralin, benzo(a)pyrene) in hepatocytes freshly isolated from rainbow trout and rat were determined using a substrate depletion approach. Atrazine was metabolized in rat hepatocytes with a CLint value of 3.81 ( 1.96 mL/h/ 106 cells, whereas in trout hepatocytes, the clearance was not significant until very high cell concentration was used and the rate was estimated to be approximately 0.002 mL/h/106 cells. Intrinsic clearance values for all other compounds were 5.5-78.5-fold lower in trout hepatocytes than those in rat hepatocytes. Trout hepatic clearance (CLH) values were extrapolated from the CLint values using a “well-stirred” liver model. Biotransformation rate constants (kMET) of the compounds in trout were subsequently estimated and used as inputs to a kinetic model for the prediction of bioconcentration factors (BCF) in fish. Compared to the BCF values predicted without consideration of fish biotransformation, the inclusion of estimated kMET values significantly improved fish BCF predictions for the reference compounds. This study demonstrates a framework for future bioaccumulation assessment of xenobiotics using combined information of the physical-chemical properties of the compounds and the biotransformation potentials of the compounds in fish.

Introduction Biotransformation is a key process in detoxification, bioactivation, and bioaccumulation of xenobiotics in fish. Adequate evaluation of fish biotransformation of xenobiotics will greatly improve our understanding of the toxicological effects (1) and bioaccumulation potentials (2) of xenobiotics in fish. * Corresponding author phone: 302-366-5214; fax: 302-366-5003; e-mail: [email protected]. 10.1021/es0626279 CCC: $37.00 Published on Web 03/22/2007

 2007 American Chemical Society

Bioaccumulation depends on the dynamics of various processes that involve fish uptake, storage, and elimination of xenobiotics, and is routinely measured by the bioconcentration factor (BCF). BCF is defined as the ratio of the chemical concentration in fish as a result of absorption via the respiratory surface to that in water at steady state. At present, bioaccumulation assessment is typically conducted with models that are driven primarily by the lipophilicity (represented by octanol-water partition coefficient, KOW) of the chemicals. A kinetic model, proposed by Arnot and Gobas (3) included a biotransformation rate constant term (kMET), but assumed kMET ) 0 in their assessment, mainly due to the lack of information of fish biotransformation capabilities. Mekenyan and others (4) proposed a “baseline” model, in which fish metabolism was not experimentally determined, but was simulated using modeled rat metabolism. It has been shown, however, that there are significant differences between rat and fish in their biotransformation capabilities (5-7). A relatively quick and reliable in vitro method to experimentally determine the rate of xenobiotic biotransformation in fish, therefore, is needed to enhance our current capabilities in bioaccumulation assessment. The in vivo metabolic stability (represented by hepatic clearance) of pharmaceutical compounds in mammals is frequently estimated via in vitro measurement of intrinsic clearance in liver microsomes or isolated hepatocytes using a substrate depletion approach and a “well-stirred” liver model (8, and references therein). It was proposed that the same concept could be adopted to evaluate fish biotransformation, which in turn would improve the prediction of bioaccumulation potentials of xenobiotics in fish (9). In this study, we selected a group of seven uncharged reference compounds (atrazine, molinate, 4,4-bis(dimethylamino)benzophenone (michler’s ketone, MK), 4-nonylphenol (4NP), 2,4-di-tert-butylphenol (DTBP), trifluralin (TF), and benzo(a)pyrene (BaP), Table 1) that have diverse chemical structures, industrial applications and metabolic pathways, and a wide range of lipophilicity (LogKOW: 2.61-6.13). We determined CLint values of these compounds in freshly isolated hepatocytes from rainbow trout and rat using the substrate depletion approach. Trout CLH and kMET values were subsequently obtained and used as inputs to Arnot and Gobas’ model for the prediction of BCF values of these compounds in fish. We showed that by including estimated fish biotransformation capabilities, we greatly improved the performance of Arnot and Gobas’ model to predict the bioaccumulation potentials of our reference compounds.

Materials and Methods Materials. Atrazine, molinate, and TF were obtained from Chem Service (West Chester, PA). MK, 4NP, and DTBP were obtained from Sigma-Aldrich (St. Louis, MO). BaP was obtained from MolTox (Boone, NC). The purities of these chemicals were between 98 and 99.5%. Perfusion media, buffers, and fetal bovine serum (FBS) were obtained from Invitrogen (Carlsbad, CA). All other chemicals, if not specified in the article, were obtained from Sigma-Aldrich. Animals. Male juvenile rainbow trout (Oncorhynchus mykiss), approximately one and a half years old and approximately 10-12 in. in length, were purchased from Limestone Springs Fishing Preserve, Richland, Pennsylvania. The fish were held for at least one week in continuously flowing well water at a water temperature of approximately 10 °C under a 16:8 light/dark cycle, fed with AquaMax Starter Fingerling 300 5D03 (PMI Nutrition International, LLC) once daily, and fasted for 24 h prior to surgical manipulation. VOL. 41, NO. 9, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3269

TABLE 1. Chemical Structures, LogKOW Values, Industrial or Agricultural Applications and Major Metabolic Pathways of Seven Reference Compounds

a The source of LogK b Metabolic pathways observed in rat. c Metabolic pathway OW values were summarized in Supporting Information, S1. observed both in rat and rainbow trout. d Metabolic pathway is assumed to be similar to 4NP. Formation of a quinone methide reactive intermediate is not likely for DTBP (34).

Male Crl:CD(SD)IGS BR rats were obtained from Charles River Laboratories (Raleigh, NC). Upon arrival, all animals were housed in quarantine for at least 4 days. Animals were provided tap water ad libitum and fed PMI Nutrition International, LLC Certified Rodent LabDiet 5002 ad libitum. At the time of hepatocyte isolation, rats were 6-8 weeks of age. Animal rooms were maintained at a temperature of 1826 °C and a relative humidity of 30-70%, and were artificially illuminated (fluorescent light) on a 12 h light/dark cycle. Hepatocytes Isolation. The procedures for isolation of trout and rat hepatocytes were described previously (6). The cells were suspended in Leibovitz L-15 medium at pH 7.8 (trout) or 7.4 (rat). Cells were counted in the presence of 0.04% trypan blue. The viabilities of the cells were greater than 95% (trout) or 80% (rat). Hepatocytes Incubation. Freshly isolated rat (1 × 106 cells/ mL) or trout (2 × 106 cells/mL) hepatocytes in Leibovitz L-15 medium were preincubated at 37 °C and pH 7.4 (rat) or 10 °C and pH 7.8 (trout) for 5 min. The reaction was initiated by adding the compound (final concentrations 2 µM for atrazine, molinate, MK, and BaP; 5 µM for 4NP and DTBP; or 15 µM for TF; in acetonitrile or methanol (MK, 4NP, and DTBP); final solvent concentration was less than 0.5%). Reactions were terminated at regular time intervals (0-120 min for rat hepatocytes; 0-240 min for trout hepatocytes) with 9-fold ice-cold acetonitrile. The samples were centrifuged briefly to remove the precipitates. The supernatants were transferred to 96-well plates for LC analysis. For control samples, the cells were first inactivated by heating at boiling temperature for 10 min and then cooled on ice. The compounds were dosed to the heat-inactivated cells and then analyzed the same way as described above. Hepatocytes Binding. Freshly isolated trout hepatocytes were left at room temperature for 24 h. Compounds were 3270

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 9, 2007

dosed to the 24-h-old hepatocytes the same way as they were in the clearance experiments. After 5-min incubation at 10 °C, aliquots were extracted with 9-fold ice-cold acetonitrile to obtain the total concentration, Ctotal. The remaining hepatocyte suspension was centrifuged and the compound concentration in the supernatant was considered the unbound concentration, Cfree. The unbound fraction, fu,h, therefore, is equal to Cfree/Ctotal. Sample Analysis. For atrazine, molinate, MK, and TF, the samples were analyzed on a LC/MS/MS system [an Applied Biosystems 4000 QTrap mass spectrometer (Foster City, CA), an Agilent 1100 HPLC (Palo Alto, CA), and a CTC PAL autosampler (LEAP technology, Carrboro, NC)] that was equipped with either an electrospray (ESI, for atrazine, molinate, and MK) or atmospheric pressure chemical (APCI, for TF) ionization interface. Positive ions were acquired with a probe temperature of 450 °C by multiple reaction monitoring (MRM) of the protonated molecular ions (M + H)+ and the fragment ions (atrazine, 216.3/174.0; molinate, 188.3/ 126.1; MK, 269.3/148.1; TF, 336.3/236.2). Compounds were separated on a Waters column (XTerra MS C18, 2.5 µm, 2.1 × 30 mm) within 7.5 min by linear gradient from 95% eluent A (0.1% formic acid in water) to 100% eluent B (0.1% formic acid in acetonitrile) at a flow rate of 250 µL/min, or from 50% A to 100% B at a flow rate of 500 µL/mL (for TF only). 4NP, DTBP, and BaP samples were analyzed on a LC/ fluorescence system (a Waters 2795 HPLC system (Milford, MA) equipped with a Waters 2475 Multi λ Fluorescence Detector). Eluent consisted of water for A and acetonitrile for B at a flow rate of 1.2 mL/min. Compounds were separated on a Agilent Zorbax SB-C18 column (4.6 × 75 mm, 3.5 µm) for 10 min linear gradient from 30% B to 100% B (4NP and DTBP) and from 50% B to 100% B (BaP). Compounds were monitored by fluorescence at excitation of 275 nm and

emission of 310 nm for 4NP and DTBP, or excitation of 270 nm and emission of 400 nM for BaP. Intrinsic Clearance. Substrate depletion data followed first-order kinetics:

LogCt ) LogC0 -

k ‚t 2.3

(2)

where Vinc is the volume (mL) of the incubation medium and Ccell is the concentration (106 cells/mL) of hepatocytes in the incubation medium, respectively. Trout hepatic clearance, CLH (mL/h/kg), was estimated based on a “well-stirred” liver model (8):

CLH )

QH × (CLintHTWL) × (fu,b/fu,h) QH + (CLintHTWL) × (fu,b/fu,h)

1 - fu,h ) 0.676LogKow - 2.215 fu,h

1 - fu,b ) 0.613LogKow - 0.569 fu,b

(4)

(5)

Prediction of BCF. Prediction of BCF was based on a kinetic model that was proposed by Arnot and Gobas (3, 12). The values of model parameters, if not explicitly referenced, were from Arnot and Gobas’ model. At steady state of the exchange of non-ionic organic chemicals between trout and water,

BCF ) k1φ/(k2 + kE + kG + kMET)

(6)

where k1 (L/kg/d), k2 (d-1), kE (kg/kg/d), kG (d-1), kMET (d-1) are the gill uptake and elimination, fecal egestion, growth dilution, and biotransformation rate constants, respectively; φ is the fraction of a chemical that is bioavailable to the fish in water.

φ ) 1/(1 + χPOCDPOCRPOCKow + χDOCDDOCRDOCKow)

where EW is the gill chemical uptake efficiency and was determined as (1.85 + (155/KOW))-1, and GV is trout ventilation rate (254.4 L/d/kg, (10)). k2 was determined as k1/KTW. KTW is trout-water partition coefficient,

KTW ) νLTKOW + νNTβKOW + νWT

(9)

where νLT (0.1, (10)), νNT (0.2), and νWT (0.7) are lipid, nonlipid organic, and water contents in trout, respectively; β is a proportionality constant reflecting the sorption capacity of non-lipid organic material relative to that of octanol (0.035). The rate constant of fecal egestion, kE, was estimated to be one-eighth of dietary uptake rate constant, kD [kg/kg/d, (12)],

kE ) 0.125‚kD ) 0.125EDGD/WB

and an analogous equation, which was developed based on published plasma protein binding data (Supporting Information, S2), was used to derive the values of fu,b,

Log

(8)

(10)

(3)

where QH is trout hepatic blood flow [536.1 mL/h/kg body weight, 25.9% of trout cardiac output, consists of both arterial and portal blood supplies to the liver, (10)], WL is trout liver weight (12.7 g/kg of body weight, (10)), HT is trout hepatocellularity (510 × 106 cells/g of liver weight, (9, 11)), and fu,b and fu,h are the unbound fractions of the chemical in blood and in hepatocyte incubation, respectively. fu,h was estimated based on an empirical equation,

Log

k1 ) EW‚GV

(1)

where t is incubation time in h; C0 and Ct are the substrate concentrations (µM) in the incubation medium at time zero and time t, respectively; and k is the first-order elimination rate constant (h-1). Intrinsic clearance, CLint (mL/h/106 cells), was normalized to the number of cells (Ncell) according to equation

CLint ) kVinc/Ncell ) k/Ccell

the similarity in phase partitioning of POC and DOC, respectively, in relation to that of octanol (RPOC, 0.35; RDOC, 0.08). k1 was calculated according to

(7)

where χPOC and χDOC are the concentrations of particulate organic carbon (POC) and dissolved organic carbon (DOC) in water, respectively (5 × 10-7 kg/L for both χPOC and χDOC); DPOC and DDOC are the disequilibrium factors for POC and DOC partitioning, respectively (1 for both DPOC and DDOC); and RPOC and RDOC are proportionality constants describing

where WB is trout body weight (0.45 kg), ED is the dietary chemical transfer efficiency and was determined as ED ) (3 × 10-7 × KOW + 2)-1; GD (kg/d) is the feeding rate and equals to 0.022 × W0.85 × e(0.06T), where T is water temperature (10 B °C). Growth dilution rate constant, kG, was determined as 0.0005 ‚W-0.2 B Biotransformation rate constant, kMET, was calculated as

kMET ) 0.024CLT/Vss

(11)

where Vss is volume of distribution at steady state (L/kg); CLT is total clearance by body metabolism (mL/h/kg) and is the sum of metabolic clearance of all organs (CLT ) CLH + CLOthers), where CLOthers is the metabolic clearance from organs other than the liver and was assumed to be zero in the current study. Factor 0.024 in the equation is used to convert the unit of mL/h/kg for CLT to L/d/kg. Vss was treated as the sorption capacity of the fish tissues other than blood relative to that of blood (9):

Vss )

νLTKOW + νNTβKOW + νWT νLBKOW + νNBβKOW + νWB

(12)

where νLB (0.014), νNB (0.147), and νWB (0.839) are lipid, nonlipid organic, and water contents in blood (3, 9, 13).

Results Intrinsic Clearance. All the compounds were stable (no loss of the parent) in heat-inactivated hepatocyte suspension for the longest incubation period (4 h, data not shown). The metabolism of the seven compounds followed first-order kinetics (eq 1 and Figure 1) in both trout and rat hepatocytes, except that atrazine was metabolized in trout hepatocytes (2 × 106 cells/mL) at a rate too low to detect. A single substrate depletion experiment for atrazine was run in trout hepatocytes at 10 × 106 cells/mL, and the CLint value was confirmed to be quite low at approximately 0.002 mL/h/106 cells. Figure 1 showed significant difference in the rate of the depletion of the parent compounds between trout (open circles and solid lines) and rat (filled circles and dashed lines) hepatocytes, where the disappearance of the parent compounds was much faster in rat hepatocytes at 1 × 106 cells/mL than in trout hepatocytes at 2 × 106 cells/mL. Table 2 showed the intrinsic clearance (CLint) values in trout and rat hepatocytes that were normalized to hepatocyte VOL. 41, NO. 9, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3271

Free fractions of 43 uncharged organic compounds in plasma from different species (human, trout, cow, sheep, and rabbit) and in the solution of bovine serum albumin (BSA) were obtained from the literature (Supporting Information, S2). The relationship between Log((1 - fu,b)/fu,b) and LogKOW for these compounds was plotted in Figure 3, assuming species difference was negligible, a blood/plasma concentration ratio of 1, and 100% representative of binding in plasma by BSA. The equation of the regression line in Figure 3 is given by eq 5. Trout Hepatic Clearance and Biotransformation Rate Constant. Trout hepatic clearance was predicted according to eq 3. fu,b and fu,h were derived from eqs 4 and 5, respectively. Table 3 showed that trout CLH values ranged from 8.85 to 155.7 mL/h/kg. The kMET values, which were calculated according to eqs 11 and 12, ranged from 0.038 to 0.672 d-1 for the reference compounds (Table 3). Prediction of BCF. The BCF values predicted according to Arnot and Gobas’ model (eq 6) with and without the kMET term are shown in Table 3. The impact of kMET to BCF prediction was significant. The largest difference was for BaP [BCF(with kMET)/BCF(kMET ) 0) ) 17.9]. For relatively low LogKOW compounds molinate and MK, BCF prediction was significantly improved by incorporating estimated kMET values (Table 3). The modeled BCF values, if calculated with the inclusion of kMET, were slightly above the range of empirical BCF values for 4NP and DTBP in fish, and were well within the range of empirical BCF values for TF and BaP (Table 3).

Discussion

FIGURE 1. Semilogarithmic plot of the substrate concentration in freshly isolated hepatocytes from rainbow trout (O and s) and rat (b and ---). Compounds (2 µM for atrazine, molinate, MK, and BaP; 5 µM for 4NP and DTPB; 15 µM for TF) were incubated with hepatocytes (rat, 37 °C and pH 7.4 at 1 × 106 cell/mL; trout, 10 °C and pH 7.8 at 2 × 106 cell/mL) for up to 2 h (rat) or 4 h (trout). The concentration of the substrate at each time point was determined by LC/MS/MS and LC/Fluorescence methods. concentrations. CLint ratios between rat and trout showed that, except for atrazine, rat hepatocytes cleared the reference compounds 5.5-78.5 times faster than trout hepatocytes. Unbound Fractions in Hepatocyte Incubation and in Blood. The unbound fractions of the compounds in trout hepatocyte incubations (fu,h) were determined to be 0.51, 0.74, 0.54, 0.015, 0.16, 0.024, and 0.02 for atrazine, molinate, MK, 4NP, DTBP, TF, and BaP, respectively. The relationship between Log((1-fu,h)/fu,h) and LogKOW for these compounds and seven neutral compounds published in ref 14 was plotted in Figure 2 (data were summarized in Supporting Information, S1.). The equation of the regression line in Figure 2 is given by eq 4. 3272

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 9, 2007

In Arnot and Gobas’ model to predict bioaccumulation potential, biotransformation rate constant (kMET) was used to represent fish biotransformation capabilities (3, 12). By definition, kMET is equal to the ratio between metabolic clearance and volume of distribution (Vss, eq 11). In our study, metabolic clearance in trout liver (CLH) was obtained via a “well-stirred” liver model (eq 3) using in vitro-determined CLint values and trout-specific model parameters, and Vss was estimated from a partition-based model (9). While the “well-stirred” liver model has been quite successful in the prediction of mammalian CLH values for large number of pharmaceutical compounds (15), its application to fish species has not been reported (9). In the following, we will discuss our rationales for the conditions of the in vitro method and for the model parameters, such as HT, fu,h, fu,b, and Vss. We will evaluate our results by comparing to existing microsomal CLint values in the literature (Table 2), and to the values obtained in vivo (CLH, kMET, and BCF). We then conclude that the “well-stirred” liver model can be used to predict fish CLH values from in vitro CLint values, and significant improvement in bioaccumulation assessment can be achieved by incorporating the measured fish biotransformation capabilities. Intrinsic Clearance. In contrast to traditional enzyme kinetic measurement, where Vmax and Km are determined for specific metabolites (CLint ) Vmax/Km), substrate depletion approach allows estimation of CLint of all metabolic pathways for a single compound. Special attention needs to be paid to the design of the substrate depletion experiment, such as the substrate and enzyme concentrations (8). In our experiments, we selected the lowest substrate concentrations (2 µM for atrazine, molinate, MK, and BaP; 5 µM for 4NP and DTPB; 15 µM for TF) that the analytical methods allowed to detect reliably to reduce the possibility of saturating the enzymes. The reported Km values for atrazine (16), molinate (17), 4NP (18), and BaP (19) were 44.5-50.4, 138-400, 250, and 14.6 µM in rat, respectively, which are all significantly larger than the substrate concentrations we used. Therefore, enzyme saturation was very unlikely to occur assuming Km values for these compounds in trout were not hugely different

TABLE 2. Comparison of Intrinsic Clearance, CLint, in Freshly Isolated Hepatocytes at 2 × 106 Cells/mL for Rainbow Trout and 1 × 106 Cells/mL for Rat compounds

atrazine molinate MK 4NP DTPB TF BaP

CLint (mL/h/106 cells)

CLint ratio

trout a

rat a

n.d.b 0.178 ( 0.036 0.650 ( 0.302 0.127 ( 0.019 0.125 ( 0.062 0.027 ( 0.016 0.044 ( 0.013

3.81 ( 1.96 4.89 ( 1.03 16.8 ( 6.11 0.70 ( 0.11 1.97 ( 0.51 2.15 ( 0.56 0.85 ( 0.12

rat/trout 27.5 25.8 5.5 15.7 78.5 19.4

literature values, hepatocyte clearance equivalent (mL/h/106 cells) trout

rat 0.738c 0.077c 2c

0.025d 0.69, 0.86, 1.67, 6.18c

a Values were averaged from at least three individual hepatocyte isolations and five animals. b Not detected. The loss of the parent compound was not significant during the incubation period at 2 × 106 cells/mL. A single substrate depletion experiment for atrazine was run in trout hepatocytes at 10 × 106 cells/mL, and the CLint value was confirmed to be quite low at approximately 0.002 mL/h/106 cells. c Sum of rat liver microsomal clearances (calculated based on Vmax/Km) via N-deethylation, N-depropylation, and isopropylhydroxylation pathways for atrazine (16), sulfoxidation and ring hydroxylation pathways for molinate (17), and clearance via glucuronidation pathway for 4NP (18). BaP clearance values were estimated in rat liver microsomes [1.67, (35); 0.86, (36); 6.18, (20)] or in rat liver homogenates (0.69, (19)]) Unit conversion was conducted assuming 1 g of liver tissues contains 45 mg microsomal proteins and 1 mg of microsomal proteins is equivalent to 3 million hepatocytes (37). d Clearance in rainbow trout liver S10 fraction via N-depropylation pathway (2), assuming this pathway accounts for 60% of the total liver metabolism (2) and 5.1 × 108 hepatocytes per g of trout liver (9, 11).

FIGURE 2. Plot of Log((1 - fu,h)/fu,h) vs LogKOW. Data of circles were from Austin et al. (14). Squares represent the data obtained in this work for atrazine, molinate, MK, 4NP, TF, and BaP. The complete data set can be found in Supporting Information, S1.

FIGURE 3. Plot of Log((1 - fu,b)/fu,b) vs LogKOW. The complete data set and references can be found in Supporting Information, S2. from those in rat. Jones and Houston (8) have found that lower enzyme concentrations (which translates to lower cell concentrations) gave better predictions of the in vivo clearance values, due to less nonspecific binding and lower inhibitory effect of the metabolites. Our rat and trout hepatocyte concentrations were at 1 and 2 × 106 cells/mL, respectively. At these concentrations, the substrate depletion profile (Figure 1) followed first-order kinetics and the

clearance data were consistent with the literature values (see discussion below). Higher hepatocyte concentrations produced lower CLint values for some of the compounds (data not shown), which supported the observations by Jones and Houston (8). The hepatocytes were incubated at 10 °C and pH 7.8 for trout. We believe this condition would best represent the conditions for trout in vivo (6). The incubation duration for rat hepatocytes were less than 2 h, following the advice from Jones and Houston (8). Trout hepatocytes were incubated at a much lower temperature (10 °C) and the incubation time was extended to 4 h to allow reliable quantification of a much slower clearance in trout. We have shown that rates of biotransformation in rainbow trout (Vmax), as evaluated using a variety of marker substrates for cytochrome P450 enzymes, were 3.4-18.4 times lower than those in rat (6). In this study, trout hepatocytes cleared a number of industrial and agricultural compounds at much slower rates than rat hepatocytes. Their ratios ranged from 5.5- to 78.5-fold (Table 2). Our findings are in good agreement with some of the earlier studies comparing rat and trout biotransformation capabilities, where 8.3-315-fold differences in microsomal clearance of a number of compounds have been reported for rat and trout with quicker clearance in rats (Supporting Information, S3). We compared the intrinsic clearance values obtained in this study to values in the literature (Table 2). Previously reported values were obtained in either liver microsomes or liver fractions using the metabolite formation approach that only accounted for the clearance of limited metabolic pathway(s). For BaP and TF, our values agreed with the literature values well (except for the value for BaP from (20)). For 4NP and atrazine, the differences were 2.6- and 5.2-fold, which are acceptable considering the significant differences in the experimental design. For molinate, our CLint value in rat was 63.5 times larger than what has been reported for the sulfoxidation and ring hydroxylation pathways of molinate (17). We postulate that this large difference was a result of other major metabolic pathways not being considered in the previous study. Hepatic Clearance. Prediction of CLH, based on eq 3, requires an accurate estimate of HT for the in vitro-to-in vivo extrapolation. The reported HT values for trout in the literature are not in good agreement, ranging from 53 to 860 × 106 cells/g liver (9, 21). Most of the values were estimated from the yield of hepatocyte isolation and ranged from 53 to 328 × 106 cells/g (9, 21). One notable exception was a study from VOL. 41, NO. 9, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3273

TABLE 3. Trout Hepatic Clearance (CLH), Biotransformation Rate Constant (kMET), Empirical BCF Values, and BCF Values Predicted with and without Estimated kMET compound

LogKOW

CLH a (mL/h/kg)

kMET (d-1)

predicted BCF (with kMET)

predicted BCF (kMET ) 0)

empirical BCF, b (range, median)

molinate MK 4NP DTBP TF BaP parathioni

2.9 3.87 4.48 5.19 5.34 6.13 3.83

87.1 155.7 37.9 37.5 8.85 15.3 168.1

0.391 0.672 0.163 0.161 0.038 0.066 0.726

67 160 652 769 2782 1528 148

85 777 2973 11427 14281 27414 710

25c 17-54, 32d 75-550, 245.9e 128-660, 360f 889-7093, 2917g 24.8-3208, 608h 63-500, 181

a Obtained based on the mean values of CL b Empirical fish BCF values are obtained from the U.S. Environmental int in Table 2 according to eq 3. Protection Agency ECOTOX database (http://cfpub.epa.gov/ecotox) or Japanese Chemical Risk Information Platform (CHRIP) biodegradation and bioconcentration database (http://www.safe.nite.go.jp/english/db.html). References of the BCF values can be found within the databases. c Value from bass. d Values from carp. e Values from carp, fathead minnow, and salmon f Values from orfe and carp g Values from sheephead minnow, fathead minnow, salmon, and trout. Trout BCF value is 2280 (28). h Values from bluegill and salmon. i CLH was estimated from clearance in trout liver S9 via desulfuration pathway (29) using a scaling factor of 37.8 mg microsomal protein/g liver (mean value in ref 9). Unbound fraction in S9 was predicted with an empirical equation developed from rat microsomes (23). LogKOW was estimated using U.S. EPA EPI Suite KOWWIN software (version 1.67). Empirical fish BCF values were from bluegill, fathead minnow, carp, rainbow trout, brown trout, and brook trout.

Hampton et al. (11), in which trout HT value was estimated from fixed tissues using a stereologic method with reported HT of 510 and 860 × 106 cells/g for male and female trout, respectively (9, 11). We consider the values from Hampton et al.’s work to be more reliable, because based on our experience, the yield from hepatocyte isolation can vary greatly (6-fold difference in the literature values) and only represents a percentage of true hepatocytes number in the liver. Also, compared to the rat, which has a well accepted HR (rat hepatocellularity) value in the literature (125 × 106 cells/g liver, (22)), the protein content per unit cell concentration in male trout hepatocytes was on average one-third to rat hepatocytes (data not shown). The average cell diameter for male trout hepatocytes was approximately half of rat hepatocytes (measured by flow cytometry, data not shown), which would result in a cell volume approximately 8 times smaller for male trout hepatocytes than rat hepatocytes, assuming spherical shape for the cells. If rat and trout have equal amounts of protein or cell volume per unit of liver mass, the male trout HT value would be 3-8 times of rat HR value. This rat-trout comparison indirectly supports the estimated trout HT values that were based on Hampton et al.’s work. The ratio of the unbound fraction of a compound in blood to that in hepatocyte incubation (fu,b/fu,h) has a significant impact on the estimation of CLH (eq 3, (14, 23)). Accurate determination of fu,h and fu,b for environmentally relevant compounds, which are normally highly lipophilic, is technically challenging and very often unachievable (20). Therefore, empirical equations have been investigated to predict their values (9, 14, 23, 24). For a nonspecific binding event, the lipophilicity and the charge state of a compound are the two most significant properties to determine the degrees of binding. We purposely chose uncharged compounds in our study to simplify our conditions. Austin et al. (14) investigated the relationship between nonspecific binding to rat hepatocytes and the lipophilicity of the compounds, and developed an empirical equation between Log((1 - fu,h)/fu,h) and LogKOW for the prediction of fu,h. However, the range of the LogKOW values for their selected neutral compounds was relatively small (1.34-3.75), and it was uncertain if this empirical relationship could be used to predict nonspecific bindings to trout hepatocytes. In our study, we experimentally determined fu,h values for our compounds using “dead” (24-h old at room temperature) trout hepatocytes and generated eq 4 by combining the data of seven neutral compounds from ref 14. Austin et al. (14) has shown that the fu,h values obtained from 24-h old “dead” rat hepatocytes were comparable to the ones obtained from “live” but metabolically inhibited hepatocytes. Figure 2 showed that the fu,h values 3274

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 9, 2007

from trout hepatocytes provide a reasonably good extension to the rat data based on the larger LogKOW values of our reference compounds and a linear correlation clearly existed between Log((1 - fu,h)/fu,h) and LogKOW. Data for fu,b for 43 uncharged industrial and agricultural compounds were obtained in the literature (Supporting Information, S2). By transforming to Log((1 - fu,b)/fu,b), we obtained a straight line relationship to LogKOW (Figure 3). The reason for this transformation (analogous to the reason for the transformation of fu,h in Figure 2) is because the form of (1 - fu,b)/fu,b is equivalent to partition coefficient determination (23) and it ensures small differences at the high end of plasma protein binding affinities (fu,b f 0) being treated with equal importance compared to larger differences in the medium range (24). Figure 3 showed a good linear correlation between Log((1 - fu,b)/fu,b) and LogKOW. Since the ratio, not the individual values of fu,b and fu,h, will influence the outcome of CLH prediction (eq 3), we used predicted fu,h and fu,b values (eqs 4 and 5) in our study (not the measured fu,h). This allows the same methodology for the estimation of fu,h and fu,b and an opportunity for cancellation of the errors that are inherent in the models. It is important to be aware that assumption of nonspecific binding formed the basis for our prediction of fu,h, fu,b, and Vss (see below) using eqs 4, 5, and 12, respectively. For compounds that bind specifically to plasma proteins or to certain tissue proteins, our method for these predictions will certainly fail. In that case, experimentally determination of fu,h and fu,b will become essential. Seubert and Kennedy (25) have reported that BaP was cleared at a rate of 40 mL/h/kg in rainbow trout (all clearance pathways combined). In our study, we predicted a BaP CLH value in trout at 15.3 mL/h/kg (Table 3). Although direct comparison is not appropriate, the in vivo data do suggest that our result is not unreasonable and support our use on model parameters in eq 3. We also estimated CLH values (mL/h) for rat using QH 4848 mL/h/kg, WL 40.4 g/kg, HR 125 × 106 cells/g, and WB 0.25 kg (22). The CLH ratios between rat and trout are 9.7, 8.2, 2.5, 6.7, 31.0, and 8.2 for molinate, MK, 4NP, DTBP, TF, and BaP, respectively. For most of the compounds, the differences were within 10-fold, which, compared to the CLint ratios shown in Table 2, might provide a better representation of the differences in trout and rat biotransformation capabilities. Prediction of kMET and BCF. Biotransformation rate constant (kMET) was obtained by dividing CLH by volume of distribution at steady state (Vss, eq 11). Vss is determined by the partition of compounds between plasma and tissues. Nichols et al. (9) provided a partition-based model for the prediction of Vss (eq 12). Based on this model, the Vss values for 4NP and TF are 5.58 and 5.59 L/kg, respectively (assuming

10% total lipid content in trout). Schult and Hayton (26) have found that the Vss values of TF in trout correlated to the body lipid content, and for the hatchery trout with an average of 8% lipid content, the Vss value was 5.73 ( 1.7 L/kg. Coldham et al. (27) reported toxicokinetic data for 4NP in rainbow trout. The volume of distribution for 4NP was estimated from the plasma concentration-time profile to be 5.28 L/kg. These in vivo data suggest that the model proposed by Nichols et al. (9) adequately predict Vss values for 4NP and TF. The elimination rate constant in trout liver for 4NP was reported to be 0.168 d-1 in a toxicokinetics study of 4NP in rainbow trout (27). In our study, we estimated 4NP kMET value in trout liver to be 0.163 d-1 (Table 3). It is worth noting that the in vivo data were obtained from total 4NP residues (parent compounds and radioactive metabolites), whereas our kMET value was estimated from the clearance of the parent compound alone. The trout elimination rate constant for TF (including all elimination pathways) was found to be highly variable and ranged between 0.057 and 0.308 d-1 (28). We estimated TF kMET value to be 0.038 d-1 (Table 3). The comparison between our predicted kMET-based BCF values and the empirical fish BCF values is encouraging and demonstrates that by incorporating kMET values, significant improvements in BCF prediction are achieved (Table 3). We also predicted the BCF value of parathion using trout liver S9 clearance value found in the literature (29) and the models described in this paper. Our prediction was in excellent agreement with the empirical BCF values of parathion in fish (Table 3). We would like to point out that metabolism capabilities have been shown to vary significantly in different fish species (2). Therefore, the comparison shown in Table 3 needs to be interpreted with caution due to the fact that the empirical BCF values were from different fish species. Contribution of metabolism from other organs was neglected (CLOther ) 0) in this study by assuming liver was the dominant organ for fish biotransformation. This will result in a conservative BCF estimation because if CLOther > 0, which is certainly the case, the predicted BCF values will be smaller than what we’ve shown in Table 3. In summary, by using in vitro hepatocyte clearance experiment, the “well-stirred” liver model, and Arnot and Gobas’ bioaccumulation assessment model, we adequately predicted trout biotransformation capabilities in vivo (CLH and kMET) and the bioaccumulation potentials (BCF) of several reference compounds in fish. We provided a framework for future bioaccumulation assessment of xenobiotics using combined information of the physical-chemical properties of the compounds and the biotransformation potentials of the compounds in fish. Future work such as further refinement of the experimental conditions, better understanding on the uncertainties of the model parameters, practice with other fish species and reference compounds with more diverse physical-chemical properties (e.g., charged vs uncharged) etc., are needed to better understand the utility, and more importantly the limitation of this in vitro method for the prediction of bioaccumulation potentials in fish.

Acknowledgments We thank Alan Samel, Barbra Ferrell, Laurie Bouchelle, and Jeff Turner for their general assistance on fish maintenance during the course of the study. In addition, we are appreciative to Drs Gary Jepson (DuPont), Robert Hoke (DuPont), Watze de Wolf (DuPont), Scott Dyer (Procter & Gamble Co.), Christina Cowen (Procter & Gamble Co.), John Nichols (U.S. Environmental Protection Agency), Jon Arnot (Trent University), and Margaret James (University of Florida) for commenting on this manuscript, and to Drs. Sue Erhardt (The Dow Chemical Company) and Annie Weisbrod (Procter & Gamble Co.) for general discussion on fish biotransformation and bioaccumulation.

Supporting Information Available A complete data set with references. This material is available free of charge via the Internet at http://pubs.acs.org.

Literature Cited (1) Stuchal, L. D.; Kleinow, K. M.; Stegeman, J. J.; James, M. O. Demethylation of the pesticide methoxychlor in liver and intestine from untreated, methoxychlor-treated, and 3-methylcholanthrene-treated channel catfish (Ictalurus punctatus): evidence for roles of CYP1 and CYP3A family isozymes. Drug Metab. Dispos. 2006, 34, 932-8. (2) Schultz, I. R.; Hayton, W. L. Interspecies scaling of the bioaccumulation of lipophilic xenobiotics in fish: an example using trifluralin. Environ. Toxicol. Chem. 1999, 18, 1440-1449. (3) Arnot, J. A.; Gobas, F. A. A food web bioaccumulation model for organic chemicals in aquatic ecosystems. Environ. Toxicol. Chem. 2004, 23, 2343-55. (4) Dimitrov, S.; Dimitrova, N.; Parkerton, T.; Comber, M.; Bonnell, M.; Mekenyan, O. Base-line model for identifying the bioaccumulation potential of chemicals. SAR QSAR Environ. Res. 2005, 16, 531-54. (5) Kramer, H. J.; van den Berg, M.; Delang, R. J.; Brandsma, L.; Dejongh, J. Biotransformation rates of Ugilec 141 (tetrachlorobenzyltoluenes) in rat and trout microsomes. Chemosphere 2000, 40, 1283-8. (6) Nabb, D. L.; Mingoia, R. T.; Yang, C. H.; Han, X. Comparison of basal level metabolic enzyme activities of freshly isolated hepatocytes from rainbow trout (Oncorhynchus mykiss) and rat. Aquat. Toxicol. 2006, 80, 52-9. (7) Ronis, M. J.; Walker, C. H. Species variations in the metabolism of liposoluble organochlorine compounds by hepatic microsomal monooxygenase: comparative kinetics in four vertebrate species. Comp. Biochem. Physiol. C 1985, 82, 445-9. (8) Jones, H. M.; Houston, J. B. Substrate depletion approach for determining in vitro metabolic clearance: Time dependencies in hepatocyte and microsomal incubations. Drug Metab. Dispos. 2004, 32, 973-82. (9) Nichols, J. W.; Schultz, I. R.; Fitzsimmons, P. N. In vitro-in vivo extrapolation of quantitative hepatic biotransformation data for fish I. A review of methods, and strategies for incorporating intrinsic clearance estimates into chemical kinetic models. Aquat. Toxicol. 2006, 78, 74-90. (10) Nichols, J. W.; McKim, J. M.; Andersen, M. E.; Gargas, M. L.; Clewell, H. J., 3rd; Erickson, R. J. A physiologically based toxicokinetic model for the uptake and disposition of waterborne organic chemicals in fish. Toxicol. Appl. Pharmacol. 1990, 106, 433-447. (11) Hampton, J. A.; Lantz, R. C.; Hinton, D. E. Functional units in rainbow trout (Salmo gairdneri, Richardson) liver: III. Morphometric analysis of parenchyma, stroma, and component cell types. Am. J. Anat. 1989, 185, 58-73. (12) Arnot, J. A.; Gobas, F. A. A generic QSAR for assessing the bioaccumulation potential of organic chemicals in aquatic food webs. QSAR Comb. Sci. 2003, 22, 337-345. (13) Bertelsen, S. L.; Hoffman, A. D.; Gallinat, C. A.; Elonen, C. M.; Nichols, J. W. Evaluation of LogKow and tissue lipid content as predictors of chemical partitioning to fish tissues. Environ. Toxicol. Chem. 1998, 17, 1447-1455. (14) Austin, R. P.; Barton, P.; Mohmed, S.; Riley, R. J. The binding of drugs to hepatocytes and its relationship to physicochemical properties. Drug Metab. Dispos. 2005, 33, 419-25. (15) Chiba, M.; Shibata, Y.; Takahashi, H.; Ishii, Y.; Sugiyama, Y. Prediction of hepatic clearance in humans from experimental animals and in vitro data. In Drug Metabolizing Enzymes, Cytochrome P450 and Other Enzymes in Drug Discovery and Development; Lee, J. S., Obach, R. S., Fisher, M. B., Eds.; Marcel Dekker, Inc.: The Netherlands, 2003; pp 453-481. (16) Hanioka, N.; Jinno, H.; Tanaka-Kagawa, T.; Nishimura, T.; Ando, M. In vitro metabolism of simazine, atrazine and propazine by hepatic cytochrome P450 enzymes of rat, mouse and guinea pig, and oestrogenic activity of chlorotriazines and their main metabolites. Xenobiotica 1999, 29, 1213-26. (17) Jewell, W. T.; Miller, M. G. Comparison of human and rat metabolism of molinate in liver microsomes and slices. Drug Metab. Dispos. 1999, 27, 842-7. (18) Daidoji, T.; Inoue, H.; Kato, S.; Yokota, H. Glucuronidation and excretion of nonylphenol in perfused rat liver. Drug Metab. Dispos. 2003, 31, 993-8. (19) Zampaglione, N. G.; Mannering, G. J. Properties of benzpyrene hydroxylase in the liver, intestinal mucosa and adrenal of VOL. 41, NO. 9, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY

9

3275

(20) (21) (22)

(23)

(24)

(25) (26) (27)

(28) (29)

untreated and 3-methylcholanthrene-treated rats. J. Pharmacol. Exp. Ther. 1973, 185, 676-85. Wiersma, D. A.; Roth, R. A. The prediction of benzo[a]pyrene clearance by rat liver and lung from enzyme kinetic data. Mol. Pharmacol. 1983, 24, 300-8. Law, F. C.; Abedini, S.; Kennedy, C. J. A biologically based toxicokinetic model for pyrene in rainbow trout. Toxicol. Appl. Pharmacol. 1991, 110, 390-402. Komura, H.; Iwaki, M. Nonlinear pharmacokinetics of propafenone in rats and humans: application of a substrate depletion assay using hepatocytes for assessment of nonlinearity. Drug Metab. Dispos. 2005, 33, 726-32. Austin, R. P.; Barton, P.; Cockroft, S. L.; Wenlock, M. C.; Riley, R. J. The influence of nonspecific microsomal binding on apparent intrinsic clearance, and its prediction from physicochemical properties. Drug Metab. Dispos. 2002, 30, 1497-503. Lobell, M.; Sivarajah, V. In silico prediction of aqueous solubility, human plasma protein binding and volume of distribution of compounds from calculated pKa and AlogP98 values. Mol. Diversity 2003, 7, 69-87. Seubert, J. M.; Kennedy, C. J. Benzo[a]pyrene toxicokinetics in rainbow trout (Oncorhynchus mykiss) acclimated to different salinities. Arch. Environ. Contam. Toxicol. 2000, 38, 342-9. Schultz, I. R.; Hayton, W. L. Influence of body fat on trifluralin toxicokinetics in rainbow trout (Oncorhynchus mykiss). Environ. Toxicol. Chem. 1997, 16, 997-1001. Coldham, N. G.; Sivapathasundaram, S.; Dave, M.; Ashfield, L. A.; Pottinger, T. G.; Goodall, C.; Sauer, M. J. Biotransformation, tissue distribution, and persistence of 4-nonylphenol residues in juvenile rainbow trout (Oncorhynchus mykiss). Drug Metab. Dispos. 1998, 26, 347-54. Schultz, I. R.; Hayton, W. L. Toxicokinetics of trifluralin in rainbow trout. Aquat. Toxicol. 1993, 26, 287-306. Wallace, K. B.; Dargan, J. E. Intrinsic metabolic clearance of parathion and paraoxon by livers from fish and rodents. Toxicol.

3276

9

ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 9, 2007

Appl. Pharmacol. 1987, 90, 235-42. (30) Struck, R. F.; Shih, T. W.; Johnston, T. P.; Kirk, M. C.; Hill, D. L. Metabolism and macromolecular binding of the carcinogen Michler’s ketone in rats. Xenobiotica 1981, 11, 569-7. (31) Doerge, D. R.; Twaddle, N. C.; Churchwell, M. I.; Chang, H. C.; Newbold, R. R.; Delclos, K. B. Mass spectrometric determination of p-nonylphenol metabolism and disposition following oral administration to Sprague-Dawley rats. Reprod. Toxicol. 2002, 16, 45-56. (32) Erkog, F. U.; Menzer, R. E. Metabolism of trifluralin in rats. J. Agric. Food Chem. 1985, 33, 1061-1070. (33) Conney, A. H.; Chang, R. L.; Jerina, D. M.; Wei, S. J. Studies on the metabolism of benzo[a]pyrene and dose-dependent differences in the mutagenic profile of its ultimate carcinogenic metabolite. Drug Metab. Rev. 1994, 26, 125-63. (34) Moridani, M. Y.; Siraki, A.; O’Brien, P. J. Quantitative structure toxicity relationships for phenols in isolated rat hepatocytes. Chem. Biol. Interact. 2003, 145, 213-23. (35) Robie, K. M.; Cha, Y. N.; Talcott, R. E.; Schenkman, J. B. Kinetic studies of benzpyrene and hydroxybenzpyrene metabolism. Chem.-Biol. Interact. 1976, 12, 285-97. (36) DePierre, J. W.; Moron, M. S.; Johannesen, K. A.; Ernster, L. A reliable, sensitive, and convenient radioactive assay for benzpyrene monooxygenase. Anal. Biochem. 1975, 63, 470-84. (37) Houston, J. B. Utility of in vitro drug metabolism data in predicting in vivo metabolic clearance. Biochem. Pharmacol. 1994, 47, 1469-79.

Received for review November 2, 2006. Revised manuscript received February 14, 2007. Accepted February 22, 2007. ES0626279