Protein and Lipid Binding Parameters in Rainbow Trout

ARTICLE pubs.acs.org/crt

Protein and Lipid Binding Parameters in Rainbow Trout (Oncorhynchus mykiss) Blood and Liver Fractions to Extrapolate from an in Vitro Metabolic Degradation Assay to in Vivo Bioaccumulation Potential of Hydrophobic Organic Chemicals Beate I. Escher,*,†,‡ Christina E. Cowan-Ellsberry,§ Scott Dyer,|| Michelle R. Embry,^ Susan Erhardt,# Marlies Halder,3 Jung-Hwan KwonO,‡ Karla Johanning,( Mattheus T. T. Oosterwijk,† Sibylle Rutishauser,†,‡ Helmut Segner,z and John Nicholsþ †

)

The University of Queensland, National Research Centre for Environmental Toxicology (Entox), 39 Kessels Road, Brisbane, Qld 4108, Australia ‡ Eawag, Swiss Federal Institute of Aquatic Science and Technology, 8600 D€ubendorf, Switzerland § CE2 Consulting, Cincinnati, Ohio, United States The Procter & Gamble Company, Miami Valley Innovation Center, 11810 East Miami River Road, Cincinnati, Ohio 45252, United States ^ ILSI Health and Environmental Sciences Institute, 1156 15th Street, NW, Washington, DC 20005, United States # Department of Entomology, Michigan State University, East Lansing, Michigan, United States 3 In Vitro Methods Unit/ECVAM, Institute for Health and Consumer Protection, Joint Research Centre, European Commission, Via E. Fermi 2749, 21027 Ispra, Italy O Department of Environmental Engineering, Ajou University, Woncheon-dong, Yeongtong-gu, Suwon 443-749, Republic of Korea ( Life Technologies, 1624 Headway Circle, Austin, Texas 78754, United States z Centre for Fish and Wildlife Health, University of Berne, Hochschulstrasse 4, 3012 Bern, Switzerland þ Mid-Continent Ecology Division, US EPA, 6201 Congdon Boulevard, Duluth, Minnesota, United States

bS Supporting Information ABSTRACT: Binding of hydrophobic chemicals to colloids such as proteins or lipids is difficult to measure using classical microdialysis methods due to low aqueous concentrations, adsorption to dialysis membranes and test vessels, and slow kinetics of equilibration. Here, we employed a three-phase partitioning system where silicone (polydimethylsiloxane, PDMS) serves as a third phase to determine partitioning between water and colloids and acts at the same time as a dosing device for hydrophobic chemicals. The applicability of this method was demonstrated with bovine serum albumin (BSA). Measured binding constants (KBSAw) for chlorpyrifos, methoxychlor, nonylphenol, and pyrene were in good agreement with an established quantitative structureactivity relationship (QSAR). A fifth compound, fluoxypyr-methyl-heptyl ester, was excluded from the analysis because of apparent abiotic degradation. The PDMS depletion method was then used to determine partition coefficients for test chemicals in rainbow trout (Oncorhynchus mykiss) liver S9 fractions (KS9w) and blood plasma (Kbloodw). Measured KS9w and Kbloodw values were consistent with predictions obtained using a mass-balance model that employs the octanolwater partition coefficient (Kow) as a surrogate for lipid partitioning and KBSAw to represent protein binding. For each compound, Kbloodw was substantially greater than KS9w, primarily because blood contains more lipid than liver S9 fractions (1.84% of wet weight vs 0.051%). Measured liver S9 and blood plasma binding parameters were subsequently implemented in an in vitro to in vivo extrapolation model to link the in vitro liver S9 metabolic degradation assay to in vivo metabolism in fish. Apparent volumes of distribution (Vd) calculated from the experimental data were similar to literature estimates. However, the calculated binding ratios (fu) used to relate in vitro metabolic clearance to clearance by the intact liver were 10 to 100 times lower than values used in previous modeling efforts. Bioconcentration factors (BCF) predicted using the experimental binding data were substantially higher than the predicted values obtained in earlier studies and correlated poorly with measured BCF values in fish. One possible explanation for this finding is that chemicals bound to proteins can desorb rapidly and thus contribute to metabolic turnover of the chemicals. This hypothesis remains to be investigated in future studies, ideally with chemicals of higher hydrophobicity.

Received: March 14, 2011 Published: May 23, 2011 r 2011 American Chemical Society

1134

dx.doi.org/10.1021/tx200114y | Chem. Res. Toxicol. 2011, 24, 1134–1143

Chemical Research in Toxicology

’ INTRODUCTION In-vitro metabolism assays using fish liver S9 fractions have the potential to provide rapid and cost-effective estimates of in vivo biotransformation by measuring the intrinsic clearance of the parent chemical.1,2 In order to estimate in vivo metabolism, however, it is necessary to extrapolate in vitro activity to the intact animal. The first step in the extrapolation procedure is to estimate in vivo intrinsic clearance by incorporating scaling factors that relate experimental test conditions from the liver S9 assay to the whole liver.3 As a second step, in vivo intrinsic clearance is combined with information on liver blood flow and an adjustment factor (fu) that corrects for free chemical fractions in blood plasma (fu,b) and the in vitro test system (fu,S9). This adjustment factor, which is calculated as the ratio fu,b/fu,S9, reflects an assumption that only chemicals freely dissolved in solution (liver S9 assay or blood plasma) are available for metabolic transformation.3 Although its application to fish is relatively new, this extrapolation procedure is based on wellestablished methods used by the pharmaceutical industry for preclinical testing of drug candidates.4 To date, direct measurements of the free chemical fraction in fish liver S9 assays have not been attempted. Han et al.5 addressed this deficiency by assuming that chemical binding in the S9 system is similar to that in rat liver microsomes, but the validity of this assumption is unknown. Blood (plasma) binding parameters have been reported for several compounds in fish.6 However, most of this information has been obtained for relatively hydrophilic compounds (log Kow values 99%) was purchased from Sigma-Aldrich (Castle Hill, Australia), methoxychlor (72-43-5, 98.8%) from TCI America (Portland, USA) and 4-n-nonylphenol (104-40-5, >98%) from Alfa Aesar (Heysham, UK). Chemical Analysis. Chlorpyrifos, FMHE, and methoxychlor were quantified using a Hewlett-Packard 5890 Gas Chromatography-Electron Capture Detector (GC-ECD) Series II with an HP-7673A auto sampler (Palo Alto, USA). Chromatographic separation was achieved using a DB-5 column (30 m 0.25 mm i.d.; J&W Scientific, Folsom, USA). Chlorpyrifos-methyl, 2,4-dichlorophenoxyacetic iso-octylester, and 1,1,1-trichloro-2,2-di(4-chlorophenyl) (DDT) were used as internal standards. Pyrene was analyzed using a Shimadzu high performance liquid chromatography (HPLC) system with an LC-20AD pump and SIL-20AHT auto sampler (Rydalmere, Australia). Separation was performed at 40 C using a Supelcosil LC-PAH column (150 mm 4.6 mm, 5 μm; Supelco, Bellefonte, USA). Pyrene was quantified with an RF-10AXL fluorescence detector using excitation and emission wavelengths of 330 and 375 nm, respectively. Nonylphenol was analyzed on the same HPLC system connected to an SPD-M20A diode array detector and was chromatographed using a Nucleodur C18 gravity column (125 mm 4.6 mm, 5 μm; Macherey-Nagel, D€uren, Germany). A wavelength of 278 nm was used for the quantification of nonylphenol. External calibrations were used for both HPLC methods. The mobile phase for pyrene analysis consisted of 5% phosphate buffer (20 mM phosphate buffer pH 3) and 95% methanol, while that used for the analysis of nonylphenol consisted of 20% buffer and 80% methanol. BSA. Bovine serum albumin (>98% purity) was purchased from Sigma-Aldrich (Castle Hill, Australia). To evaluate the effect of heat treatment for the inactivation of S9, partition experiments with BSA were performed both with native and heat-denatured material. The suspension was boiled for 5 min to denature the BSA. Liver S9 Preparation. Trout liver S9 fractions were prepared by Life Technologies (Carlsbad, USA) from livers that were cleared and frozen at the PNNL/Battelle Marine Research Laboratory in Sequim, WA USA. Male rainbow trout (Oncorhynchus mykiss; Shasta/Kamloops hybrids) were obtained from a regional private hatchery (Nisqually Trout Farm, Lacey, WA, USA), and were acclimated and maintained under a simulated natural photoperiod regime (16 h of light to 8 h of darkness) and temperature (11.8 ( 0.8 C). Fish were fasted for at least 24 h prior to sacrifice and anesthetized with 100 mg/L MS-222 (tricaine methanesulphonate). Their average weight was 451 ( 15 g. A longitudinal incision was made from the cloacae to the jaw area, and the fish was cut transversally along the gills to expose the liver. The hepatic portal vein was cannulated with a needle attached to a syringe and flushed with ice-cold, isotonic saline (0.9% NaCl) until the liver was pale in color (i.e., the blood was removed). The liver was immediately excised, patted dry, weighed (9.51 ( 0.66 g), and flash frozen in liquid nitrogen. Livers were stored at 80 C until processing. Whole livers were pooled, thawed out on ice, homogenized in ice-cold 50 mM Tris-HCl, pH 7.8, 150 mM KCl, 2 mM ethylenediaminetetraacetic acid (EDTA), and 1 mM dithiothreitol (DTT), and spun at 13,000g for 20 min. The resulting supernatant (S9 fraction) was aliquoted out and stored at 80 C for further analysis. S9 was heat-inactivated by boiling for 1 min at 100 C to avoid concomitant degradation of chemicals during the partition experiment. Rainbow Trout Blood Plasma. Rainbow trouts used for blood sampling were 2-year-old fish originating from the stock of the Centre for Fish and Wildlife Health, University of Bern, Switzerland. Fish were starved for at least 24 h prior to sampling for blood. Blood was sampled from the caudal vein using a heparinized syringe and 18 gauge needle. Samples were transferred to 2 mL Eppendorf tubes, and the cells were 1135

dx.doi.org/10.1021/tx200114y |Chem. Res. Toxicol. 2011, 24, 1134–1143


ARTICLE

pelleted by centrifugation at 3000g for 5 min at 4 C. The resulting supernatant (plasma) was then aliquoted and stored at 20 C until use. Protein Determination. Protein content was determined using the Lowry method15 adapted to 96 well plates. FolinCiocalteu’s phenol reagent was obtained from Sigma-Aldrich (Castle Hill, Australia). Lipid Determination. Lipids were extracted using the Bligh and Dyer method16 modified by Smedes et al.17 as follows: To 1 mL of suspension (liver S9, fish plasma, control) were added 0.8 mL of 2-propanol, 1 mL of cyclohexane, and 0.1 mL of Milli-Q water. The mixture was vortexed for 30 s and then sonicated for 5 min. After centrifugation at 2000 rpm for 5 min, the upper layer was removed and the extraction process repeated. Cyclohexane extracts were combined, evaporated to dryness in a nitrogen stream, and further dried over silica gel in a vacuum desiccator. Lipid content was determined gravimetrically. A liposome suspension (10 glip L1) served as the positive control. Liposomes were prepared using palmitoyl-oleyl-phosphatidylcholine (POPC, Avanti Polar Lipids, Alabaster, AL, USA) according to ref 18. The results were recovery-corrected by the recovery of POPC, which was 76.4%. BSA was used as the negative control. Only 0.35% of BSA weight was extracted from the negative controls, which was deemed satisfactory.

Determination of PDMSWater Partition Coefficients.

The PDMSwater partition coefficient (KPDMSw ; Lw LPDMS1) is defined as the ratio of chemical concentrations in PDMS (CPDMS) and water (Cw) under equilibrium conditions. KPDMSW ¼

CPDMS CW

ð1Þ

KPDMSw values for nonylphenol, methoxychlor, and chlorpyrifos were determined using a dynamic permeation method.19 PDMS disks (70 μL; prepared with Soxhlet cleanup as described in ref 19) were loaded with individual test chemicals using a methanol/water (60/40) solution. Initial concentrations of test chemicals in PDMS ranged from 0.4 to 4 g L1 (resulting in maximum aqueous concentrations of 20 nM to 0.8 μM). All experiments were performed in glass vials to minimize the sorption of hydrophobic chemicals. The KPDMSw value for pyrene was measured previously by this method.19 Because FMHE hydrolyzes abiotically with a relatively short half-life (3.2 d at pH 920), it was not possible to use the dynamic permeation method for this compound. We therefore developed a modified depletion method similar to that described in ref 19. Here, each preloaded PDMS disk was placed in a glass vial filled with 1.2 mL of buffer and a stainless steel stirring disk (7.95 mm diameter, 0.635 mm thick, V&P Scientific, Inc., San Diego, CA, USA). A tumble stirrer (VP710, V&P Scientific, Inc.) was used to stir the solution at 300 rpm for 4 h. The aqueous solution (1 mL out of the 1.2 mL) was then transferred to an extraction vial containing a clean 70 mg PDMS disk and extracted for 4 h. Although this extraction time may not be sufficient to reach equilibrium between water and PDMS, complete extraction is assumed because of the high PDMS volume. Finally, the PDMS disks were extracted by shaking with 1 mL n-hexane for 15 h.

Determination of Lipid and ProteinWater Partition Coefficients. Blood protein sorption is generally regarded as a specific binding process with a limited number of binding sites on each protein molecule. However, since liver S9 fractions and fish blood plasma contain a large number of different proteins as well as significant amounts of lipid, binding was expressed not in terms of a binding constant but rather as a partition coefficient. Partition coefficients between phase i (surrogate protein, BSA (native and heat-treated), trout liver S9 (inactivated by heat treatment), or trout blood plasma and water, Kiw (Lw kgi1), were determined using a PDMS depletion method.11 PDMS disks (1.69 μL; prepared with Soxhlet cleanup as described in ref 19) were loaded with the test chemicals using a methanol/water (60/40) solution. Initial concentrations of test chemicals in PDMS were the same as those in the KPDMSw experiment described above, and the mass of phase i to volume of water ratios mi/Vw ranged from 0.1 to 12 g L1. Equilibrium between all phases

was achieved by shaking the system for 24 h at 25 C.11 PDMS disks were then removed and extracted by shaking for 2 h with 500 μL of hexane containing the appropriate internal standard for GC analysis or 500 μL methanol for HPLC analysis. The partition coefficient between PDMS and the suspension including water and dispersed organic phase i (KPDMSsus; Lsus LPDMS1) was determined from the chemical concentration in PDMS at the beginning of an experiment (CPDMS,initial) and the concentration after equilibrium had been achieved (CPDMS,equilibrium) for a known ratio of suspension volume (Vsus ≈ Vw) and the volume of PDMS (VPDMS). KPDMSsus ¼

CPDMS, equilibrium CPDMS, equilibrium ni ni þ nw Vw Vw

CPDMS, equilibrium ¼ ¼ Csus

Vw VPDMS CPDMS, initial CPDMS, equilibrium

1

ð2Þ

Csus refers to the concentration of chemical in the suspension. Equation 2 holds if the mass of chemical in phase i associated with organic material (ni) is much larger than that associated with the aqueous phase (nw) so that in the mass balance the fraction in the aqueous phase and the fraction sorbed to the vessel is negligible. It was established in a previous study with PAHs11 that the amount of chemical lost from the PDMS disk is equivalent to the amount in suspension. When preloaded PDMS disks were incubated with buffer alone, as expected, there were no differences between measured concentrations in PDMS and CPDMS,initial. The partition coefficient between phase i and water (Kiw) was derived by combining eqs 1 and 2, where mi is the mass of phase i in the system to yield eq 3. Ci refers to the concentration of chemical in phase i. Note that i can also refer to a composite phase, e.g., lipids plus proteins or total dry material. Ci KPDMSw Vw ¼ 1 ð3Þ Kiw ¼ Cw KPDMSsus mi A minimum of three independent measurements with each of the 12 vials with phase i in variable ratios of mi/Vw and 6 control vials (containing only preloaded PDMS and buffer) was performed. From the chemical concentrations in PDMS of each vial and the average of chemical concentration in PDMS of the controls, a partition coefficient was calculated with eqs 2 and 3, and the mean and 95% confidence intervals of these values are reported. Mass-Balance Modeling. As will be demonstrated in Results and Discussion, partitioning to lipids plays a significant role in chemical binding to liver S9 fractions and plasma. For this reason, partition coefficients were not normalized to protein content, but rather to the sum of lipid and protein content (mi = mp þ mlip). Additional nonaqueous components of S9 and plasma such as carbohydrates and salts are thought to have insignificant affinity for hydrophobic compounds and were therefore neglected. We have previously shown that liposomebuffer and protein buffer partitioning does not differ from liposomewater and protein water partitioning for neutral compounds of high hydrophobicity.11 For simplicity, therefore, we speak of phasewater partition coefficients even though experiments were conducted in different buffers or with different background electrolytes. From mass-balance considerations, the S9water partition coefficient KS9w and the blood plasmawater partition coefficient Kbloodw (both in units of L (kglipþ kgp)1) also may be predicted from chemical partitioning to protein and lipid components of either phase as follows: Kiw ¼

flip Clip þ fp Cp ¼ flip Klipw þ fp Kpw Cw

ð4Þ

where Cw (mol L1) is the chemical concentration in the aqueous phase, Clip (mol kglip1) is the concentration of chemical sorbed to lipids, and 1136



ARTICLE

Cp (mol kgp1) is the concentration of chemical sorbed to proteins. Klipw refers to the lipidwater partition coefficient and Kpw to the proteinwater partition coefficient. The fraction of lipid (flip) and protein (fp) are given by eqs 5 and 6. mlip ð5Þ flip ¼ mlip þ mp fp ¼

mp mlip þ mp

ð6Þ

In Vitro to in Vivo Extrapolation of Hepatic Metabolism. Procedures used to extrapolate in vitro intrinsic clearance to an in vivo whole-body metabolism rate constant (kMET; h1) were described by Nichols et al.3 These methods were subsequently used by Cowan-Ellsberry et al.13 and Han et al.5 to extrapolate trout liver S9 clearance data to the whole animal as a means of predicting metabolism impacts on chemical bioaccumulation. In an S9 metabolic assay, log-transformed substrate depletion data are used to calculate the value of a first-order rate constant, km (or the metabolic degradation half-life t1/2 = ln(2) km1), which is related to the intrinsic in vitro clearance (CLS9; L h1 gP1) by the protein concentration in the assay. While only a fraction of total protein present in the system contributes to this metabolism, it is not possible to differentiate active from nonactive proteins. The CLS9 is therefore normalized to total protein concentration (mP/Vw), which has units of gP L1. km CLS9 ¼ mp Vw

ð8Þ

In the second step of the extrapolation procedure, CLin vivo,int is used as an input to a well-stirred liver model21,22 that accounts for potential rate limitations imposed by the rate of hepatic blood flow.13 CLh ¼

LF 3 Q C 3 fu 3 CLinvivo, int LF 3 Q C þ fu 3 CLinvivo, int

ð9Þ

Hepatic blood flow is calculated as the product of cardiac output (QC; L h1 kg1) and the fraction of total blood flow that perfuses the liver (LF; unitless). The fraction of QC perfusing the liver is generally thought to represent the contribution of both arterial and portal blood flows, although other variations of the model are possible.13 The remaining term in this model (fu) accounts for chemical binding effects on metabolism and is calculated as the ratio of the free fraction in blood (fu,blood) to the free fraction in the liver S9 assay (fu,S9).3 fu ¼

fu, blood fu, S9

ð10Þ

The parameter relevant for the in vitro to in vivo extrapolations is thus the fraction unbound in the suspension that contains phase i, fu,i. fu, i ¼

nw nw 1 ¼ ¼ mi ntot nw þ ni 1 þ Kiw Vw

The final step in estimating kMET (d1) is to translate CLh, referenced to the total chemical concentration in blood, into an elimination rate constant with units of inverse time that is referenced to the whole-body concentration. This is accomplished by dividing CLh by the compound’s apparent volume of distribution, Vd (L kg1). kMET ¼

ð11Þ

A commonly used notation of the unbound fraction is given in eq 12. The ratio of fraction bound (1 fu,i) to fraction unbound fu,i is related to

CLh Vd

ð13Þ

Vd relates the amount of chemical in the body to the concentration in the blood at steady state. Nichols et al.3 described an approach for estimating Vd as the ratio of a partitioning-based theoretical bioconcentration factor (BCFp; assuming no metabolism) and an equilibrium bloodwater partition coefficient Pbloodw.3 Vd ¼

BCF p Pbloodw

ð14Þ

BCFp may be estimated using a mass-balance approach that accounts for equilibrium partitioning into whole-organism lipid (modeled by Kow), nonlipid organic matter (modeled by Kpw using the experimental data for denatured BSA obtained in this study) and water content.23 The density of all phases was assumed to be 1 kg L1. BCFp ¼ υlb Kow þ υnlb Kpw þ υw

ð7Þ

The first step in the extrapolation procedure is to estimate in vivo intrinsic clearance (CLin vivo,int; L h1 gbody weight1) by incorporating scaling factors that relate experimental conditions from the S9 assay to the whole animal. Since the in vitro clearance rate is expressed as L h1 gp1 in the experimental vessel, the scaling factors are the fraction of the whole body weight represented by the liver (LW; gliver gbody weight1) and the amount of S9 protein in fish liver tissue (PL; gP gliver1). CLinvivo, int ¼ LW 3 PL 3 CLS9

the partition coefficients via the sum of the concentrations of lipid and protein. mlip þ mp 1 fu, i ð12Þ ¼ Kiw fu, i Vw

ð15Þ

The mass fraction of whole-organism lipid (υlb) was assumed to be 5%. This is less than the 10% proposed by Arnot and Gobas14 but is more appropriate for small fish used as test organisms in current bioaccumulation assessment protocols (e.g., OECD Guidelines for testing of chemicals No. 30524). The fraction of whole-organism nonlipid organic matter (υnlb) was assumed to be 20% of body weight. This fraction is dominated by protein but also contains carbohydrates and inorganic salts. Partitioning to this fraction was approximated in previous work by 0.035 3 Kow3,23 and more recently by 0.05 3 Kow.25 In the present study, however, we used measured partitioning to denatured BSA as a surrogate for partitioning to nonlipid organic material. The remaining 75% was assigned as water content (υw). Pbloodw can be defined analogously to BCFp and may be visualized as a partition coefficient between an aqueous phase separated from plasma by a membrane. Pbloodw ¼ υl Kow þ υp Kpw þ υw

ð16Þ

The mass fractions of blood plasma comprising lipids (υl) and protein (υp) were quantified experimentally. Other solid fractions such as carbohydrates and inorganic salts were neglected. The fraction's υ differs from the fraction's f (eqs 5 and 6) in that υ refers to the full mass balance including water (υl þ υp þ υw =1), while f refers only to the relevant partitioning phases (fp þ flip =1). Alternatively, Pbloodw can be calculated by rescaling the experimentally determined Kbloodw with eq 17. Pbloodw ¼ ðυl þ υp ÞKbloodw þ υw

ð17Þ

Finally, the BCF was calculated using the steady-state solution for a one-compartment bioaccumulation model given by Arnot and Gobas.14 BCF ¼

k1 k2 þ kMET þ kE þ kG

ð18Þ

For this exercise, we assumed that the exposure was limited to the chemical in the surrounding water (as in a standard BCF test). To implement the model, kinetic parameters (k1, the gill uptake rate constant; k2, the gill elimination rate constant; ke, the elimination rate 1137



ARTICLE

Table 1. Protein and Lipid Content of Trout Liver S9 and Blood Plasma Samples Used in This Study in Comparison with Literature Data mP/Vw (gP L1)

mlip/Vw (glip L1)

mp/mlip

rainbow trout S9

2

0.51 ( 0.01

3.92

this study

rainbow trout blood plasma rainbow trout blood

41.2 ( 7.5 35.9

18.4 ( 0.6

2.29

this study 42

rainbow trout blood

14.0

lake trout blood

5.9 to 12.9

27 26

fetal bovine serum

24

1.92

12.50

30

human blood serum

38

7.22

5.26

30

Table 2. All Partition Coefficients Determined in This Study log Kowa chlorpyrifos methoxychlor 4-n-nonylphenol pyrene a

4.96 5.08 5.76 4.88

log Kow range log in literatureb KPDMSw 4.105.46 4.745.62 5.456.07 4.495.31

4.47 4.49 4.51 4.36c

95% CI

log KBSAw (native) (L kgp1)

4.424.52 4.454.54 4.464.56 4.26 4.45

3.83 4.20 4.80 3.77

95% CI

log KBSAw (heat-denatured) (L kgp1)

3.783.87 4.124.26 4.734.86 3.663.87

3.47 3.72 4.53 3.64

95% CI

log KS9w (heat-dena-tured) (L kgpþlip1)

3.403.53 3.653.78 4.484.58 3.543.72

3.70 3.92 4.62 4.66

95% CI

log Kbloodw (native) (L kgpþlip1)

95% CI

3.583.80 3.774.03 4.474.74 4.504.77

4.46 4.94 5.60 4.92

4.414.50 4.885.00 5.575.64 4.874.96

Experimental values from the Physprop database.43 b Range of predicted and experimental values reported in ref 44. c Ref 19.

constant for fecal egestion; and kG, the rate constant that accounts for growth dilution) were estimated as described by Arnot and Gobas,14 assuming a 100 g fish with 5% lipid content and an ambient temperature of 12 C.

’ RESULTS AND DISCUSSION Protein and Lipid Determination. The liver S9 fraction obtained from trout was diluted to a protein content of 2 gP L1, which is typical for in vitro metabolism assays. When analyzed for lipid content, these samples contained 0.51 glip L1 (Table 1). This means that of the overall dry mass of lipids plus proteins (2.51 g L1), 20% was in the form of lipids. As lipids are expected to have a higher sorptive capacity than proteins, especially for very hydrophobic chemicals,25 the contribution of lipid to sorption of hydrophobic organic chemicals in the S9 fraction is likely to be substantial. Blood plasma contained 41.2 gP L1 and 18.4 glip L1. Although the lipid content of blood plasma is approximately half the protein content, sorption to lipid is expected to dominate overall partitioning. While the measured protein concentration agreed well with published data (Table 1), lipid content values reported in the literature are more variable. For example, Lien et al.26 reported blood lipid values ranging from 0.59 to 1.29% in lake trout (Salvelinus namaycush), while Bertelsen et al.27 reported values ranging from 1.3 to 2.3% in four different species. Previous work has shown that lipids in fish plasma are largely in the form of lipoproteins and that lipoprotein content changes dynamically in response to food composition and quantity,28 as well as temperature.29 Moreover, lipid content values tend to be operationally determined by the extraction method, which varies among studies. The value determined in this study is in good agreement with the 1.4% for rainbow trout given by Bertelsen et al.27 For comparison to reported values for trout plasma, Table 1 also gives protein and lipid values for human serum and fetal bovine serum. It is interesting to note that the protein to lipid ratio mp/mlip in trout plasma is half that of human serum and one-fifth that of fetal bovine serum, which is used for mammalian cell culture work (Table 1). This difference is relevant as it is often assumed that

chemical binding in blood is dominated by chemical binding to blood proteins.30 While this assumption may be valid for binding in human serum and fetal bovine serum, it may not be valid for binding in fish blood plasma with its higher lipid content. PDMSWater Partitioning. Measured KPDMSw (Table 2) values were in good agreement with the QSAR developed by Kwon et al.19 for PDMSwater partitioning, indicating the validity of these measurements. Because of its instability, the PDMSwater partition coefficient for FMHE was measured using a simplified method that reduced total experimental time. The resulting value (log KPDMSw = 4.49) also was consistent with the QSAR prediction. Three-Phase Partition Method Is Not Suitable for Compounds with Relatively Fast Abiotic Hydrolysis Rates. Measured Kiw values for FMHE (where i refers to BSA, liver S9 fraction, or blood plasma nonaqueous fraction) were consistently higher than the QSARs and mass-balance models discussed below would have predicted. The most plausible explanation for this finding is that loss of chemical from the system (be it through degradation or volatilization) resulted in artificially low CPDMS,equilibrium values (eq 2). Because the reported hydrolysis half-life for FMHE (3.2d at pH 9 and slower at lower pH20) is substantially longer than the total exposure time (26 h) for these partitioning studies, it was assumed that abiotic hydrolysis would not compromise measured values. This assumption was apparently violated, perhaps because the added protein acted as a catalyst for the hydrolysis reaction. On the basis of these observations, we elected to exclude FHME as an outlier throughout the remainder of this article. Further, we recommend caution when applying the three-phase partition method to compounds prone to relatively fast abiotic degradation because, apart from consistency checks, this method does not allow one to differentiate between degradation and partitioning if there is no degradation observed in control cells without added phase. ProteinWater Partitioning (Native BSA). Measured KBSAw values for native BSA (Table 2) were 7.6 to 13.5 times lower than the corresponding Kow for each tested compound and were consistent with the QSAR collected by De Bruyn et al.25 (Figure SI-1, Supporting Information). With the amount of chemical held 1138


Chemical Research in Toxicology constant, there was no dependence of measured KBSAw values on protein concentration, which ranged from 0.5 to 12 g L1. This finding suggests that sorption isotherms were linear in the protein concentration range tested. The number of molecules bound to BSA was calculated using a molecular weight of 60 kDa. Under the experimental conditions of this study, 0.02 to 0.16 mol pyrene was bound per mol BSA. Similar calculations for chlorpyrifos (0.02 to 0.11 mol per mol BSA) and methoxychlor (0.02 to 0.09 mol per mol BSA) indicated that saturation of binding for these three compounds was highly unlikely. In contrast, 13 mol nonylphenol were bound per mol BSA. Albumin is considered to have one to two primary binding sites and a number of low affinity binding sites.31 The number of binding sites on BSA for chlorpyrifos is 1 to 1.25.32,33 For bisphenol A, it was postulated that there is no saturation,34 and for cypermethrin, the number of binding sites per molecule was given as 0.78.33 Exploring the saturation of BSA binding was clearly not the focus of the present study. Nevertheless, we attempted to keep chemical concentrations as low as possible to avoid saturation. For comparison, Sultatos et al.32 measured a protein dissociation constant of 3.2 μM for chlorpyrifos. This translates to a log Kpw value of 3.71, which is similar to the value measured in this study (3.83). Kramer et al.35 determined a protein association constant for pyrene to BSA that is equivalent to a log Kpw of 5.14,35 which is higher than the log KBSAw of 3.77 determined here. ProteinWater Partitioning (Denatured BSA). Because chemical partitioning to liver S9 fractions cannot be determined with native material due to its enzymatic activity, we examined how heat-treatment affects the partitioning properties of BSA. Heat treatment causes BSA to denature and become gelatinous. As a consequence, it was more difficult to handle the suspension and ensure that it was homogeneous. Nevertheless, chemical partitioning to BSA only changed slightly with heat treatment (see also Figure SI-1, Supporting Information), resulting in KBSAw values that were a factor of 1.3 (chlorpyrifos) to 3.0 (methoxychlor) times lower than corresponding values for native BSA (Table 2). Binding to Liver S9. KS9w values (L kgpþlip1) for the tested compounds were higher than the corresponding KBSAw values (heat-treated; L kgp1) (Table 2 and Figure SI-3, Supporting Information), presumably because the S9 fractions contain lipid, which has a higher sorptive capacity than proteins. Using the protein and lipid data for liver S9 fractions given in Table 1, we calculated flip and fp values of 0.2 and 0.8, respectively. These parameters were then used as inputs to the mass-balance equation shown in eq 4 to predict KS9w values for each chemical, using Kow as a surrogate of lipid partitioning and KBSAw (heat-treated) as a surrogate for protein binding. The resulting modeled KS9w values match the experimental data neatly (Figure 1), with a ratio of experimental to modeled values ranging from 0.3 to 3. With the exception of pyrene, modeled KS9w values tended to overestimate measured values. This could have been due to the fact that BSA and octanol are imperfect surrogates for S9 protein and lipid, respectively. Underprediction of the measured KS9w value for pyrene may have been due to an experimental artifact caused by partial degradation of chemical during the experiment. Independent of the source of the variability, the deviations from expectations were small. Binding to Trout Blood Plasma. Blood plasmawater partition coefficients (Kbloodw; L kgpþlip1) were substantially greater than corresponding KS9w values for each chemical and more than 10 times higher than corresponding KBSAw (heat-treated) values (Table 2, Figure SI-4, Supporting Information). Using the massbalance model (eq 4), we predicted chemical binding to blood

ARTICLE

Figure 1. Comparison between experimental and modeled partition coefficients between liver S9 and water (KS9w; in L kglipþp1) and between fish blood plasma and water (Kbloodw; in L kglipþp1). Predictions obtained with the mass-balance model were generated using experimentally determined fractions of lipids and proteins (Table 1). Kow was used as a surrogate for lipidwater partitioning. Measured log KBSAw values for heat-treated BSA were used to predict partitioning to the protein fraction in liver S9 samples and blood. The solid line indicates the 1:1 line, and the dashed lines are 1 order of magnitude in either direction.

plasma using measured protein and lipid data from Table 1. The flip and fp values calculated using these data were 0.31 and 0.69, respectively. Modeled and experimental Kbloodw values were in generally good agreement (Figure 1). The ratio of experimental to modeled Kbloodw values ranged from 1.0 to 3.1. Partitioning to lipids dominated the overall partition coefficient with 88 to 93% of Kbloodw determined by lipid partitioning and 7 to 12% by protein binding. Fraction Unbound in Liver S9. The S9 assay is typically performed with a protein concentration of 2 g L1.13 Using measured KS9w values for each compound as an input to eq 11, we estimated the fraction of chemical unbound in heat-denatured liver S9 (fu,S9) which ranged from 0.9 to 7.3% (Table 3). Thus, more than 90% of each test chemical was bound to proteins and lipids. The ratios of fraction bound (1 fu,S9) to fraction unbound (fu,S9) ranged from 13 to 114 (Figure 2). S9 binding data for all compounds except pyrene were in good agreement with a QSAR developed by Han et al.5 to predict chemical binding in trout liver microsomes and S9 fractions (eq 19). Why pyrene was an outlier could not be explained. Although this algorithm was developed using data from binding studies with rat microsomes, it may be applicable to fish liver microsomes and S9 fractions. ! 1 fu, S9 ð19Þ ¼ 0:694 3 log Kow 2:158 log fu, S9 Fraction Unbound in Fish Blood Plasma. The fraction of chemical unbound in blood plasma also can be calculated from eq 11 using measured partition coefficients and assuming fixed levels of lipid and protein. Calculated fu,blood values for the four test chemicals ranged from 0.004 to 0.058%, and the ratios of fraction bound (1 fu,blood) to fraction unbound (fu,blood) ranged from 2400 to 24000 (Table 3). Thus, more than 99.94% of each chemical tested was predicted to be bound to plasma proteins and lipids. Previously, Han et al.5,36 presented a QSAR equation for chemical binding to blood plasma (eq 20), which was based on data from various literature sources (largely human studies, but 1139



ARTICLE

Table 3. Free Fraction of the Test Compounds in Trout Plasma and Liver S9 Fractions, Including Comparison of Experimental Values with Model Predictions

chlorpyrifos methoxychlor

a

(1- fu,blood)/ fu,blood

(1- fu,blood)/ fu,blood

(1-fu,S9)/ fu,S9)

(1-fu,S9) fu,S9)

fu,blood

exp.

mass-balance modela

fu,S9

experimental

QSAR modelb

0.058% 0.019%

1724 5215

1803 2433

7.32% 4.59%

13 21

16 19

nonylphenol

0.004%

23935

12008

0.94%

106

43

pyrene

0.020%

4932

1578

0.87%

114

15

eq 4. b QSAR developed by Han et al. for microsomes.36

Figure 2. Logarithms of the ratio of fraction bound (1 fu,i) to fraction unbound (fu,i) for trout blood plasma (0) and liver S9 fractions (() experimentally determined in this study in comparison with QSAR equations for blood (solid lines)36 and rat liver microsomes (dashed lines). The bold solid line refers to the full mass-balance model developed in this study, while the thin solid line is from ref 36. The thick dashed line is based on an algorithm given by Austin et al.37 The thin dashed line was generated using an algorithm given by Han et al.,5 which was developed from a subset of the data given by Austin et al.37

including data for several other species including trout). ! 1 fu, blood log ¼ 0:613 3 log Kow 0:569 fu, blood

ð20Þ

A comparison of binding values measured in this study to values predicted by eq 20 suggests that eq 20 underestimates binding in trout blood plasma (Figure 2). This discrepancy can be explained by the fact that the lipid content of trout plasma is greater than that of human serum (Table 1). When the mass-balance model given in eq 4 was used to predict (1 fu,blood)/fu,blood) from Kow (acting as a surrogate for Klipw) and measured KBSAw values (heat treated), then eq 21 was obtained. It should be noted that eq 21 was derived for comparison purposes. Because it is based on only four data points, it is not a valid QSAR equation. Nevertheless, experimentally determined blood binding values were found to be highly consistent with eq 21 (Figure 2). ! 1 fu, blood ð21Þ ¼ log Kow 1:69 log fu, blood Volume of Distribution. The volume of distribution can be estimated as the ratio of the partitioning-based bioconcentration factor (BCFp) to the bloodwater partition coefficient (Pbloodw). Both parameters have hitherto been calculated using Kow based models.3 In the present effort, BCFp values were estimated using measured KBSAw values as surrogates for chemical binding to

nonlipid organic matter (Table 4). Bioconcentration factors calculated in this manner were generally close to those determined using the log Kow-based model (data not shown). This result can be attributed to the dominant role of lipid partitioning in both sets of calculations. Using experimental data measured in the present study, it was possible to directly calculate Pbloodw from Kbloodw by rescaling for the dry weight content of plasma (lipids plus proteins eq 17). Blood plasmawater partitioning values determined in this manner were 1.0 to 3.4 times higher than the values obtained using the Kow/KBSAw modeling approach (i.e., eq 16; Table 4). When the Kow/KBSAw based model was used to calculate both BCFp and Pbloodw, estimated volumes of distribution were close to 3.0 for all test compounds. Scaling Factors for in Vitro to in Vivo Extrapolation. The well-stirred liver model (eq 9) includes a scaling factor (fu) that corrects for binding to blood plasma and the in vitro system used to measure metabolism. To date, it has not been possible to directly estimate fu from measured binding values for fish plasma and liver S9 fractions. Cowan-Ellsberry et al.13 addressed this deficiency by setting fu values for six test compounds equal to 1, while Han et al.36 estimated fu values for five compounds using QSARs for chemical binding in blood plasma (largely based on human data) and rat liver microsomes. Using the QSARs given by Han et al.,36 fu values for compounds of moderate-to-high hydrophobicity (log Kow values from 3 to 7) are predicted to vary between 0.05 and 0.1.13,36,37 In the present study, we calculated fu from experimentally determined fractions unbound in trout plasma and liver S9 fractions (eq 10). This approach has the advantage that data obtained under different conditions (e.g., different protein or lipid content, different ratio of S9 to water) can be compared. For an S9 assay with 2 g L1 protein content, calculated fu values ranged from 0.0042 to 0.0233 (Table 4), indicating that in blood the unbound chemical fraction is 40 to 260 times lower than that in the S9 assay. It is clear, therefore, that fu values determined in this effort differ substantially from values used in previous studies. The effect of fu on modeled BCF values is illustrated in Figure 3 for (A) chlorpyrifos, and (B) nonylphenol. The previously reported in vitro metabolic half-life of chlorpyrifos was 640 min (equal to 0.325 μmol of parent lost gP h1).13 In another study, Arnot et al.38 used a mass balance model to estimate kMET values from measured BCF data. The in vivo metabolic biotransformation half-life for chlorpyrifos determined in this manner for QSAR development was 5.9 d when normalized to a 10 g fish.39 Using parameter values (fu and Vd) developed in the present study, and assuming total hepatic blood flow, the modeled in vivo kMET was 0.00142 d1 and 0.17 d1 if fu was set to 1. These values correspond to in vivo half-lives of 1140



ARTICLE

Table 4. Scaling Factor (fu) and Volume of Distribution (Vd) Values Calculated Using the Kow Based Model from the Literature3 as well as Experimental Data Measured in This Study Kow and KBSAw based model fua

BCFpb

1

(L kg )

Pbloodwc

1

(L L )

experimental 1

d

Vd (L kg )

Pbloodwe

1

(L L )

Vdf (L kg1)

chlorpyrifos

0.0079

5154

1801

2.86

1725

2.99

methoxychlor

0.0042

7067

2430

2.91

5216

1.35

nonylphenol pyrene

0.0045 0.0233

35589 4669

11993 1577

2.97 2.96

23935 4933

1.49 0.95

a Calculated with eq 10 using the data in Table 3 and an S9 protein content of 2 mg/mL. b Calculated with eq 15. c Calculated with eq 16 using denatured BSA as the surrogate for proteins. d Calculated with eq 14 from the modeled BCFp and modeled Pbloodw. e Rescaled from experimental Kbloodw (Table 2). f Calculated with eq 14 from the modeled BCFp and experimental Pbloodw.

Figure 3. Influence of metabolism half-life (t1/2) and the scaling factor, fu, on modeled BCF values for (A) chlorpyrifos13 and (B) nonylphenol.5 Measured rates of intrinsic in vitro clearance were obtained from literature sources.5,13 Chemical-specific volumes of distribution (Vd) were generated using measured Pbloodw values and a Kow/Kpw model to predict the partitioning-based bioconcentration factor, BCFp (Table 4). The dotted lines refer to experimental data, and the gray area indicates the range of reported experimental BCF data.

489 and 4 d, respectively. The modeled BCF with metabolism was 4123 L kgww1, while the partitioning-based BCF (BCFp; Kow/KBSAw model) was 5154. The modeled BCF without metabolism (kMET = 0) was 4221 L kgww1. This is somewhat lower than the partitioning-based value of 5154 L kgww1 because the kinetic BCF model was parametrized with rate constants for passive uptake and elimination and includes additionally the kinetic rate constant of dilution by growth. When fu was set to 1, the predicted BCF was 1093 L kgww1. For comparison to these modeled values, a review of 35 publicly available BCF values from various freshwater fish species indicates a range from 300 to 3400 L kgww1.40 Figure 3A shows the interaction between fu

and changes in metabolic clearance, represented by the in vitro halflife, and indicates the range of experimental BCF values with a gray box. As the figure demonstrates, modeled BCF values are highly responsive to changes in both fu and in vitro metabolism rate. Qualitatively similar relationships were observed for nonylphenol (Figure 3B), whose clearance in the trout S9 assay was measured by Han et al.5 The measured CLS9 of 0.11 L h1 gP1 (equivalent to t1/2 of 190 min) translates to a kMET of 0.0054 d1 (in vivo half-life of 128 d), which is in stark contrast to the in vivo half-life of 0.59 d deduced from BCF data38 and used to parametrize the QSAR model of Arnot et al.39 If fu was set to 1, the predicted in vivo half-life would be 1.03 d. The kMET modeled with the experimentally determined fu of 0.0045 lowers the BCF from 19170 to 13570 L kg1. Han et al.5 used an fu value of approximately 0.1, which yielded a predicted BCF of 1900 L kg1. When an fu value of 1 was incorporated into the model developed in this study, the predicated BCF was 242 L kg1. For comparison, reported experimental BCF values (from studies with carp, fathead minnows, and salmon) range from 75 to 550 with a median value of 246 L kg1.5 Figure 3B again illustrates that both fu and the metabolism rate constant have large impacts on predicted BCF values, and the gray area represents the variability of measured BCF values. In both examples described above, the use of experimentally derived fu and Vd values resulted in BCF estimates that were much higher than measured values. In contrast, BCFs predicted using an assumed fu of 1.0 (i.e., the Cowan-Ellsberry et al. approach13) resulted in much better consistency between model results and empirical BCF values. The reason for this apparent discrepancy is unknown but is unlikely to be due to the quality of partitioning data obtained here as the data were consistent with mass-balance calculations as well as QSARs for similar matrices. One possible explanation for this finding challenges the initial assumption of the extrapolation model that only freely dissolved chemicals are available for metabolic degradation. The assumption that only freely dissolved substrate is available for metabolic degradation is valid when (1) the absorption and desorption between bound and free form are occurring instantaneously, and (2) the substrate can only be delivered as a freely dissolved form to the active binding site of the enzyme. If the rate of chemical delivery to the active site of an enzyme is fast relative to the rate of enzymatic degradation, then equilibrium considerations represented by bound/free relationships may play a limited role, and the algorithms currently used to extrapolate in vitro hepatic metabolism data for fish to the whole animal could underestimate true metabolism impacts on chemical bioaccumulation. This 1141


Chemical Research in Toxicology hypothesis is substantiated by kinetic studies of Kramer et al.35 who demonstrated that BSA facilitates the uptake of pyrene into a polymer. When 5% of fetal bovine serum was added to the aqueous phase, uptake was accelerated by a factor of 8. A similar dramatic effect has been demonstrated by passive dosing of very hydrophobic chemicals in bioassays, whose time to reach equilibrium was decreased by orders of magnitude in the presence of serum proteins and membrane vesicles.11,41 A second factor that may contribute to the observed discrepancy between predicted and measured BCFs is representativeness of the liver S9 fraction of liver enzymatic activity. During preparation of the S9 fraction, enzymes might loose their activity, the added cofactors in the in vitro system might not be optimal, and/or the enzymatic activity in vivo might be inducible, while it is constant in the S9 extract. A third source of uncertainty is the assumption that metabolism occurs only in the liver. Current in vitro to in vivo metabolism extrapolation procedures generally ignore the possibility of significant extrahepatic metabolism as is discussed in more detail in ref 3. If extrahepatic activity contributes substantially to whole-body clearance of chemicals, BCFs predicted using measured levels of hepatic clearance will tend to overestimate actual values.

’ ASSOCIATED CONTENT

bS

Supporting Information. Comparison of experimental data with literature QSARs, relationship between log KBSAw of native and heat-treated BSA as well as log KS9w and log Kbloodw. This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*National Research Centre for Environmental Toxicology (Entox), The University of Queensland, 39 Kessels Rd, Brisbane, Qld 4108, Australia. Phone: þ61 7 3274 9180. Fax: þ61 7 3274 9003. E-mail:[email protected]. Funding Sources

This work was financially supported by CEFIC/LRI project no LRI-ECO6.2-ILSIHESI-0804 and European Commission Joint Research Centre Institute for Consumer Health Protection Contract CCR.IHCP.C434207.X0.

’ DISCLOSURE The views expressed in this article are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission. ’ ACKNOWLEDGMENT We thank the members of the HESI Bioaccumulation Project Committee for their helpful input and guidance. We thank Nynke Kramer and Satoshi Endo for reviewing an earlier version of the manuscript. The invaluable suggestions of two anonymous reviewers are gratefully accepted. ’ ABBREVIATIONS BCF, bioconcentration factor; BCFp, partitioning-based bioconcentration factor (calculated from mass-balance model); BSA,

ARTICLE

bovine serum albumin; Ci, concentration of test chemical in phase i; i = PDMS, BSA, fish liver S9, fish blood plasma, etc.; Csus, concentration of test chemical in the aqueous suspension; CLS9, intrinsic clearance rate of liver S9 fraction; CLin vivo,int, in vivo intrinsic clearance by the liver; CLh, total hepatic clearance; Cw, freely dissolved (aqueous) concentration of test chemical; flip, fractional lipid content (related to dry weight or sum of proteins and lipids); FMHE, fluoxypyr-methyl-heptyl ester; fp, fractional protein content (related to dry weight or sum of proteins and lipids); fu, free fraction correction term calculated as the ratio of fu,blood to fu,S9; fu,blood, fraction unbound in blood; fu,i, fraction unbound in phase i; fu,S9, fraction unbound in the liver S9 preparation; k1, the gill uptake rate constant; k2, the gill elimination rate constant; KBSAw, BSAwater partition coefficient; Kbloodw, trout blood plasmawater partition coefficient; KE, the elimination rate constant for fecal egestion; kG, the rate constant that accounts for growth dilution; Kiw, partition coefficient between phase i and water; Klipw, lipidwater partition coefficient; km, in vitro first-order metabolic rate constant; kMET, in vivo metabolic transformation constant; Kow, octanolwater partition coefficient; Kpw, proteinwater partition coefficient; KS9w, partition coefficient between liver S9 fraction and water; KPDMSw, partition coefficient between PDMS and water; KPDMSsus, partition coefficient between PDMS and the colloidal suspension; LF, fraction of total cardiac output perfusing the liver; lip, lipid; mBSA/Vw, BSA concentration in the assay; mi/Vw, concentration of phase i in the assay; ni, molar amount of chemical in phase i; p, protein; Pbloodw, equilibrium bloodwater partition coefficient that takes all compart8w?>ments into account (and thus differs from Kbloodw); Qc, cardiac output; QSAR, Quantitative structureactivity relationship; t1/2, metabolic degradation half-life; υlb, fraction of whole organism lipid; υnlb, fraction of whole organism nonlipid organic matter; Vd, volume of distribution which refers to the relative affinity of a chemical for the whole body and for blood.

’ REFERENCES (1) de Wolf, W., Comber, M., Douben, P., Gimeno, S., Holt, M., Leonard, M., Lillicrap, A., Sijm, D., van Egmond, R., Weisbrod, A., and Whale, G. (2007) Animal use replacement, reduction, and refinement: development of an integrated testing strategy for bioconcentration of chemicals in fish. Integr. Env. Assess. Manag. 3, 3–17. (2) Nichols, J. W., Bonnell, M., Dimitrov, S. D., Escher, B. I., Han, X., and Kramer, N. I. (2009) Bioaccumulation assessment using predictive approaches. Integr. Env. Assess. Manag. 5, 577–597. (3) Nichols, J. W., Schultz, I. R., and Fitzsimmons, P. N. (2006) 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. 78, 74–90. (4) Kwon, Y. (2002) Pharmacokinetics, Pharmacodynamics and Drug Metabolism for Industrial Scientists; Kluwer Academic Publishers, New York. (5) Han, X., Nabb, D. L., Yang, C. H., Snajdr, S. I., and Mingoia, R. T. (2009) Liver microsomes and S9 from rainbow trout (Oncorhynchus mykiss): comparison of basal-level enzyme activities with rat and determination of xenobiotic intrinsic clearance in support of bioaccumulation assessment. Environ. Toxicol. Chem. 28, 481–488. (6) Schmieder, P., and Henry, T. (1988) Plasma binding of 1-butanol, phenol, nitrobenzene and pentachlorophenol in the rainbow trout and rat: a comparative study. Comp. Biochem. Physiol. 91(C), 413–418. (7) Tsai, T. H. (2003) Assaying protein unbound drugs using microdialysis techniques. J. Chromatogr., B 797, 161–173. (8) Wang, H. L., Wang, Z. X., Liu, M. L., and Zou, H. F. (2008) Microdialysis sampling method for evaluation of binding kinetics of small molecules to macromolecules. Anal. Chem. 80, 2993–2999. 1142


Chemical Research in Toxicology (9) Vuignier, K., Schappler, J., Veuthey, J. L., Carrupt, P. A., and Martel, S. (2010) Drug-protein binding: a critical review of analytical tools. Anal. Bioanal. Chem. 398, 53–66. (10) Artola Garicano, E., Vaes, W. H. J., and Hermens, J. L. M. (2000) Validation of negligible depletion solid-phase microextraction as a tool to determine tissue/blood partition coefficients for semivolatile and nonvolatile organic chemicals. Toxicol. Appl. Pharmacol. 166, 138–144. (11) Kwon, J.-H., W€uthrich, T., Mayer, P., and Escher, B. I. (2009) Development of a dynamic delivery method for in vitro bioassays. Chemosphere 76, 83–90. (12) Kramer, N. I. (2010) In Utrecht University, The Netherlands Measuring, modeling and increasing the free concentration of test chemicals in cell assays, PhD dissertation, Utrecht University, The Netherlands. (13) Cowan-Ellsberry, C., Dyer, S., Erhardt, S., Bernhard, M. J., Roe, A. L., Dowty, M. E., and Weisbrod, A. V. (2008) Approach for extrapolation in-vitro metabolism data to refine bioconcentration factor estimates. Chemosphere 70, 1804–1817. (14) Arnot, J. A., and Gobas, F. (2003) A generic QSAR for assessing the bioaccumulation potential of organic chemicals in aquatic food webs. QSAR Comb. Sci. 22, 337–345. (15) Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) Protein measurement with the folin phenol reagent. J. Biol. Chem. 193, 265–275. (16) Bligh, E. G., and Dyer, W. J. (1959) A rapid method of total lipid extraction and purification. Can. J. Biochem. Physiol. 37, 911–917. (17) Smedes, F. (1999) Determination of total lipid using nonchlorinated solvents. Analyst 124, 1711–1718. (18) Kaiser, S. M., and Escher, B. I. (2006) The evaluation of liposome-water partitioning of 8-hydroxyquinolines and their copper complexes. Environ. Sci. Technol. 40, 1784–1791. (19) Kwon, J.-H., W€uthrich, T., Mayer, P., and Escher, B. I. (2007) Dynamic permeation method to determine partition coefficients of highly hydrophobic chemicals between poly(dimethylsiloxane) (PDMS) and water. Anal. Chem. 79, 6816–6822. (20) FOOTPRINT (2007) The FOOTPRINT Pesticide Properties DataBase, Database collated by the University of Hertfordshire as part of the EU-funded FOOTPRINT project (FP6-SSP-022704), http://www. eu-footprint.org/ppdb.html. (21) Rowland, M., Benet, L. Z., and Graham, G. G. (1973) Clearance concepts in pharmacokinetics. J. Pharmacokinet. Biopharm. 1, 123–136. (22) Wilkinson, G. R., and Shand, D. G. (1975) A physiological approach to hepatic drug clearance. Clin. Pharmacol. Ther. 18, 377–390. (23) Arnot, J. A., and Gobas, F. (2004) A food web bioaccumulation model for organic chemicals in aquatic ecosystems. Environ. Toxicol. Chem. 23, 2343–2355. (24) OECD (1996) Test guideline 305. Bioaccumulation: Flowthrough Fish Test, Organization for Economic Co-Ordination and Development, Paris, France. (25) De Bruyn, A., and Gobas, F. (2007) The sorptive capacity of animal protein. Environ. Toxicol. Chem. 26, 1803–1808. (26) Lien, G. J., McKim, J. M., Hoffman, A. D., and Jenson, C. T. (2001) A physiologically based toxicokinetic model for lake trout (Salvelinus namaycush). Aquat. Toxicol. 51, 335–350. (27) Bertelsen, S. L., Hoffman, A. D., Gallinat, C. A., Elonen, C. M., and Nichols, J. W. (1998) Evaluation of log K-ow and tissue lipid content as predictors of chemical partitioning to fish tissues. Environ. Toxicol. Chem. 17, 1447–1455. (28) Babin, P. J., and Vernier, J. M. (1989) Plasma lipoproteins in fish. J. Lipid Res. 30, 467–490. (29) Wallaert, C., and Babin, P. J. (1994) Effects of temperaturevariations on dietary-lipid absorption and plasma-lipoprotein concentrations in trout (Oncorhynchus mykiss). Comp. Biochem. Physiol. B 109, 473–487. (30) G€ulden, M., Diederickx, P., and Seibert, H. (2006) Validation of a prediction model for estimating serum concentrations of chemcials which are equivalent to toxic concentrations in vitro. Toxicol. in Vitro 20, 1114–1124.

ARTICLE

(31) Bertucci, C., and Domenici, E. (2002) Reversible and covalent binding of drugs to human serum albumin: methodological approaches and physiological relevance. Curr. Med. Chem. 9, 1463–1481. (32) Sultatos, L. G., Basker, K. M., Shao, M., and Murphy, S. D. (1984) The interaction of the phosphorothioate insecticides chlorpyrifos and parathion and their oxygen analogs with bovine serum-albumin. Mol. Pharmacol. 26, 99–104. (33) Cui, Y., Guo, J. F., Xu, B. J., and Chen, Z. Y. (2006) Binding of chlorpyrifos and cypermethrin to blood proteins. Pest. Biochem. Physiol. 85, 110–114. (34) Zhang, Y. L., Zhang, X. A., Fei, X. C., Wang, S. L., and Gao, H. W. (2010) Binding of bisphenol A and acrylamide to BSA and DNA: Insights into the comparative interactions of harmful chemicals with functional biomacromolecules. J. Haz. Mat. 182, 877–885. (35) Kramer, N. I., van Eijkeren, J. C. H., and Hermens, J. L. M. (2007) Influence of albumin on sorption kinetics in solid-phase microextraction: Consequences for chemical analyses and uptake processes. Anal. Chem. 79, 6941–6948. (36) Han, X., Nabb, D. L., Mingoia, R. T., and Yang, C.-H. (2007) Determination of xenobiotic intrinsic clearance in freshly isolated hepatocyxtes from rainbow trout (Oncorhynchus mykiss) and rat and its application in bioaccumulation assessment. Environ. Sci. Technol. 41, 3269–3276. (37) Austin, R. P., Barton, P., Cockroft, S. L., Wenlock, M. C., and Riley, R. J. (2002) The influence of nonspecific microsomal binding on apparent intrinsic clearance, and its prediction from physicochemical properties. Drug Metab. Dispos. 30, 1497–1503. (38) Arnot, A. J., Mackay, D., and Bonnell, M. (2008) Estimating metabolic biotransformation rates in fish from laboratory data. Environ. Toxicol. Chem. 27, 341–351. (39) Arnot, A. J., Meylan, W., Tunkel, J., Howard, P. H., Mackay, D., Bonnell, M., and Boethling, R. S. (2009) A Quantitative-Structure Activity Relationship for predicting metabolic biotransformation rates for organic chemicals in fish. Environ. Toxicol. Chem. 28, 1168–1177. (40) Arnot, J. A., and Gobas, F. (2006) A review of bioconcentration factor (BCF) and bioaccumulation factor (BAF) assessments for organic chemicals in aquatic organisms. Environ. Rev. 14, 257–297. (41) Kramer, N. I., Busser, F. J. M., Oosterwijk, M. T. T., Schirmer, K., Escher, B. I., and Hermens, J. L. M. (2010) Development of a Partition-Controlled Dosing System for Cell Assays. Chem. Res. Toxicol. 23, 1806–1814. (42) Manera, M., and Britti, D. (2006) Assessment of blood chemistry normal ranges in rainbow trout. J. Fish Biol. 69, 1427–1434. (43) Syracuse Research Physprop database, http://esc.syrres.com/ physprop/ (2010). (44) Virtual Computational Chemistry Laboratory, ALOGPS2.1, http://www.vcclab.org (2009).

1143


Protein and Lipid Binding Parameters in Rainbow Trout

Recommend Documents