Environmental Toxicology and Chemistry, Vol. 33, No. 5, pp. 1064–1071, 2014 # 2013 SETAC Printed in the USA
Environmental Toxicology A PHYSIOLOGICALLY-BASED PHARMACOKINETIC MODEL FOR DISPOSITION OF 2,3,7,8-TCDD IN FATHEAD MINNOW AND MEDAKA ZAHRA PARHIZGARI* and JAMES LI Program of Environmental Applied Science and Management, Ryerson University, Toronto, ON, Canada (Submitted 5 April 2013; Returned for Revision 18 June 2013; Accepted 16 December 2013) Abstract: A physiologically-based pharmacokinetic model was developed for the disposition of 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) in 2 fish species—fathead minnow and medaka. The model was developed based on the empirical data on disposition of dioxins in fish tissues, as well as existing knowledge on the mechanisms of uptake, distribution, storage, and elimination of dioxins in various species (other than fish). The present study examined the applicability of mechanisms known to occur in other species for fish and concluded that the same mechanisms defined for disposition of 2,3,7,8-TCDD in (mostly) rodents can be applicable for the 2 fish species examined as well. Parameter values for the model were selected and/or calibrated using available databases. Model compartments included the gill, kidney, liver, and other richly-perfused tissues, as well as fat and other slowly-perfused tissues. The model was calibrated using 2 independent datasets for exposure of fathead minnow and medaka to 2,3,7,8-TCDD in water. The initial values of the model parameters were selected from several sources, and calibrated to represent the 2 exposure datasets. With very few exceptions, the estimated parameter values for the 2 species were comparable, and the final predictions were in strong agreement with the observations. The model developed in the present study can therefore be used in the prediction of the body burden of 2,3,7,8-TCDD in fathead minnow and medaka. Uncertainty in the model prediction as a result of variability in input parameters is discussed for the parameters with the highest impacts on the model outcome. Environ Toxicol Chem 2014;33:1064–1071. # 2013 SETAC Keywords: Dioxin
Pharmacokinetics
Fish
Aryl hydrocarbon
Dose-response
et al. [7] improved previous models by using time-course distribution data obtained from PBPK studies on rats. One of the main concerns in the assessment of exposure to dioxins and furans is their interactions with hepatic cytochrome P450 enzymes. PBPK models have successfully explained these interactions for a number of test animals (mostly rodents). They have modeled both the dioxin binding to the Ah receptor and the binding of the Ah receptor-dioxin complex to promoter sites on DNA. The latter form of binding can enhance the rate of gene transcription. These findings have been consistent across species, which increase the confidence in extrapolating the model between species [8]. Although the PBPK models have been successfully implemented to extrapolate kinetic data between mammals until early 1990s, the applications to fish were limited to models adapted from mouse models describing the kinetics of compounds injected intravenously [9]. Gill description and elimination mechanisms were included in later models. In 1990, Nichols et al. [9] developed a physiologically based toxicokinetic model for uptake and disposition of waterborne organic chemicals in fish. The model was parameterized using information from published data, and accurately simulated the uptake of pentachloroethane in rainbow trout. One study developed a PBPK model for disposition of dioxins in a fish species. Nichols et al. [10] proposed a model for maternal transfer of 2,3,7,8-TCDD in brook trout. The model considered dietary uptake and was evaluated through the comparison to measured data. This comparison suggested that whole body residue measurements could be used to estimate residues in developing ovaries within a factor of 2. The present study proposes a model for toxin accumulation of 2,3,7,8-TCDD in 2 fish species in the form of a PBPK model. The model considers the gill pathway and has been developed based on available information on the uptake, distribution, storage, and elimination mechanisms of dioxins in various
INTRODUCTION
Polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs)—commonly known as dioxins and furans—are toxic, persistent, and bioaccumulative chemicals that are predominantly a result of human activities such as the production of chemicals and the metallurgical industry. The most potent congener in the series is 2,3,7,8-tetrachlorodibenzop-dioxin (2,3,7,8-TCDD). The long half-life of dioxins means that even a low rate of exposure will lead to the accumulation of dioxins in the body and continual long-term exposure might result in an extremely high body burden. The biochemical and toxicological effects of dioxins and furans are more closely related to their concentration in the target tissue than to the daily dose [1,2]. Evaluation of effective or tissue dose under a variety of conditions is possible using the understanding of xenobiotic modes of actions, which provides insight into both the sites/ mechanisms of action and the xenobiotic form causing the response. Physiologically based pharmacokinetic (PBPK) models can describe chemical disposition in the body based on the fundamental information on physicochemical properties, transport, metabolism, and various excretory mechanisms [3], and the predictive capabilities of PBPK models make it possible to estimate the behavior of chemicals in species over time without the need for direct experimentation, by using the knowledge of more fundamental physiological and biochemical processes [4]. Several authors have developed PBPK models for disposition of dioxins and furans in rodents [5–8]. Early PBPK models examined the disposition of PCDFs (and PCBs) by taking into account the role of lipid solubility in tissue uptake. Wang * Address correspondence to [email protected]. Published online 25 December 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2504 1064
A PBPK Model for Disposition of 2,3,7,8-TCDD in Fathead Minnow and Medaka
species (other than fish). In the present study, model parameters not available in the literature were calibrated using available measurement data. MATERIALS AND METHODS
Model development Exposure pathways. The relative contributions of uptake from food and water have been a subject of discussion. Loonen et al. [11] studied the total bioaccumulation of PCDDs and PCDFs in guppies and concluded that the uptake from food was small compared with the uptake from water (0.9–4.9%). Rifkin and Lakind [12] reported that fish accumulate dioxins by ingestion (biomagnifications) rather than water (bioconcentration). Karl et al. [13] found a correlation between the dioxin concentration of commercial feed and the resulting concentration in the rainbow trout muscle fat. Based on Randall et al. [14], the uptake of persistent lipophilic toxicants in fish occurs by transfer across the gills. They argue that flux across the gills is rapid and in order for feeding to have a significant effect on toxicant concentration in the body, the fish must eat at quantities much higher than those typical of fish. A review of available studies on the relative importance of contaminant uptake from water versus food by Nowell et al. [15] revealed that most of the studies observed greater accumulations from water as opposed to food. Uptake. The movement of water and blood through the gill results in a counter-current exchange between the 2. The rate of uptake of a chemical by the gill is defined by uptake clearance, which is proportional to the concentration difference between the external water and the blood plasma water. Nichols et al. [9], in their physiologically based toxicokinetic model for uptake of organic chemicals by fish, used a general expression for chemical flux across gills and related it to an exchange coefficient, concentration gradient between the incoming water and the venous blood, and a blood to water partition coefficient. Lien and McKim [16] incorporated the diffusion limitation into a more detailed gill model and suggested a numerical solution for the exchange coefficient through parameters such as molecular diffusivity (or the gill permeability coefficient) for specific chemicals, total lamellar surface area, and the average thickness of the diffusion path length. Lien et al. [17] proposed a model for countercurrent exchange of chemicals at the gill surface, considering the free concentrations in water and blood. They calculated an exchange coefficient for 3 chloroethanes in a way that it represented limitations by 3 barriers including flow of water and blood, diffusion barriers, and blood-water partitioning. Another parameter taken into account is the binding to organic carbon in water, which reduces the free chemical concentration that is available to diffuse across the gills. Arnot and Gobas [18] related the dissolved (bioavailable) concentration of chemicals in water to the dissolved organic carbon, particulate organic carbon, and the octanol:water partition coefficient of the chemical. In the present study, uptake through the gill is controlled by partitioning between blood and water, and the gill permeability for dioxin, also called the molecular diffusivity of dioxins. The model is as follows
K xg C vb C b ¼ C vb þ C wd QT Pbw
ð1Þ
where Cb is blood concentration, Cvb is venous blood concentration, kx-g is an exchange coefficient, QT is the cardiac
Environ Toxicol Chem 33, 2014
1065
output, Cwd is the dissolved concentration in water, and Pbw is blood-to-water partition coefficient. The exchange coefficient was calculated using a diffusion rate constant for bronchial flux, with an initial value selected at 5.6 106 cm2/s based on the molecular diffusivity of 2,3,7,8-TCDD in water [19]. A complete list of selected initial values is provided in Table 1. Distribution and bioaccumulation. In terms of the arterial blood flow into the organ, fish tissues can be classified into richly perfused and slowly perfused tissues. Chemical flux in the tissue can be blood-flow limited or diffusion limited. When the flux is limited by blood flow, a chemical equilibrium exists between the Table 1. Initial values for model parameter Parameter Physiological Growth coefficient Power term in growth Cardiac output coefficient Ventilation volume coefficient Effective ventilation Gill surface area coefficient Average thickness of diffusion path Uptake Blood-water partitioning (gills) Gill permeability for dioxin Partitioning Richly-perfused tissue: blood Slowly-perfused tissue: blood Fat:blood Kidney:blood Liver:blood Permeability Slowly-perfused tissue Fat Molecular toxicity Total AhR in liver Basal CYP1A in liver AhR-dioxin binding affinity CYP1A-dioxin binding affinity DNA-dioxin binding affinity Basal synthesis rate Maximum fold in synthesis rate Hill coefficient Degradation rate Elimination Elimination rate
Unit
Value
Source
/d – L/d L/d
0.00076a–0.00194b 2.911a–3a 0.151c 2.4a–1.15b
[40] [40] [41] [16,41]
–
0.6
cm
8.65
M
8 106
Alometric equatione Alometric equationf [16]
L/kg.d
39366
Calculatedg
cm2/s
5.6 106
[19]
–
15
Calculatedg
–
3.9
– – –
133 5.3 2.6
Of tissue blood Of tissue blood
0.25
[8]
0.25
[10]
nM nmol/g nM
0.012 1.6 0.4
[7,28] [7] [28]
nM
9
[27]
nM
130
[7]
pmol/min/mg –
17 4.7
[25] [25]
– /h
0.6 0.035
[8] [27]
/d
0.12
[32]
2
a
Fathead minnow. Medaka. c Fathead minnow; assumed the same for medaka. e Qv ¼ Qv0 ðBWÞ0:75 where Qv0 is the reference ventilation (0.2 L/h for a 5 g fish for fathead minnow [41] and 0.014 L/h for a 202 mg fish for medaka [16]); and BW is body weight. f Agill ¼ 0:000865 ðBWÞ0:785. F g Fw tw Pt ¼ PPbw where Pbw orPtw ¼ 10logK ow 100f þ 1 þ 100 , KOW is octanolb
water partition coefficient (value from [15]), Ff is the sample’s nonpolar lipid content and Fw is the sample’s water content (values based on Lien et al [17] and Bertelsen et al [22]).
1066
Environ Toxicol Chem 33, 2014
Z. Parhizagari and J. Li
tissue in the compartment and the blood exiting it [9]. Tissues with blood flow limited flux include gills, liver, kidney, and other richly perfused tissues (red muscle, digestive tract, and spleen). In addition, the diffusion-limited flux can be represented by a tissue blood subcompartment, where transport of xenobiotics from tissue blood to tissue is proportional to a mass transfer coefficient [20]. Tissues with diffusion-limited flux include fat and other slowly perfused tissues (white muscle, fins, skin, and bones). Since most of dioxin congeners are highly fat soluble but practically water insoluble, they are preferentially retained in adipose (fatty) tissue [1], as well as the liver, where high concentration of dioxins have been observed. It has been suggested that at low doses (
Environmental Toxicology A PHYSIOLOGICALLY-BASED PHARMACOKINETIC MODEL FOR DISPOSITION OF 2,3,7,8-TCDD IN FATHEAD MINNOW AND MEDAKA ZAHRA PARHIZGARI* and JAMES LI Program of Environmental Applied Science and Management, Ryerson University, Toronto, ON, Canada (Submitted 5 April 2013; Returned for Revision 18 June 2013; Accepted 16 December 2013) Abstract: A physiologically-based pharmacokinetic model was developed for the disposition of 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) in 2 fish species—fathead minnow and medaka. The model was developed based on the empirical data on disposition of dioxins in fish tissues, as well as existing knowledge on the mechanisms of uptake, distribution, storage, and elimination of dioxins in various species (other than fish). The present study examined the applicability of mechanisms known to occur in other species for fish and concluded that the same mechanisms defined for disposition of 2,3,7,8-TCDD in (mostly) rodents can be applicable for the 2 fish species examined as well. Parameter values for the model were selected and/or calibrated using available databases. Model compartments included the gill, kidney, liver, and other richly-perfused tissues, as well as fat and other slowly-perfused tissues. The model was calibrated using 2 independent datasets for exposure of fathead minnow and medaka to 2,3,7,8-TCDD in water. The initial values of the model parameters were selected from several sources, and calibrated to represent the 2 exposure datasets. With very few exceptions, the estimated parameter values for the 2 species were comparable, and the final predictions were in strong agreement with the observations. The model developed in the present study can therefore be used in the prediction of the body burden of 2,3,7,8-TCDD in fathead minnow and medaka. Uncertainty in the model prediction as a result of variability in input parameters is discussed for the parameters with the highest impacts on the model outcome. Environ Toxicol Chem 2014;33:1064–1071. # 2013 SETAC Keywords: Dioxin
Pharmacokinetics
Fish
Aryl hydrocarbon
Dose-response
et al. [7] improved previous models by using time-course distribution data obtained from PBPK studies on rats. One of the main concerns in the assessment of exposure to dioxins and furans is their interactions with hepatic cytochrome P450 enzymes. PBPK models have successfully explained these interactions for a number of test animals (mostly rodents). They have modeled both the dioxin binding to the Ah receptor and the binding of the Ah receptor-dioxin complex to promoter sites on DNA. The latter form of binding can enhance the rate of gene transcription. These findings have been consistent across species, which increase the confidence in extrapolating the model between species [8]. Although the PBPK models have been successfully implemented to extrapolate kinetic data between mammals until early 1990s, the applications to fish were limited to models adapted from mouse models describing the kinetics of compounds injected intravenously [9]. Gill description and elimination mechanisms were included in later models. In 1990, Nichols et al. [9] developed a physiologically based toxicokinetic model for uptake and disposition of waterborne organic chemicals in fish. The model was parameterized using information from published data, and accurately simulated the uptake of pentachloroethane in rainbow trout. One study developed a PBPK model for disposition of dioxins in a fish species. Nichols et al. [10] proposed a model for maternal transfer of 2,3,7,8-TCDD in brook trout. The model considered dietary uptake and was evaluated through the comparison to measured data. This comparison suggested that whole body residue measurements could be used to estimate residues in developing ovaries within a factor of 2. The present study proposes a model for toxin accumulation of 2,3,7,8-TCDD in 2 fish species in the form of a PBPK model. The model considers the gill pathway and has been developed based on available information on the uptake, distribution, storage, and elimination mechanisms of dioxins in various
INTRODUCTION
Polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs)—commonly known as dioxins and furans—are toxic, persistent, and bioaccumulative chemicals that are predominantly a result of human activities such as the production of chemicals and the metallurgical industry. The most potent congener in the series is 2,3,7,8-tetrachlorodibenzop-dioxin (2,3,7,8-TCDD). The long half-life of dioxins means that even a low rate of exposure will lead to the accumulation of dioxins in the body and continual long-term exposure might result in an extremely high body burden. The biochemical and toxicological effects of dioxins and furans are more closely related to their concentration in the target tissue than to the daily dose [1,2]. Evaluation of effective or tissue dose under a variety of conditions is possible using the understanding of xenobiotic modes of actions, which provides insight into both the sites/ mechanisms of action and the xenobiotic form causing the response. Physiologically based pharmacokinetic (PBPK) models can describe chemical disposition in the body based on the fundamental information on physicochemical properties, transport, metabolism, and various excretory mechanisms [3], and the predictive capabilities of PBPK models make it possible to estimate the behavior of chemicals in species over time without the need for direct experimentation, by using the knowledge of more fundamental physiological and biochemical processes [4]. Several authors have developed PBPK models for disposition of dioxins and furans in rodents [5–8]. Early PBPK models examined the disposition of PCDFs (and PCBs) by taking into account the role of lipid solubility in tissue uptake. Wang * Address correspondence to [email protected]. Published online 25 December 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2504 1064
A PBPK Model for Disposition of 2,3,7,8-TCDD in Fathead Minnow and Medaka
species (other than fish). In the present study, model parameters not available in the literature were calibrated using available measurement data. MATERIALS AND METHODS
Model development Exposure pathways. The relative contributions of uptake from food and water have been a subject of discussion. Loonen et al. [11] studied the total bioaccumulation of PCDDs and PCDFs in guppies and concluded that the uptake from food was small compared with the uptake from water (0.9–4.9%). Rifkin and Lakind [12] reported that fish accumulate dioxins by ingestion (biomagnifications) rather than water (bioconcentration). Karl et al. [13] found a correlation between the dioxin concentration of commercial feed and the resulting concentration in the rainbow trout muscle fat. Based on Randall et al. [14], the uptake of persistent lipophilic toxicants in fish occurs by transfer across the gills. They argue that flux across the gills is rapid and in order for feeding to have a significant effect on toxicant concentration in the body, the fish must eat at quantities much higher than those typical of fish. A review of available studies on the relative importance of contaminant uptake from water versus food by Nowell et al. [15] revealed that most of the studies observed greater accumulations from water as opposed to food. Uptake. The movement of water and blood through the gill results in a counter-current exchange between the 2. The rate of uptake of a chemical by the gill is defined by uptake clearance, which is proportional to the concentration difference between the external water and the blood plasma water. Nichols et al. [9], in their physiologically based toxicokinetic model for uptake of organic chemicals by fish, used a general expression for chemical flux across gills and related it to an exchange coefficient, concentration gradient between the incoming water and the venous blood, and a blood to water partition coefficient. Lien and McKim [16] incorporated the diffusion limitation into a more detailed gill model and suggested a numerical solution for the exchange coefficient through parameters such as molecular diffusivity (or the gill permeability coefficient) for specific chemicals, total lamellar surface area, and the average thickness of the diffusion path length. Lien et al. [17] proposed a model for countercurrent exchange of chemicals at the gill surface, considering the free concentrations in water and blood. They calculated an exchange coefficient for 3 chloroethanes in a way that it represented limitations by 3 barriers including flow of water and blood, diffusion barriers, and blood-water partitioning. Another parameter taken into account is the binding to organic carbon in water, which reduces the free chemical concentration that is available to diffuse across the gills. Arnot and Gobas [18] related the dissolved (bioavailable) concentration of chemicals in water to the dissolved organic carbon, particulate organic carbon, and the octanol:water partition coefficient of the chemical. In the present study, uptake through the gill is controlled by partitioning between blood and water, and the gill permeability for dioxin, also called the molecular diffusivity of dioxins. The model is as follows
K xg C vb C b ¼ C vb þ C wd QT Pbw
ð1Þ
where Cb is blood concentration, Cvb is venous blood concentration, kx-g is an exchange coefficient, QT is the cardiac
Environ Toxicol Chem 33, 2014
1065
output, Cwd is the dissolved concentration in water, and Pbw is blood-to-water partition coefficient. The exchange coefficient was calculated using a diffusion rate constant for bronchial flux, with an initial value selected at 5.6 106 cm2/s based on the molecular diffusivity of 2,3,7,8-TCDD in water [19]. A complete list of selected initial values is provided in Table 1. Distribution and bioaccumulation. In terms of the arterial blood flow into the organ, fish tissues can be classified into richly perfused and slowly perfused tissues. Chemical flux in the tissue can be blood-flow limited or diffusion limited. When the flux is limited by blood flow, a chemical equilibrium exists between the Table 1. Initial values for model parameter Parameter Physiological Growth coefficient Power term in growth Cardiac output coefficient Ventilation volume coefficient Effective ventilation Gill surface area coefficient Average thickness of diffusion path Uptake Blood-water partitioning (gills) Gill permeability for dioxin Partitioning Richly-perfused tissue: blood Slowly-perfused tissue: blood Fat:blood Kidney:blood Liver:blood Permeability Slowly-perfused tissue Fat Molecular toxicity Total AhR in liver Basal CYP1A in liver AhR-dioxin binding affinity CYP1A-dioxin binding affinity DNA-dioxin binding affinity Basal synthesis rate Maximum fold in synthesis rate Hill coefficient Degradation rate Elimination Elimination rate
Unit
Value
Source
/d – L/d L/d
0.00076a–0.00194b 2.911a–3a 0.151c 2.4a–1.15b
[40] [40] [41] [16,41]
–
0.6
cm
8.65
M
8 106
Alometric equatione Alometric equationf [16]
L/kg.d
39366
Calculatedg
cm2/s
5.6 106
[19]
–
15
Calculatedg
–
3.9
– – –
133 5.3 2.6
Of tissue blood Of tissue blood
0.25
[8]
0.25
[10]
nM nmol/g nM
0.012 1.6 0.4
[7,28] [7] [28]
nM
9
[27]
nM
130
[7]
pmol/min/mg –
17 4.7
[25] [25]
– /h
0.6 0.035
[8] [27]
/d
0.12
[32]
2
a
Fathead minnow. Medaka. c Fathead minnow; assumed the same for medaka. e Qv ¼ Qv0 ðBWÞ0:75 where Qv0 is the reference ventilation (0.2 L/h for a 5 g fish for fathead minnow [41] and 0.014 L/h for a 202 mg fish for medaka [16]); and BW is body weight. f Agill ¼ 0:000865 ðBWÞ0:785. F g Fw tw Pt ¼ PPbw where Pbw orPtw ¼ 10logK ow 100f þ 1 þ 100 , KOW is octanolb
water partition coefficient (value from [15]), Ff is the sample’s nonpolar lipid content and Fw is the sample’s water content (values based on Lien et al [17] and Bertelsen et al [22]).
1066
Environ Toxicol Chem 33, 2014
Z. Parhizagari and J. Li
tissue in the compartment and the blood exiting it [9]. Tissues with blood flow limited flux include gills, liver, kidney, and other richly perfused tissues (red muscle, digestive tract, and spleen). In addition, the diffusion-limited flux can be represented by a tissue blood subcompartment, where transport of xenobiotics from tissue blood to tissue is proportional to a mass transfer coefficient [20]. Tissues with diffusion-limited flux include fat and other slowly perfused tissues (white muscle, fins, skin, and bones). Since most of dioxin congeners are highly fat soluble but practically water insoluble, they are preferentially retained in adipose (fatty) tissue [1], as well as the liver, where high concentration of dioxins have been observed. It has been suggested that at low doses (
