Quantitative low-dose assessment of seafood toxin, domoic acid, in the rat brain: application of physiologically-based pharmacokinetic (PBPK) modeling1

Environmental Toxicology and Pharmacology 6 (1998) 49 – 58

Quantitative low-dose assessment of seafood toxin, domoic acid, in the rat brain: application of physiologically-based pharmacokinetic (PBPK) modeling1 Chung S. Kim a,*, Ivan A. Ross a, Jennifer A. Sandberg 2,b, Edward Preston c a

Di6ision of Toxicological Research (HFS-506), Center for Food Safety and Applied Nutrition, Food and Drug Administration, Washington, DC 20204, USA and Chemical Industry Institute of Toxicology, Research Triangle Park, NC 27709, USA b Di6ision of Neurotoxicology (HFT-132), National Center for Toxicological Research, Jefferson, AR 72079, USA c Institute for Biological Sciences, National Research Council of Canada, Ottawa, Ontario KIA OR6, Canada Received 5 December 1997; received in revised form 4 March 1998; accepted 9 March 1998

Abstract The purpose of this study was to construct a physiologically based pharmacokinetic model and demonstrate its ability to predict low-dose uptake of domoic acid, a seafood contaminant, in discrete areas of the rat brain. The model we used was derived from the generic PBPK model of our previous studies with 2,4-dichlorophenoxyacetic acid (Kim et al., 1994. Pharmacokinetic modeling of 2,4-dichlorophenoxyacetic acid (2,4-D) in rats and in rabbits brain following single dose administration. Toxicol. Lett. 74, 189; Kim et al., 1995. Development of a physiologically based pharmacokinetic model for 2,4-dichlorophenoxyacetic acid dosimetry in discrete areas of the rabbit brain. Neurotoxicol. Teratol. 17, 111), to which physiological- and chemical-specific parameters for domoic acid were applied. It incorporates two body compartments along with compartments for venous and arterial blood, cerebrospinal fluid, brain plasma and seven brain regions. Uptake of the blood-borne toxin is membrane-limited by the blood-brain barrier with clearance from the brain provided by cerebrospinal fluid ‘sink’ mechanisms. This model generated predicted profiles of toxin level in brain and blood over a 1-h period that compared reasonably well with concentrations calculated from in vivo data of rats that had been given [3H]domoic acid intravenously (Preston and Hynie, 1991. Transfer constants for blood–brain barrier permeation of the neuroexcitatory shellfish toxin, domoic acid. Can. J. Neurol. Sci. 18, 39). This PBPK model should be an effective tool for evaluating the target doses that produce the potential neurotoxicity of domoic acid found in foods. © 1998 Elsevier Science B.V. All rights reserved. Keywords: Pharmacokinetic modeling; Domoic acid; Neurotoxicity risk assessment

1. Introduction

* Corresponding author. Division of Toxicological Research (HFS506), Center for Food Safety and Applied Nutrition, Food and Drug Administration, Washington, DC 20204, USA. Tel.: + 1 301 5945004; fax: + 1 301 5941426. 1 A portion of this research was carried out while the senior author was a visiting scientist at the Chemical Industry Institute of Toxicology. 2 Present address: Ribozyme Pharmaceutical, Inc., 2950 Wilderness Place, Boulder, CO 80301, USA. 1382-6689/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S1382-6689(98)00019-2

A seafood contaminant domoic acid was found in cultured blue mussels from a localized area of eastern Prince Edward Island in Canada in the late 1980s and has caused incidents of severe neurotoxicity and death in humans (Iverson et al., 1989; Quilliam and Wright, 1989; Work et al., 1991). Domoic acid is a naturally occurring neurotoxin that is structurally similar to kainic acid (Robertson et al., 1991) and acts as a potent glutamate receptor agonist capable of producing excitotoxic effects (Collins, 1987). Neurological manifesta-

50

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

tions in humans include vertigo, ataxia, confusion, disorientation and short-term memory loss (Perl et al., 1990), and seizures and scratching behavior in rats (Robertson et al., 1991; Truelove and Iverson, 1994). Physiologically-based pharmacokinetic (PBPK) models are considered as potentially useful for toxicity risk assessment of food contaminants and other chemicals, in that they can be used to characterize pharmacokinetic and pharmacodynamic information obtained in animals and extrapolate these to humans. In an earlier study we developed a PBPK model to describe the brain uptake of blood-borne 2,4-dichlorophenoxyacetic acid (2,4-D) in the rabbit (Kim et al., 1995). The magnitudes of transfer constants for blood to brain passage of this substance and those for domoic acid (Preston and Hynie, 1991) indicate that the latter penetrates much more slowly across the mammalian blood– brain barrier. We thought it would be of value therefore to test the ability of the PBPK model to simulate the disposition of intravenously-delivered domoic acid in the rat, as it compares to the calculated values of parenchyma from previous in vivo study (Preston and Hynie, 1991).

2. Materials and methods

2.1. Model parameters Physiological-, biochemical- and chemical-specific information were taken in part from the literature as detailed in the footnotes to the Tables, estimated from the best fit of the model to in vivo data, or based on in vitro measurements as described below. Mass balance differential equations used to describe each body compartment were numerically integrated by using SimuSolv software (Dow Chemical, Midland, MI), and the model simulations were compared with experimental data. A more detailed description for the relevant mass-balance differential equations underlying the model is provided in the Appendix A.

2.2. Measurement of domoic acid tissue partition coefficients The brain tissue partitioning of domoic acid in both artificial cerebrospinal fluid (CSF) (Kim et al., 1980) and Krebs phosphate buffer were measured by using the ratio of in vitro tissue to medium at the steady state as previously described (Kim et al., 1996b). Briefly, brain slices were prepared and incubated in artificial CSF or Krebs phosphate buffer containing 1 mM domoic acid. Incubation vials were agitated at 37°C until steady-state uptake was reached (45 – 60 min). At the end of incubation, brain tissue slices were processed for domoic acid measurement as described below. The

domoic acid brain/blood or brain/CSF partition coefficients were estimated by dividing brain tissue concentration by incubation medium concentration in Krebs phosphate buffer or artificial CSF. The brain/Krebs phosphate buffer domoic acid partition coefficients were then divided by the brain/CSF partition coefficient to estimate the CSF/blood partition coefficient. The value of PB (Table 3) was estimated by fitting this model to the data sets. The details for the chemical analyses for the tissue partitioning are described as follows:

2.2.1. Chemicals Domoic acid purified by ion exchange chromatography of aqueous extracts of contaminated anchovies collected by the Food and Drug Administration was used for the measurement of domoic acid tissue partitioning. The purity of the dosing solution was assessed by HPLC and mass spectrometry and found to be greater than 97%. Proton and carbon NMR and combustion analysis indicated purity greater than 99%. Additional domoic acid for HPLC standards, dihydrokainic acid (DHKA), and 9-fluorenylmethyl chloroformate (FMOC-CL) were obtained from Sigma (St. Louis, MO). All other chemicals were HPLC or reagent grade. 2.2.2. Sample preparation Varian 50 mg C18 BondElute columns were prepared by rinsing columns with 1 ml MeOH, 1 ml H2O, and 1 ml 0.01 M NaH2PO4 (pH 2.1). Incubation medium samples (25 ml) were mixed with internal standard 25 ml DHKA (5 mg/ml), 150 ml H2O, and then acidified with 5 ml concentrated HCl. After vortexing, aliquots were drawn through a C18 BondElute column. Column effluents were rinsed with 1 ml 0.01 M NaH2PO4 (pH 2.1) buffer followed by 100 ml n-hexane and then eluted with 250 ml of 2% acetic acid (v/v) in MeOH. The MeOH was evaporated at 45°C under nitrogen, and the residues were reconstituted in 200 ml H2O. Brain tissues were homogenized in 10× volume of 3% trichloroacetic acid (v/v). Then, 200 ml of the brain homogenate was mixed with 10 ml DHKA and extracted as described above for incubation medium. Brain extracts were reconstituted in 200 ml H2O. 2.2.3. Deri6atization After reconstitution, analytes were derivatized according to the method of Milley et al. (1990). Briefly, 50 ml of 1 M borate buffer (pH 7.5) was added to the reconstituted incubation medium or brain extract and 250 ml of 15 mM FMOC-CL in acetonitrile was added. After 45 s of incubation, 3× 500 ml ethyl acetate washes were added to remove excess reagent. An aliquot of the lower aqueous layer was removed, filtered through a 0.45 mm nylon Micro-Spin filter

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

51

Fig. 1. Schematic of the physiologically based pharmacokinetic model used for domoic acid. The PBPK model consists of brain, body, and venous and arterial compartments. The brain compartment consists of brain plasma, brain tissue, and CSF. The brain tissue is further divided into seven different subcompartments including the frontal cortex (FC), striatum (ST), hippocampus (HIP), occipital cortex (OC), diencephalon-mesencephalon (DM), pons-medulla (PM), and cerebellum (CB). Each compartment is defined by a volume, a blood flow, and a partition coefficient. Domoic acid enters the body intravenously (IV), and is eliminated by renal excretion from the central compartment. Domoic acid enters the brain from blood by diffusion across endothelial membranes and is cleared by CSF sink mechanism including choroidal excretion.

(LabSource, Janesville, WI), and 10 ml injected onto the HPLC system.

2.2.4. HPLC analysis The chromatographic system consisted of a Waters model M-6000A pump, model 474 scanning fluorescence detector (wavelengths on the fluorescence detector were set at lex = 264 nm and lem =313 nm), and a model 746 data module integrator (Waters Associates, Milford, MA). Separation was achieved by using a 5 mm, 250× 4.6 mm Supelcosil LC-18 (Supelco, Belle-

fonte, PA), preceded by a C-130B 20× 2 mm guard column (Upchurch Scientific, Oak Harbor, WA) packed with Bondapak C18/CORASIL, 37–50 mm particle size (Waters Associates, Milford, MA). The mobile phase (38% CH3CN, v/v, in 0.01 M NaH2PO4, pH 2.1) was run at a flow rate of 1.6 ml/min.

2.3. Determination of domoic acid kinetic constants The domoic acid clearance kinetic constants (Km1 and Vmax1) for choroid plexus (Table 3) were initially

52

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

Table 1 Abbreviations and symbols used in the PBPK model QBr QB CPLa CPLv CVBrP CVB QC Km1 Vmax1 Km2 Vmax2 D1 D2 K12/K21

Table 3 Chemical-specific parameters used in the PBPK model

Blood flow to the brain (l/h) Blood flow to the body (l/h) Concentration in the arterial plasma (mg/l) Concentration in the venous plasma (mg/l) Concentration in the venous plasma leaving the brain plasma (mg/l) Concentration in the venous plasma leaving the body (mg/l) Cardiac output (l/h) Michaelis-Menten constant for saturable clearance by choroid plexus (mg/l) Maximum velocity of saturable clearance by choroid plexus (mg/h) Michaelis-Menten constant for saturable binding by plasma protein (mg/l) Maximum velocity of saturable binding by plasma protein (mg/h) Transfer constant between brain plasma and brain tissue (l/h) Transfer constant between brain tissue and CSF (l/h) Transfer constant between body and deep compartment (l/h)

used with published values for glutamate (Kim et al., 1996a), and domoic acid clearance rate constant (CLr) for kidney (Table 3) was obtained from the literature (Suzuki and Hierlihy, 1993). These values were converted to units relevant to the model. The values of Km2 and Vmax2 (Table 3) were estimated by fitting this model to the data sets. Transfer rate constants for D1 and D2 were originally used with previously published values of Ki for domoic acid (Preston and Hynie, 1991) and D2 for 2,4-D (Kim et al., 1995), respectively, and reestimated by optimization against in vivo experimental data sets to give best-fit simulations of domoic acid Table 2 Physiological parameters used in the PBPK model 0.35a 14b,c 0.005a 0.845a

Body weight (kg) Cardiac output (l blood/h/kg) Fraction brain tissue (brain weight/BW) Fraction rest of the body (body without brain/ BW) Fraction brain plasma (l brain plasma/BW) Fraction CSF (l CSF/BW)

BW QCC VBrC VBC

Fractional plasma flow to brain (l plasma/h/ QC) Hematocrit Fraction plasma (l plasma/BW) Fraction plasma flow to body (l plasma/h/QC) Fraction arterial plasma Fraction venous plasma Fraction blood volume (l blood/BW)

QBrC

0.02d

HCT VPLC QBC VPLaC VPLvC B VC

0.45d 0.04d 0.98d 0.35d 0.65d 0.06d

a

VBrPC 0.0008d VCSFC 0.0017d

Determined in this study. Stott et al. (1982), Delp et al. (1991). c Arms and Travis (1988), Ramsey and Andersen (1984). d Kim et al. (1994, 1995). b

Brain/plasma partition coefficient Brain plasma/plasma partition coefficient CSF/plasma partition coefficient Body/plasma partition coefficient Molecular weight Maximum velocity of saturable clearance by choroid plexus (mg/h) Maximum velocity of saturable binding by plasma protein (mg/h) Michaelis-Menten constant for saturable clearance by choroid plexus (mg/l) Michaelis-Menten constant for saturable binding by plasma protein (mg/l) Transfer rate constant between body and deep compartment (l/h) Transfer rate constant between deep compartment and body (l/h) Renal clearance rate (l/h/kg)

PBr PBrP PCSF PB MW Vmax1

0.45a 1.0a 1.0a 0.226b,d 311 3.4c

Vmax2 9.23d Km1

39c

Km2

4.17d

K12

0.017b

K21

0.05b

CLr

0.55e

a

Determined in this study. Kim et al. (1995). c Kim et al. (1996a). d Adjusted in this study. e Suzuki and Hierlihy (1993). b

disposition over time. Estimates of rate constants were obtained by iterative fitting of the data; the formal optimization was conducted by maximization of the log likelihood function by the Nelder-Meade Search optimization method using the Simusolv software (Dow Chemical, Midland, MI).

3. Results

3.1. In 6i6o study Temporal profiles of domoic acid level in brain parenchyma and plasma, as predicted by the computer model, were compared with published and auxiliary data taken from an earlier study in rats (Preston and Hynie, 1991). Pentobarbital-anesthetized Sprague-DawTable 4 Mass transfer rate constants in discrete areas of the rat brain

Frontal cortex Striatum Hippocampus Occipital cortex Diencephalon-mesencephalon Pons-medulla Cerebellum

D1 (×10−6)

D2 (×10−4)

3.7 1.6 1.2 1.4 2.0 2.0 2.8

1.0 0.8 0.2 0.2 0.2 0.2 1.0

Initial estimates of D1 and D2 were obtained from the previously published data and were adjusted in fitting the data sets here. D1, mass transfer constant between brain plasma and brain (l/h); D2, mass transfer constant between brain and CSF (l/h).

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

53

Fig. 2. Final plasma concentrations of domoic acid measured in anesthetized rats which received an i.v. dose of 7 mg/kg at time zero, and were decapitated at 3, 10, 20, 30 or 60 min after injection (Preston and Hynie, 1991). Symbols are for individual rats except for 1-h point which is the mean from four rats (range: 3.3–5.4 ng/ml). The line is a computer simulation.

ley rats (male, 312– 402 g) had received an i.v. dose of 7 mg/kg [3H]domoic acid of known specific activity. In one experiment ten rats had been allotted tracer circulation times of 3, 10, 20, or 30 min before decapitation and measurements obtained of the plasma concentration-time integral, final plasma level and brain concentration (inclusive of intravascular tracer) in seven anatomical regions. Multiple-time graphical analysis (Patlak et al., 1983) of the regional increase in tracer distribution volume with circulation time had been used to publish transfer constants (Kis) for regional BBB passage (Preston and Hynie, 1991). For the present study, we entered these Kis in the equation of Ohno et al. (1978) to estimate the regional parenchymal concentrations (tissue counts devoid of intravascular tracer) achieved in individual rats as follows: parenchymal concentration=Ki × plasma concentration −time integral. In the second experiment (Preston and Hynie, 1991), the [3H]domoic acid had been allowed to circulate 60 min and intravascular tracer had been perfusion-cleared from brain before decapitation (n = 4 rats) allowing for a direct measurement of parenchymal (extravasated) tracer. Mean regional parenchymal levels in these 60-min rats were also used with the above 3–30 min estimates to plot a profile of brain domoic acid uptake with time. From all of the foregoing experiments, values for final plasma concentration of individual rats at the time of sacrifice provided a temporal profile of toxin level that would be expected in plasma over a 1-h circulation period.

3.2. Description of the PBPK model Fig. 1 shows the PBPK model for disposition of domoic acid as derived from the membrane-limited brain model developed earlier for organic anionic toxicants and specifically 2,4-D. The abbreviations and symbols for brain and body compartments are explained in Table 1. The transport of domoic acid from plasma into brain is conceived as a membrane-limited process; transport across the choroidal epithelial cell membranes from CSF to the blood occurs by a saturable mechanism described by Michaelis-Menten kinetics. Domoic acid is eliminated from the plasma compartment primarily by urinary elimination; metabolic elimination from the central compartment is negligible (Suzuki and Hierlihy, 1993). The renal excretory mechanism involves only glomerular filtration without tubular reabsorption or secretion processes (Suzuki and Hierlihy, 1993). Tables 2 and 3 summarize physiological-, biochemical- and chemical-specific information that were incorporated into the model. Organ volumes and blood flows are expressed as a fraction of body weight or cardiac output as based on previous studies (see footnotes). Table 4 shows the values for D1 and D2, the mass transfer rate constants which characterize bloodto-brain and brain-to-CSF diffusion of toxin. Figs. 2 and 3(A–G) show the output of the computer model as it compares with in vivo data. Fig. 2 shows the profile of domoic acid level in plasma as was

54

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

Fig. 3. (A – G) Regional concentrations of domoic acid accumulated in brain parenchyma (extravasated tracer) of rats sacrificed at 3, 10, 20, 30 or 60 min after an i.v. injection of 7 mg/kg at time zero. Values at 1 h are the mean from four rats in which intravascular tracer was perfusion cleared from brain before measurement (Preston and Hynie, 1991). Points plotted at other times are from individual rats and calculated from measurements of the transfer constant for BBB passage (Ki) and the time integrated plasma levels: parenchymal uptake = Ki ×plasma integral (Ohno et al., 1978). The lines are computer simulations based on the PBPK model.

expected after an intravenous bolus injection (7 mg/kg). The falling curve initially reflects the distribution to various compartments. Levelling off and steady decline thereafter reflect the equilibration between toxin level in these compartments, toxin in plasma and steady elimination of the latter provided by kidney excretion. Fig. 3(A –G) shows the computer model simulation of domoic acid concentrations in discrete areas of brain parenchyma with time, as might be anticipated to result from the plasma toxin profile just described. These curves compare reasonably well with data calculated

from in vivo measurements of domoic acid in the rat brain.

4. Discussion The PBPK modeling should serve as an important tool for improving the accuracy of chemical risk assessment. As shown with domoic acid in this study, the model can closely simulate the minute amount of toxin entering brain with time, even though the BBB passage

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

55

Fig. 3. (Continued)

is limited. This PBPK model (Fig. 1) was derived from the previous model (Kim et al., 1995) that specifically describes the pharmacokinetics and brain uptake of systemically delivered 2,4-D, an organic acid. Specific chemical parameters for domoic acid were applied to this generic model. Probably the major difference with the present study is the fact that domoic acid permeation of the BBB is very limited as compared to 2,4-D. This is reflected by the magnitudes of mass transfer constants for blood to brain movement (Table 4). Values for D1 were initially calculated from the K1 values (Preston and Hynie, 1991) and subsequently re-estimated by optimization against an in vivo kinetic data set from the rat to give

best-fit simulations of domoic acid disposition over time. The D1 values (Table 4) were over three orders of magnitude lower than those reported for 2,4-D (Kim et al., 1995). Although domoic acid is in fact comparable to sucrose in its limited penetrability across the blood– brain barrier, the ability to predict its pharmacokinetics is a valid concern given that the BBB is not an absolute barrier, and the toxin is very potent. The PBPK model also incorporated the features of brain-to-CSF diffusion of domoic acid, the so-called CSF ‘sink’ effect, including CSF to blood clearance of toxin by the choroid plexus organic anion pump. While we have no direct measurements of choroidal transport of domoic acid, its incorporation in the model has

56

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

Fig. 3. (Continued)

indirect support. For example, it was shown that pretreatment with probenecid, which inhibits the organic acid transport by the choroid plexus (Kim and Pritchard, 1993) greatly increased the expression of c-fos in the rat brain by domoic acid as compared to rats without pretreatment, presumably due to increased toxin level in the brain (Robertson et al., 1991). In considering these estimates of parenchymal tracer concentration (pg/g brain) 3 – 60 min after a given tracer dose, allowance should be made that this extravasated toxin may not penetrate appreciably into brain cells and be primarily distributed in the extracellular compartment. Assuming distribution, for example, in an extracellular fluid space of 150 ml/g brain, the toxin concentration affecting neuronal membranes

would be approximately 7-fold higher than those estimated in Fig. 3(A–G). This would be an important consideration in evaluating the significance of such parenchymal estimates since the primary site of action (membrane glutamate receptors of the kainate variety) would be accessed from the extracellular compartment. The suggestion of limited cell membrane permeation and primarily an extracellular distribution would be in keeping with the fact that the toxin’s passage across BBB endothelial cells is very poor, characterized by transfer constants little different from those for sucrose (Preston and Hynie, 1991). In this model, the renal clearance of toxin was featured as the sole source for its elimination. This is supported by the fact that total clearance of domoic

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

57

Fig. 3. (Continued)

acid is not significantly different from the urinary clearance (Suzuki and Hierlihy, 1993). The dominant role of clearance by the renal glomerular filtration is further suggested by the fact its total clearance is comparable to that of inulin, less than that of para-aminohippuric acid (PAH) and is not affected by probenecid, indicated absence of saturable tubular reabsorption or secretion mechanisms. For these reasons, an equation for the clearance of domoic acid is as glomerular filtration from the central plasma compartment. The correlation between the PBPK-simulated and calculated values of parenchyma from in vivo study for domoic acid uptake by the rat brain indicates that the present model provides a reliable description of low-dose domoic acid pharmacokinetics to describe the relationship between i.v. dosage, plasma profile and brain parenchymal uptake. Unlike the findings with imaging technology in which no water-soluble isotope is found in the brain tissue (Wilcox et al., 1984), the computer modeling is sensitive enough to trace reliably a low dose of water-soluble domoic acid that penetrates into the brain over time. This PBPK model should be an effective tool for evaluating the target doses that produce the potential neurotoxicity of domoic acid found in foods.

Acknowledgements This article was presented in part at the Annual Meeting of the American Society for Pharmacology and Experimental Therapeutics, San Diego, CA, 1997. Special thanks are due to Drs David Dorman, Paul M. Schlosser and Roger O. McClellan for their helpful suggestions and support.

Appendix A The rate of change in the amount of domoic acid in the brain tissue (dABr/dt) was described by a generalized mass balance differential equation of the following form for the brain subcompartments: dABr/dt =D1(CVBrP − CBr/PBr) +D2(CCSF/PCSF − CBr/PBr) where D1 = Transfer rate constant between brain plasma and brain tissue (l/h) D2 = Transfer rate constant between brain tissue and CSF (l/h) CVBrP = Concentration of domoic acid in the venous plasma leaving the brain plasma (mg/l) CBr = Concentration of domoic acid in the brain (mg/kg) PBr = Brain/plasma partition coefficient CCSF = Concentration of domoic acid in the CSF (mg/l) PCSF = CSF/plasma partition coefficient The rate of change in the amount of domoic acid in the rest of the tissue compartment (dAi/dt) was described by a generalized mass balance equation of the following form: dAi/dt = Qi(Ca − Cvi) where Qi = Rate of blood flow to tissue i (l/h) Ca = Concentration of domoic acid in the arterial plasma (mg/l)

58

C.S. Kim et al. / En6ironmental Toxicology and Pharmacology 6 (1998) 49–58

Cvi =Concentration of domoic acid in the venous plasma leaving the tissue i (mg/l) The rate of the amount of domoic acid cleared per unit time from the CSF or by plasma protein binding was described by generalized mass balance equation of the following form: dAcl/dt = VmaxC/(Km +C) where Vmax =maximum transport velocity of a saturable clearance or binding by choroid plexus or plasma protein (mg/h) Km = Michaelis-Menten constant for a saturable clearance or binding by the choroid plexus or plasma protein (mg/l) C =Concentration of domoic acid in the CSF or plasma (mg/l) The rate of the amount of domoic acid cleared per unit time from the central compartment was described as follows: dAclt/dt = CLrCPLv where CLr =Total plasma clearance rate by kidney (l/h/kg) CPLv = Concentration of domoic acid in the venous plasma (mg/l)

References Arms, A.D., Travis, C.C., 1988. Reference Physiological Parameters in Pharmacokinetic Modeling. Office of Health and Environmental Assessment. US Environmental Protection Agency, Washington, DC, NTIS PB88-196019. Collins, R.C., 1987. Neurotoxins and the selective vulnerability of brain. In: P. Jenner (Ed.), Neurotoxins and Their Pharmacological Implications, Raven Press, New York, p. 1. Delp, M.D., Manning, R.O., Bruckner, J.V., Armstrong, R.B., 1991. Distribution of cardiac output during diurnal changes of activity in rats. Am. J. Physiol. 261, H1487. Iverson, F., Truelove, J., Nera, E., Tryphonas, L., Campbell, J., Lok, E., 1989. Domoic acid poisoning and mussel-associated intoxication: Preliminary investigations into the response of mice and rats to toxic mussel extract. Food Chem. Toxicol. 27, 377. Kim, C.S., Pritchard, J.B., 1993. Transport of 2,4,5-trichlorophenoxyacetic acid across the blood-cerebrospinal fluid barrier of the rabbit. J. Pharmacol. Exp. Ther. 267, 751. Kim, C.S., O’Tuama, L.A., Cookson, S.L., Mann, J.D., 1980. The effects of lead poisoning on calcium transport by brain in 30-day old albino rabbits. Toxicol. Appl. Pharmacol. 52, 491. Kim, C.S., Gargas, M.L., Andersen, M.E., 1994. Pharmacokinetic modeling of 2,4-dichlorophenoxyacetic acid (2,4-D) in rats and in

.

rabbits brain following single dose administration. Toxicol. Lett. 74, 189. Kim, C.S., Slikker, W. Jr., Binienda, Z., Gargas, M.L., Andersen, M.E., 1995. Development of a physiologically based pharmacokinetic model for 2,4-dichlorophenoxyacetic acid dosimetry in discrete areas of the rabbit brain. Neurotoxicol. Teratol. 17, 111. Kim, C.S., Virella, A., Braunberg, R.C., Ross, I.A., Matthews, R.N., Johnson, W., Friedman, L., 1996a. Kinetic analysis of glutamate transport by the miniswine choroid plexus in vitro. Brain Res. 709, 59. Kim, C.S., Binienda, Z., Sandberg, J.A., 1996b. Construction of a physiologically based pharmacokinetic model for 2,4-dichlorophenoxyacetic acid dosimetry in the developing rabbit brain. Toxicol. Appl. Pharmacol. 136, 250. Milley, J.E., Bates, S.S., Bird, C.J., De Freitas, A.S.W., Quilliam, M.A., 1990. Trace determination of domoic acid in seawater and phytoplankton by high-performance liquid chromatography of the fluorenylmethoxycarbonyl (FMOC) derivative. Int. J. Environ. Anal. Chem. 38, 351. Ohno, K., Pettigrew, K.D., Rapoport, S.I., 1978. Lower limits of cerebrovascular permeability to nonelectrolytes in the conscious rat. Am. J. Physiol. 235, 299. Patlak, C.S., Blasberg, R.G., Fenstermacher, J.D., 1983. Graphical evaluation of blood to brain transfer constants from multiple-time uptake data. J. Cereb. Blood Flow Metab. 3, 1. Perl, T.M., Bedard, L., Kosatsky, T., Hockin, J.C., Todd, E.C.D., Remis, R.S., 1990. An outbreak of toxic encephalopathy caused by eating mussels contaminated with domoic acid. N. Engl. J. Med. 322, 1775. Preston, E., Hynie, I., 1991. Transfer constants for blood-brain barrier permeation of the neuroexcitatory shellfish toxin, domoic acid. Can. J. Neurol. Sci. 18, 39. Quilliam, M.A., Wright, J.L.C., 1989. The amnesic shellfish poisoning mystery. Anal. Chem. 61, 1053A. Ramsey, J.C., Andersen, M.E., 1984. A physiologically based description of the inhalation pharmacokinetics of styrene in rats and humans. Toxicol. Appl. Pharmacol. 73, 159. Robertson, H., Renton, K., Kohn, J., White, T., 1991. Patterns of fos expression suggest similar mechanisms of action for the excitotoxins domoic acid and kainic acid. In: Langston, J.W., Young, A.B. (Eds.), Neurotoxins and Neurodegenerative Disease, vol. 648. Annals of the New York Academy of Sciences, p. 330. Stott, W.T., Quast, J.F., Watanabe, P.G., 1982. The pharmacokinetics and macromolecular interactions of trichloroethylene in mice and rats. Toxicol. Appl. Pharmacol. 62, 137. Suzuki, C.A.M., Hierlihy, S.L., 1993. Renal clearance of domoic acid in the rat. Food Chem. Toxicol. 31, 701. Truelove, J., Iverson, F., 1994. Serum domoic acid clearance and clinical observations in the cynomolgus monkey and SpragueDawley rat following a single i.v. dose. Bull. Environ. Contam. Toxicol. 52, 479. Wilcox, J., Sage, M.R., Evill, C.A., 1984. Effect of intravenous contrast material on the integrity of the blood-brain barrier: Experimental study. Am. J. Neuroradiol. 5, 41. Work, T.M., Beale, A.M., Fritz, L., Quilliam, M.A., Silver, M., Buck, K., Wight, J.L.C., 1991. Domoic acid intoxication of brown pelicans (Pelecanus occidentalis) in California. 5th Int. Conf. Toxic Marine Phytoplankton. Newport, Rhode Island, Oct. 28 – Nov.1, pp. 33 [abstract].