ToxSci Advance Access originally published online on June 12, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Toxicological Sciences 74, 416-436 (2003)
Copyright © 2003 by the Society of Toxicology
REPRODUCTIVE AND DEVELOPMENTAL TOXICOLOGY |
Predicting Neonatal Perchlorate Dose and Inhibition of Iodide Uptake in the Rat during Lactation Using Physiologically-Based Pharmacokinetic Modeling
,2
* Geo-Centers, Inc., Wright-Patterson AFB, Ohio 45433;
AFRL/HEST, Wright-Patterson AFB, Ohio 45433;
Mantech Environmental Technology, Inc., Dayton, Ohio 45437; and
Operational Technologies Corp., Dayton, Ohio 45432
Received March 5, 2003; accepted May 1, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Perchlorate (ClO4-), a contaminant in drinking water, competitively inhibits active uptake of iodide (I-) into various tissues, including mammary tissue. During postnatal development, inhibition of I- uptake in the mammary gland and neonatal thyroid and the active concentration ClO4- in milk indicate a potentially increased susceptibility of neonates to endocrine disruption. A physiologically based pharmacokinetic (PBPK) model was developed to reproduce measured ClO4- distribution in the lactating and neonatal rat and predict resulting effects on I- kinetics from competitive inhibition at the sodium iodide symporter (NIS). Kinetic I- and ClO4- behavior in tissues with NIS (thyroid, stomach, mammary gland, and skin) was simulated with multiple subcompartments, Michaelis-Menten (M-M) kinetics and competitive inhibition. Physiological and kinetic parameters were obtained from literature and experiment. Systemic clearance and M-M parameters were estimated by fitting simulations to tissue and serum data. The model successfully describes maternal and neonatal thyroid, stomach, skin, and plasma, as well as maternal mammary gland and milk data after ClO4- exposure (from 0.01 to 10 mg/kg-day ClO4-) and acute radioiodide (2.1 to 33,000 ng/kg I-) dosing. The model also predicts I- uptake inhibition in the maternal thyroid, mammary gland, and milk. Model simulations predict a significant transfer of ClO4- through milk after maternal exposure; approximately 50% to 6% of the daily maternal dose at doses ranging from 0.01 to 10.0 mg ClO4-/kg-day, respectively. Comparison of predicted dosimetrics across life-stages in the rat indicates that neonatal thyroid I- uptake inhibition is similar to the adult and approximately tenfold less than the fetus.
Key Words: PBPK model; lactation; perchlorate; iodide; inhibition; milk.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Perchlorate (ClO4-), the soluble anion of ammonium and potassium perchlorate, is a thyroid iodide uptake inhibitor known to be present in several United States drinking water sources, ranging in concentrations from less than 4 ppb to more than 3700 ppm in some Las Vegas water samples (Motzer, 2001
). Human health concerns arise from the fact that ClO4-, being similar in size and shape to iodide (I-), is able to bind to the sodium-iodide symporter (NIS), thus reducing the amount of iodide taken up into the thyroid. Thyroid hormones are synthesized from I- in the thyroid and are responsible for regulating metabolism. In the adult, lack of iodide causes reduced thyroxine (T4) and triiodothyronine (T3) levels and can eventually lead to hypothyroidism (Wolff, 1998). Furthermore, since these hormones are required for normal physical and mental development, exposure to thyroid inhibitors during the period of rapid growth in late gestation and early infancy could result in long-term consequences (Howdeshell, 2002; Porterfield, 1994). Congenital hypothyroidism and gestational iodide deficiency are known to cause delayed development and, in severe cases, lowered IQ, mental retardation, and even cretinism (Delange, 2000; Haddow et al., 1999; Howdeshell, 2002; Klein et al., 1972; Porterfield, 1994). In the case of perchlorate contamination of drinking water sources, the question is whether ClO4- is capable of inducing these same developmental effects and at what level of exposure the fetus or infant may be at risk of adverse effects.
In order to help answer these questions, several studies have been conducted in rats involving chronic and short-term perchlorate exposure during gestation, lactation, and in adult males at a variety of doses (Bekkedal et al., 2001; Mahle et al., 2002, 2003; York et al., 1999, 2001; Yu et al., 2002). None of the available studies show the same extent of adverse effects from perchlorate exposure as are known to occur in iodide deficiency. However, consolidating these various data sets into a quantitative measure of risk to the perchlorate-exposed infant is quite difficult due to the variations in study design, as well as the rapid physical and biochemical changes taking place during lactation and infancy. To incorporate these kinetic, physiological, and biochemical data into a predictive tool for perchlorate, iodide, and perchlorate-induced inhibition kinetics, a physiologically based pharmacokinetic (PBPK) model was developed in the lactating and neonatal rat. Together with the concurrent PBPK models developed for the male rat (Merrill et al., 2003), pregnant and fetal rat (Clewell et al., 2003), and the adult human (Merrill et al., 2001), the models can be used to compare internal dose metrics, such as ClO4- concentration in the serum across developmental life stages and species. Thus, the models provide a means for extrapolating predicted kinetics and measures of dose to the potentially more sensitive and often overlooked subpopulation, the human fetus and infant (Clewell and Gearhart, 2002a).
Although the hormone feedback system of the postnatal rat is independent of the mother (Howdeshell, 2002; Potter et al., 1959; Vigouroux and Rostaqui, 1980), the effect of maternal perchlorate exposure is intricately tied to neonatal risk. In fact, there are several unique factors that must be accounted for when attempting to quantify risk to the nursing neonate. For example, the lactating mammary gland contains active NIS that concentrates iodide in the milk, ensuring an adequate supply of iodide to the newborn. However, since ClO4- competitively inhibits iodide binding to NIS, maternal exposure during lactation also inhibits iodide transfer in the milk, as has been noted in several species including the rat, goat, rabbit, and cow (Brown-Grant, 1957; Cline et al., 1969; Grosvenor, 1963; Lengemann, 1965; Potter et al., 1959). It is also possible that this binding of perchlorate to NIS, which inhibits iodide uptake, is responsible for concentration of ClO4- in milk. Intra-laboratory studies have shown milk ClO4- levels to be consistently higher than the maternal plasma, and the neonate was found to have significant blood ClO4- concentrations after nursing from the exposed dams (Yu et al., 2001). Thus, the infant would be at risk not only from the diminished iodide intake from the milk, but also from the significant doses of ClO4- received from the milk, and the resulting additional inhibition of iodide uptake at the neonatal thyroid NIS. The PBPK model described here is able to account for the changing physiology of the lactating dam and pup, as well as the resulting impact of these physiological changes on experimentally determined iodide and perchlorate kinetics in order to provide a meaningful, quantitative estimate of two previously uncharacterized determinants of neonatal risk: the relative ClO4- dose to maternal and neonatal rats, and the dose-response relationship between perchlorate exposure and iodide inhibition in the maternal thyroid and milk. At this stage, the model includes only a rudimentary description of endogenous I- kinetics and incorporation into thyroid hormones, sufficient to reproduce the kinetics of radioiodide.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Supporting Experiments
All supporting experiments were performed on timed-pregnant rats and pups of the Sprague-Dawley strain (Crl: CD, Charles River Laboratory, Raleigh, NC). Animals were housed in individual light, heat, and humidity controlled cages and were kept on a 12 h light/dark cycle with access to water and food ad libitum. Euthanization was performed by CO2 asphyxiation on either postnatal day (PND) 5 or 10 and tissues were collected for analysis of ClO4- or radioiodide content using the methods described in Narayanan et al.(2003)
, respectively. In all experiments, pup serum was pooled by sex within individual litters due to small sample volumes. Pup skin, gastrointestinal (GI) tract, GI contents, and maternal tissues were analyzed individually. The animals used in in-house studies were handled in accordance with the principles stated in the Guide for the Care and Use of Laboratory Animals, National Research Council, 1996, and the Animal Welfare Act of 1966, as amended.
Perchlorate drinking water study.
Pregnant dams (n = 12 per group) were given either deionized water or water containing perchlorate from gestational day (GD) 2 through the day of sacrifice. Daily measurements of body weight and water intake were taken to ensure consistent dosing at levels of 0.0, 0.01, 0.1, 1.0, and 10.0 mg ClO4-/kg-day. Maternal serum, thyroid and milk, and neonatal serum were collected from the PND 10 groups and analyzed for ClO4- content. On PND 5, six of the dams from each dosing regimen were given a tail vein injection of 33 µg/kg 125I- 2 h prior to sacrifice. Maternal and neonatal serum, skin, GI contents and GI tract, as well as maternal thyroid and mammary gland, were collected from these PND 5 rats and analyzed for ClO4- and 125I-. Thyroid hormones (free and total T4 and T3) and TSH were also measured in the serum of PND 5 rats that were not dosed with radioiodide.
Radioiodide and inhibition kinetic studies.
Lactating dams (six control, six inhibition) were dosed via tail vein injection to 125I- (average dose = 2.10 ng/kg) on PND 10 and euthanized at 0.5, 2, 4, 8, and 24 h postdosing. Two h prior to the administration of 125I-, dams from the inhibition group were given a 1.0 mg/kg ClO4- iv and euthanized at 0.5, 1, 2, 4, 8, 12, and 24 h post-125I- dosing. This particular ClO4- dose was chosen to be large enough dose to significantly affect iodide uptake, based on inhibition of thyroid iodide uptake in the male rat (Merrill et al., 2003), while being lower than the dose required to saturate the symporter, based on the drinking water study results (see Results section below). Maternal and neonatal serum, skin, GI contents and GI tract, and maternal thyroid and mammary gland tissue were collected from both control and perchlorate dosed groups and analyzed for 125I- with a gamma counter.
Direct pup radioiodide dosing.
PND 10 Sprague-Dawley pups (n = 6) were given a po gavage of 125I- (0.001mg/kg) in water. Pups and nursing dams were euthanized at 0.5, 1, 2, 4, 8, and 24 h postdosing. Neonatal serum, thyroid, skin, GI contents and GI tract, as well as maternal serum and thyroid were harvested and analyzed for radioiodide content.
Model Structure
All model code was written in ACSL (Advanced Continuous Simulation Language, Aegis Technologies Group, Inc., Huntsville, AL). The model structure is based on those of the male rat and gestation models for iodide and perchlorate kinetics developed concurrently by Merrill et al.(2003) and Clewell et al.(2003). The maternal model (Fig. 1A) consists of compartments for plasma, thyroid, skin, GI, kidney, liver, fat, mammary gland and milk, plus compartments for the combined slowly and richly perfused tissues. The thyroid, GI and mammary gland are described with three subcompartments representing the stroma, follicle and colloid in the thyroid, the capillary bed, GI tissue and GI contents in the GI, and the capillary blood, tissue, and milk in the mammary gland. For iodide, additional compartments representing the organified (hormone-bound) iodine were included in the thyroid and serum (Fig. 1B). Skin is described with two subcompartments, representing the capillary bed and tissue. Active uptake into the thyroid follicle and colloid, as well as the skin tissue, mammary gland, milk, and GI contents, was described with Michaelis-Menten (M-M) type kinetics for saturable processes (bold arrows in Fig. 1). Permeability area cross products and partition coefficients were used to describe the passive movement of the anions (I- and ClO4-) between the capillary bed, tissue, and inner compartments (small arrows in Fig. 1), which results from the inherent electrochemical gradients within these tissues (Chow et al., 1969). The flow-limited kidney, liver, and fat compartments were described with partitions and blood flows. Plasma binding of the inorganic anions (I- and ClO4-) was simulated using a saturable term for association of the anions to binding sites in the plasma and a first order clearance rate for dissociation from plasma binding sites. Urinary clearance and transfer of anions between the dam and pup were represented by first order clearance rates.
|
Due to the short time frame of the radioiodide experiments, the kinetic behavior of total radioiodine was assumed to behave as free iodide in all compartments other than the thyroid and plasma. Inorganic iodide and perchlorate were then modeled in the same manner, based on the similar size and charge of the ions and their shared affinity for NIS (Wolff, 1998
). The thyroid, skin, GI and mammary gland contain active NIS and were therefore defined separately in the structure of the model (Kotani et al., 1998; Spitzweg et al., 1998). The thyroid, mammary gland, milk, skin and GI contents have also been found to maintain higher concentrations of ClO4- and I- than the plasma (Brown-Grant, 1961; Brown-Grant and Pethes, 1959; Chow et al., 1969; Halmi and Stuelke, 1959; Wolff, 1998; Zeghal et al., 1995). Although other tissues, such as the salivary gland, ovary, and choroid plexus, are also known to sequester iodide and perchlorate in the rat and human (Brown-Grant, 1961; Honour et al., 1952; Spitzweg et al., 1998), the small amounts of anions present in these tissues do not affect plasma concentrations. These tissues were therefore combined in the richly or slowly perfused compartments.
In order to describe free thyroidal iodide, it was necessary to account for the incorporation of iodide into hormones in the thyroid and the secretion of this incorporated iodine into the blood. Our goal was to describe this process with enough detail to predict time course data, while keeping the model as parsimonious as possible. Thus, thyroid hormone production, or the incorporation of iodide into hormones, is described using a first order clearance (ClProdci) of inorganic iodide from the thyroid follicle to the incorporated thyroid iodine compartment. Secretion of the incorporated iodine into the plasma is also described with a first order clearance (ClSecrci) from the incorporated thyroid iodine compartment to the incorporated plasma iodine compartment (Fig. 1B
). It was not necessary to include a description of hormone incorporation in the perchlorate model, since ClO4- is not organified or metabolized in vivo (Anbar et al., 1959).
The incorporated plasma iodine compartment represents combined plasma hormonal iodine, including free T3 and T4 as well as protein bound T3 and T4, but does not attempt to predict individual hormone kinetics. Studies of iodine distribution from dietary intake suggest that the majority (>80%) of serum iodine is, in fact, incorporated into hormones or bound to plasma proteins (Stolc et al., 1973b). In contrast, studies in our laboratory found that inorganic iodide accounts for approximately 80% of the total measured plasma radioiodine up to 24 h after an administered 125I- dose (Mahle et al., 2002). This apparent discrepancy may be explained by the slow incorporation of administered or ingested iodide into hormones over time. Endogenous serum iodine data primarily reflect hormone-incorporated iodine, since the system is at steady state. This normal iodine turnover is well established in the animal when the radioiodide is introduced. However, the kinetics reflected in the radiolabeled iodide time course is primarily due to binding of the free anion to plasma proteins, uptake of the anion into various tissues, and urinary clearance. Hormone incorporation would have little effect on these radioiodide kinetics, due to the long time-frame required to incorporate the radioiodide into hormones in the thyroid and the slow secretion of these newly produced radiolabeled hormones into the serum. The model is able to reconcile these data by including an incorporated iodine compartment in the plasma, into which the hormone-incorporated iodine enters as it is secreted from the thyroid. A generic first order rate (ClDeiodci) is then used to describe the overall deiodination accomplished in the various tissues, allowing the inorganic iodide to re-enter the free plasma compartment.
Some description of plasma binding was required for both anions in order to adequately reproduce the available data. The inclusion of binding to plasma proteins is especially important in the case of perchlorate. In fact, at low serum concentrations (
100 µg/l), approximately 99% of the anion is bound to plasma proteins and at higher concentrations (
500 µg/l), 50% is bound (J. W. Fisher, personal communication). Binding of perchlorate to plasma proteins has also been measured in both human and bovine serum (Carr, 1952; Scatchard and Black, 1949). Iodine also binds to plasma proteins, but to a lesser extent than ClO4-. Therefore, the model includes a description for the binding of I- and ClO4- to plasma proteins. Competition of the two anions for plasma binding sites was also included in the model.
In addition to the reported presence of NIS in the mammary gland (Spitzweg et al., 1998), studies of perchlorate-induced inhibition of iodide uptake in milk and mammary tissue support the conclusion that a transport mechanism similar to that of the thyroid exists in the mammary gland (Brown-Grant, 1957; Grosvenor, 1963; Potter et al., 1959). Furthermore, hormones produced during lactation, such as prolactin, regulate the mammary gland NIS activity (Tazebay et al., 2000). Shennan and Peaker (2000) also found evidence of a second anion transport mechanism in the secretory cells of the mammary gland, suggesting that this transporter is also able to move iodide and perchlorate against a concentration gradient. This second anion channel is represented in the model as active uptake into the milk compartment.
The kidney and liver were separately defined within the structure of the model in order to describe the rapid urinary clearance of the anions and to allow for future elaboration of the model that would address hormone metabolism in the liver. A fat compartment was also included to account for the possible effect of changing fat volume on the kinetics due to the hydrophilic nature of both anions. Since kidney, liver, and fat do not maintain tissue:plasma ratios greater than one for either anion, these tissues were described as single, flow-limited compartments and do not contain terms for active uptake. Effective partitioning into these compartments is thought to result from the electrochemical gradient that moves ClO4- from serum to tissue (Chow and Woodbury, 1970).
The basic structure of the neonatal model (Fig. 1A) is similar to that of the lactating rat, excluding the mammary gland. In order to simplify the model, all pups from a single litter were combined together within the model structure. Neonatal dose is described as a first order transfer rate between the maternal milk and pup GI contents. The anions are then recirculated to the mother through the pup urine based on the work of Samel and Caputa (1965), showing that lactating dams ingest approximately 60% of the neonate’s iodine dose while grooming their pups.
Perchlorate-induced inhibition of iodide uptake was included in the maternal and neonatal thyroid follicle and colloid, GI contents and skin, as well as the maternal mammary gland and milk. Literature sources have reported inhibition of iodide uptake into gastric juice of the male rat (Halmi and Stuelke, 1959) and the milk of the lactating rat (Brown-Grant, 1957; Grosvenor, 1963; Potter et al., 1959). Studies in our laboratory have also shown consistent evidence of significant inhibition of iodide uptake in neonatal GI and skin, and slight inhibition in the maternal skin and mammary gland (Mahle et al., 2002).
Dosing Procedures
In order to simulate the daily dosing regimen of the perchlorate drinking water experiment, a pulse function in ACSL was used to introduce drinking water to the GI contents of the lactating dam at a constant rate for 12 h per day (1800 to 0600 h). The neonate was dosed continuously throughout the day from the maternal milk. Both the pup milk dose and the oral bolus dose were introduced into the GI contents of the neonate utilizing pulse functions. Tail vein injections were simulated by introducing the anions into the iv serum compartment. Dosing for the dietary iodine studies (Stolc et al., 1973a,b) was based on the iodine content of the feed and supplemented water, as well as published water, milk, and dietary intake data (Stolc et al., 1966) in both the maternal and neonatal rat throughout the postnatal time period, assuming a constant (24 h/day) intake for both dam and neonate.
Model Parameters
Model equations are described in the Appendix. Whenever possible, physiological and kinetic parameters were obtained from literature or experiments. Allometric scaling was generally employed to account for differences in parameters due to variations in body weight of male, female, and neonatal rats. Tissue volumes were scaled linearly by body weight (BW). Blood flows, maximum velocities, permeability area cross products (PA), and clearance values were scaled by BW0.75. Pup values were scaled in a similar manner to the maternal parameters. Physiological and chemical-specific parameters were scaled first by the body weight of an individual pup, as described above, and were then multiplied by the total number of neonates to represent the value for the entire the litter.
Physiological parameters.
The physiological description of maternal and neonatal rats during lactation is based on the work of Fisher et al.(1990), measurements from our laboratory and published physiological data. Due to the nonuniform changes in tissue volume and body weight during lactation, it was necessary to include gender and life stage-specific physiological descriptions whenever possible. Parameters that were not available specifically for the lactating female or neonate were described by adjusting male rat values by body weight. Final values for the physiological parameters and sources from which they were obtained are listed in Table 1.
|
Lactation-specific changes in tissue volumes and suckling rates were included in the model using ACSL TABLE functions, which employ linear interpolation between available data points. Maternal body weight was assumed to increase by 12% between PND 1 and 10, based on the daily measurements of dams in the perchlorate drinking water study described previously. Since there was no significant difference in the average daily body weights between the five dose groups, the average body weight from all doses were used. The relative volume of the mammary tissue increased from 4.4% on PND 2 to 5.6, 6.3, and 6.6% of the maternal body weight on PND 7, 14, and 21, respectively (Knight et al., 1984
). Maternal body fat increased from 12.4 to 15.2% of the body weight between parturition and PND 2, with a subsequent decrease to 6.9% of the body weight from PND 2 to 16 (Naismith et al., 1982). Values for body fat content on PND 2 and 16 were taken from the measured values of Naismith et al.(1982). Maternal body fat at parturition was calculated from the previously developed PBPK model for the pregnant rat (Clewell et al., 2003). Thus, the relative volume of the maternal body fat in the model increases slightly between PND 1 and 2 and then decreases after PND 2. The rate of milk production was assumed to be equal to the suckling rate described below.
As in the maternal tissues, TABLE functions were used to interpolate between reported data points for the changing body weight, suckling rate, and relative tissue volumes in the neonate. Growth of the neonate is directly dependent on the ingestion of milk.Stolc et al.(1966) measured both pup body weight and the milk ingestion of suckling rats from birth through weaning. Although the author used a different strain of rats than was used in our studies, the pup body weights were nearly identical between the two studies. Thus, the more comprehensive, published data set of Stolc et al.(1966) was used for these physiological parameters. Values for neonatal body fat were based on the data of Naismith et al.(1982), showing a rapid increase from 2.7 to 11% BW between PND 2 and 16 and a subsequent decrease to the adult value of 4.61% (Brown et al., 1997).
Increase in neonatal thyroid volume was based on the work of Florsheim et al.(1966), who reported relative thyroid volumes of 0.013, 0.015, 0.012, 0.014, 0.013, 0.013, and 0.013% body weight for neonates on PND 1 through 5, 7, and 11, respectively. The model also describes changing thyroid stroma, follicle, and colloid fractions over time based on the work of Conde et al.(1991), who measured the fractional volumes on postnatal days 0, 5, 10, 15, 20, 25, 30, 60, and 120. Relative tissue volumes of the skin, GI tract, liver, and kidney were modeled based on measured body and tissue weights on PND 1, 5, 10, 20, 30, and 64 (Palou et al., 1983). Relative skin volume increased from 19.3 to 20.8% BW from PND 1 to 20 and then decreased to 19% BW by PND 30. Similar trends were also seen in the GI, kidney, and liver, with increasing volume (with respect to body weight) peaking at PND 30, 20, and 30, respectively.
All maternal blood flows that were not directly affected by the changes induced by lactation were scaled allometrically from the adult male rat parameters. TABLE functions were used to describe the changing in cardiac output and fractional blood flow to the mammary tissue throughout lactation, according to the data of Hanwell and Linzell (1973) and the neonatal cardiac output, hematocrit, and regional blood flows, based on the data of Rakusan and Marcinek (1973). Among other tissues, Rakusan and Marcinek (1973) measured fractional blood flows to the kidney, liver, skin, stomach, and large and small intestines in 1, 30, 60, and 140 day-old rats. Relative blood flow to the neonatal kidney, liver, and total GI are modeled as increasing from 3.6, 4.5, and 4.6 to 15.5, 5.2, and 6.8% cardiac output, respectively, in the first 60 days. Blood flow to the skin remains relatively constant after birth, at approximately 11% of the cardiac output.
Chemical-specific parameters.
Chemical-specific parameters (Table 2) for perchlorate and iodide in tissues other than the mammary gland were kept as similar as possible to those used in the PBPK models for the male and pregnant rat (Clewell et al., 2003; Merrill et al., 2003) in order to facilitate comparison of the models and extrapolation between life-stages. As in the previous models, binding of I- to NIS was given a Km of 4.0 x 106 ng/l based on the work of Gluzman and Niepomniszcze (1983) in human thyroid slices. This value for the follicular Km (KmTFi) remained constant across species (Gluzman and Niepomniszcze, 1983) and tissues (Wolff, 1998) and was therefore applied to all compartments in the model with NIS. The second transporter located at the apical membrane in the thyroid was studied by Golstein et al.(1992), who measured a Km of approximately 4.0 x 109 ng/l for iodide (KmTLi) in bovine thyroid. In the model, a somewhat lower value than that measured by Golstein et al. of 1.0 x 109 was used for KmTLi, based on the ability of the model to fit the later (>8 h) time points. The Km for the second active transport mechanism in the mammary gland (KmMk), for milk uptake, was set by fitting the model simulation to available mammary gland and milk data.
|
The Km value for ClO4- transport by NIS was given a value of 1.5 x 105 ng/l for all relevant compartments. This value is based on the assumption that perchlorate acts as a competitive inhibitor of iodide uptake and is, in fact, transferred into the tissues via NIS (Clewell and Gearhart, 2002b
). Therefore, the Km value for perchlorate transport by NIS would be equal to its Ki value. Kosugi et al.(1996) measured the Ki for ClO4- at 1.5 x 105 ng/l. This value was adjusted slightly to obtain the best fit of thyroid perchlorate to the drinking water data, resulting in a Km of 2.0 x 105ng/l. This value is further supported by various literature sources suggesting that ClO4- actually has as much as an order of magnitude greater affinity for NIS than I– itself (Chow et al., 1969; Halmi and Stuelke, 1959; Harden et al., 1968; Lazarus et al., 1974). Likewise, the Km values for perchlorate transport by the second anion channels in the thyroid and mammary gland were also nearly a factor of 10 less than that of iodide, resulting in values of 1.0 x 108 and 1.0 x 106 ng/l for KmTLp and KmMkp, respectively.
The values for Vmax vary significantly across species and tissues with NIS (Gluzman and Niepomniszcze, 1983; Wolff, 1998) and were, therefore, determined by fitting the model simulations to available data in the various tissues of PND 5 and 10 rats. In the perchlorate drinking water study, the nonlinearity of tissue ClO4- concentrations across doses in compartments with NIS suggests that the symporter is saturated between the 1.0 and 10.0 mg/kg-day doses. Thus at doses below saturation (1.0 mg ClO4-/kg-day), the active transport via NIS would drive tissue concentrations and were therefore used to set Vmax values. For iodide, kinetic data were taken at doses well below the saturation of NIS. Thus, the time course data from the kinetic studies were used to determine values for Vmax in tissues with active uptake.
Partitioning of ClO4- and I- into tissues results from the electrochemical potential present across tissue membranes (Chow and Woodbury, 1970). Theoretical effective partition coefficients were calculated from measured electrical potentials presented by Chow and Woodbury (1970) using the equations given in Kotyk and Janacek (1977). Calculations are described in detail in the male rat perchlorate model (Merrill et al., 2003). Ranges for the partition coefficients corresponding to the stroma:follicle and follicle:lumen membrane diffusion were estimated to be 0.11 to 0.15 and 6.48 to 8.74, respectively. Based on the fit of the model simulation to the data and the calculated values above, values of 0.15 and 7.0 were used for PTFi and PTLi, respectively.
As mentioned previously, the perchlorate drinking water data indicate that NIS transport is saturated between the 1.0 and 10.0 mg ClO4-/kg-day doses. Therefore, at 10.0 mg/kg-day, ClO4- uptake into the various tissues would be predominantly determined by the passive diffusion parameters. Thus, parameters describing partitioning of ClO4- into the tissues were obtained by fitting the model simulation to the highest dose group data from the drinking water study. In the cases where data were not available in the lactating or neonatal rat, such as the muscle (slowly perfused), liver (richly perfused), kidney, and red blood cells, values were obtained from those used in the adult male rat model (Merrill et al., 2003). The partition coefficient for perchlorate in fat was measured in the laying hen (Pena et al., 1976). Other tissues in the hen, such as the muscle and kidney, were found to have similar partition coefficients to those of the rat.
Iodide partition coefficients and PA values were calculated from the tissue:blood ratios measured during the clearance phase of data for the tissue of interest either from literature or experimental data in the rat. The partitioning parameters for the muscle (slowly perfused), liver (richly perfused), kidney, and red blood cells were given the same values measured in the male rat (Merrill et al., 2003) and the value for partitioning of iodide into the fat was given the same value as ClO4-.
Parameters for plasma binding were determined by fitting the model to time course data in the case of iodide and the 0.01 and 0.1 mg/kg-day drinking water data in the case of ClO4-, due to the fact that binding was most prevalent at the lower doses. Urinary clearance of ClO4- was determined from fitting the serum at the 10.0 mg/kg-day dose group, where binding had little effect on serum concentrations. First-order clearances for incorporation of iodide into thyroid hormones and hormone secretion were determined by the fit of the model to the incorporated and free thyroid iodide time course data. Parameters for binding of inorganic iodide to plasma proteins were determined by the fit of the model simulation to initial portion of the serum radioiodine data. Later time points were assumed to be more affected by hormone secretion and deiodination rates, as sufficient time had passed to allow for incorporation of the administered radioiodide into hormones. Urinary iodide clearance was determined from the fit of the model to serum inorganic iodide time course data.
Upregulation of thyroid NIS activity.
At the time of data collection in the drinking water study, rats had been exposed to ClO4- throughout gestation and up to the day of sacrifice. At this point, upregulation of the thyroid activity is evidenced by decreased T4 and elevated TSH levels in the serum at all doses, as well as a lack of noticeable thyroid iodide uptake inhibition (Yu et al., 2001). Since increased TSH upregulates thyroid iodide uptake by increasing the number and activity of NIS (Wolff, 1998), the value for VmaxcTFi, which corresponds to the maximum capacity of active transport at the basolateral membrane, was increased to fit the measured radioiodide concentrations in the upregulated thyroids for each dose. The resulting values for VmaxcTFi were then plotted versus the corresponding concentrations of serum free ClO4- and the data were fitted with an M-M equation. This equation then was used in the model to describe the induction of NIS upregulation with dose, in a similar manner to the description used by Andersen et al.(1984) to describe enzyme induction.
Upregulation of thyroidal NIS also affects thyroid ClO4- uptake, and hence the measured thyroid ClO4- concentrations in the drinking water study, as both ClO4- and I- are transported by the same symporter (Wolff, 1964, 1998). Thus, increased thyroid ClO4- uptake was modeled in the same manner as I-, increasing the value for VmaxcTFp with dose and applying the resulting M-M fit to the model.
Dietary iodine model.
The dietary iodine model is identical to that of the radioiodine model. Dosing is accomplished as a constant intake of iodine through water and diet. All chemical-specific parameters are assumed to be the same as those determined from the radioiodide kinetics. The radioiodide, endogenous iodine and perchlorate models operate independently in tissues with passive diffusion, and are linked through competitive uptake in tissues with NIS (e.g., thyroid follicle). Thus, in tissues governed by passive diffusion, tissue:blood ratios are identical for the radiolabeled and dietary iodine models. The interaction of the three models through competitive inhibition at NIS allows us to explore the effect of varying dietary intake on both radiolabeled iodine kinetics and perchlorate-induced inhibition of radioiodide uptake in the thyroid.
Sensitivity analysis of chemical-specific parameters.
A sensitivity analysis was run after finalizing the model parameters as to examine the relative influence of each of the chemical-specific parameters on model predictions. The model was run to determine the change in the average serum ClO4- concentration (AUC: area under the curve) and total thyroid iodide uptake at 8 h postdosing, resulting from a 1% change in the value of each kinetic parameter. In an effort to determine the effect of NIS saturation on relative parameter importance, the sensitivity analysis was performed at two ClO4- doses, presumably representing unsaturated and saturated symporter states (0.1 and 10.0 mg-kg-day, respectively). Since the iodide doses used in this model are not expected to saturate the NIS, thyroid iodide sensitivity analysis was run only at the dose used in the kinetic experiments (2.1 ng/kg 125I-). The following equation shows the calculation of the sensitivity coefficient for each parameter.
 |
where A is the serum AUC with 1% increased parameter value, B is the serum AUC at the starting parameter value, C is the parameter value after 1% increase, and D is the original parameter value.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Model Parameterization
The ClO4- data from the drinking water study were used to determine kinetic parameters for ClO4- in the lactating and neonatal rat. Upregulation of NIS transport of ClO4- into the thyroid was accounted for as described in the Methods section. Figure 2
shows the model simulations for ClO4- concentrations in the maternal and neonatal serum, maternal thyroid, GI contents, mammary gland, and milk, versus measured data from the drinking water study on PND 5 and 10 at 0.01, 0.1, 1.0, and 10.0 mg ClO4-/kg-day. The upregulated values for Vmaxc_TFp were used in the simulation of the thyroid as described in the Methods section. In these and subsequent plots, solid lines indicate the model prediction and cross-bars indicate the mean ± SD of measured data.
|
Neonatal iodide kinetic parameters were determined by the fit of the model to the data obtained from directly dosing the pup, while maternal parameters were primarily determined from the data obtained by dosing the dam via iv. Transfer from the neonate to the dam was established by fitting the maternal kinetic parameters first and then utilizing maternal data from the pup dosing study to determine the magnitude of iodide transfer to the maternal stomach via pup urine. In the same way, milk transfer was parameterized by first fitting the neonatal parameters from the pup dosing study and then utilizing the pup data from the maternal dosing kinetic study to estimate pup exposure via milk. PA values were adjusted to describe the behavior of iodide data, where increasing PA values toward 1.0 l/h-kg generally increased the rate at which uptake and clearance in a particular tissue occurred, and decreasing PA slowed uptake and clearance. Model simulations of radioiodide kinetics in maternal and neonatal tissues versus the data obtained from dosing the dams are shown as the control group in the inhibition study (Fig. 3
). The fit of the model to iodide levels in maternal and neonatal tissues from the direct oral dosing of PND 10 pups are shown in Figure 4. The model was able to reproduce data in both maternal and neonatal tissues, whether exposure occurred via maternal iv or oral bolus to the pup. Thus, the model is able to describe both maternal to neonatal and neonatal to maternal iodide transfer, and is also able to reproduce data across exposure routes.
|
|
Model validation.
Once model parameters were established as described above, the robustness of the model was tested against a variety of data sets taken across different laboratories, rat strains, exposure routes, and time points in lactation. Several published kinetic studies using various isotopes of iodide were used to test the model description of iodide kinetics in the dam and neonate (see below). Accuracy and usefulness of the model perchlorate and iodide descriptions were further validated against studies of perchlorate-induced inhibition of iodide uptake in the maternal thyroid, mammary gland, milk, and neonatal tissues based on the description of competitive binding to NIS.
Radioiodide kinetics on PND 10.
Validation of PND 10 iodide kinetics was performed with the data of Iino and Greer (1961), Samel and Caputa (1965), Vigouroux (1976), and Vigouroux and Rostaqui (1980). In order to simplify comparison of the different data sets, which were originally performed at slightly different dose levels, the data of each study were normalized to a dose of 1.0 ng. The model simulation was then run versus the combined data sets (Fig. 5). Figures 5A and 5B show the maternal and neonatal thyroid radioiodide levels on PND 10 following an acute 131I- dose to the dam. The model is able to predict the maternal thyroid iodide, but underpredicts pup thyroid iodide levels at the 24 h time point. However, the model prediction is within a factor of two of the measured data. Figures 5C and 5D show the model-predicted maternal and neonatal thyroid iodide levels after an ip dose to the pup. The model simulation is able to describe both maternal and neonatal thyroid uptake, again within a factor of two.
|
The data of Samel and Caputa (1965)
also allowed validation of the model parameters for urinary output in the pup. Difficulty in separating maternal and pup urine precluded the collection of these data in our own studies. Therefore, urinary clearance values in the model were determined by the fit of the serum to the time course data. Using the previously determined kinetic parameters, the model-predicted urinary iodide, 4 h after an iv dose of 131I- in the PND 10 rat, was 14% of the pup dose, which is close to the range of values (0.61 to 10.9%) given by Samel and Caputa (1965).
Radioiodide kinetics on PND 5.
In order to determine whether the model would provide reasonable predictions of iodide and perchlorate kinetics in younger pups (<PND 10), the model was tested against data obtained in our laboratory after a single iv injection of 125I- on PND 5 of the ClO4- drinking water study. Since the control group did not receive perchlorate during the study, tissue 125I- data from these animals can be used to confirm the model’s ability to predict kinetics in PND 5 rats. Figure 6 shows that the model-predicted radioiodide concentrations in the maternal serum, thyroid and mammary gland and neonatal serum are in good agreement with the available data on PND 5.
|
Upregulation of thyroid NIS activity was modeled against the thyroid iodide levels measured after 23 days of exposure to perchlorate in drinking water at doses of 0.01, 0.1, 1.0, and 10.0 mg ClO4-/kg-day as was described in the Methods section. Using this equation for the upregulation of the follicular Vmaxc (VmaxcTFi), it was possible to describe the increase in iodide uptake based on the perchlorate dose in chronic exposure scenarios. Neither the measured data nor the model showed any inhibition in iodide concentrations in the maternal thyroid 2 h postdosing with 125I- after 18 days of exposure to 0.01, 0.1, 1.0, and 10.0 mg/kg ClO4- in drinking water. Thus, the model was able to describe the upregulation of thyroid NIS activity resulting from subchronic perchlorate exposures.
Radioiodide kinetics in late lactation.
The model’s ability to simulate iodide kinetics at later time points in lactation (>PND 10) was tested against the normalized data of several literature studies. Figure 7 shows the model-predicted radioiodide milk:plasma ratio versus the data of Brown-Grant (1957), Grosvenor (1963), and Potter et al. (1959) collected on PND 14, 17–20, and 18, respectively. Although the simulation shown in Figure 7 was run on PND 14, there is no noticeable change in the model-predicted milk:plasma ratio when run at different days in lactation. Despite the changing kinetics (body composition, suckling rate, etc.), the model prediction of relative iodide concentration in the milk remains constant. Although the amount of milk provided to the infant may change during the course of lactation, the concentration of iodide does not. This assumption is supported by the available data, which do not show any increasing or decreasing trend in milk iodide concentrations. Dietary iodine studies by Stolc et al. (1966; 1973a,b) also showed a relatively unchanging milk iodine concentration through the different stages of lactation. The model predictions are in reasonable agreement with the trend suggested by the total composite of the different data sets.
|
The maternal urinary iodide was tested against the data of Grosvenor (1960)
. For the reasons mentioned previously, urine was not collected in our own studies. Thus, model parameters for maternal urinary clearance (ClUci) were determined by fitting the model simulation to serum time course data, while maintaining fits to data in the thyroid, mammary gland, and pup. Using these previously set parameters, the model predicts that 40.8% of the 131I- dose will be excreted within 8 h post-131I- dosing, which is within the range of 36.7 to 42.1% reported in Grosvenor’s 1960 study.
Perchlorate-induced inhibition of iodide uptake.
The ability of the model to predict inhibition of iodide uptake into the thyroid, milk, and other tissues is not only important in calculating risk to the dam and neonate, but is also indicative of the model’s ability to accurately describe both perchlorate and iodide kinetics. Thus, the available data for perchlorate-induced inhibition of iodide kinetics in lactation was used as the final validation of the current model structure. Using the conditions for the inhibition time course study described in the Methods section, the model simulation was run to predict the effect of the administration of 1.0 mg ClO4-/kg on the kinetic behavior of an iv dose of radioiodide (given 2 h post-ClO4- dosing). Figure 8 shows that the model accurately simulates inhibition of iodide uptake in the maternal thyroid and mammary gland. The simulations of the iodide inhibition kinetics in the mammary gland also suggest that the model is able to predict the affect of perchlorate exposure on the availability of iodide to the neonate. Although NIS inhibition would occur in both maternal GI and skin, neither the data nor the model showed significant difference between the control and ClO4- dosed animals, suggesting that other factors (i.e., diffusion, partitioning) are offsetting the effects of symporter inhibition.
|
The model-predicted inhibition of both milk and thyroid iodide was validated against the data of Potter et al. (1959)
. Potter and coauthors measured the amount of radioiodide (131I-) taken up into the thyroid gland and the combined mammary gland and expressed milk 24 h after dosing with iodide in PND 18 rats. The total effect on iodide uptake was measured after two ClO4- doses given 1 h prior to and 30 min after the radioiodide dose. At these large doses of ClO4-, thyroid iodide uptake is essentially blocked (approximately 99% inhibition). Therefore, it is likely that the NIS transport of iodide into the mammary gland is also completely inhibited. However, unlike the thyroid, the milk still showed significant amounts of iodide (approximately 10% of the control value at 24 h). It is possible that the second mammary transport mechanism is responsible for this difference in tissue response to ClO4- exposure, because it depletes iodine content in the mammary gland to the extent that passive diffusion between the gland and the blood becomes significant. In contrast, passive diffusion of iodine into the thyroid is restricted by the slow clearance via secretion of hormones into the blood, as well as the lower permeability and partitioning of the anions suggested by the model. The ability of the model to predict the combined mammary gland and milk control and ClO4- dosed data of Potter et al. (1959; Fig. 9) supports the accuracy of the description of iodide partitioning in the mammary gland and the second active transporter. The model predicts the data remarkably well despite significant differences between this study and those used for model development, such as different time points in lactation (PND 18 vs. 10) and a different dosing regimen.
|
Dietary iodine.
In three separate experiments, Stolc et al. (1966
, 1973a,b) studied the distribution of endogenous iodine in various tissues of the maternal and neonatal rat resulting from a controlled intake of iodine through the diet and water. Model-predictions of the maternal and neonatal tissue iodine concentrations are shown in Figure 10 versus data collected at water and feed concentrations of 60 ng I/g and 500 ng I/ml, respectively (Stolc et al., 1973a), which corresponds to daily doses of 4.8 mg I/kg-day. Of the three dosing levels presented by Stolc and coauthors, this highest dose was chosen to validate the model prediction of normal iodine distribution, due to the fact that it is within the range of normal dietary intake for rats and it is closest to dietary intake expected for our experimental studies (800 ng I/mg food). The two lower dose groups would be considered moderately deficient and deficient iodine diets, resulting in changes in the H-P-T axis in order to affect upregulation as is evidenced by the measured tissue iodine concentrations. Despite a fivefold difference in water iodide concentrations, the serum and thyroid concentrations remain unchanged. Other tissues, including the skin and GI, do show a dose-dependent change in tissue concentration. Thus, since this model does not yet include the pharmacodynamic response of the thyroid axis to dietary insufficiency, it is premature to attempt to predict such data. Using kinetic parameters obtained from acute data, the model is generally able to predict endogenous data within a factor of two of the measured data for tissue concentrations ranging over more than four orders of magnitude (Fig. 10).
|
While admittedly only a preliminary description, this endogenous iodine model could be exercised to assess the affect of changing dietary intake on predicted acute radioiodide kinetics, as well as predicted perchlorate-induced inhibition of radioiodide kinetics in the perinatal and adult rats. Since the radioiodide, endogenous iodine and perchlorate models operate independently in tissues with passive diffusion, and are linked through competitive uptake in tissues with NIS (e.g., thyroid follicle), only tissues with active uptake are expected to show any change in tissue concentrations with varying dietary intake. From this modeling exercise, it was determined that the effects of changes in dietary iodine intake on acute radioiodide kinetics are likely to be negligible. In fact, significant differences in predicted thyroid concentrations were not observed at feed concentrations as high as 8.0 ppm (more than an order of magnitude higher than standard laboratory diet). Indeed, the first apparent change in predicted inhibition of thyroidal radioiodide uptake is seen at feed concentrations 100 times greater than standard laboratory rat chow.
Sensitivity analysis.
Sensitivity analysis performed at 0.1 and 10.0 mg ClO4-/kg-day drinking water revealed a dose-dependent difference in model sensitivity to several of the chemical-specific parameters (Fig. 11). At 0.1 mg/kg-day, the maternal serum is primarily dependent on serum binding, showing less sensitivity to urinary clearance. All other parameters had calculated sensitivity coefficients less than 0.1. At the 10.0 mg/kg-day dose, binding parameters are no longer important determinants of predicted serum levels. Only the urinary clearance remains significant, with a sensitivity coefficient of –0.87. Neonatal serum ClO4- levels are influenced by several model parameters at the 0.1 mg/kg-day dose, including the parameters defining passive diffusion and active transfer in the mammary gland, milk, and neonatal GI. However, similar to dam, neonatal serum AUC shows the greatest sensitivity to serum binding parameters at this lower dose. At the higher dose (10.0 mg/kg-day), where active uptake into the mammary gland and serum binding are likely saturated, partitioning into mammary gland and milk and urinary clearance (both maternal and neonatal) show the greatest influence on pup serum ClO4- levels. Results of the sensitivity analysis for thyroid iodide uptake (not shown) were similar, in that the magnitude of all of the parameter sensitivities were less than one, although model predictions for this metric were sensitive to a much larger number of input parameters. This result is not unexpected due to the fact that the uptake prediction is for a specific point in time after administration of the radioiodide, and the rate of distribution into all tissues can effect the time-dependent result [as compared to an average, or AUC (area under the curve), measure, which reflects steady-state behavior]. Thus, the validation of the model with kinetic thyroid iodine uptake and inhibition data provides a reasonable test of the model parameterization.
|
Calculation of Internal Dose Metrics
The model was used to calculate internal dose metrics corresponding to both acute and subchronic perchlorate exposure in the lactating and neonatal rat. These internal measures of dose include: AUC for ClO4- in the maternal and neonatal serum, relative neonatal dose (% maternal dose) and inhibition of thyroid iodide uptake after acute dosing. Table 3
shows the predicted neonatal dose on PND 10 as % maternal dose and as an amount adjusted for milk intake and pup body weight. The model predicts a significant transfer of maternal ClO4- to the neonate on PND 10 at low maternal doses. In fact, per kg body weight, the PND 10 pup receives a greater dose than the dam (0.07 vs. 0.01 mg/kg BW). Tables 4 and 5 show the dose metric comparisons