ToxSci Advance Access originally published online on April 15, 2003
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Toxicological Sciences 73, 256-269 (2003)
Copyright © 2003 by the Society of Toxicology
BIOTRANSFORMATION AND TOXICOKINETICS |
PBPK Predictions of Perchlorate Distribution and Its Effect on Thyroid Uptake of Radioiodide in the Male Rat
,2
* Operational Technologies Corporation, Dayton, Ohio 45432;
GeoCenters, Inc., Wright-Patterson AFB, Ohio 45433;
ManTech Environmental Technology, Inc., Dayton, Ohio 45437; and
AFRL/HEST Operational Toxicology Branch, Wright-Patterson AFB, Ohio 45433
Received November 19, 2002; accepted March 3, 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Due to perchlorate’s (ClO4-) ability to competitively inhibit thyroid iodide (I-) uptake through the sodium-iodide symporter (NIS), potential human health risks exist from chronic exposure via drinking water. Such risks may include hypothyroidism, goiter, and mental retardation (if exposure occurs during critical periods in neurodevelopment). To aid in predicting perchlorate’s effect on normal I- kinetics, we developed a physiologically-based pharmacokinetic (PBPK) model for the adult male rat. The model structure describes simultaneous kinetics for both anions together with their interaction at the NIS, in particular, the inhibition of I- uptake by ClO4-. Subcompartments and Michaelis-Menten (M-M) kinetics were used to describe active uptake of both anions in the thyroid, stomach, and skin. Separate compartments for kidney, liver, plasma, and fat were described by passive diffusion. The model successfully predicts both 36ClO4- and 125I- kinetics after iv doses of 3.3 mg/kg and 33 mg/kg, respectively, as well as inhibition of thyroid 125I- uptake by ClO4- after iv doses of ClO4- (0.01 to 3.0 mg/kg). The model also predicts serum and thyroid ClO4- concentrations from 14-day drinking water exposures (0.01 to 30.0 mg ClO4-/kg/day) and compensation of perchlorate-induced inhibition of radioiodide uptake due to upregulation of the thyroid. The model can be used to extrapolate dose metrics and correlate observed effects in perchlorate toxicity studies to other species and life stages, such as rat gestation (Clewell et al., 2003
). Because the model successfully predicts perchlorate’s interaction with iodide, it provides a sound basis for future incorporation of the complex hypothalamic-pituitary-thyroid feedback system. Key Words: perchlorate; radioiodide; thyroid; inhibition; sodium iodide symporter; PBPK; model.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Ammonium perchlorate is an oxidizing agent used in solid rocket and missile fuels, pyrotechnics, and air bag inflators. Past use and disposal practices of the ammonium salt have resulted in accidental groundwater contamination by its extremely stable dissociation product, perchlorate (ClO4-). Groundwater concentrations have been detected in ranges as high as 630,000 to 3,700,000 ppb near Las Vegas, Nevada, but most groundwater detections throughout other states have been below 20 ppb (Motzer, 2001
). The current detection limit for perchlorate in water is 1 ppb (EPA Method 314).
Perchlorate’s similarity to iodide (I-) in charge and size allows the anion to competitively inhibit I- uptake by the sodium-iodide symporter (NIS) into the thyroid (Wolff, 1998). NIS actively transports both Na+ and I- from extracellular fluid into the thyroid epithelial cell and other tissues (Ajjan et al., 1998). Decreased intrathyroidal iodide eventually results in a drop in circulating iodide-containing thyroid hormones, triggering a series of compensatory mechanisms by the hypothalamus-pituitary-thyroid axis. These mechanisms include pituitary secretion of thyroid releasing hormone, which signals increased production of thyroid-stimulating hormone (TSH) by the hypothalamus. In turn, TSH increases the synthesis of NIS in the thyroid, thereby restoring intrathyroidal iodide levels. TSH also directly promotes hormone production by increasing the synthesis of thyroid peroxidase, an enzyme required for iodide organification (Spitzweg et al., 1998). Because ClO4- is a potent inhibitor of thyroid iodide uptake, there is concern that chronic exposure to the low levels of ClO4- found in some drinking water supplies may result in adverse human health effects, such as hypothyroidism, leading to goiter or even impaired neurodevelopment from gestational or neonatal exposure (Delange, 2001).
Perchlorate does not appear to be metabolized in the body. Anbar et al. (1959) administered 36Cl18O4- to rats and reported that less than 0.1% of the dose appeared in urine as 36Cl- and 36Cl18O3-. Iodine is readily absorbed from the upper GI tract, distributed throughout extracellular fluid, and cleared mainly through the thyroid, where it is incorporated into hormones or by the kidneys, where it is excreted (Hays and Wegner, 1965). In rats, Yu et al. (2002) reported 99.5% 36ClO4- and 78% 125I- excreted in urine within 48 and 24 h after administered doses, respectively. DiStefano and Sapin (1987) reported between 1% and 7% of activity from administered radiolabeled thyroid hormones excreted in rat feces. Iodine excretion is not regulated by any iodine-conserving feedback system; therefore, regular dietary intake is important.
Literature regarding the toxicity of ClO4- from chronic low-level exposures and its effect on the hypothalamus-pituitary-thyroid axis is limited. High-level drinking water exposures (0.2–10.0 mg ClO4-/kg/day for 14 and 90 days) have shown increased TSH and decreased thyroid hormone levels in rats (Siglin et al., 2000). Yu et al. (2002) reported elevated TSH in rats after 14 days of drinking water exposure, ranging from 0.1 to 10 mg/kg/day, and decrements in T4 and free thyroxine (fT4) from 1.0 to 10 mg/kg/day. In adult humans, short-term studies have shown little or no significant change in TSH and fT4 levels after 2 weeks of drinking water exposure from 0.007 to 0.05 mg/kg/day (Greer et al., 2002). A significant drop in fT4, intrathyroidal iodine and an increase in thyroglobulin (Tg) were reported from subchronic high exposures (900 mg/day for 4 weeks) (Brabant et al., 1992). Yet the length of chronic low level perchlorate exposure required to cause significant hormone deficiencies is definitely not known. A retrospective epidemiologic study on children in three Chilean cities with groundwater levels of <4, 5–7, and 100–120 mg ClO4-/l revealed significantly higher fT4 but normal TSH in the two cities with highest concentrations (Crump et al., 2000).
Currently, the U.S. Environmental Protection Agency (EPA) is conducting a risk assessment to determine a safe reference dose (RfD) and drinking water level for ClO4-. To assist in predicting perchlorate’s effect on normal I- kinetics, we have developed a physiologically based pharmacokinetic (PBPK) model for the adult male rat. The model describes active uptake of I- and ClO4-, as well as ClO4- induced inhibition of I- uptake into NIS-containing tissues (thyroid, gastric mucosa, and skin). The model also simulates distribution of both anions in the kidney, liver, fat, serum, and remaining richly and slowly perfused tissues, as well as elimination from the kidneys.
Understanding the impact of chronic displacement of I- during prolonged exposure to ClO4—contaminated drinking water is the focus of this and other ongoing research efforts. Currently, the model is used in extrapolating dose metrics to reproductive stages (Clewell et al., 2001) and to other species, including adult humans (Merrill et al., 2001). The model structure also forms the basis for future model developments that will include subsequent effects on thyroid hormone homeostasis.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
The kinetic studies by Yu et al.(2002)
served as the main data sets used in the development of this model. A brief summary of their experiments is provided below. Each study utilized adult male Sprague-Dawley rats (330 ± 35 g, N = 6 rats per group), which were dosed as described and euthanized by CO2 asphyxiation at selected time points for tissue collection. All time-course data described were utilized in estimating model parameter values, with the exception of I- inhibition data, which were used for model validation. The 36ClO4- and 125I- doses used in the kinetic studies (Yu et al., 2002) were selected as tracer studies. Due to the low specific activity of 36ClO4-, a dose 100 times that used for the 125I- studies was required for reliable measurements.
Experiments
Radiolabeled perchlorate (36ClO4-) time-course kinetics.
Naive rats received a tail vein iv dose of 3.3 mg/kg 36ClO4- and were euthanized at 0.5, 6, 12, 24, 32, and 48 h post dosing. The thyroid, stomach tissue and contents, intestinal tissue and contents, muscle, skin, liver, kidney, plasma, and red blood cells were analyzed for 36ClO4- using a liquid scintillation counter (LSC). Urine was collected from the 24- and 48-h time point animals.
Radiolabeled iodide (125I-) time-course kinetics.
Rats were administered a tail vein iv dose of physiological saline (control group) or 33 mg/kg 125I- (with carrier) in physiological saline. At 5, 15, and 30 min and 1, 2, 6, 9, 24, 32, 48, and 96 h post dosing, rats were euthanized; thyroids and serum were collected. Urine voids were collected from the 24-h group. The same study was repeated to obtain additional tissues (thyroid, serum, skin, and stomach contents) at 30 min and 2 and 6 h post dosing.
Perchlorate-induced inhibition of thyroid 125I- uptake time-course kinetics.
Rats received one of five tail vein iv doses of ClO4- (0.0, 0.01, 0.1, 1.0, and 3.0 mg/kg), followed by an iv challenge of 125I- (33 mg/kg with carrier) 2 h later. Thyroids were collected at 5, 15, and 30 min and 1, 2, 6, 9, and 24 h post dosing with 125I- and counted for 125I- activity.
Inhibition of thyroid 125I- uptake after drinking water exposure to ClO4-.
Rats were exposed via drinking water for 1, 5, and 14 days to ClO4- doses of 0.0, 1.0, 3.0, and 10.0 mg/kg/day. In addition to the dose groups described in Yu et al.(2002), a 30.0 mg/kg/day group was exposed in the same manner (unpublished data). Daily doses were verified by measuring the amount of water consumed. At the end of day 14, all dose groups were challenged with a single iv dose of 33.0 mg/kg 125I- (with carrier) and euthanized 2 h post 125I- dosing. Serum and thyroid glands were collected for ClO4- analyses at the end of days 1, 5, and 14 for the 3.0, 10.0, and 30.0 mg/kg dose groups (N = 6 per group) and at the end of day 14 for the 0.0 and 1.0 mg/kg dose groups. Thyroids, collected on day 14 from all dose groups, were also analyzed for 125I-.
Analytical Methods
Analyses of 36ClO4- and 125I- in urine and tissues were performed using a liquid scintillation counter (LSC) and g-counter, respectively, as explained in Yu et al.(2002). Cold ClO4- concentrations in urine were analyzed using high-pressure liquid chromatography (HPLC), as described elsewhere (Fisher et al., 2000). To determine whether any of the ClO4- was being metabolized, Yu et al.(2002) performed both ClO4- and chlorate (ClO3-) analyses on thyroid and serum samples, using isocratic and gradient chromatographic conditions. Chlorate was not detected.
Model Structure
Early model development was based on work by Fisher et al.(2000). Several compartmental models have been developed for iodide metabolism (Berman et al., 1968; Degroot et al., 1971; Hays and Wegner, 1965). However, because these models are not physiologically based, their utility for extrapolating across species and life stages is limited. In addition, this work represents part of a series of new PBPK models that predict both perchlorate and iodide kinetics, and the anion interaction at the NIS. Nearly identical model structures were used to describe the distribution of both anions (Fig. 1), given their similar size and ionic charge. Multiple subcompartments were used to describe both active and passive transport of both anions in NIS-containing tissues (thyroid, stomach, and skin), whereas single compartments were used to describe passive diffusion through non-NIS tissues (kidneys, liver, fat, and the remaining lumped rapidly and slowly diffused tissues).
|
Model structures of the stomach and skin were identical between anions. The stomach contains the capillary bed, stomach wall, and contents. Skin contains two subcompartments, representing the capillary bed and the skin tissue. Other tissues also exhibit NIS, such as the salivary glands and choroid plexus (Brown-Grant, 1961
; Spitzweg et al., 1998). However, the anion pools in these other tissues are not large enough to affect serum levels significantly and, therefore, were not included in the model.
Differences in model structure between anions exist in the thyroid and blood compartments. In the thyroid, I- is organified (thyroid hormone production), whereas ClO4- is unreactive and eventually diffuses from the thyroid back into the systemic circulation (Anbar et al., 1959; Wolff, 1998). Therefore, for iodide, the thyroid consists of three subcompartments, representing the stroma, the follicle, the colloid, as well as a separate compartment representing all bound (organified) iodide in the thyroid. For perchlorate, only the stroma, follicle, and colloid were included in the thyroid structure. Thyroid perchlorate time-course data reveal an initial rapid phase in thyroid uptake and equilibrium, presumably between the stroma and follicle, and a slower phase of equilibrium and clearance between the follicle and lumen (Chow and Woodbury, 1970; Yu et al., 2002). A two-compartmental thyroid failed to fit this behavior (see Results section). Similarly, several compartmental models for iodide have also described the necessity for a three-compartmental thyroid (Lee et al., 1982), which captures the early phase of iodide uptake, consisting of trapping and organification processes and a slower phase of hormone accumulation and secretion.
The major difference in the blood compartments is that there is considerable reversible binding between plasma proteins and ClO4- but not I-. Passive diffusion between plasma and red blood cells (RBCs), however, occurs for both anions. Equilibrium dialysis studies in rat plasma performed at the University of Georgia show greater than 99% binding of ClO4- in plasma at concentrations 
100 mg/l and approximately 50% bound at concentrations 
500 mg/l (see Table 1) (J. W. Fisher, personal communication). Michaelis-Menten (M-M) kinetics were used in the model to describe the association of the free ClO4- fraction to unspecific plasma binding sites, and a first-order rate was used for the dissociation. In the case of iodide, shortly after an injection of radiolabeled iodide, 88% of plasma iodide radioactivity exists as free unorganified iodide (Yu et al., 2002). The remaining fraction is incorporated into either thyroid hormones or nonhormonal iodinated proteins and is analytically indistinguishable from the "free" iodide. Unlike ClO4-, however, both the free and incorporated iodine fractions are taken up into tissues. Therefore, for simplifying the interpretation of such radiolabeled iodine data, the model lumps the fractions of free and bound iodide into a single pool for "total iodide" in the plasma. Thus, it was not necessary to include binding of I- to serum proteins in order to describe radioiodine kinetics at this point. In the thyroid itself, however, it is important to distinguish between free and organified (incorporated) iodide, due to the fact that approximately 90% is incorporated (Yu et al., 2002).
|
Active uptake of ClO4- and I- by NIS in the thyroid, skin, and gastric mucosa and by the apical iodide channel in the thyroid was described using Michaelis-Menten (M-M) kinetics for nonlinear processes (Fig. 1
, bold arrows) (Chow et al., 1969; Kotani et al., 1998; Wolff, 1998). Permeability area cross-products and partition coefficients were used to describe the movement of the anions (ClO4- and I-) between the capillary bed, tissue, and inner compartments (Fig. 1, small arrows), which results from the inherent electrochemical gradient within the tissues. First-order clearance rates were used to describe the organification of iodide in the thyroid and the subsequent release of the bound iodide into systemic circulation.
Passive diffusion using partitions and blood flows were used to describe movement of both anions into the kidney, liver, and fat, because these tissues do not contain NIS. Urinary clearances were modeled using first-order clearance rates from the kidneys. Fecal excretion of both anions was not significant (Hays and Wegner, 1965; Yu et al., 2002). Although the current model does not encompass thyroid hormone regulation, the liver is the major site of extrathyroidal deiodination and, therefore, was maintained as a separate compartment for future model development. Fat was included as an essentially exclusionary compartment because large variations in fat content among humans, as well as dramatic changes during growth and reproduction in females, could alter ClO4-/I- kinetics. This enhanced the model’s capacity for extrapolations to other species, genders, or reproductive stages.
Physiological parameters.
Tissue volumes, V(s), and blood flows, Q(s), for most compartments were derived from Brown et al.(1997), except for the cellular dimensions of the thyroid, which were obtained from Malendowicz and Bednarek (1986) (Table 2). Tissue volumes were scaled linearly with BW, and Q(s) was multiplied by BW3/4.
|
Chemical-specific parameters.
The values and sources of chemical-specific parameters used in the model are provided in Table 3
. Partition coefficients (P) and blood flows (Q) were used to describe compartments with flow-limited diffusion. Partition coefficients for both anions were calculated from tissue/plasma ratios at 24-h time points from Yu et al. (2002), when available. Tissue/plasma concentrations that were not available from Yu et al. (2002), such as those for iodide partitioning into kidney, liver, skin, and RBCs, were obtained from other literature sources (see Table 3). In the case of I- partitioning into fat, data were not available. Therefore, due to similar polarity, the fat/blood value reported by Pena et al.(1976) for ClO4- was used. Clearance values (Cl) for first-order rates of urinary excretion, dissociation rates from serum binding proteins, rates of organification of thyroid iodide, and secretion of bound iodine from the thyroid into systemic circulation were determined by fitting model simulations to available time-course data at various doses.
|
For compartments with nonlinear uptake of the anions, effective partition coefficients (P) (reflecting either approximate tissue/serum ratios or electrical potential gradients) and permeability area cross-products, PA(s), were used to describe passive diffusion. Simulations reveal that alterations in PA values have a direct impact on the uptake portion of the curve, with lower values indicating slower uptake. Therefore, PA(s) were visually optimized to the uptake portion of the time-course data after setting P to the experimentally determined values in Table 3
.
Measured anion concentration gradients between the cellular subcompartments of the thyroid were not available. Therefore, effective partitions of the anions between the cellular subcompartments of the thyroid (stroma/follicle and follicle/lumen) were based on electrochemical gradients measured by Chow and Woodbury (1970) at three doses of ClO4-. The difference between the stroma and follicle can be interpreted as an effective P for charged moieties, such as ClO4- and I-, hindering the entry of negatively charged ions into the follicle. The approximately equal and opposite potential from the follicle to the colloid enhances passage of the anions into the colloid, creating an effective P > 1. The equivalence between electrical potential differences, fi–fo, and ionic concentrations, ci and co, were estimated in the manner of Kotyk and Janacek (1977), as follows:
 |
where R is the gas constant, T is the absolute temperature, F is the Faraday constant, and subscripts i and o refer to "inside" and "outside" the cellular membranes. At 37°C (2.303 RTF = 61.6 mV) for a singly charged ion (z = 1), this becomes:
 |
In general, the ratio of species concentrations between two media is given by the partition coefficient (Kp), where Kp = coci. Thus, Equation 2 becomes:
 |
From Chow and Woodbury (1970), the potential difference for the stroma/follicle interface ranges from -58 to -51 mV. Therefore, from the above equation, Kp for a monovalent, negatively charged ion is between 0.114 and 0.149. Similarly, for the follicle/lumen interface, fi-fo ranges from +50 to +58 mV, rendering the effective Kp between 6.48 and 8.74 (Table 3). These values were also used to describe the effective partitioning of I-, because both anions have the same ionic charge.
An affinity constant (Km) of 4.0 ± 1.2 x 106 ng/l for I- at the NIS was derived by Gluzman and Niepomniszcze (1983) from thyroid slices of five normal humans. They noted little variation in Km(s) between thyroid specimens from other species. For example, average Km(s) for bovine and porcine specimens were 5.0 and 3.9 x 106 ng/l, respectively (Gluzman and Niepomniszcze, 1983) and Km(s) from rat thyroid specimens varied from 3.17 to 10.2 x 106 ng/l (Wolff, 1964). Wolff and Maurey (1963) found that iodide’s affinity for NIS varied little across different tissues. This is supported by Kosugi et al. (1996), who reported a Km for iodide of 4.4 x 106 ng/l in Chinese hamster ovary cells. Therefore, the Km value for I- measured by Gluzman and Niepomniszcze (1983) was used in all NIS compartments.
Both anions compete with each other for the NIS receptor. However, ClO4- is known to have a much greater affinity for NIS than I- and, thus, is preferentially transported (Wolff, 1998; Wolff and Maurey, 1963). Several studies support this. Halmi and Stuelke (1959), found ClO4- to be 10 times as effective as I- in depressing thyroid and gastric juice to blood I- ratios in the rat. Similarly, Harden et al.(1968) found human saliva/plasma radioiodide concentrations to be seven times lower after equimolar doses of ClO4- than after I-. Lazarus et al.(1974) also demonstrated this effect on saliva/plasma iodide in mice. Both the salivary glands and stomach are useful for estimating inhibitory effects on iodide uptake by monovalent anions at NIS, because organic binding of iodide does not occur in these tissues. Lastly, Kosugi et al. (1996) actually measured a ClO4- Km of 1.5 x 105 ng/l in Chinese hamster ovary cells. Based on these findings, the measured Km reported by Kosugi et al. (1996) was initially used. However, visual optimization of thyroid ClO4- data indicated that a slightly higher Km value of 1.8 x 105 ng/l was appropriate (Table 3). Therefore, due to perchlorate’s considerably lower Km and, thus, its greater affinity for NIS than that of I-, inhibition of ClO4- uptake by I- is insignificant and not included in the model.
The apical membrane of the thyroid follicle exhibits an additional I- transport channel. A Km of approximately 4.0 x 109 ng/l, describing I- transport from the bovine thyroid follicle to the colloid (KmTLi in our model notation) was estimated by Golstein et al.(1992). Using this value, however, the model underestimates thyroid iodide concentrations at and beyond 8 h post iv-dosing. A slightly lower Km of 1.0 x 109 ng/l was required to fit thyroid iodide data. As with NIS, this apical channel also appears to be sensitive to inhibition by ClO4- (Golstein et al., 1992). Model simulations of both thyroid I- inhibition and thyroidal ClO4- levels supported a KmTLp value of 1.0 x 108 ng/l, approximately 10 times less than that of I- (KmTLi). Radiolabeled perchlorate data presented by Chow and Woodbury (1970) also supports this value. After ip administration of 200 mg 36ClO4-/kg in rats, Chow and Woodbury measured thyroid saturation at approximately 1.8 x 108 ng/l, indicating saturation of the lumen.
Whereas the Km is similar across NIS-containing tissues and species, maximum velocity (Vmax) varies significantly (Gluzman and Niepomniszcze, 1983; Wolff, 1998; Wolff and Maurey, 1961). Simulations revealed Vmax(s) having a greater effect on tissue clearance than on uptake. Tissue clearance is also affected by substrate concentration and blood flow; however, measured values are used for these parameters. Therefore, due to the lack of reported values, ClO4- Vmaxc(s) for the thyroid, stomach, and skin were estimated by visually optimizing the clearance portion of the 36ClO4- time-course data and from thyroid measurements obtained from the drinking water data by Yu et al.(2002), at dose levels exhibiting saturation (Figs. 2A, 2D, 2E; also Fig. 6B in Results section). ClO4- demonstrated saturability between 1.0 and 3.0 mg/kg/day. Iodide Vmaxc(s) for the same compartments were also visually fit to the clearance portion of iodide time-course kinetics (Figs. 3 and 4).
|
|
|
|
TSH-induced upregulation has been shown to increase the Vmax at thyroidal NIS by increasing both its activity and half-life (Riedel et al., 2001
). Upregulation of thyroid NIS was occurring during the ClO4- drinking water studies, as revealed by elevated TSH levels across doses and compensation of thyroid iodide inhibition (Yu et al., 2002). To fit the upregulated thyroid uptake, thyroid ClO4- and I- levels were first visually optimized by increasing the Vmax(s) at the NIS of the thyroid follicular membrane (VmaxTFp and VmaxcTFi) to fit measured values at each dose. The "upregulated" VmaxTF(s) were plotted against the predicted corresponding free ClO4- levels in serum. The equations from the resulting M-M type curves were used to calculate "upregulated" VmaxTFp and VmaxcTFi in the model.
Sensitivity analysis.
An analysis of model parameter sensitivity on the predicted area under the curve (AUC) of serum ClO4- concentrations was performed. Using a 1% increase in each chemical-specific parameter value, the model was run at two drinking water ClO4- doses, above and below NIS saturation (0.1 and 10 mg/kg/day). The sensitivity coefficient for each parameter was then calculated, using Equation 4.
 |
Where A equals serum ClO4- AUC with 1% increased parameter value, B equals serum ClO4- AUC using original parameter value, C equals parameter value increased 1% from original value, and D equals the original parameter value.
Allometric scaling and equations.
For ongoing extrapolations of the model to humans, allometric scaling was applied by multiplying all Vmaxc(s), PA(s), and Cl(s) by BW3/4. Simultaneous differential equations in the model code were written and computed using ACSL (Advanced Continuous Simulation Language) software (AEgis Technologies, Huntsville, AL). Example equations demonstrating both diffusion-limited uptake, using P(s) and PA(s), first-order clearance values, Cl(s), and saturable uptake using M-M parameters are presented in the Appendix.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Parameterization
Radiolabeled perchlorate 36ClO4- time-course kinetics.
Model simulations of various tissues and serum concentrations were derived using the physiologic and chemical-specific parameters provided in Tables 1
and 2. Figure 2 shows simulations of 36ClO4- in various tissues after an iv dose (3.3 mg/kg). Using partitions (P) and blood flows (Q) derived from the literature, excellent reproductions of the 36ClO4- kinetics in all flow-limited tissues were achieved, without any fitting required. The stomach, skin, and thyroid concentrations were simulated by using literature-derived Km and effective P values and by visually optimizing the PA(s) and Vmaxc(s), as described above. The model reproduced each data set with only slightly overpredicting data in the clearance portion of the thyroid time-course data (Fig. 2A).
Radiolabeled iodide (125I-) time-course kinetics.
As previously stated, the amount of I- in the simulations represents total iodine in all tissues except the thyroid, where both free I- and bound iodine are shown (where data are available). Model-predicted I- kinetics (from 0.033 mg/kg 125I- iv) in the thyroid, serum, stomach contents, skin, and urine are shown versus measured data in Figures 3 and 4. Although stomach and skin data displayed wide variation, the simulated amount of 125I- in stomach contents and skin indicated rapid uptake and gradual reabsorption, suggesting that these tissues play an important role in iodide turnover. The model accurately simulates the amount of 125I- in the thyroid over 96 h and in serum over 24 h after dosing (Figs. 3A and 4B).
Perchlorate drinking water kinetics.
To simulate drinking water exposure, the model was programmed to provide a continuous oral dose for 12 h/day, assuming rats drink throughout their waking hours. Without taking plasma binding into account, the model underpredicts serum concentrations at 0.1 mg/kg/day and lower. Therefore, binding parameters (KmBp, VmaxcBp, and Clunbp) were estimated by fitting serum simulations to data from 0.1 and 1.0 mg/kg/day, in accordance with the unpublished binding data by Fisher (personal communication). Using these binding parameters with the other ClO4- parameters describing acute 36ClO4- kinetics, the model successfully predicted other serum concentrations from subchronic exposure, ranging from 0.01 to 10.0 mg/kg/day (Fig. 5). Upregulation of the thyroid was evident from elevated TSH levels in the drinking water studies across doses (Yu et al., 2002). However, this upregulation of the thyroid does not significantly affect serum concentrations, due to the small size of the gland. Additionally, TSH does not regulate NIS activity in any tissues other than the thyroid (Brown-Grant, 1961; Cavalieri, 1997; Spitzweg et al., 2000). Although not of consequence in this model, upregulation of mammary gland NIS expression has been noted in lactating rats with normal TSH levels and appears to be regulated by prolactin and/or oxytocin (Spitzweg et al., 2000), inferring that other agents may stimulate NIS expression and function. Thus, parameters for the extrathyroidal tissues should not change from those determined from the kinetic data. This is consistent with our model; had VmaxGJ and VmaxSk changed between acute and subchronic exposures, significant reductions in serum levels, which are not observed, would be noted.
|
As stated above, TSH-induced upregulation of thyroid NIS was apparent in the drinking water studies. As expected, using the acute thyroid parameters, the model underpredicted thyroid ClO4- concentrations at 3.0 mg/kg/day and higher (Fig. 6A
). By using the "upregulated" VmaxTFp, described in the Materials and Methods section, the model successfully predicted thyroid ClO4- concentrations across doses at day 14 (336 h) (Fig. 6B). However, the dose-dependent equation does not account for the time lag of induction, demonstrated by the gradual increase in ClO4- uptake between days 1 and 5 at 3.0 and 10.0 mg/kg/day, suggesting that NIS induction by ClO4- is both time- and dose-dependent.
Model Validation
Using the parameters estimated from the various time-course data presented above, the model’s ability to predict thyroid inhibition data from Yu et al.(2002) was tested. In addition, model predictions were compared with available 36ClO4- and radioiodide data from the literature.
Inhibition of thyroid I- uptake after ClO4- iv dosing.
The model successfully simulated inhibition of thyroid 125I- uptake following iv doses of 0, 0.01, 0.1, 1.0, and 3.0 mg/kg ClO4- (Fig. 7) using parameters established from the acute 36ClO4- and subchronic drinking water data. The measured means ± SD for 125I- inhibition in the thyroid were 13.0 ± 18.7, 23.6 ± 12.6, 69.7 ± 9.5, and 87.5 ± 2.0% at 2 h post 125I- dosing (4 h post ClO4-) (Yu et al., 2002) versus model-predicted reductions of 1.7%, 18%, 76%, and 91% from the predicted radioiodide uptake in controls (without ClO4-) at the same time point for the 0.01, 0.1, 1.0, and 3.0 mg/kg dose groups, respectively. The model simulations lie within the range of measured data, with exception of slight overprediction of inhibition at 3.0 mg/kg at 4 h post ClO4-.
|
Inhibition of thyroid 125I- uptake after drinking water exposure to ClO4-.
TSH-induced upregulation of NIS compensated for competitive inhibition of thyroid I- uptake by ClO4- across doses in the drinking water study. Only slight inhibition was seen at 10 and 30 mg ClO4-/kg/day on day 14 (Yu et al., 2002
). The increased NIS activity was modeled using the upregulated VmaxTFi and VmaxTFp, calculated using M-M equations dependent on serum-free ClO4- concentrations, as described in the Materials and Methods section. Using the upregulated VmaxTFi and VmaxTFp values, the model restores thyroid uptake of radioiodide across doses to control levels.
Other ClO4- and radioiodide studies used for model validation.
Using the parameters developed with the data primarily from Yu et al.(2002), the model predicted various tissue concentrations of ClO4- and radioiodide from other studies. Urinary ClO4- levels measured by Eichler (1929) were successfully predicted at doses of 1.6, 8.0, and 49 mg/kg (Fig. 8). Chow and Woodbury (1970) gave 0.5, 10.0, and 200 mg 36ClO4-/kg by ip administration to rats and measured radioactivity in their thyroids and serum. The rats in this study were functionally nephrectomized by ligation of the renal pedicle of both kidneys. Therefore, to simulate this condition, the model’s urinary ClO4- excretion (ClUp) was set to zero. The model adequately predicted thyroid concentrations across doses (Fig. 9A). Predicted serum concentrations at 0.5 and 10.0 mg/kg were within a factor of 2 from the data (Fig. 9B).
|
|
Finally, the model also reproduced iodide data from Perlman et al.(1941)
and Wolff (1951). Perlman et al. studied iodide distribution in rats after administering 2.63 mg/kg labeled iodine via stomach tubes. The model predicted their time-course data for liver accurately and their kidney data within a factor of less than 1.3 (Fig. 10). Wolff (1951) measured thyroid iodide clearance in male rats up to 210 h after an ip injection of 33.0 mCi 131I- (approximately 0.88 ng/kg). Again, the model successfully predicted the kinetic behavior of I- in the thyroid (Fig. 11).
|
|
Sensitivity Analysis
Sensitivity analysis on serum ClO4- AUC levels at 0.1 mg/kg/day indicated plasma binding parameters for maximum capacity (VmaxBp) and the first-order dissociation rate (Clunbp) had the greatest influence, with sensitivity coefficients of 0.5 and -0.32, respectively. Skin parameters exhibited the next greatest effect on serum ClO4- AUC, with sensitivity coefficients of -0.29, -0.22, and 0.21 for the skin/plasma partition coefficient (PSkp), VmaxSkp, and PASkp, respectively. Urinary clearance (ClUp) followed, with a sensitivity coefficient of -0.12. Sensitivity coefficients of all other parameters were at or below an absolute value of 0.01.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
Model predictions of internal doses of ClO4- show excellent correspondence with measured tissue concentrations. Thus, we have confidence in the chosen model description of ClO4- and I- kinetics. The model also stipulates a specific mechanism for the interaction of perchlorate with iodide at the thyroid and other tissues, particularly its competition for transport by the NIS. In this case, too, the model has been found to agree with the available experimental data related to iodide uptake inhibition after acute ClO4- dosing, ranging from 0.01 to 3.0 mg/kg, and after subchronic exposure to ClO4- via drinking water.
The closeness of parameter values describing passive diffusion indicates that the kinetics of I- and ClO4- are generally similar. In addition, urinary clearances were very close and agreed with Eichler and Hackenthal (1962), in that urinary clearance of ClO4- was found to be slightly higher than that of I-. Values for saturable uptake, however, are somewhat different between anions. Perchlorate exhibits a greater affinity for NIS than does I-, with a Km approximately 10 times lower than that of I-. Although both anions can compete for NIS, perchlorate’s lower Km renders inhibition of ClO4- uptake by iodide insignificant, as described by the M-M equations and as supported by the inhibition data. Vmax(s) and PA(s) for the skin and stomach, however, were similar between anions, with the exception of VmaxG(s), for anion transport into stomach contents, which was an order of magnitude higher for iodide than for ClO4-. However, this value was based on limited and highly variable data. Thus, it is possible that additional data would suggest that the VmaxG(s) (for gastric uptake of I- and ClO4-) are actually closer.
Thyroid parameters showed the greatest difference between the two anions, where VmaxcTF (for transport into the thyroid follicle) and VmaxcTL (for transport into the colloid or lumen) were approximately one and three orders of magnitude greater for I- than for ClO4-, respectively. In addition, a lower permeability area cross-product (PATL) was required to describe diffusion between the follicle and lumen for I- than for ClO4- (0.0001 vs. 0.01 l/h/kg). This slower uptake and clearance of iodide is expected, due to I- organification in the follicle and hormone storage in the colloid. The storage of organified I- also results in much higher thyroid/serum concentrations than those of ClO4-. The rate of organification of I- (Clhormi) is relatively fast (0.1/h), whereas the rate of secretion of organified I- into systemic circulation (Clsecri) is very slow (9.5 x 10-7/h), thus resulting in a buildup of organic iodine that would not occur with ClO4-.
Serum kinetics were quite different between the anions, due to the large amount of nonspecific plasma binding of ClO4- at low doses. The literature suggests that albumin and prealbumin are the major serum binding proteins for ClO4-. Shishiba et al.(1970) found that ClO4- interferes with binding of free thyroxine (fT4) to prealbumin and albumin but not to thyroid binding globulin (TBG) in human blood. As stated earlier, T4 was also displaced in rat serum after ClO4- administration (Yamada, 1967; Yamada and Jones, 1968). The ClO4- data suggest that more than one binding site in serum is present. Model-predicted serum ClO4- values at 0.1 mg/kg/day were at the low end of the data, and it is possible that multiple plasma binding sites are responsible.
To fit the TSH-upregulated thyroid ClO4- concentrations resulting from high drinking water exposures, it was necessary to use a dose-dependent increase in VmaxcTFp. However, it is evident from the increase in thyroid ClO4- between days 1 and 5 at 3.0 mg/kg/day and higher that the dose-dependent VmaxTFp should also be time-dependent in order to capture changing NIS activity. The fact that full compensation of thyroidal 125I- inhibition was not seen on day 14 in the 10.0 and 30.0 mg/kg/day of ClO4- also suggests that a similar time-dependent saturable function is also required for VmaxTFi. Other subtle changes known to result from TSH stimulation include increased follicle size (Conde et al., 1991; Ginda et al., 2000), total protein, and RNA and DNA content (Pisarev and Kleiman de Pisarev, 1980). Bagchi and Fawcett (1973) suggest that variations in Vmax are also affected by intrathyroidal I- and the magnitude of I- efflux.
The slight underprediction of Chow and Woodbury’s (1970) serum ClO4- levels, from the nephrectomized rats (Fig. 9A), may suggest increased plasma binding under those conditions. Measured serum concentrations at each dose level exceed the plasma binding KmBp of 1.1 x 104 ng/l. However, it does not necessarily suggest that increased plasma binding is required under normal physiologic conditions because the model adequately simulates 36ClO4- concentrations in various tissues and serum at lower doses. Analytical differences in measurement between Chow and Woodbury and Yu et al.(2002) may be responsible. Another possibility is that blocking kidney discharge, as done by Chow and Woodbury, creates physiologic effects, such as increased extracellular Na+ (Tietz et al., 1990), which may possibly affect binding to an extent that the model cannot account for by simply eliminating urinary clearance.
Chow and Woodbury’s thyroid uptake data displayed an interesting decrease in thyroid saturation with dose (Fig. 9B). A 20-fold increase in dose, from 0.5 to 10.0 mg/kg, yielded a small increase (factor of 2) in thyroid 36ClO4- concentration, whereas an additional 20-fold increase in dose (200 mg/kg) resulted in a 10-fold increase in thyroid concentration. The model can physiologically describe this phenomenon, due to the three-compartmental thyroid, which allows sequestration of the anions tens of times greater than serum concentrations. At 200 mg/kg, the thyroid reaches an approximate 36ClO4- concentration of 1.8 x 108 ng/l. Although this concentration is three orders of magnitude higher than the Km of NIS (KmTFp), 1.8 x 105 ng/l, it approximately equals the Km of the apical channel (KmTLp), 1 x 108 ng/l. Therefore, the jump in thyroid concentration measured by Chow and Woodbury reflects saturation of the lumen compartment and supports our estimated KmTLp. The model, however, slightly underpredicts the rapid uptake during the first hour post dosing at 0.5 and 10.0 mg/kg. It is possible that nephrectomization may also have had an effect on the rate of 36ClO4- uptake in the thyroid.
Accounting for free and bound iodide in the thyroid and inhibition of NIS iodide transport by perchlorate are initial steps toward modeling a complex regulatory process for thyroid hormone production. Free and bound thyroidal radioiodide were accurately predicted from an iv dose of 33 µg/kg. In addition, the model accurately predicted total thyroidal iodide levels from an ip dose as low as 0.88 ng/kg. The current model, based on the measured biologic and physiologic characteristics of the organism, is readily extrapolated across species and gender (Clewell et al., 2001). Most importantly, the model accurately predicts serum and thyroid perchlorate levels from subchronic drinking water exposures (the most common route of human exposure) from 0.01 to 30 mg/kg/day (90 to 270,000 ppb). Perchlorate-induced inhibition in humans after 2 weeks of exposure via drinking water, from 0.007 to 0.5 mg/kg/day (250 to 17,500 ppb, assuming water consumption of 2.0 l/day) (Greer et al., 2002), has been shown to be more similar to the acute inhibition seen in the rat (post iv-dosing with ClO4-, from 0.01 to 3.0 mg/kg), which this model predicts satisfactorily. However, after 2 weeks of ClO4- exposure, upregulation and hormone perturbations were not yet seen in the humans. Clearly, the human’s iodide economy is more efficient than the rat’s, likely due to greater plasma binding of thyroid hormones (Dohler et al., 1979), but the extent of low-level perchlorate exposure required for upregulation and hormone perturbations in humans is unknown. Therefore, model-based predictions of serum perchlorate from subchronic exposure and inhibition from acute exposures may be highly valuable in risk assessment, especially in identifying dose levels that could potentially cause adverse effects, given prolonged exposure.
Similar models, incorporating physiologic changes in rats during pregnancy and neonatal growth, have been developed concurrently by Clewell et al. (in press). These models were validated with both acute radioiodide doses, ranging from 0.005 to 33 mg/kg, and ClO4- drinking water data within the same dose range as those used to validate this model. Overall, model parameters were consistent with those of this model, instilling further confidence in the parameters established here. Differences found between sex and/or reproductive stages were all physiologically logical. For example, during pregnancy, estrogen levels rise sharply (Iino and Greer, 1961), resulting in increased serum proteins and greater plasma binding of ClO4- in the pregnant rat. The models also account for subtle, yet important sex differences in thyroid morphology. In the male rat, the follicle and colloid comprise approximately 60% and 24% of the thyroid volume, whereas in the pregnant dam, the follicle and colloid both comprise approximately 45% of the thyroid volume (Conde et al., 1991; Malendowicz and Bednarek, 1986), reflecting the capacity for increased hormone production during gestation. Together, these models enable life-stage differences to be quantified and predictions of internal ClO4- dose (and its effect on iodide incorporation into the thyroid) to be made (Clewell et al., 2001).
This model provides the foundation for development of a more complex physiologically based dynamic model that would account for hormone control systems and provide a basis for the design of further experimental investigations. Finally, by extrapolating to specific human exposure situations (including particularly susceptible individuals, such as pregnant women or infants), such extended models will provide the basis for more accurate and reliable risk assessments of the effects of perchlorate on thyroid hormone production and homeostasis.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONAPPENDIXREFERENCES |
|---|
EXAMPLE OF DIFFERENTIAL EQUATIONS USED IN THE PBPK MODEL TO DESCRIBE CLO4- AND I- DISTRIBUTION
1. Rate of change (ng/h) in the amount (RAXi) of I- or ClO4- in non-NIS tissues (X):
 |
where Q is the blood flow to Xth tissue, and CAi is the concentration of either I- or ClO4- (identified by subscripts i or p, respectively) in the arterial blood. CXi is concentration of iodide (i) or perchlorate (p) in Xth tissue.
2. Rates of change (ng/h) in amount of I- in the thyroid stroma, follicle, and colloid (lumen) (RATSi, RATFi and RATLi, respectively) plus the amount of bound thyroid I-, RAbndi:
 |
 |
 |
 |
 |
 |
where subscripts i and p identify the anion I- or ClO4-, QT is thyroid blood flow (l/h), CAi is the arterial blood concentration (ng/l), CVTSi,p is the thyroid stroma concentration (ng/L) and CTFi,p is the follicular concentration (ng/l) of iodide or perchlorate. PTFi, PTLi, PATFi, and PATLi are the partition coefficients and permeability cross-products describing passive diffusion of I- across the basal (follicle/stroma) and apical (lumen/follicle) membranes. M-M equations used to describe the rates of active uptake of I- in the follicle and colloid (RupTFi and RupTLi, respectively), include inhibition by ClO4- at NIS and the apical iodide channel. VmaxTFi, VmaxTLi, KmTFi,p, and KmTLi,p are the maximum velocities (ng/h/kg) and affinity constants (ng/l) for transport of I- or ClO4- into the follicle and lumen. Clhormi and Clsecri are first-order clearance values (h-1) for the organification of iodide into thyroid hormones and the secretion of organified iodide into systemic circulation. Thyroid transport of ClO4- is calculated similarly, with the exception that organification and inhibition of ClO4- uptake by iodide are not included. As described earlier, due to iodide’s lesser affinity (10-fold higher Km) than ClO4-, it does