Competitive Inhibition of Thyroidal Uptake of Dietary Iodide by Perchlorate Does Not Describe Perturbations in Rat Serum Total T4 and TSH

Background Perchlorate (ClO4−) is an environmental contaminant known to disrupt the thyroid axis of many terrestrial and aquatic species. ClO4− competitively inhibits iodide uptake into the thyroid at the sodium/iodide symporter and disrupts hypothalamic–pituitary–thyroid (HPT) axis homeostasis in rodents. Objective We evaluated the proposed mode of action for ClO4−-induced rat HPT axis perturbations using a biologically based dose–response (BBDR) model of the HPT axis coupled with a physiologically based pharmacokinetic model of ClO4−. Methods We configured a BBDR-HPT/ClO4− model to describe competitive inhibition of thyroidal uptake of dietary iodide by ClO4− and used it to simulate published adult rat drinking water studies. We compared model-predicted serum thyroid-stimulating hormone (TSH) and total thyroxine (TT4) concentrations with experimental observations reported in these ClO4− drinking water studies. Results The BBDR-HPT/ClO4− model failed to predict the ClO4−-induced onset of disturbances in the HPT axis. Using ClO4− inhibition of dietary iodide uptake into the thyroid, the model underpredicted both the rapid decrease in serum TT4 concentrations and the rise in serum TSH concentrations. Conclusions Assuming only competitive inhibition of thyroidal uptake of dietary iodide, BBDR-HPT/ClO4− model calculations were inconsistent with the rapid decrease in serum TT4 and the corresponding increase in serum TSH. Availability of bound iodide in the thyroid gland governed the rate of hormone secretion from the thyroid. ClO4− is translocated into the thyroid gland, where it may act directly or indirectly on thyroid hormone synthesis/secretion in the rat. The rate of decline in serum TT4 in these studies after 1 day of treatment with ClO4− appeared consistent with a reduction in thyroid hormone production/secretion. This research demonstrates the utility of a biologically based model to evaluate a proposed mode of action for ClO4− in a complex biological process.

Perchlorate (ClO 4 -), a water-soluble chemical distributed extensively in the environment, has caused public health concerns because of widespread exposures to populations by ingestion of ClO 4 --contaminated food and water. The ClO 4 anion has recently been classified as a ubiquitous environmental contaminant throughout the United States, with detectable concentrations in many drinking water supplies (Motzer 2001) and food and beverage products (e.g., milk, lettuce, grains) (El Aribi et al. 2006), and is also found as a contaminant in dietary supplements (Snyder et al. 2006). The occurrence of ClO 4 in the environment is most often attributed to anthropogenic uses of ClO 4 salts as oxidizers in solid rocket propellants and application of Chilean nitrate fertilizers (Motzer 2001); however, ClO 4 is also formed naturally by atmospheric processes (Dasgupta et al. 2005). The primary health concern for environmental exposures to ClO 4 is adverse (and irreversible) neuro developmental outcomes mediated by disruption of the hypothalamic-pituitarythyroid (HPT) axis in the immature fetus or child. Adequate levels of thyroid hormones are essential for proper growth, develop ment, reproduction, and metabolism.
ClO 4 blocks thyroidal uptake of radiolabeled iodide and alters thyroid hormone homeostasis [an increase in serum thyroidstimulating hormone (TSH) concentrations and a decline in serum thyroxine (T 4 ) concentrations] (National Research Council 2005;U.S. Environmental Protection Agency 2002). Dietary iodide is an essential nutrient used in the formation of thyroid hormones, and if in short supply, from the blocking effect of ClO 4 -, an iodide-deficiency-induced decrease in thyroid hormone production is thought to occur. For humans, under clinical conditions, ClO 4 clearly inhibits thyroidal uptake of radio labeled iodide (Greer et al. 2002); however, disruption of the HPT axis by ClO 4 remains to be clearly demonstrated in euthyroid adults. Interestingly, ClO 4 was used to treat thyrotoxicosis (overactive thyroid gland) with large doses, ranging from 17-29 mg/kg, until several cases of fatal aplastic anemia occurred in the early 1960s. The mechanism of action for this adverse effect of ClO 4 is unknown. ClO 4 was also used clinically to test thyroid function, referred to as the ClO 4 discharge test. For this test, radio labeled tracer iodide is administered a few hours before dosing with 2-13 mg/kg ClO 4 - (Wolff 1998), causing blocking of active uptake of radiolabeled iodide for several hours. By 2 hr after dosing, the thyroid gland has sequestered radio labeled iodide, but circulating levels of the iodide remain in the body and are available for sequestration into the thyroid gland. If the thyroid gland is defective in binding (trapping) iodide, rapid and excessive efflux of radio labeled iodide from the thyroid is observed in the presence of ClO 4 -. The primary mechanism by which ClO 4 blocks thyroidal uptake of iodide is competitive inhibition at the sodium/iodide symporter (NIS) protein. The NIS protein is responsible for active translocation of both iodide and ClO 4 from circulating blood to the thyroid follicle. Bidirectional passive diffusion of iodide and ClO 4 also occurs between the thyroid gland and the blood supply perfusing the thyroid. When the ClO 4 discharge test is administered, a possible mechanism of action for efflux of radiolabeled iodide from defective thyroid glands may be simple diffusion of radiolabeled iodide down a concentration gradient from the thyroid gland into the blood supply because the NIS "pump" that maintains an iodide gradient between the blood and the thyroid gland is shut down by ClO 4 -. A rapid efflux of radiolabeled iodide has been observed in cell preparations deficient in the binding protein, thyroglobulin (Kosugi et al. 1996).
Several decades ago, in vitro studies were conducted in Sprague-Dawley rat thyroid Background: Perchlorate (ClO 4 -) is an environmental contaminant known to disrupt the thyroid axis of many terrestrial and aquatic species. ClO 4 competitively inhibits iodide uptake into the thyroid at the sodium/iodide symporter and disrupts hypothalamic-pituitary-thyroid (HPT) axis homeostasis in rodents. oBjective: We evaluated the proposed mode of action for ClO 4 --induced rat HPT axis perturbations using a biologically based dose-response (BBDR) model of the HPT axis coupled with a physiologically based pharmacokinetic model of ClO 4 -.

Methods:
We configured a BBDR-HPT/ClO 4 model to describe competitive inhibition of thyroidal uptake of dietary iodide by ClO 4 and used it to simulate published adult rat drinking water studies. We compared model-predicted serum thyroid-stimulating hormone (TSH) and total thyroxine (TT 4 ) concentrations with experimental observations reported in these ClO 4 drinking water studies.

results:
The BBDR-HPT/ClO 4 model failed to predict the ClO 4 --induced onset of disturbances in the HPT axis. Using ClO 4 inhibition of dietary iodide uptake into the thyroid, the model underpredicted both the rapid decrease in serum TT 4 concentrations and the rise in serum TSH concentrations. conclusions: Assuming only competitive inhibition of thyroidal uptake of dietary iodide, BBDR-HPT/ClO 4 model calculations were inconsistent with the rapid decrease in serum TT 4 and the corresponding increase in serum TSH. Availability of bound iodide in the thyroid gland governed the rate of hormone secretion from the thyroid. ClO 4 is translocated into the thyroid gland, where it may act directly or indirectly on thyroid hormone synthesis/secretion in the rat. The rate of decline in serum TT 4 in these studies after 1 day of treatment with ClO 4 appeared consistent with a reduction in thyroid hormone production/secretion. This research demonstrates the utility of a biologically based model to evaluate a proposed mode of action for ClO 4 in a complex biological process. homogenates to examine the effects of antithyroid compounds on iodide peroxidase activity (Alexander 1959). For each antithyroid compound tested, thyroid homogenates were incubated with 1 µmol potassium iodide (KI) and 2 × 10 -3 M antithyroid compound for 2 hr. Thiocyanate, thiouracil, and cyanide were several of the antithyroid compounds shown to inhibit incorporation of 131 I into l-tyrosine, decreasing the formation of monoiodotyrosine (MIT) and diiodo tyrosine (DIT). However, ClO 4 did not signi ficantly alter the synthesis of MIT and DIT. Thus, the data suggest that ClO 4 does not alter organification of iodide, which is a key step in thyroid hormone synthesis (Alexander 1959). Several years later, Greer et al. (1966) provided contradicting results, demonstrating a direct effect of ClO 4 on the rat thyroid gland in vitro. These authors reported that ClO 4 reduced formation of MIT and DIT in rat thyroid lobes when incubated with ClO 4 -. They observed changes in MIT and DIT formation for ClO 4 media concentrations starting at 10 mg/L, with a reported 50% effective reduction of DIT at 250 mg/L. In addition to ClO 4 acting at the NIS protein of the thyroid gland, a few other studies have demonstrated another potential mode of action (MOA) for high doses of ClO 4 to disrupt the HPT axis. Using an in vitro system and very high concentrations of ClO 4 -(≥ 1,700 mg/L), Okabe and Hokaze (1993) reported that ClO 4 displaces T 4 bound to bovine serum albumin. This observation implies that free T 4 (fT 4 ) dislodged from serum proteins would then be cleared from the body more quickly, leading to a disruption of the HPT axis. In earlier in vivo studies, Yamada (1967), who was interested in this same MOA, removed the thyroid or gave methimazole [methylmercapto imidazole (MMI)] to rats to inhibit thyroid function (and treated them with T 4 ) and then evaluated the ClO 4 dose-dependent decline in serum protein-bound iodide (PBI; primarily radiolabeled T 4 ). The decline in serum PBI was interpreted as displacement of protein bound T 4 by ClO 4 -. The daily intake rates of ClO 4 were estimated to be 10-1,000 mg/kg/day. Historically, pharmacokinetic or computational analyses of the HPT axis have played a prominent role in efforts to quantify and understand the complex relationships between biological action of thyroid hormones and their production, metabolism, transport, distribution, and interaction with receptors (DiStefano and Landaw 1984;Oppenheimer 1983). Recently, physiologically based pharmaco kinetic (PBPK) models were developed to describe the kinetics of ClO 4 and radio labeled iodide and the interaction of ClO 4 on the thyroidal uptake of radiolabeled iodide in the adult rat (Fisher et al. 2000;Merrill et al. 2003). The rodent PBPK model described transport of ClO 4 into the thyroid gland by the NIS protein because high concentrations of ClO 4 were measured in thyroid tissue relative to serum (Yu et al. 2002). High concentrations of ClO 4 in the thyroid gland had also been observed in laboratory animals administered ClO 4 - (Chow et al. 1969;Chow and Woodbury 1970).
However, using an indirect electrochemical technique to infer movement of ClO 4 by the NIS into thyroid cells, Riedel et al. (2001) concluded that ClO 4 was not taken up into the thyroid gland because an electrical gradient was not created. Evidence for ClO 4 movement into the thyroid gland was reviewed by Wolff (1998) and Clewell et al. (2004). Finally, a recent study directly meas ured the TSH-mediated movement of ClO 4 into NIS-expressing FRTL-5 rat thyroid cells. This research provided conclusive evidence that the NIS protein did trans locate ClO 4 into the thyroid cells (Tran et al. 2008). At the same time, Dohan et al. (2007) concluded that ClO 4 is transferred by the NIS based on indirect lines of evidence using another anion (perrhenate) that is also transported with an electroneutral stoi chiome try into thyroid cells.
In the study we report here, using computational analysis, we tested the hypothesis that the primary MOA of ClO 4 on the HPT axis is competitive inhibition of uptake of thyroidal iodide. We combined our recently published biologically based dose-response (BBDR) model for the HPT axis (McLanahan et al. 2008), which we calibrated to describe perturbations resulting from dietary iodide deficiency, with a simple PBPK model for ClO 4 -. The published ClO 4 dose-response data sets for the adult rat HPT axis (Männistö et al. 1979;Yu et al. 2002) were simulated with our combined BBDR-HPT/ClO 4 models. Additionally, we conducted limited in vitro experiments at relevant ClO 4 plasma concentrations to evaluate the role of ClO 4 in displacement of T 4 from serum-binding proteins. The model provides for recycling of dietary iodide released from metabolism of thyroid hormones (dashed lines), as well as the stimulation and regulation of the HPT axis by TSH. Numbers indicate several key processes: 1, loss of thyroidal-bound iodide secreted as thyroid hormones; 2, TSH stimulation of NIS iodide uptake, organification of iodide, and stimulation of thyroid hormone production; 3, T 4 negative feedback on TSH production (thick solid lines); 4, formation of free iodide from serum and liver T 3 and T 4 metabolism; 5, phase II metabolism of T 4 and excretion into feces; and 6, fecal elimination of T 3 . For additional details on the BBDR-HPT axis model, see McLanahan et al. (2008). Bold arrows represent active transport and double thin arrows represent passive diffusion processes. Thin solid lines represent blood flows to or from the thyroid (Q T ) or the liver (Q L ).

Rate of TSH elimination
Volume of distribution

Materials and Methods
Laboratory experiments. Displacement of T 4 from serum proteins. Radiolabeled L-[3´,5´-125 I]thyroxine ( 125 I-T 4 ; specific activity, 1,500 µCi/µg; Perkin Elmer Life Sciences, Waltham, MA) was prepared by diluting 25 µCi (0.5 mL of 50 µCi/mL solution) with 1.0 mL 10% bovine serum albumin solution and predialyzing it against 2 mL 0.15 M phosphate buffer solution to reduce the amount of free 125 I in the solution. We spiked male rat and male human serum (both obtained from Bioreclamation Inc., Hicksville, NY) with ClO 4 -(Sigma Aldrich, Milwaukee, WI; 0.01 mL ClO 4 -/mL serum) diluted in saline to give a serum concentration of 0, 1.0, 10.0, 50.0, 100, 200, or 300 µg/mL, vortexed the solution, and allowed it to stand for 30 min. We then added 0.02 mL (200,000-300,000 cpm) predialyzed 125 I-T 4 solution/mL serum, vortexed the solution, and allowed it to stand for 30 min. One mL of spiked serum was dialyzed across a 6,000-kDa membrane overnight against 1.0 mL 0.15 M phosphate buffer (pH 7.3) at 37°C in a gently shaking water bath. Serum and dialysate were removed from each half-cell with a Pasteur pipette precoated with carrier solution. We immediately placed 750 µL dialysate into 750 µL carrier solution (3.0 mg/mL T 4 and 3.72 mg/mL NaI in 0.5 M NaOH) and vortexed it. Volume recovered was assessed in order to account for possible fluid shifts across the membrane. The dialysate/carrier solution was further processed by adding 1.5 mL magnesium precipitating solution (6.05 g/L Trizma base, 5.85 g/L NaCl, and 100 g/L MgCl 2 •6H 2 O, pH 9.274); the solution was vortexed and allowed to stand for 10 min. The precipitate was centrifuged for 5 min at 2,000 rpm and the supernatant was removed. The pellet was washed three times by resuspending it in 1.5 mL washing solution (precipitate solution with pH adjusted to 8.772), centrifuging it at 2,000 rpm for 5 min, and removing the supernatant. We assessed the tube with 750 µL serum and the pellet precipitated from 750 µL dialysate by gamma counting and compared the results with control incubations.
Model development. HPT axis. We constructed the BBDR-HPT axis model in acslXtreme, version 2.4 (AEgis Technologies, Huntsville, AL) and solved it using the Gear algorithm for stiff systems. The model for dietary iodide and the thyroid axis was used as previously described (McLanahan et al. 2008). Briefly, this BBDR-HPT axis model includes submodels for dietary iodide, TSH, and the thyroid hormones T 4 and T 3 (Figure 1). The submodels combine to form a simplified, quantitative description of the thyroid axis in the adult rat. Regulatory and compensatory effects of TSH were empirically described for several thyroidal processes, including TSH stimulation of a) NIS thyroidal iodide uptake, b) formation of thyroid hormone precursors, and c) thyroid hormone secretion. The TSH/ T 4 negative feedback loop is described using the relationship of serum total T 4 (TT 4 ) and TSH, as well as the metabolism of thyroid hormones with recycling of iodide.
We developed the BBDR-HPT axis model (McLanahan et al. 2008) to predict perturbations in serum TT 4 , total T 3 (TT 3 ), TSH, and total thyroidal iodide content that resulted from a decrease in dietary iodide intake. Normal laboratory rat iodide intake is approximately 20 µg/day. Figure 2 shows a model simulation up to 28 days for Simonsen Albino rats placed on a diet containing 0.33 µg iodide per day from day 0 through day 28 (Okamura et al. 1981).
Perchlorate. Instead of the more elaborate modeling approach implemented by Merrill et al. (2003), a simple model structure for ClO 4 was constructed (Figure 3), similar to the approaches reported by Crump and Gibbs (2005) and Lorber (2008). This PBPK model consisted of three compartments: plasma, thyroid, and the rest of the body. ClO 4 is rapidly absorbed after oral administration and is distributed throughout the body but excreted unchanged through the urine (Wolff 1998). Urinary excretion of ClO 4 is described by first-order clearance from the plasma. The thyroid gland is described with a blood and tissue compartment using diffusion limitation and active uptake into the thyroid via the NIS. Several studies have shown that ClO 4 and 36 ClO 4 are transported into the thyroid via the NIS, and an increased uptake has been observed in rats administered TSH and ClO 4 - (Anbar et al. 1959;Chow et al. 1969;Chow and Woodbury 1970;Dohan et al. 2007;Goldman and Stanbury 1973;Tran et al. 2008;Yu et al. 2002).

BBDR-HPT axis and ClO
where rTNIS i is the rate of iodide actively transported into the thyroid via NIS (nanomoles per hour), Cvt i is the free concentration of iodide in thyroid blood (nanomoles per liter), Km i is the affinity constant of iodide for the NIS (nanomoles per liter), Cvt p is the concentration of ClO 4 in the thyroid blood (nanomoles per liter), Ki p is the inhibition constant of ClO 4 for NIS iodide transport (nanomoles per liter), V max T i is the maximum rate of NIS iodide uptake (nanomoles per hour), CA TSH is the serum concentration of TSH (nanomoles per liter), and K T N S I H S is the concentration of TSH that gives rise to half-maximal rate of NIS transport of iodide (nanomoles per liter). We previously simultaneously optimized V max T i Figure 2. BBDR-HPT axis model simulation (lines) of HPT axis alterations in serum TT 4 , TT 3 , and TSH and thyroid iodide up to 28 days after administration of a low-iodide diet (0.33 µg/day). Data (mean ± SD) are adapted from Okamura et al. (1981) and are offset at days 14 and 28 to aid in visualization of data.   Eng et al. (1999) and McLanahan et al. (2007), respectively. The other parameter, NIS affinity constant for iodide (Km i , nanomoles per liter) in Equation 1 was obtained from Gluzman and Niepomniszcze (1983) as reported by McLanahan et al. (2008).
We used an equation similar to Equation 1 to model iodide's inhibition of the uptake of ClO 4 via the NIS; however, the effect of iodide on transport of ClO 4 into the thyroid is minimal because ClO 4 has a higher affinity for the NIS protein compared with iodide (1,500 nmol ClO 4 -/L vs. 31,519 nmol I/L). Model parameter values. We obtained physiologic parameters (Table 1) for tissue volumes and blood flows from previous studies (Brown et al. 1997;Everett et al. 1956;Malendowicz and Bednarek 1986;McLanahan et al. 2007McLanahan et al. , 2008. The BBDR-HPT axis model parameters had been calibrated for the euthyroid adult rat previously and used to describe iodide-deficient conditions (McLanahan et al. 2008). Table 2 shows model parameters for ClO 4 -. We set the affinity constant, Ki p , for the inhibition of iodide uptake by ClO 4 equal to Km p , the measured affinity constant for ClO 4 and the NIS. A Km p value of 1.5 µM was used, which was reported as an inhibition affinity constant for uptake of radiolabeled iodide in thyroid cells (Kosugi et al. 1996). We calculated a ClO 4 tissue:blood partition coefficient for the body compartment (minus the thyroid and plasma) by weighting the partition coefficients based on tissue volume for tissue:blood partition coefficient values reported in Merrill et al. (2003) for the adult male rat. Optimization was carried out using acslXtreme parameter estimation tool kit, using 36 ClO 4 kinetic data reported by Yu et al. (2002), who administered a single intravenous bolus dose of 3.3 mg/kg 36 ClO 4 -. Optimized values were obtained for V max Tc p , the maximal rate of uptake of 36 ClO 4 into the thyroid and urinary clearance constant, and for ClUc p , a term used to describe the rate of excretion of 36 ClO 4 in urine. The ClUc p value was obtained by fitting both the urinary excretion of 36 ClO 4 and serum concentrations of 36 ClO 4 -. Experimental simulations. Yu et al. (2002) administered 0, 0.1, 1, 3, and 10 mg ClO 4 -/kg/day in drinking water to adult male Sprague-Dawley rats and determined serum TT 4 , fT 4 , TT 3 , TSH, and ClO 4 concentrations after exposures of 1, 5, and 14 days. We did not consider the Yu et al. (2002) data for fT 4 for this study because the increase in fT 4 with ClO 4 concentration is considered to be erroneous owing to a nondialysis method. The problem with this method has been previously discussed (Fisher et al. 2006). In addition, Männistö et al. (1979) administered ClO 4 at a rate of 15 mg/kg/day in drinking water to adult male Sprague-Dawley rats and measured serum TT 4 , TT 3 , and TSH after exposure for 0, 2, 4, 6, 9, and 14 days. The BBDR-HPT/ ClO 4 model was configured to simulate these experiments. Ingestion of ClO 4 in drinking water took place over a 12-hr period.

Results
Laboratory binding and displacement experi ments. ClO 4 had little or no effect on displacing 125 I-T 4 from serum proteins for both human and rats [see Supplemental Material, Figure 1 (http://www.ehponline.org/members/ 2009/0800111/suppl.pdf)] across the relevant range of ClO 4 concentrations tested (1-300 µg/mL). The highest concentration tested, 300 µg/mL, is 30 times greater than the range of interest for ClO 4 . Thus, older studies that examined high doses of ClO 4 probably caused a displacement in proteinbound T 4 , but for the concentration range of interest, 0.01-1.0 µg/mL (corresponding to a ClO 4 dose of 0.01-10 mg/kg/day), there appears to be little displacement of proteinbound T 4 by ClO 4 -. We found no significant differences by analysis of variance for either rat (p = 0.13) or human (p = 0.96) serum. Therefore, this MOA was not explored in the computational analysis. Figure 4 shows model-predicted and laboratory-observed serum and thyroid 36 ClO 4 concentrations and cumulative urinary excretion of 36 ClO 4 after a single intravenous bolus dose of 3.3 mg/kg 36 ClO 4 -. We obtained these successful predictions of thyroidal and serum concentrations of 36 ClO 4 and urinary clearance using an optimized V max Tc p value of 1,800 nmol/hr/kg 0.75 and an optimized ClUc p value of 0.007 L/hr/kg 0.25 ( Table 2). Simulations of ClO 4 in serum and the thyroid gland were similar to previous model findings of Merrill et al. (2003) for the drinking water dose groups of 0.1, 1.0, 3.0, and 10 mg/kg/day that were originally reported by Yu et al. (2002) (data not shown).    HPT response predictions: testing the hypothesis. We linked the ClO 4 model with the BBDR-HPT axis model that we previously calibrated to predict serum TT 4 , TT 3 , TSH, and total thyroid iodide for sufficient and insufficient dietary iodide intakes and evaluated the ability of the combined model to predict ClO 4 --induced HPT axis disturbances. The hypothe sis was that the BBDR-HPT/ClO 4 model would predict the HPT axis disturbances based on competitive inhibition of thyroidal uptake of dietary iodide (Equation 1) for ClO 4 drinking water dose rates reported by Männistö et al. (1979) (15 mg/kg/day) and Yu et al. (2002) (0.1, 1, 3, 10 mg/kg/day). In this case, the BBDR-HPT/ClO 4 model failed to predict the HPT axis responses. Dose-dependent increases in serum TSH concentrations were severely underpredicted, and dose-dependent decreases in serum TT 4 concentrations were also underpredicted ( Figure 5A,B). The small decrease in serum TT 3 in response to 10 mg/kg/day ClO 4 was better predicted by the model [see Supplemental Material, Figure 2 (http:// www.ehponline.org/members/2009/0800111/ suppl.pdf)]. Yu et al. (2002) reported dose-dependent serum TSH increases ranging from 60% for the lowest dose to 174% for the highest dose compared with controls after 1 day of ClO 4 treatment. For all dose groups, serum TSH level continued to increase by day 5 of treatment, with the exception of the 0.1 mg/kg/day dose group. By day 14 of treatment, TSH concentrations were in decline relative to the peak measured TSH concentrations. Männistö et al. (1979) reported that 15 mg/kg/day caused serum TSH levels to increase by 18% on day 2 of treatment and by 96% on day 14 ( Figure 5C). These measured TSH concentrations are in stark contrast to the model-predicted values ( Figure 5C). The model-simulated serum TSH concentrations were predicted to increase only by 2-6% after 1 day of treatment and by 21-36% after 14 days of treatment with the highest ClO 4 doses (10 and 15 mg/kg/day; Figure 5C).

PBPK model predictions.
Similar to the TSH findings, the dosedependent decline in serum TT 4 concentrations reported by Yu et al. (2002) ranged from 6% for the lowest dose to 24% for the highest dose after 1 day of ClO 4 treatment. Consistent with the increases in serum TSH on day 5, all treatment groups, except for the 0.1 mg/kg/ day ClO 4 dose group, displayed a continued decline in serum TT 4 concentrations. By day 14 of ClO 4 treatment, TT 4 levels were recovering ( Figure 5A). Model predictions of serum TT 4 concentrations were also in disagreement with observations. The model-predicted decreases in serum TT 4 concentrations were only 2-6% by day 1 of treatment and 13-26% by day 14 for the 10-and 15-mg/kg/day ClO 4 dose groups. Although the model-predicted and observed serum TT 4 concentrations were in agreement for the 10-mg/kg/day group by day 14, the temporal aspect of the TT 4 perturbations was not predicted ( Figure 5C).
The model-predicted interaction of ClO 4 -(competitive inhibition) on thyroidal uptake of dietary iodide (Table 3) demonstrated the nonlinear uptake behavior of dietary iodide that is described by Michaelis-Menten kinetics for each anion. NIS-mediated thyroidal uptake of dietary iodide (micrograms per day) was predicted to be similar on days 1 and 14 (Table 3), with a slight increase on day 14 because of TSH stimulation of the NIS for the two highest ClO 4 doses (10 and 15 mg/kg/day). Although thyroidal iodide stores were not reported by Yu et al. (2002) and Männistö et al. (1979), only a 6% and 12% depletion was predicted to occur after 1 day and 2 weeks of treatment of ClO 4 for dose rates of 10 and 15 mg/kg/day, respectively (Table 3). The model-predicted depletion of thyroidal iodide stores was 33% and 46% by day 14 for these two doses, respectively. The model-predicted percent inhibition in thyroidal uptake of dietary iodide was predicted to range from 26% for the 0.1 mg/kg ClO 4 dose group to 96% and 97% inhibition for the two highest dose groups (Table 3).
Is ClO 4 altering thyroid hormone secretion? The best approach to answer this question would be to design studies focused on evaluating the status of the thyroid gland for a known intake of dietary iodide, such as meas uring the ratios of MIT and DIT, and thyroid hormones in animals treated with ClO 4 -. Another equally important and time-intensive effort would be to develop a mechanistic-based model of the thyroid gland that would include explicit descriptions of the synthesis and secretion pathways for thyroid hormones, for which some data are still lacking. Instead, we simply imposed "nonmechanistic" conditions on the thyroid gland by limiting thyroid hormone production and asking whether the systemic concentrations of TT 4 and TSH can be described by simply limiting thyroid hormone production and secretion. To account for inhibition of thyroid hormone production, we modified equation 14 from McLanahan et al. (2008), which describes in drinking water. Data (mean ± SD) for T 4 and TSH for 10 mg/kg/day (~ 350 g rat) from Yu et al. (2002), and for 15 mg/kg/day (~ 200 g rat) from Männistö et al. (1979). The model does not predict the rapid changes in serum T 4 and TSH observed in these studies.  Table 3. Model predictions of daily NIS uptake of iodide into thyroid (µg/day) and total thyroid iodide stores (µg) for a 300-g adult rat, after 1 and 14 days of ClO 4 exposure in drinking water at various doses.
where k IB TSH (square liters/nanomole per hour) is a linear rate term, CA TSH (nanomoles per liter) is the concentration of TSH in the serum, and CTB i (nanomoles per liter) is the bound concentration of iodide in the thyroid as thyroid hormone precursors. The production rate of T 4 and T 3 was described as a fraction of the total production rate, and the proportion of T 4 and T 3 changes as a result of depletion of thyroidal iodide stores (McLanahan et al. 2008). Figure 6A shows the 10-mg/kg/day dose group (Yu et al. 2002) model predictions for TT 4 when a 25% (INH p = 0.75) or 50% (INH p = 0.50) inhibition of thyroid hormone synthesis is tested using the model. A 50% decrease in thyroid hormone production rate appears to describe the decreased TT 4 concentrations for the first 5 days of exposure, but for serum TSH ( Figure 6B) 90% inhibition of thyroid hormone synthesis did not describe the kinetic behavior for TSH. Serum TT 3 was predicted fairly well without imposing inhibition of thyroid hormone production; however, a 25% decrease in thyroid hormone production provided a slightly better fit to the Yu et al. (2002) 10 mg/kg/day dose group [see Supplemental Material, Figure 2 (http:// www.ehponline.org/members/2009/0800111/ suppl.pdf)]. This simple simulation exercise appears to suggest that thyroid hormone production may be partially inhibited on a transient basis. However the relationship between TSH production and serum TT 4 is complex and very different from dietary-iodide-induced hypothyroidism.

Discussion
In this article, we report on the use of computational modeling analyses to examine the biological processes associated with disruption of an endocrine system, the HPT axis in the adult rat, by ClO 4 -. The BBDR-HPT/ClO 4 model simulation results suggest that ClO 4 administered in drinking water (or by other routes of administration) interacts with the rat thyroid gland itself, potentially altering thyroid hormone synthesis or secretion. This interaction is in addition to blocking thyroidal uptake of iodide. Current presumptions of a single MOA for ClO 4 on the HPT axis do not appear tenable based on these results.
Changes in serum TSH and TT 4 after only 1 day of ClO 4 treatment were surprising because thyroidal iodide depletion was expected to be minimal over this time period, although 26-96% inhibition of NIS thyroidal iodide uptake was predicted by the model, which is in close agreement with Yu et al. (2002). The only reported effects of ClO 4 on thyroidal iodide content we found were those of Ortiz-Caro et al. (1983), who reported a decrease in thyroidal 127 I of nearly 8% in rats administered 1,000 ppm ClO 4 in drinking water for 2 days and provided a low-iodide diet. A 10-15% decrease over 1 day is predicted by our model, if ClO 4 caused a 100% block of thyroidal uptake of dietary iodide. In comparison, for a severely iodide-deficient diet (0.3 µg I/day), it would take 7 days to deplete thyroid stores by 43% and result in a 35% decrease in serum T 4 (McLanahan et al. 2008). This degree of perturbation in the HPT axis compares with the 10 mg/kg/day ClO 4 dose group after only 1 day of exposure (Yu et al. 2002). In the iodide-deficient BBDR-HPT axis model (McLanahan et al. 2008), the thyroidal iodide stores were closely related to iodide deficiency perturbations in the HPT axis. With ClO 4 -, it is apparent that the HPT axis is disturbed before sufficient model-predicted depletion of thyroidal iodide stores. For iodide deficiency, slow depletion of thyroidal iodide stores was governed by iodide available for thyroidal uptake and the secretion and metabolism rates of thyroid hormones. McNabb et al. (2004aMcNabb et al. ( , 2004b) reported on the dose-dependent thyroidal depletion of T 4 from bobwhite quail chicks ingesting ClO 4 -. This depletion in thyroidal T 4 may have occurred because of the blocking effect of ClO 4 on thyroidal uptake of iodide, ClO 4 acting on the thyroid gland synthesis and secretion of T 4 , or a combination of both modes of action. The idea that ClO 4 may affect thyroid hormone synthesis/secretion has previously been discussed (Hildebrandt and Halmi 1981;Wolff 1998). Greer et al. (1966) provided experimental evidence for a direct effect on ClO 4 on the rat thyroid gland in vitro. These authors reported that ClO 4 reduced formation of MIT and DIT in rat thyroid lobes when incubated with ClO 4 -. They observed changes in MIT and DIT forma tion for ClO 4 media concentrations starting at 10 mg/L with a reported 50% effective reduction of DIT at 250 mg/L. By comparison, in vivo, rats ingesting 1, 3, and 10 mg/kg/day ClO 4 in drinking water had measured serum ClO 4 concentrations of approximately 0.3, 1.4, and 4.9 mg/L and thyroidal ClO 4 concentrations of 10, 50, and 176 mg/kg, respectively (Yu et al. 2002). Although difficult to compare, the in vitro media and in vivo serum ClO 4 concentrations suggest that the in vitro ClO 4 --induced changes in organification may be a high-dose effect of ClO 4 -. Nevertheless, the data reported by Greer et al. (1966) demon strate that interactions of ClO 4 and thyroid gland processes are possible. Interestingly, Yu et al. (2002) also reported that when rats were administered an intravenous dose of 3 mg/kg of ClO 4 -, serum TSH and TT 4 concentrations were at control levels 8 hr postdosing, but by 12 hr serum TSH concentrations increased and serum TT 4 concentrations decreased. The secretion rate of iodide (as thyroid hormones) is about 0.5-1 µg over a 12-hr period, which suggests that thyroidal iodide stores may be depleted by only 5-8% when HPT axis perturbations were observed. Männistö et al. (1979) administered propylthiouracil (PTU), MMI, potassium perchlorate (KClO 4 ), and KI to rats in drinking water and measured serum TSH and TT 4 at several time points over 14 days of treatment. KI and KClO 4 both rapidly depleted serum TT 4 concentrations by day 2 of treatment. KI animals recovered by day 4 of treatment, whereas the KClO 4 rats did not, and serum TT 4 concentrations remained low, similar to MMI and PTU animals, over 14 days of treatment. Excess iodide is well known for this Wolff-Chaikoff effect, which is associated with a transient "shutdown" in the thyroid gland (Wolff and Chaikoff 1948). MMI and PTU interact with thyroid peroxidase in vitro, but ClO 4 does not (Magnusson et al. 1984). A recent in vitro study conducted by Tonacchera et al. (2004) in Chinese hamster ovary cells transfected with human NIS affirm that ClO 4 inhibits iodide uptake in a competitive fashion, with potencies 15, 30, and 240 times greater than the related compounds thiocyanate (SCN -), iodide, and nitrate (NO 3 -), respectively. The relative concentrations of these monovalent anions determine the ability of NIS to transport iodide, and because SCNand NO 3 are present in human sera at greater concentrations than ClO 4 -, several researchers (De Groef et al. 2006;Gibbs 2006) have suggested that SCNand NO 3 are more likely than ClO 4 to adversely affect (e.g., correlate with) thyroid function. However, Blount et al. (2006) reported correlations between ClO 4 and serum TT 4 and TSH, no correlation with nitrate, but a negative association of TT 4 with thiocyanate. That a decrease in serum TT 4 and increase in serum TSH correlated with ClO 4 but not with the other monovalent anions suggests that ClO 4 may act through another MOA in addition to simple competitive inhibition.
TSH secretion from the pituitary is controlled by thyroid-releasing hormone (TRH) from the hypothalamus, and secretion of both TSH and TRH is partially controlled by thyroid hormones derived from serum. Using a quantitative evaluation of the pulsatile nature of the HPT axis, Li et al. (1995) suggest that the feedback effects of thyroid hormones on the hypothalamus and control of TRH secretion are much smaller than effects on the pituitary for control of TSH secretion. However, if TRH secretion is high, TRH may dominate the control of TSH secretion. In the case of ClO 4 -, the current BBDR-HPT axis model, calibrated for iodide deficiency, would need to be calibrated to explicitly include differential control of TSH secretion by both TRH and thyroid hormones.
Hypothalamic TRH does drive pituitary TSH secretion; however, details in rats, humans, and other species are not sufficient to distinguish between hypothalamic-and pituitary-derived TSH secretion. Measuring pituitary portal blood TRH is not realistic. A few attempts in rodents, using extreme surgical techniques, have been carried out, but not in reference to perturbations of the HPT axis. This aspect of HPT axis regulation remains a data gap. Eisenberg et al. (2008) recently published a control systems simulation model of the HPT axis feedback system in humans. They also did not include the hypothalamic aspect in their feedback simulator because of a lack of information. In our BBDR-HPT axis model (McLanahan et al. 2008), the TRH aspect of TSH production is embedded or lumped with the equation that describes the relationship between serum TSH and TT 4 concentrations (Equation 3). Eisenberg et al. (2008) also lumped the hypothalamic contribution to describe TSH secretion in humans.
In conclusion, our computational analysis indicates that ClO 4 alters the HPT axis by blocking iodide transport and by interacting in some fashion with hormone production/release from the thyroid. The present description of synthesis and secretion of thyroid hormones (Equation 3) is simple, and if the thyroid gland itself is a target of ClO 4 -, as the modeling suggests, then a more complex description of the thyroid hormone synthesis and secretion process will be required. Our goal in this effort was to impose "non mechanistic" outcomes to better understand what conditions may help explain the observed ClO 4 dose response. A significant correlation of effects of ClO 4 with serum thyroid hormones in humans has yet to be confirmed, so the importance of this potential MOA in humans is not known. Computational models can provide quantitative tools to evaluate possible MOAs of ClO 4 on the HPT axis. This evaluation would be supported by conducting specific studies to develop a quantitative mathematical description of the MOA, along with studies that measure temporal perturbations in several HPT axis end points. Studies are needed that evaluate the molecular biology of the genes within the thyroid gland, coupled with measure ment of thyroid hormone precursors and thyroid hormones using modern analytical techniques such as liquid chromatography/ mass spectrometry. These studies need to be completed for known iodine intake rates along with sufficient doses and time points to construct a temporal dose-response curve after ClO 4 exposure via drinking water.