June 28, 2007

Current and Potential Rodent Screens and Tests for Thyroid Toxicants

By Zoeller, R Thomas Tyl, Rochelle W; Tan, Shirlee W

This article reviews current rodent screens and tests to detect thyroid toxicants. Many points of disruption for thyroid toxicants are outlined and include: (a) changes in serum hormone level; (b) thyroperoxidase inhibitors; (c) the perchlorate discharge test; (d) inhibitors of iodide uptake; (e) effects on iodothyronine deiodinases; (f) effects on thyroid hormone action; and (g) role of binding proteins (e.g., rodent transthyretin). The major thyroid endpoints currently utilized in existing in vivo assay protocols of the Organization for Economic Cooperation and Development (OECD), Japanese researchers, and U.S. Environmental Protection Agency (EPA) include thyroid gland weight, histopathology, circulating thyroid hormone measurements, and circulating thyroid-stimulating hormone (TSH). These endpoints can be added into the existing in vivo assays for reproduction, development, and neurodevelopment that are outlined in this chapter. Strategic endpoints for possible addition to existing protocols to detect effects on developmental and adult thyroid endpoints are discussed. Many of these endpoints for detecting thyroid system disruption require development and additional research before they can be considered in existing assays. Examples of these endpoints under development include computer-assisted morphometry of the brain and evaluation of treatment-related changes in gene expression, thyrotropin-releasing hormone (TRH) and TSH challenge tests, and tests to evaluate thyroid hormone (TH)-dependent developmental events, especially in the rodent brain (e.g., measures of cerebellar and cortical proliferation, differentiation, migration, apoptosis, planimetric measures and gene expression, and oligodendrocyte differentiation). Finally, TH-responsive genes and proteins as well as enzyme activities are being explored. Existing in vitro tests are also reviewed, for example, thyroid hormone (TH) metabolism, receptor binding, and receptor activation assays, and their restrictions are described. The in vivo assays are currently the most appropriate for understanding the potential effects of a thyroid toxicant on the thyroid system. The benefits and potential limitations of the current in vivo assays are listed, and a discussion of the rodent thyroid system in the context of human health is touched upon. Finally, the importance of understanding the relationship between timing of exposure, duration of dose, and time of acquisition of the endpoints in interpreting the results of the in vivo assays is emphasized. Keywords Cerebellar development (proliferation, apoptosis, migration, differentiation), in vitro screens, in vivo screens, Regulatory Screens and Tests, Serum Thyroid Hormone (TH) Levels, TH Receptor Binding and Activation, Thyroid Gland Weight and Histopathology, Thyrotropin-Releasing Hormone (TRH), Thyroid- Stimulating Hormone (TSH), Thyroid System Toxicants

Table of Contents

Table of Contents

The goal of this article is to review current screens and tests performed in rodents for thyroid toxicants, to discuss their underlying strengths and weaknesses, and to propose additional endpoints for thyroid toxicants that are now available or are in the research and development phase, and that can overcome existing weaknesses. We first review the modes of action of known thyroid toxicants and potential in vitro and in vivo assays associated with these modes of action. We then review endpoints normally employed in screens/tests for thyroid toxicants. Finally, we review current assays-developed by the Organization for Economic Cooperation and Development (OECD), by Japanese investigators, and by the U.S. Environmental Protection Agency (EPA)-with particular attention to ways in which these protocols can be strengthened for their ability to identify potential thyroid toxicants.


As discussed earlier, all chemicals classified as thyroid toxicants to date have been defined by their ability to reduce circulating levels of thyroid hormone (Brucker-Davis, 1998). Thus, these chemicals alter the relationship between thyroid hormone biosynthesis and elimination such that the steady-state levels of hormones are reduced. The mode of action by which these chemicals can influence circulating levels of thyroid hormone are either focused on effects on thyroid hormone biosynthesis, or on thyroid hormone metabolism (Table 1 ). The modes of action reviewed below are germane to other taxa and are relevant to humans.


Points of disruption of thyroid hormone synthesis and of chemicals known to exert this action

A workshop was held at Duke University in June 1997 to bring together international experts on thyroid toxicology to review methods for screening putative thyroid toxicants (DeVito et al., 1999). The workshop focused on more than 20 assays or test systems that have been used to examine chemicals that alter synthesis, storage, transport, and catabolism of T4 and/or T3, assays that evaluate ligand binding and activation of the TRs, and in vivo assays that examine the effects of antithyroid agents and thyromimetics in mammalian and nonmammalian wildlife models. The purpose of the workshop was not to recommend a screening battery or to deal with policy issues pertaining to the use of such screens; the product of the workshop was intended to describe and evaluate the methods that are currently available or could be developed in the near future for screening and testing. To date, the paper by DeVito et al. ( 1999) likely remains one of the most cogent and concise descriptions of the extant assays for thyroid toxicity (at least in rodents) and speculates on some potentially new assays. The following subsections represent modes of action by which toxicants influence thyroid endocrinology.


Changes in serum concentrations of thyroid hormones (T4, Ta, and TSH) can be caused by chemicals that inhibit thyroid hormone synthesis, release, and transport, and by chemicals that increase metabolism of various thyroid hormones (e.g., deiodinases, UDPGTs). If a chemical decreases serum hormone concentrations, specific assays can be used to determine the mechanism by which these hormone concentrations are decreased. Moreover, the specific profile of these changes in hormone concentrations may be informative. For example, inhibition of Dl may preferentially reduce circulating levels of T3, which may not be accompanied by a reduction in serum T4 or TSH. In contrast, inhibition of iodine uptake is predicted to cause a reduction of T4, leading to a decrease in both T4 and T^ and an increase in serum TSH. However, it is important to keep in mind that interpreting changes in hormone levels in terms of mechanisms of toxicant action or potential adverse effects is quite complex. For example, if thyroid hormones are decreased and TSH is elevated, it is important to avoid assumptions about compensatory actions. As described later, recent studies in rats demonstrate that goitrogens can produce effects on the fetal brain at concentrations below that which affects maternal serum TSH. Moreover, exogenous thyroid hormone can influence fetal brain measurements before it downregulates maternal serum TSH.

Thyroperoxidase Inhibitors

6-Propyl-2-thiouracil (PTU) is a thioamide drug that has been intensively studied in animals and in humans and is used therapeutically to treat patients with Graves' disease (Zoeller and Crofton, 2005b; Cooper, 2003). As a drug, it does not exist in nature and there are no environmental sources of PTU. However, as a chemical class, it represents compounds found in the environment that can affect thyroid function. PTU is well known to reduce circulating levels of T4 and T.i and to increase circulating levels of TSH (e.g., Frumess and Larsen, 1975; Sato et al., 1976) and has been extensively used in mechanistic research focused on identifying the role of thyroid hormone in brain development. The ability of PTU to reduce circulating thyroid hormone levels has been exploited in the treatment of hyperthyroidism in humans, including in pregnant and lactating women (Mestman, 1998). PTU (Figure 1) is generally believed to produce deleterious effects in animals by causing a dosedependent reduction in circulating levels of thyroid hormone. This reduction is caused by the ability of PTU to directly inhibit the function of the thyroperoxidase enzyme (Engleret al., 1982), which is responsible for iodination of the tyrosine residues on thyroglobulin (Taurog, 2000), a key step in thyroid hormone synthesis. In addition, PTU inhibits the type 1 5'-deiodinase (Ortega et al., 1996), which converts T4 to Tj in peripheral tissues. As such, PTU reduces the synthesis of nascent thyroid hormone including both T^sub 4^ and T^sub 3^, causing a dose- dependent decrease in circulating levels of thyroid hormone (St. Germain and Croteau, 1989).

FIG. 1. Propylthiouracil.

Thyroperoxidase is a multisubstrate enzyme, which reacts first with hydrogen peroxide, forming an oxidized enzyme. This species then oxidizes iodide, the second substrate, to an enzymebound "active iodine," transferable to tyrosyl residues on thyroglobulin (TG) (Davidson et al., 1978). The thioureylene drugs, including PTU, methimazole (MMI), and thiouracil, can inhibit TPO's ability to activate iodine and transfer it to TG (Davidson et al., 1978). However, these drugs act by different mechanisms. Specifically, PTU interacts with the "activated" iodine, producing a reversible inhibition of TPO (Nagasaka and Hidaka, 1976; Davidson et al., 1978), whereas MMI interacts directly with the TPO enzyme and irreversibly inhibits it. The key event of TPO inhibition by PTU leads to a series of events within the hypothalamic-pituitary- thyroid (HPT) axis that may directly produce adverse effects or which may be surrogate markers of adverse effects. No other modes of action have been proposed for the ability of PTU to reduce circulating levels of thyroid hormone or to affect thyroid histopathology. However, a recent study indicates that PTU can exert direct actions on the activity of the neuronal isoform of nitric oxide synthase (Wolff and Marks, 2002). Considering the importance of neuronal NOS in brain development and in neuronal plasticity (Blackshaw et al., 2003), it is possible that this direct action may influence brain development. A good example of TPO inhibitors are the isoflavones, especially those found in soy protein (e.g., genistien, coumesterol) (reviewed by (Doerge and Sheehan, 2002). In humans, goiter has been reported in infants fed soy formula (Labib et al., 1989; Chorazy et al., 1995; Jabbar et al., 1997). In addition, teenage children diagnosed with autoimmune thyroid disease were found to have twice the rate of occurrence if they had consumed soy formula as infants (Fort et al., 1990). Boker et al. (2002) recently reviewed the dietary sources of a variety of isoflavones (see Table 2), showing that these are common dietary components. These isoflavones are also so-called "phytoestrogens," which are highly enriched in some commercial preparations.

The TPO assay itself involves monitoring the iodination reaction using bovine serum albumin or tyrosine as substrates (Divi and Doerge, 1996). In addition, the oxidation of guaiacol can be used as an indicator of thyroid peroxidase activity (Divi and Doerge, 1994). All chemicals that inhibit the iodination reaction also inhibit the coupling reaction (Divi and Doerge, 1994). The coupling reaction can be assayed using human low iodine thyroglobulin or preiodinated casein as substrates.

A disadvantage of the TPO assay is that purified TPO is not readily available commercially. It was previously reported that porcine TPO is the only purified preparation available (DeVito et al., 1999). Moreover, a recent online search of possible commercial products revealed none. However, if this assay were an important component of a chemical screening program, recombinant enzymes from different species could be developed. In fact, a strength of the TPO assay is that the sensitivity to chemical inhibition of TPO from human and experimental animals can be directly examined. In vitro studies have shown that TPOs from different mammals exhibit differences in their sensitivity to inhibition by propylthiouracil (PTU) and sulfamethazine (Takayama et al., 1986). Comparisons of the relative sensitivity of TPO across species to various toxicants may assist in risk assessment for chemicals that inhibit TPO activity, although differences in the pharmacokinetics/dynamics in various species would not be captured by this in vitro approach. The iodination and coupling assays are specific for chemicals that inhibit TH synthesis and are unlikely to produce false positives. However, used alone as a screen, these assays have high potential for false negatives, as chemicals that alter TH concentrations through other mechanisms would not be detected. These assays have been performed for many years, are well established in the scientific literature, and numerous chemicals have been tested using these assays. Although there are no published methodologies that can be defined as high throughput screens, modification of this assay into a high throughput screen is under development in several laboratories (DeVito et al., 1999).

Perchlorate Discharge Test

Perchlorate competes with iodide for thyroid uptake and also promotes the efflux of iodide from follicular cells (Atterwill et al., 1987). The perchlorate discharge test has been used for decades in both animals and humans to detect iodide organification defects (Wolff, 1998; Meller and Becker, 2002). In this assay, animals are exposed to a test chemical and then administered ^sup 125^I followed by perchlorate. Accumulation of ^sup 125^I in the thyroid is determined before and after administration of perchlorate. Perchlorate promotes the release of iodine that has not been incorporated into thyroglobulin. Therefore, if a chemical inhibits or deactivates thyroid peroxidase, there would be a brisk decrease in the accumulation of ^sup 125^I in the thyroid gland. This assay has the potential for providing mechanistic information on the actions of chemicals that alter thyroid function, but it does not necessarily meet the requirements of a screen (DeVito et al., 1999). A modification of the perchlorate discharge test that would test for chemicals that interfere with iodine uptake would be the use of thyroid scintigraphy (e.g., Schellingerhout et al., 2002). This technique is essentially that of radioactive iodine uptake inhibition used by Greer et al. ( 2002).


Intakes of phytoestrogen by food groups by Dutch women

Inhibitors of Iodide Uptake

Sodium/Iodide Symporter

A variety of complex anions can inhibit iodide uptake through the sodium/iodide symporter (NIS) (Wolff, 1998). The defining characteristic of iodide transport is its very high specificity for iodide with respect to the chloride ion, which is abundant in biological systems. However, despite this, iodide is not the only ion selected by the NIS, nor is it the most avid (Wolff, 1998). The following potency series for anions capable of blocking iodide uptake was constructed by Wolff and reviewed later (Wolff, 1998): TcO^sub 4^ > ClO^sub 4^ > ReO^sub 4^ > SCN > BF^sub 4^ > I > NO^sub 3^ > Br > Cl. Although nitrate is actually less potent than iodide at the NIS, environmental contamination with nitrate has nevertheless been associated with goiter (Gatseva et al., 1998; Vladeva et al., 2000). Perchlorate (ClO^sub 4^) contamination also has been studied for its effects on thyroid function, especially considering its potency at inhibiting iodide uptake into the thyroid gland (Urbansky, 2002; Strawson et al., 2004). The only epidemiological study focused on nonneonates (Crump et al., 2000) indicates that exposure to perchlorate in drinking water, in combination with elevated iodine intake, tends to increase circulating levels of thyroid hormone rather than decrease it. This observation was also observed in mice (Thuett et al., 2002). Genetic defects in NIS result in congenital iodide deficiency and congenital hypothyroidism (De La Vieja et al., 2004, 2005). Interestingly, the specific defects do not simply block the ability of NIS to take up iodide, but block its targeting to the plasma membrane, presumably by causing protein misfolding. Finally, a recent report (Breous et al., 2005) shows that environmental chemicals (phthalates) increase NIS expression by directly acting on the NIS promoter in a defined expression system.

Inhibitors ofPendrin

Pendrin is a protein identified by positional cloning, revealing a genetic defect resulting in Pendred syndrome. This syndrome is one of the most common causes of profound sensorineural hearing loss and thyroid goiter (Pendred, 1896; Reardon et al., 1977; Taylor et al., 2002). Interestingly, the pendrin protein is expressed in a highly specific manner: in the thyroid gland, the kidney and in the inner ear (Everett et al., 1997; Everett and Green, 1999). It is not completely clear how this expression pattern accounts for the symptoms of the syndrome. The Pendrin protein transports iodide from the apex of the thyroid follicular cells into the colloid (Zoeller, 2006, Figure 2), and it also appears to account for the iodide efflux from the thyroid gland upon perchlorate administration (i.e., the perchlorate discharge test) (Yoshida et al., 2002), which is why iodide efflux is exacerbated in Pendred's syndrome (Reardon et al., 1977; Reardon et al., 1999). Pendrin has a high degree of structural similarity to known sulfate transporters, but it transports iodide and chloride, not sulfate (Fugazzola et al., 2001). Although it is possible that Pendrin is a site of action for some xenobiotic chemicals, there is no information on this.

FIG. 2. Study design for OECD Test Guideline 421 (reprinted by permission of OECD).

Inhibitors of TSH Effectiveness

There is limited, but important, evidence that some toxicants can influence the potency of TSH on its receptor in vitro (Santini et al., 2003). This may be especially important in situations of low iodine because it is clear that changes in iodine status can influence the efficacy of TSH on its receptor.

Xenobiotic Effects on lodothyronine Deiodinases

Few studies have focused on the ability of environmental toxicants to interfere with thyroid hormone metabolism by deiodinases. However, this may be an important mechanism by which environmental chemicals could interfere with thyroid hormone action on tissues considering recent evidence that these enzymes play an important role in controlling tissue sensitivity to thyroid hormone, especially during development.

The development of the mammalian brain is characterized by an orderly sequence of events (Cowan et al., 1997). Moreover, the relative timing of maturational events within the brain is quite similar among mammalian species (Clancy et al., 2001 ). Recent work in both humans and experimental animals demonstrates that thyroid hormone exerts effects on the developing brain throughout a broad period of fetal and neonatal development (Chan and Rovet, 2003a), and that the developmental events and brain structures affected by thyroid hormone differ as development proceeds. Therefore, it is possible that the human brain uses a strategy for "timing" thyroid hormone sensitivity of different brain regions that is similar to that used by Xenopus (reviewed below). The work by Kester et al. (2004) represents a key observation suggesting that this is indeed the case. Kester et al. (2004) report that in several brain regions in humans-especially the cerebral cortex-levels of T^sub 3^ increase during fetal development and this is correlated with an increase in the activity of type 2 deiodinase (D2) while the activity of the type 3 deiodinase (D3) is low to undetectable. Type 2 deiodinase controls the conversion of T^sup 4^ to the hormonally active T^sub 3^, but D3 controls the conversion of T^sub 4^ to the hormonally inactive reverse T^sub 3^. Because T^sub 3^ levels in the fetal cerebral cortex increased to an extent that could not be accounted for simply on the basis of the age-dependent increase in T^sub 4^, it indicates that D2 is causing the age-dependent increase in T^sub 3^ from 14 to 20 weeks gestation. Importantly, during this same period, the fetal cerebellum has high levels of D3 and low levels of T^sub 3^. Finally, at later gestational ages, D^sub 3^ activity in the cerebellum declines and T^sub 3^ levels increase.

Deiodinase expression responds to changes in circulating levels of TH (Burmeister et al., 1997). Thus, thyroid toxicants may affect the ability of tissues to compensate for changes in circulating levels of thyroid hormone (Hood and Klaassen, 200Ob; Meerts et al., 2002). Moreover, deiodinase activities may be regulated in a complex manner that is related to both T^sub 4^ and T^sub 3^ availability in the serum (Burmeister et al., 1997). Thus, the shape of the dose- response curve defining the effect of toxicant on serum TH levels may not be parallel to the dose-response curve defining the effect of toxicant on endpoints of TH action in tissues.

In mammals, approximately 80% of the circulating T^sub 3^ is derived from peripheral deiodination of T^sub 4^ (St Germain and Gallon, 1997). As reviewed earlier, the deiodinases may control tissue sensitivity to thyroid hormone. For example, a recent report indicates that the human fetal cortex contains high levels of T^sub 3^ associated with high D2 activity and low D3 activity (Auso et al., 2004). In contrast, the human cerebellum exhibited low levels of T^sub 3^ before birth, and this was associated with low D2 and high D3. Thus, it is possible that xenobiotic chemicals that alter deiodinase activity may affect thyroid hormone signaling in the developing brain or in adult tissues, thereby producing an adverse effect, but may not produce changes in serum hormone concentrations. Deiodinase assays have been used for decades to understand the metabolism of thyroid hormones and may be amenable to high throughput assays. However, because the activity of these enzymes is dependent on the serum concentrations of these hormones, these assays would be sensitive toward chemicals that alter serum TH concentrations. Moreover, alterations in deiodinase activity also may alter serum TH concentrations. If serum TH concentrations are changed by deiodinase inhibitors, it may be easier to measure serum TH concentrations than it is to determine deiodinase activity. Similar to many of the assays described above, these assays have greater utility in understanding the mechanism of action of a chemical rather than as an initial screen.

Toxicant Effects on Thyroid Hormone Clearance

Oppenheimer's group was among the first to examine the ability of chemicals (phenobarbital and chlordane) to enhance biliary secretion of thyroxine (Bernstein et al., 1968). These seminal studies were the first to show that chemicals could activate the liver to trap thyroid hormones, enhancing their elimination through the bile and decreasing their serum half-life. Research in this area has focused on the ability of chemicals to induce liver enzymes that conjugate T^sub 4^ or T^sub 3^ to glucuronide, and/or the ability of chemicals to displace thyroid hormones from their serum binding proteins. However, there is not a consensus about the mechanism by which these chemicals, which do not act on the thyroid directly, can reduce circulating levels of thyroid hormones.

Role of Liver UDPGTs

Thyroid hormones (T^sub 4^ and T^sub 3^) are handled by the liver the way organic ions are handled-they are glucuronidated and sulfated, secreted into the biliary canaliculus, and concentrated into bile (Sellin and Vassilopoulou-Sellin, 2000). The microsomal enzymes responsible for this activity are the UDP-glucuronosyl transferases (UDPGTs). These phase II inducible enzymes are functionally heterogeneous. This functional heterogeneity is classically revealed in the different substrates they modify- 4dintrophenol compared to bilirubin (Chowdhury et al., 1983). In addition, different enzyme activities are directed toward T^sub 4^ and T^sub 3^ (Hood and Klaassen, 200Oa), indicating the possible differential regulation of excretion of these two iodothyronines. However, there is very little information about the role of iodothyronine metabolism by liver in the regulation of serum thyroid hormone levels under normal circumstances. Moreover, there is a paucity of information about the role of these enzymes in the production of thyroid disease (hypo- or hyperthyroidism). In contrast, there is a very large literature about the role of UDPGTs in the pathway by which various microsomal enzyme inducers can cause changes in circulating levels of thyroid hormones (Barter and Klaassen, 1992; Liu et al., 1995; Kolaja and Klaassen, 1998; Hood et al., 1999; Hood and Klaassen, 200Oa, 200Ob; Klaassen and Hood, 2001 ; Zhou et al., 2001,2002,2003).

An example of the key questions regarding the role of UDPGTs in mediating toxicant effects on serum thyroid hormone levels is provided by the effect of polychlorinated biphenyls (PCBs) on serum thyroid hormone. The chlorinated biphenyl 3,3',4,4',5- pentachlorobiphenyl, Aroclor 1254, and 2,3,7,8-tetrachlorodibenzo-/ 7-dioxin in rats are all known to reduce circulating T^sub 4^ (Barter and Klaassen 1992; van Birgelen et al., 1994; Schuur et al., 1998), perhaps because of their ability to induce T^sub 4^-UDPGT (Saito et al., 1991 ; Barter and Klaassen 1992; de Sandro et al., 1992). However, the degree to which these chemicals reduce serum T^sub 4^ is not correlated with the increase in T^sub 4^-UDP-GT activity (de Sandro et al., 1992; Hood et al., 2003). In addition, Kenechlor-500 reduces circulating levels of T^sub 4^ in both rats and mice, but induces UDP-GT in rats but not mice (Kato et al., 2003). In addition, Kenechlor-500 induces a decrease in circulating levels of T^sub 4^ in Gunn rats, a strain that is deficient in UDPGTlA isoforms (Kato et al., 2004). Thus, there is an argument that UDPGT induction alone is not a uniform marker of the ability of chemicals to cause a reduction in serum thyroid hormone. Nonetheless, the ability of chemicals to reduce circulating levels of thyroid hormone can be associated with UDPGT induction and an increase in fecal elimination of T^sub 4^ (de Sandro et al., 1992; Vansell and Klaassen, 2001).

Glucuronidation Assays

Glucuronidation, sulfation and sulfonation followed by biliary elimination of T^sub 4^ is one of the major pathways for removing T^sub 4^ from the circulation. In both humans and rodents, there is evidence of all three mechanisms of T^sub 4^ elimination. There are at least three isoforms of UDPGT in mammals that glucuronidate T^sub 4^ (Visser et al., 1993). Several classes of chemicals induce UDPGTs responsible for the glucuronidation of T^sub 4^ (Matsumoto et al., 2002; Meerts et al., 2002; Wade et al., 2002; Zhou et al., 2002; Hood et al., 2003; Kato et al., 2003). Induction of T^sub 4^ glucuronidation increases clearance and decreases serum concentrations of T^sub 4^. Induction of T^sub 4^ glucuronidation is typically determined in hepatic microsomes from animals treated with test chemicals. These assays have been performed for decades in numerous laboratories throughout the world. These ex vivo assays require several days of dosing of the test chemical. The advantage of this type of assay is that it is responsive to metabolic activation of the test chemical because exposure occurs in vivo. The activity of hepatic microsomal T^sub 4^ glucuronidation is not as sensitive to stress and circadian rhythms as are measurements of serum TH concentrations. The disadvantage is that these assays are not developed for use as high throughput screening tests and at present are laborious. Additionally, although these assays provide data useful in understanding the mechanisms of action, not all chemicals that affect the thyroid produce alterations in T^sub 4^ glucuronidation. Finally, measuring serum T^sub 4^ half-life would be a more direct measure of the adverse effect of increasing T^sub 4^ clearance.

Role of Binding Proteins

Another prevailing theory proposed to explain the mechanism by which some chemicals can reduce circulating levels of thyroid hormone is that of displacing the hormone from serum binding proteins-especially transthyretin in rodents (Brouwer et al., 1998). This hypothesis is supported by the observation that certain biphenyls can displace T^sub 4^ from transthyretin with great affinity (Chauhan et al., 2000). Although provocative, TTR-null mice are euthyroid as are humans with a TTR deficiency (Palha, 2002a, 2002b). Thus, it does not appear that TTR is a requirement for normal thyroid hormone homeostasis. However, it is likely to be important to measure serum binding proteins as a way of interpreting changes in serum total T^sub 4^/T^sub 3^.

In mammals, the serum-binding proteins for thyroid hormones are thyroid-binding globulin (TBG), transthyretin (TTR), and albumin (see review above). T^sub 4^ exhibits a greater affinity for TBG and TTR than does T^sub 3^ (25). Although TBG is present both in humans and rodents, the role of TBG in thyroid physiology in rodents is less well understood than in humans. However, TTR is present in humans, rodents, and nonhuman primates (Schussler, 2000). In addition, there is speculation that xenobiotics can alter circulating levels of thyroid hormone by displacing T^sub 4^ from TTR (Brouwer et al., 1998; Chauhan et al., 2000). Although this hypothesis is plausible, it is by no means proven. Thus, effects of xenobiotics on serum protein binding are not known to produce adverse effects. It has also been suggested that xenobiotic binding to TTR is predictive of interactions with other TI binding proteins such as the deiodinases and sulfotransferases as well as chemicals with potential for high fetal accumulation (Brouwer et al., 1998). These assays have been performed in several laboratories examining xenobiotics for several decades (e.g., Brouwer and van den Berg, 1986). Although these assays can be modified for high throughput screening, they are specific for chemicals that compete with 125I- T4 for serum binding proteins and will not detect chemicals that act through other mechanisms. In addition, the use of either TBG or TTR may not be relevant for nonmammalian species such as teleosts. However, one of the strengths of this assay is that it may be predictive of chemicals that alter fetal concentrations of TH and may provide for a useful screen in this capacity.

No comprehensive examination of the ability of individual toxicants or their metabolites to interfere with TH signaling has been reported. However, several recent reports have begun to characterize the ability of various xenobiotics to bind to the TR and/or to affect TH signaling in vitro and in vivo. Bogazzi et al. (2003) reported that the PCB mixture Aroclor 1254 can displace T^sub 3^ from rat TRbeta1 at 10 [mu]M. Ten micromolar Aroclor also inhibited T3-induced CAT activity from a DR4-tk-CAT reporter construct in COS-7 cells co-transfected with the rat TRbeta1 receptor in their study, indicating that individual congeners, or the mixture, can act as TRbeta1 antagonists. They did not observe that Aroclor affected the ability of the TR to bind to a TH response element (TRE) using electrophoretic mobility shift assay (EMSA), but they did report that in the presence of Aroclor, the TRbeta1 protein is more sensitive to protease digestion than in the presence of T3, indicating that Aroclor binds to the TRbeta1 protein and affects its conformation differently than does T^sub 3^.

These findings of Bogazzi et al. conflict to some degree with those of Koibuchi's group (Iwasaki et al., 2002; Miyazaki et al., 2004). Specifically, Iwasaki et al. (2002) reported that a specific hydroxylated PCB (4-OH-CB106) inhibited relative luciferase activity driven by a DR4-tk-Luc reporter in the presence of TRbeta1 and T^sub 3^ in CV-1,293, and TE671 cells. This inhibition occurred at a very low dose (10^sup -10^ M). Miyazaki et al. (2004) followed this observation with an experiment indicating that 4-OH-CB106 causes the TR to become partially dissociated from the DR4 TRE using EMSA. Differences in study design between the Bogazzi et al. and the Koibuchi et al. reports (e.g., the use of Aroclor by Bogazzi et al. and 4-OH-PCB106 by Koibuchi's group) do not clearly reconcile their unconcordant results. An important issue may be the use of parent polychlorinated biphenyls (PCBs) versus a hydroxylated metabolite.

Kitamura et al. (2005) have surveyed a large number of hydroxylated PCBs, finding nine that bind to the TR. However, in GH3 cells, all of these nine hydroxylated PCBs acted as TR agonists, increasing GH production and GH3 cell growth. In Kitamura's work, the source of TR for their binding studies was nuclear extracts from MtT/E-2 cells, so it was not possible to discriminate between binding to TRalpha and TRbeta. Moreover, no parent PCBs tested singly exhibited TH-like effects in GH3 cells. Finally, none of the compounds that exhibited TH-like effects in Kitamura's experiments were found to exert estrogenic activity in MCF-7 cells; thus, there was a clear separation of thyroid hormone and estrogenic activities.

Arulmozhiraja and Morita (2004) studied the structureactivity relationships of 24 different hydroxylated PCBs on thyroid hormone "activity" in a yeast-two hybrid system. In this system, the ligand- binding domain of the human TRa receptor was linked to the coactivator TIF2 and inserted into the S. cervisiae Y190 cells that carried the beta-galactosidase reporter gene. All of the 24 hydroxylated PCBs evaluated were low in chlorine number (tetra-, tri- , and dichloro). The most active ligand was 6-OH-CB49 with 3.7% of T^sub 3^ activity on the ligand binding domain of the human TRa 1.

Best characterized as a weak estrogen (Staples et al., 1998), bisphenol A (BPA) binds to the estrogen receptor (ER) with a K^sub i^ of approximately 10^sup -5^M (Krishnan et al., 1993; Gaido et al., 1997). However, BPA binds to and antagonizes T^sub 3^ activation of the TR (Kitamura et al., 2002; Moriyama et al., 2002) with a K1 of approximately 10^sup -4^ M, but as little as 10^sup - 6^ M. At these concentrations, BPA significantly inhibits TR- mediated gene activation (Moriyamaet al., 2002). Additionally, Moriyama et al. found that BPA reduced T3-mediated gene expression in culture by enhancing the interaction with the co-repressor NCoR (Moriyama et al., 2002). Developmental exposure of BPA in rats produces an endocrine profile similar to that observed in thyroid resistance syndrome (Cheng, 2005). Specifically, T4 levels were elevated during development in the pups of BPAtreated animals, but TSH levels were not different from controls (Zoeller et al., 2005a). This profile is consistent with BPA inhibition of TRbeta-mediated negative feedback. However, the thyroid hormone-response gene RC3 was elevated in the dentate gyrus of these BPA-treated animals (Zoeller et al., 2005a). Because the TRa isoform is uniquely expressed in the dentate gyrus, the conclusion was that BPA may be a selective TRbeta antagonist in vivo.

If BPA acts as a TR antagonist in vivo, it is predictable that specific developmental events and behaviors would be affected by developmental exposure to BPA. In this regard, Seiwa et al. (2004) have shown that BPA blocks T^sub 3^-induced oligodendrocyte development from precursor cells (OPCs). In addition, there may be an association between the thyroid resistance syndrome and attention deficit hyperactivity disorder (ADHD) in humans (Hauser et al., 1998; Vermiglio et al., 2004; Siesser et al., 2005) and in rats (Siesser et al., 2005); therefore, it is potentially important that BPA-exposed rats exhibit ADHD-I ike symptoms (Ishido et al., 2004).

Despite the antagonistic effects of BPA on the TRbeta, halogenated BPAs may act as TR agonists (Kitamura et al., 2002). Both tetrabromo- and tetrachlorobisphenol A (TBBPA and TCBPA, respectively) can bind to the thyroid hormone receptor and induce GH3 cell proliferation and growth hormone production (Kitamura et al., 2002). In contrast, Ghisari et al. (2005) found that both BPA and its halogenated derivatives increased GH3 cell proliferation. Thus, these compounds may exert agonistic effects on the TR and this could be important during early brain development. For example, thyroid hormone of maternal origin can regulate gene expression in the fetal brain (Dowling et al., 2000; Dowling and Zoeller, 2000; Dowling et al., 2001); one of these genes codes for Hesl (Bansal et al., 2005). Considering the role of HES proteins in fate specification in the early cortex (Gaiano and Fishell, 2002; Schuurmans and Guillemot, 2002; Wu et al., 2003), the observation that industrial chemicals can activate the TR and increase HES expression (Bansal et al., 2005) may indicate that these chemicals can exert effects on early differentiative events.

Incorporating Thyroid Endpoints into Preexisting Assays

To date, all known thyroid toxicants in mammals have been identified by their ability to change measures of thyroid function, specifically circulating levels of thyroid hormone and changes in thyroid weight and histology (Brucker-Davis, 1998). Based on this observation, DeVito et al. (1999) recommended that serum T4 be used as a central feature of screening and testing programs for thyroid toxicity with concurrent evaluation of thyroid histology to support this screen. Preexisting mammalian in vivo assays are designed to accomplish specific goals, whether it be to evaluate developmental toxicity or reproductive toxicity. Therefore, endpoints of thyroid function can be incorporated into these preexisting assays, but the timing and duration of toxicant exposure and timing of endpoint acquisition must be carefully designed considering the information presented as background in this article. In addition, subchronic and long-term (chronic toxicity and/or carcinogenicity) studies may also provide evidence of thyroid toxicity. In particular, these regularly required tests to screen pesticides will provide relevant information about whether a given compound is a thyroid toxicant.


The existing in vivo mammalian assays were developed by efforts within the OECD, Japan, and the U.S. EPA.

The OECD in vivo mammalian assays include the following designs in rats:

1. OECD Test Guideline 407: Repeated Dose 28-Day Oral Toxicity Study.

2. OECD Test Guideline 414: Prenatal Developmental Toxicity Study.

3. OECD Test Guidelines 415/416: One- and Two-Generation Reproductive Toxicity Studies.

4. OECD Test Guidelines 421/422: Reproduction/Developmental Toxicity Screening Test and Combined Repeated Dose Toxicity Study with the Reproduction/Developmental Toxicity Screening Test.

5. OECD Draft Test Guideline 426: Developmental Neurotoxicity Study.

Japanese researchers are developing computer-based screening models, in vitro cell lines, and a "one life-span test" in rodents. The current in vivo mammalian assays developed by efforts within the U.S. EPA include the following designs in rats:

6. One-generation study (also see OECD 415).

7. Two-generation study (also see OECD 416).

8. 20-day pubertal female study.

9. 20-day pubertal male study.

10. 15-day adult male study.

The U.S. EPA Office of Pesticide Programs has issued "Guidance for Thyroid Assays in Pregnant Animals, Fetuses and Postnatal Animals, and Adult Animals," which recommends when/how the endpoints described in the assays just listed should be collected, depending on the assay and chemical being studied.

The study designs for these assays vary, as described later, but in every case, the endpoints included were originally designed to capture measures of reproductive or general toxicity. Thus, the goal of this article is to demonstrate how the addition of endpoints that will capture thyroid toxicity can be included. Specific endpoints are discussed later. These endpoints need not be added to all assays; rather, strategic additions should be made to include a combination of developmental and adult thyroid endpoints.

Endpoints for Thyroid Toxicity in Rodent Developmental


The current endpoints proposed for thyroid toxicity in the in vivo OECD and U.S. EPA mammalian assays listed above include thyroid weight and histopathology, as well as hormone measurements (T4 and TSH, and perhaps Tj). Thyroid weight provides a measure of its stimulation by TSH over time; thus, if thyroid hormone levels are altered slightly for some duration, thyroid weight may reflect this change before technical assays can detect changes in serum hormone levels. Likewise, thyroid histopathology may provide a more sensitive indicator of this process and may be interpreted as a potential cancer endpoint. McNabb et al. (2004b) recently employed a thyroid endpoint that may be more sensitive and simpler to recruit than thyroid weight and histopathology, which should be considered for development and possibly validation. Specifically, they measured the T^sub 4^ content of the thyroid gland in response to ammonium perchlorate exposure in bobwhite quail and found that this measure was far more sensitive to perchlorate exposure than was either serum T^sub 4^ concentration or thyroid weight (McNabb et al., 2004b). Although this measure cannot be taken to indicate cancer, this measure (intrathyroidal T4) may be an important and easily captured endpoint for thyroid toxicity.

It is important to recognize that changes in circulating levels of thyroid hormones (and thyroid histology) represent important precursor events to adverse outcomes. This issue has been addressed most clearly by Crofton (Crofton, 2004; Crofton et al., 2005). In a seminal series of studies, Crofton found a linear relationship between PCB-induced reductions in circulating levels of thyroid hormone and adverse outcome (i.e., hearing) (Crofton, 2004; Crofton et al., 2005). Thus, at least for one thyroid hormone effect, the relationship between hormone level and hormone action has been characterized (Zoeller and Crofton, 2005b).

The endpoints of thyroid gland weight and histology, serum T^sub 4^, T^sub 3^, TSH, and, potentially, intra-thyroidal T^sub 4^ are measures of thyroid function, which are and should remain key endpoints of thyroid toxicity studies. Importantly, not all toxicants that cause a reduction in serum thyroid hormone levels produce the same profile of hormonal and thyroid changes. For example, PTU exposure produces a decrease in circulating levels of T^sub 4^, T^sub 3^, and a concomitant increase in circulating levels of TSH (appendix A, OECD, 2006). In contrast, PCB exposure causes a decrease in serum T^sub 4^, but this is not usually associated with an increase in serum TSH (appendix A, OECD, 2006). It is possible, though not well studied, that the specific profile of hormonal changes represents a "fingerprint" for the mechanism of action that mediates the antithyroid effect. In the example above, PTU is a TPO inhibitor and PCBs likely reduce serum T^sub 4^ by inducing UDPGTs, and perhaps by displacing T^sub 4^ from serum binding proteins.

However, regardless of the mechanism by which circulating levels of thyroid hormone are reduced, it may be important to determine how a decrement in serum thyroid hormones specifically affects tissues that require thyroid hormone (especially in development).

Therefore, to resolve whether a thyroid toxicant affects brain development, for example, measures of thyroid hormone action in the developing brain would be informative. Similar logic would be applied to resolve whether a thyroid toxicant affects heart or lung development, or adult physiological functions. Recommended rodent developmental endpoints known to be sensitive to thyroid hormone insufficiency will be discussed later in the context of potential inclusion in existing in vivo screens and tests. These endpoints may be considered to be measures of specific effects as well as generalized endpoints of thyroid disruption. From this point of view, endpoints of thyroid hormone action in nonmammalian vertebrates (e.g., frog metamorphosis) may also be considered to be generalized endpoints of thyroid disruption. The concept that one endpoint of TH disruption could be emblematic of TH disruption in general is complex and requires additional information. For example, the ability to metabolize xenobiotics is different among the vertebrate taxa and this may affect the generalizability of specific endpoints of TH action. In addition, there may be sufficient difference among vertebrates in the various proteins that govern the thyroid system (e.g., receptors, peroxidases, transport proteins, etc.) that xenobiotics will not interact with all of these proteins in the same way across the vertebrates.

The assays proposed for screening and testing for endocrine disrupters are reviewed below, and exposure times and endpoints are described to familiarize the reader with current thyroid analyses, as well as to help the reader visualize how well new endpoints or assays may fit into or alter the current rodent methods for analysis of the thyroid system.

Current ways of detecting thyroid toxicity include measures of hormone levels by radioimmunoassay, characterization of thyroid function using histopathological techniques, and in some cases computer assisted morphometry. Although radioimmunoassays are commonly used in current assays, there should be some standardization of the kits used or the analysis methods employed. Some groups evaluate RIAs by extrapolating from the lowest standard to the "zero" tube because some or many of their samples are below the lowest standard, possibly leading to inaccurate measures of hormone levels. This situation likely arises in part because total T4 in rat serum runs near the low end of the standard curve that is calibrated for humans. Direct endpoints of toxicant effects on thyroid hormone action would require validated methods of measurement that could be calibrated across labs. Real-time PCR techniques would likely provide the greatest degree of quantitative reliability because a standard curve could be generated that would provide a basis for comparison across labs. However, if endpoints of thyroid hormone action were to be included in screens and tests for thyroid toxicants, techniques required to capture these endpoints (in situ hybridization, northern blots, etc.) would have to be in some way standardized. While many gene expression changes are being developed as potential endpoints for screens, these endpoints are still in the research and development phase and most are not yet ready for validation.

OECD Test Guidelines

Thyroid endpoints for the OECD Test Guidelines can be proposed as additional endpoints to add on to the existing assay protocols. The thyroid endpoints currently under consideration for the OECD Test Guidelines are the same as those in the EDSP-namely, thyroid weight, hormone analysis (T^sub 4^, T^sub 1^, TSH), and thyroid gland histopathology. In the material presented next, the term "screening assay" refers to a protocol designed to obtain initial information about the ability of a compound to interfere with the thyroid system. The goal of this test is to maximize the number of true positives and minimize the number of false negatives. In contrast, the term "test" or "testing assay" would establish whether a substance could cause effects through the thyroid system, determine the consequences to the organism studied, and would establish a dose response relationship between the substance and the effects observed in the test.

OECD TG 407-Repeated Dose 28-Day Oral Toxicity Study in Rodents

This is a 28-day assay to evaluate a test chemical's oral toxicity using repeated daily doses in adult animals. The preferred rodent species is the rat, although other rodent species may be used. Females should be nulliparous and nonpregnant; dosing should begin as soon as possible after weaning and, in any case, before the animals are 9 weeks old. The route of administration should be by gavage, dosed feed, or dosed water. This study will indicate the potential health hazards of a test chemical after repeated exposure for a relatively short duration, especially immunological and neurological effects as well as reproductive toxicity. The TG 407 protocol was recently enhanced to include the thyroid endpoints mentioned earlier. This assay is considered to identify a test chemical's effects through clinical observation, hematology, clinical biochemistry of the blood serum and urine, pathology, and histology on organs that are chosen according to the user's needs. Results from this assay will inform the chemical testing community on how to proceed with further tests.

OECD TG 414-Prenatal Developmental Toxicity Study

OECD TG 414 tests for the effects of prenatal toxicant exposure (normally by intubation) on both the pregnant test animal and the developing offspring. Animals are dosed with the test chemical from implantation (around 5 days after mating) to 1 day before the planned cesarean section. This test will usually include the entire period of gestation, but can be shortened depending on the needs of the administering scientist. The assay is designed to observe effects on organogenesis. Suggested endpoints include: clinical observations; analysis of the dams including a complete examination of the uterus; and analysis of the fetus including sex, external alterations, and skeletal and soft tissues analysis. No specific thyroid endpoints are in included in this assay. This assay corresponds to the U.S. EPA Developmental Toxicity Assay and the U.S. Food and Drug Administration (FDA) Segment II study. OECD TG 415-One-Generation Reproduction Toxicity Study

OECD TG 415 tests for a chemical's effects on male and female reproductive performance (i.e., gonadal function, estrus cyclicity, mating behavior, conception, parturition, lactation, and weaning). The one-generation assay also identifies developmental toxicity (i.e., neonatal morbidity, mortality, behavioral abnormalities, teratogenesis). It corresponds to the U.S. EPA onegeneration assay, but doses animals earlier than the EDSP proposed one-generation reproduction assay.

The experimental schedule for this assay doses the parental generation prior to mating (at least 10 weeks for male rats and 2 weeks for female rats) and then throughout mating. The dams are then dosed throughout gestation and lactation until weaning of the Fl generation. Dosing and necropsy of the Fl generation are adjusted according to the intended use for this assay (see EDSP on the one- generation assay later in this article). The endpoints included in the test guideline include physical observations, and histopathology of the ovaries, uterus, cervix, vagina, testes, epididymides, seminal vesicles, prostate, coagulating gland, and the pituitary gland. Other target organs may be added as necessary. Thyroid endpoints, including those mentioned earlier, could easily be added to this assay.

OECD TG 416-Two-Generation Reproductive Toxicity Studies

The OECD TG 416 corresponds to the U.S. EPA TwoGeneration Reproductive Toxicity Test as described later. The EDSP proposed two- generation test differs from TG 416 in that the dosing does not begin prior to mating, whereas the TG 416 begins dosing the male rats at least 10 weeks prior to mating and the female rats at least 2 weeks prior to mating. In both guidelines, the dosing begins with the parental generation, continuing throughout mating, pregnancy, and lactation to weaning of the Fl generation. The Fl offspring, once weaned, are dosed throughout development, mating, pregnancy, and lactation, to weaning of the F2 generation. Results from this assay are used to assess whether additional studies are required.

Endpoints outlined for this assay are very similar to those described for the one-generation assay (TG 415), but also include estrus cycle and sperm evaluations, extensive observation of the offspring in the Fl and F2 generations, and organ weights of dosed animals (uterus, ovaries, testes, epididymides, prostate, seminal vesicles and coagulating glands and fluids, brain, liver, kidneys, spleen, pituitary, thyroid, and adrenal glands). Other target organs can also be added on as needed. Histopathology of the parental and Fl generation are also required for certain organs (vagina, uterus with cervix, ovaries, 1 testis, 1 epididymis, seminal vesicles, prostate, and coagulating gland), and additional ones can be examined if necessary.

OECD TG 421 and422-The Reproduction/Developmental Toxicity Screening Test and the Combined Repeated Dose Toxicity Study with the Reproduction/ Developmental Toxicity Screening Test

OECD TG 421 and 422 are both screening assays designed to provide the initial information on the effects of a test chemical on male and female reproduction (Figure 2). Both screens offer limited information on whether a test substance causes abnormal postnatal effects after prenatal exposure, or whether the effects are due to postnatal exposure. Because these are both considered screens, negative data do not indicate that a chemical is safe. TG 422 also focuses on neurological endpoints.

Exposure schedules for these test guidelines are approximately 54 days long with dosing for ~14 days premating, ~14 days mating (or less), 22 days during gestation, and then 4 days of lactation.

Endpoints for this assay include clinical observations of adults, body weight, and food consumption changes throughout the study, pathology, and histology (for reproductive organs and accessory sex glands). TG 422 also includes hematology, clinical biochemistry on blood plasma or serum samples and urine. Histopathology includes organs other than the reproductive organs such as the brain (cerebellum, cerebrum, pons), spinal cord, stomach, small and large intestines, liver, kidneys, adrenals, spleen, heart, thymus, thyroid, trachea, lungs, uterus, urinary bladder, lymph nodes, peripheral nerve, and bone marrow. Pups are observed after necropsy for external gross abnormalities. Thyroid hormone serum analysis and thyroid histopathology are included as potential endpoints for these two test guidelines.

No currently proposed EDSP assay is similar to these two test guidelines in the dosing schedule or the proposed endpoints.

OECD Test Guideline 426: Developmental Neurotoxicity Study (Draft 2003)

This test guideline was initially developed based on the U.S. EPA guideline for developmental neurotoxicity testing. OECD expert consultation meetings were held in 1998 and in 2000 to develop this draft. The protocol is designed to be performed as an independent study. However, observations and measurements could also be incorporated into a perinatal developmental toxicity study, or added on to a one- or twogeneration study. Thyroid endpoints could be added onto Tg 426, or onto a fusion of Tg 426 and a one- or two- generation study. The recommendation is that dams be exposed to at least 3 doses of compound from gestational day (G) 6 to weaning at PND 21. Preweaning measures would occur on PND 22 and would include behavioral ontogeny, brain weight, neuropathology and morphometry (optional). Postweaning investigations would occur around PND 60 and would include behavioral/functional tests, motor activity, sexual maturation, auditory startle, brain weight, neuropathology and morphometry (optional).

Summary of OECD Test Guideline Thyroid Endpoints

The thyroid endpoints being considered as "add-ons" for these test protocols include thyroid gland weight, histology, and serum thyroid hormone measurements including TSH. As discussed previously, these endpoints can provide information about whether a chemical alters serum thyroid hormone levels. For example, endpoints of developmental neurotoxicity that are specific to thyroid hormone action may be considered to provide information that is important in interpreting the consequences of changes in thyroid hormone that were identified either by radioimmunoassay or by changes in thyroid weight/histopathology.

Japanese Screening and Testing Program

Overall, the screening and testing program for EDCs being developed by Japanese researchers is comprised of three components: ( 1 ) in silico screening using a 3D-S AR docking model (but not for TR), (2) in vitro assays using mammalian cell lines including a TR expression system, and (3) several in vivo assays for estrogen and androgen. A "rodent one-life-span test" that will include endpoints for thyroid toxicants is also being developed as the Tier 2 "definitive" rodent test. The presumptive one-life-span test protocol would monitor the major stages of one life-span of rodents, including conception, in utero development, growth, maturation, and senescence. The exposure period may be perinatal and the monitoring periods would be not only around puberty but also in adulthood and/ or early senescence. Currently, the endpoints under consideration will cover reproductive endpoints as well as endpoints for neurotoxicity and the immune system, with an emphasis on functional endpoints including acceleration of senescence-related phenotypes. Toxicogenomics approaches may be incorporated for monitoring the molecular events underlying the adverse effects. It will be important to incorporate endpoints of thyroid toxicants in this one life-cycle test.

FIG. 3. Exposure and endpoint collection in EPA two-generation reproductive toxicity assay.

U.S. EPA EDSP Rodent Assays

As with the OECD test guidelines, the proposed assays in the EDSP battery contain thyroid endpoints that were added to assays for reproductive and developmental toxicity. These thyroid endpoints, generally proposed as "add-ons" in the EDSP battery of assays, are thyroid gland weight and histology, and serum thyroid hormone measures (T^sub 3^, T^sub 4^, TSH).

Two-Generation Study (Similar to OECD TG 416)

One of the tests being considered for inclusion in the EDSP is a rat two-generation reproductive toxicity test that could be modified for thyroid toxicity. It is similar to the OECD TG 416. The basic two-generation test is described by the U.S. EPA Office of Prevention, Pesticides, and Toxic Substances Health Effects Test Guideline 870.3800: Reproduction and Fertility Effects (U.S. EPA, 1998): http://www.epa.gov/scipoly/ oscpendo/docs/edmvs/ ptu2gendraftforedmvs.pdf. The assay is illustrated in Figure 3.

Thyroid endpoints under consideration for this test protocol include thyroid weight, histology, and thyroid hormone analysis of T^sub 4^ and TSH. This test has completed prevalidation with the thyroid endpoints included. The two-generation assay is a Tier 2 test that identifies functional disruption of the estrogen, androgen, and thyroid systems during exposure to a chemical over two generations. One-Generation Assay (Similar to OECD TG 415)

Although the basic two-generation study design was developed to provide information on insult to the reproductive tract, there is concern that certain effects may be missed because the reproductive tract has not had sufficient time to develop before the observations are made. In the standard two-generation test, most Fl animals are sacrificed and examined at PND 21 ; only one animal per sex per litter is usually allowed to continue to maturity, and these animals are then used to breed the F2 generation. An alternative to the two- generation study is a one-generation study that would allow for examination of the Fl males past puberty at postnatal day (pnd) 90 +- 2. The study design tests whether continuing toxicant exposure in the Fl generation males through puberty to adulthood will provide additional information for detection of endocrine-mediated effects. The one-generation study has been proposed as an alternative to the two-generation study. In addition, the U.S. EPA EDSP conducted a special study of a one-generation test that was added on as an extension to a two-generation assay and continued the Fl male generation out to pnd 95 +- 5 (Gray et al., 2003). It should be noted that the original study design described only utilized Fl males, but future assays will likely also include females. The study design is illustrated in Figure 4.

The objectives of the one-generation study and the onegeneration extension study are to determine the following: ( 1 ) Can some of the effects of perinatal exposure to thyroid toxicants be missed if the timing of endpoint acquisition is structured to identify reproductive toxicants in postweanling animals, and (2) do some of these effects occur at an incidence that would go undetected if only one male per litter were retained past puberty and examined in adulthood?

FIG. 4. Study design to examine effects in Fl offspring.

Retaining a greater number of the Fl males to examine at or after puberty may allow for greater distinction of the thyroid endpoints such as thyroid growth and histology. It is not yet clear whether the EDSP will proceed in validating the basic onegeneration study to use as an alternative to the two-generation assay or the one- generation extension of the two-generation study.

Endpoints of thyroid function are included in the one- and two- generation assays. As more endpoints are developed and incorporated into these regulatory assays, more specific effects of toxicants will be identified, decreasing the possibility that these regulatory assays will produce false negatives.

Female Pubertal Assays

The EDSTAC, assembled by the U.S. EPA in 1996, recommended the use of a female 20-day pubertal assay with endpoints to evaluate test materials for effects on the thyroid, the hypothalamic- pituitary-gonadal (HPG) axis, and aromatases. The EPA, at the recommendation of the EDSTAC, has proposed to include a female pubertal assay in an endocrine disruptor screening program. This assay (Figure 5) is the most comprehensive assay in the proposed Tier 1 battery of assays, as it is capable of detecting substances that alter thyroid function, inhibit aromatase, act as estrogens or antiestrogens, and interfere with the hypothalamus-pituitary-gonad/ thyroid axis (EDSTAC Report, 1998, Vol. 1, pp. 5-26 to 5-27). The female pubertal assay is currently being validated by several labs. The protocol for the female pubertal assay measures the following thyroid endpoints: serum T4 and TSH concentrations, thyroid gland histology, thyroid gland weight, and body weight changes. Results from other, shorter assays and/or with the use of intraperitoneal (ip) injection as the route of administration, have also been reported (O'Connor et al., 1996,1999). EDSTAC also recommended that the male 20-day pubertal assay in rodents be evaluated as an alternate assay (EDSTAC, 1998, Vol. 1, pp. 5-30; see section

FIG. 5. Female pubertal assay.

In the female pubertal assay, toxicant exposure begins on the day of weaning (pnd 21 ). Thus, many of the developmental endpoints sensitive to thyroid hormone (see Figure 5) have passed. Two potential endpoints of thyroid hormone to be considered during puberty and potentially in future versions of this assay protocol include measures of myelination and toxicant effects on BrdU- labeled cells in the hippocampus. Addition of these endpoints will require research and development, followed by protocol standardization and validation. Although this process is not immediate, it is important to consider new thyroid endpoints as the state of thyroid research expands, and in recognition that the current assay endpoints will need to be improved as more information on the thyroid system becomes available.

Male Pubertal Assay

The EDSTAC also recommended that a 20-day male pubertal assay in rodents be evaluated as an alternative assay (EDSTAC, 1998, Vol. 1, pp. 5-30). This assay is the most comprehensive assay in the proposed Tier 1 battery of assays, as it is capable of detecting substances that alter thyroid function, inhibit aromatase, act as androgens or antiandrogens, and interfere with the hypothalamus- pituitary-gonadal axis (EDSTAC, 1998, Vol. 1, pp. 5-30 through 5- 32). The study design for the male pubertal assay (Figure 6) is similar to the female pubertal assay. The male pubertal assay has been lengthened to a 30-day pubertal assay that covers PNDs 22-52 and is currently being validated. It includes the following thyroid- related endpoints: body weight, thyroid gland weight, thyroid gland histology, and T4 and TSH plasma concentrations at necropsy. Therefore, the EDSP is pursuing the validation of a male pubertal assay as a potential alternative to other assays in the Tier 1 battery.

FIG. 6. Male pubertal assay.

15-Day Adult Male Screen

One of the assays recommen