Detection of Thyroid Toxicants in a Tier I Screening Battery and Alterations in Thyroid Endpoints Over 28 Days of Exposure

Detection of Thyroid Toxicants in a Tier I Screening Battery and Alterations in Thyroid Endpoints over 28 Days of Exposure

John C. O’Connor,1 Steven R. Frame, Leonard G. Davis, and Jon C. Cook2 DuPont Haskell Laboratory for Toxicology and Industrial Medicine, P.O. Box 50, Elkton Rd., Newark, Delaware 19714

Received February 5, 1999; accepted April 28, 1999 follicular cell proliferation (1- and 2-week time points). Histolog- Phenobarbital (PB), a thyroid hormone excretion enhancer, and ical effects in PB-treated rats were limited to mild colloid deple- propylthiouracil (PTU), a thyroid hormone-synthesis inhibitor, tion at the 2- and 4-week time points. At all three time points, PTU have been examined in a Tier I screening battery for detecting increased relative thyroid weight, increased serum TSH, decreased endocrine-active compounds (EACs). The Tier I battery incorpo- serum T3 and T4, increased thyroid follicular cell proliferation, rates two short-term in vivo tests (5-day ovariectomized female and produced thyroid gland hyperplasia/hypertrophy. Thyroid battery and 15-day intact male battery using Sprague-Dawley gland histopathology, coupled with decreased serum T4 concen- rats) and an in vitro yeast transactivation system (YTS). In addi- trations, has been proposed as the most useful criteria for identi- tion to the Tier I battery, thyroid endpoints (serum hormone fying thyroid toxicants. These data suggest that thyroid gland concentrations, liver and thyroid weights, thyroid histology, and weight, coupled with thyroid hormone analyses and thyroid his- UDP-glucuronyltransferase [UDP-GT] and 5؅-deiodinase activi- tology, are the most reliable endpoints for identifying thyroid ties) have been evaluated in a 15-day dietary restriction experi- gland toxicants in a short-duration screening battery. The data ment. The purpose was to assess possible confounding of results further suggest that 2 weeks is the optimal time point for identi- due to treatment-related decreases in body weight. Finally, several fying thyroid toxicants based on the 9 endpoints examined. Hence, thyroid-related endpoints (serum hormone concentrations, hepatic the 2-week male battery currently being validated as part of this UDP-GT activity, thyroid weights, thyroid follicular cell prolifer- report should be an effective screen for detecting both potent and ation, and histopathology of the thyroid gland) have been evalu- weak thyroid toxicants. ated for their utility in detecting thyroid-modulating effects after Key Words: screening; Tier I battery; rats; thyroid gland; hor- 1, 2, or 4 weeks of treatment with PB or PTU. In the female battery, changes in thyroid endpoints following PB administra- mone concentrations; cell proliferation; UDP-glucuronyltransferase; phenobarbital; propylthiouracil; time course; dietary re- tion, were limited to decreased serum tri-iodothyronine (T3) and striction. thyroxine (T4) concentrations. There were no changes in thyroid stimulating hormone (TSH) concentrations or in thyroid gland histology. In the male battery, PB administration increased serum TSH and decreased T3 and T4 concentrations. The most sensitive Under the current legislative requirements of the Food Qual- indicator of PB-induced thyroid effects in the male battery was ity Protection Act of 1996 and the Safe Drinking Water Act of thyroid histology (pale staining and/or depleted colloid). In the 1996, the United States Environmental Protection Agency female battery, PTU administration produced increases in TSH (U.S. EPA) was mandated, starting in August 1999, to imple- concentrations, decreases in T3 and T4 concentrations, and microscopic changes (hypertrophy/hyperplasia, colloid depletion) in the ment a testing strategy for screening chemicals and pesticides thyroid gland. In the male battery, PTU administration caused for endocrine activity. In response, the EPA established the thyroid gland hypertrophy/hyperplasia and colloid depletion, and Endocrine Disruptor Screening and Testing Advisory Commit- the expected thyroid hormonal alterations (increased TSH, and tee (EDSTAC) to provide screening strategies. Presently, the decreased serum T3 and T4 concentrations). The dietary restriction EDSTAC is recommending screening compounds for their study demonstrated that possible confounding of the data can potential to act as agonists or antagonists to the estrogen occur with the thyroid endpoints when body weight decrements receptors (ER) or androgen receptors (AR), steroid biosynthe- are 15% or greater. In the thyroid time course experiment, PB produced increased UDP-GT activity (at all time points), in- sis inhibitors, or for their ability to alter thyroid function (EDSTAC, 1998). As part of our research program to develop creased serum TSH (4-week time point), decreased serum T3 (1- and 2-week time points) and T4 (all time points), increased relative effective methods to screen for endocrine-active compounds thyroid weight (2- and 4-week time points), and increased thyroid (EACs), we have initiated a validation of a tiered-testing

1 scheme using 15 model EACs (Cook et al., 1997; O’Connor et To whom correspondence should be addressed. Fax (302) 366-5003. al., 1998a,b). Our Tier I-testing scheme incorporates two short- Email: [email protected]. 2 Present address: Pﬁzer, Inc., Central Research, Eastern Point Rd., Groton, term in vivo tests (5-day ovariectomized female battery; 15-day CT 06340. intact male battery) and an in vitro yeast transactivation system

54 RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 55

(YTS) for identifying compounds that alter endocrine ho- Many factors can affect thyroid hormone concentrations in- meostasis. The two main goals of these studies are to test the cluding diet, stress, age, and circadian rhythm (Capen, 1997; hypothesis that distinct “fingerprints” of endocrine activity can Hill et al., 1989). Furthermore, detection of small changes in be identified for specific EACs, and to evaluate which end- thyroid hormone concentrations can be difficult due to normal points should be included in a final Tier I type screen. We have variability between animals (Davies, 1993). For these reasons, previously used 6 positive controls to examine the usefulness care must be taken in the interpretation of hormonal data to of the Tier I screening battery for identifying EACs with differentiate between changes in thyroid hormone homeostasis diverse endocrine activities (O’Connor et al., 1998a,b). In this that represent modulation (compensated changes that produce report, we examined phenobarbital (PB), a thyroid hormone- no tissue structural changes) and those that represent disruption excretion enhancer, and 6-n-propyl-2-thiouracil (propylthio- (changes that cannot be compensated for and produce tissue uracil; PTU), a thyroid hormone-synthesis inhibitor, in order to structural changes) of the thyroid axis. characterize the ability of the Tier I screening battery to detect At the 1997 Duke University Thyroid Workshop (Durham, known thyroid toxicants. NC) entitled Screening Methods for Chemicals That Alter Considerable data from studies in experimental animals in- Thyroid Hormone Action, Function and Homeostasis, thyroid dicate the existence of a relationship between sustained dis- gland histopathology was judged to be the most sensitive ruption of the hypothalamic-pituitary-thyroid (HPT) axis and parameter for the detection of compounds that adversely affect the progression of thyroid follicular cells to hypertrophy, hy- thyroid function. One concern regarding the inclusion of thy- perplasia, and eventually neoplasia (Hill et al., 1989; Paynter et roid gland histopathology is that it is unclear whether a 2-week al., 1988). In particular, hypersecretion of thyroid stimulating or even a 4-week exposure is long enough to detect weak- hormone (TSH) has been recognized as a common mechanism acting compounds that disrupt thyroid economy (i.e., com- for the induction of thyroid follicular cell proliferative lesions pounds whose primary action is in the periphery). To date, only by xenobiotics that disrupt thyroid hormone homeostasis (re- one published report has observed that PB can induce his- viewed in Capen, 1997; Davies, 1993; Hill et al., 1989). All topathological changes in the thyroid gland after 20 days of known thyroid tumorigens (except those that are direct muta- treatment (Japundzic, 1969). A comprehensive evaluation of a gens), as well as iodine deficiency, induce thyroid proliferative battery of thyroid endpoints in a single study design over 4 lesions through a chronic compensatory secretion of TSH, as a weeks of treatment has yet to be reported. result of perturbations in thyroid hormone economy (reviewed In the current study, two model thyroid toxicants, PB and in Capen, 1997; Davies, 1993; Hill et al., 1989). However, the PTU, have been examined using a Tier I screening battery. PB increased secretion of TSH may be prompted by several mech- and PTU were used to characterize the ability of our proposed anisms, such as direct action on the thyroid gland, either by Tier I screening battery to detect thyroid toxicants with differ- inhibition of thyroid hormone synthesis or secretion, or indi- ent modes of action, indirect versus direct, respectively. In rectly through inhibition of 5Ј-deiodinase or induction of he- addition, we have examined the effect of diet restriction for 15 patic microsomal enzymes (reviewed in Capen, 1997; Leonard days on several thyroid endpoints (serum hormone concentra- and Roseburg, 1980; McClain et al., 1989; Pazos-Moura et al., tions, hepatic UDP-glucuronyltransferase (UDP-GT) and 5Ј- 1991). It is noteworthy to add that while receptor binding is deiodinase activities, and thyroid gland weight) in male rats. common among many EACs (i.e., ER agonists bisphenol A The purpose was to help identify possible confounding due to and methoxychlor; AR antagonists linuron and p,p’-DDE), decreased body weight. Finally, serum thyroid hormone con- there is no evidence of environmental chemicals inducing centrations, UDP-GT activity, thyroid gland weight, thyroid thyroid effects by binding to the thyroid hormone receptor. follicular cell proliferation, and histopathology of the thyroid In developing a screening battery for identifying thyroid gland have been evaluated for their utility in detecting thyroid toxicants, many models are available including both in vitro effects after 1, 2, or 4 weeks of treatment, using PB or PTU. and in vivo assays. However, two endpoints that are routinely This comprehensive evaluation of thyroid endpoints was initi- used for identifying compounds that alter thyroid function are ated to examine whether a 2-week time point, similar to that thyroid hormone measurements and histopathology of the thy- currently undergoing validation, is useful for detecting thyroid roid gland (Davies, 1993). In addition, since all known thyroid toxicants. toxicants affect thyroid hormone concentrations (i.e., increase TSH) (Capen, 1997; Davies, 1993; Hill et al., 1989) and the MATERIALS AND METHODS ultimate measure of a true thyroid toxicant is histopathological changes of the thyroid gland, these two endpoints may be the Test materials. The following materials were obtained from the Sigma most reliable for a short-duration screening model. With the Chemical Co. (St. Louis, MO): acetic acid, adenine, ammonium sulfate, bovine increasing number of commercially available rat hormone as- serum albumin (BSA), Brij 58, 5-bromo-2Ј-deoxyuridine (BrdU), neutralized charcoal, copper sulfate, diethylstilbestrol, dextran, dithiothreitol, EDTA (di- say kits, examining thyroid hormone homeostasis is a quick sodium salt), glycerol, HEPES, L-histidine, L-lysine, leupeptin, magnesium and easy method to identify compounds that alter thyroid chloride, 2-mercaptoethanol, p-nitrophenol, o-nitrophenyl ␤-o-galactopyrano- function; however, these data require cautious interpretation. side (ONPG), PB, potassium phosphate (monobasic), progesterone, PTU, 56 O’CONNOR ET AL. sodium bicarbonate, sodium dodecyl sulfate (SDS), sodium EDTA, sodium for thyroid-modulating activity, the male rat is a more appropriate model than fluoride, sodium hydroxide, sodium molybdate, sodium phosphate (monobas- the female. Male rats were given a range of PMI௡Feeds, Inc., Certified Rodent ic), sucrose, testosterone, trichloroacetic acid (TCA), Trizma-base, Tris–HCl, Diet #5002 meal in order to produce different levels of growth retardation. The L-tryptophan, and uridine 5Ј-diphosphoglucuronic acid (UDPGA). All other levels of diet restriction were: ad libitum control, 22 g feed/day, 19 g feed/day, materials were obtained from the following manufacturers: Certified Rodent 16 g feed/day, and 13 g feed/day. Levels of diet restriction were based on 125 3 Diet #5002, PMI௡Feeds, Inc. (St. Louis, MO); [ I] reverse T3 (rT3), [ H] calculated food consumption from a range-finder study and were targeted to 17␤-estradiol, [3H] promegestone (R5020), and unlabelled promegestone achieve decreases in body weight to 95, 90, 85, and 80% of the ad libitum (R5020), New England Nuclear (Boston, MA); Bio-Rad Poly-Prep columns control. Serum hormone concentrations, UDP-GT and 5Ј-deiodinase activities, AG50W-X8, 200–400 mesh, Bio-Rad Corp. (Hercules, CA); osmotic liver and thyroid weights, and thyroid gland histopathology were evaluated. minipumps model 2 ML1, Alza Corporation (Palo Alto, CA); ethanol, Electron Cell proliferation. Six female rats from each group were designated for Microscopy Science (Fort Washington, PA); pefabloc, Boehringer Mannheim uterine stromal cell proliferation analyses. The day prior to study start (test day (Indianapolis, IN); methylcellulose and potassium chloride, Fisher Scientific Ϫ 1), rats were anesthetized using isoflurane and implanted (subcutaneously) (Springfield, NJ); dextrose, magnesium sulfate, and methanol, J.T. Baker, Inc. with Alzet osmotic pumps loaded with 20 mg/ml BrdU dissolved in 0.5 N (Phillipsburg, NJ); oxalyticase, Enzogenetics, Inc. (Corvallis, OR); yeast ex- sodium bicarbonate buffer. On test day ϩ5, the rats were sacrificed and the tract, Difco, Inc. (Detroit, MI); luteinizing hormone (LH; catalog #RPA.552), uteri were trimmed, affixed to dental wax, fixed for2hinBouin’s solution, prolactin (PRL; catalog #RPA.553), thyroid stimulating hormone (TSH; cata- routinely processed, paraffin embedded, sectioned, and stained for immuno- log #RPA.554), and follicle stimulating hormone (FSH; catalog #RPA.550) histochemical analysis of BrdU incorporation into DNA, or sectioned and radioimmunoassay (RIA) kits, Amersham Corp. (Arlington Heights, IL); tes- evaluated for epithelial cell height. Uterine stromal cell proliferation and tosterone (T; catalog #TKTT5), estradiol (E2; catalog #KE2D5), tri-iodothy- uterine epithelial cell height evaluation were performed as previously de- ronine (T3; catalog #TKT35), and thyroxine (T4; catalog #TKT45) RIA kits, scribed (O’Connor et al., 1996). Diagnostic Products Corp. (Los Angeles, CA); rT RIA kits (catalog #10834), 3 Estrous conversion. Rats assigned to the biochemical/hormonal subset Polymedco Inc. (Cortlandt Manor, NY); and, dihydrotestosterone (DHT; cat- were evaluated for the stage of estrous by vaginal cytology. Vaginal washes alog #DSL-9600) RIA kit, Diagnostic Systems Laboratories (Webster, TX). were collected once daily by repeated pipetting of 75 ␮l of 0.9% sterile saline Test species. Male and female Sprague-Dawley (Crl:CD௡(SD)IGS BR) into the vagina. Slides were air dried, stained by the Wright-Geimsa method, rats were acquired from Charles River Laboratories, Inc. (Raleigh, NC). Male and evaluated for conversion out of diestrus (Davis, 1993). rats were approximately 63 days old upon receipt. Female rats, approximately 42 days old, were ovariectomized on the day of shipment (41 days old). Upon Pathological Evaluations arrival, rats were housed in stainless steel, wire-mesh cages suspended above cage boards and were fed irradiated PMI௡Feeds, Inc., Certified Rodent Diet Female rats. Between 0700 h and 1000 h on the morning of test day ϩ5 #5002 and provided with tap water (United Water Delaware) ad libitum. (approximately 24 h after the last administered dose), rats were anesthetized

Animal rooms were maintained on a 12-h light/dark cycle (fluorescent light), using CO2 and euthanized by exsanguination. Blood was collected from the a temperature of 23 Ϯ 2°C, and a relatively humidity of 50% Ϯ 10%. descending vena cava and serum prepared for hormonal analyses. The pres- Upon arrival, rats were quarantined for at least 1 week and released on the ence of fluid in the uterine horns was recorded as a gross observation. Uteri bases of adequate body weight gain and freedom from clinical signs of disease from the biochemical/hormonal subset were dissected, weighed, and processed or injury. After release from quarantine, rats were divided by computerized, to uterine cytosol for measurement of uterine progesterone and estrogen stratified randomization into treatment groups so that there were no statistically receptor content (O’Connor et al., 1996). The thyroid glands from the bio- significant differences among group body weight means. During testing, all chemical/hormonal subset were placed in formalin fixative and examined rats were weighed daily and cage-side examinations were performed to detect microscopically. Uteri from the cell proliferation subset were dissected, moribund or dead rats. At each weighing, rats were individually handled and weighed, and fixed in Bouin’s solution. Processed tissues were embedded in examined for abnormal behavior or appearance. All test compounds were paraffin, sectioned at 5 ␮m, and stained with hematoxylin and eosin (HE) for prepared in 0.25% methylcellulose vehicle and administered by intraperitoneal microscopic evaluation. Uterine stumps (the site of the ovarian dissection) injection at approximately 0900 h daily. Doses were selected in order to obtain from all rats were saved in 10% neutral-buffered formalin, processed as the maximal pharmacologic effect for each compound and not to exceed the described above, and examined microscopically to confirm the absence of maximum tolerated dose (MTD) as determined from the scientific literature. ovarian tissue. Blood was collected from the descending vena cava and serum The dose volume was 2.0 ml/kg body weight. prepared for hormonal analyses. Male rats. On the morning of test day ϩ15, rats were injected with the final dose of test compound approximately two h prior to sacrifice. Between In Vivo Tier I Battery Studies 0700 h and 1000 h, rats were anesthetized using CO2 and euthanized by Study design. The study design consisted of 5 groups of 20 females and 5 exsanguination. Blood was collected from the descending vena cava and serum groups of 15 males. Within each group of 20 female rats, 14 animals were prepared for hormonal analyses. The liver, testes, epididymides, prostate, designated for biochemical/hormonal evaluation and 6 rats were designated for seminal vesicles, and accessory sex gland (ASG) unit (consisting of ventral and cell proliferation/morphometric evaluation. All male rats were designated for dorsal lateral prostate, seminal vesicles with fluid, and coagulating glands) biochemical/hormonal evaluation. Male rats were dosed for 15 days and were weighed and organ weights calculated relative to body weight. The euthanized on the morning of test day ϩ15. Female rats were dosed for 4 days epididymis and thyroid gland were placed in formalin fixative, the testes were and euthanized on the morning of test day ϩ5. Refer to the report by Cook and placed in Bouin’s fixative, and all were processed as described above and coworkers (1997) for an explanation of the study design. The following examined microscopically. In addition, selected sections of thyroid gland were compounds were used in the male and female in vivo battery: PB (5, 25, 50, or stained with periodic acid-Schiff (PAS) to further assess colloid changes. 100 mg/kg/day), and PTU (0.025, 0.25, 1, or 10 mg/kg/day). Because of the subtle nature of some changes in the thyroid gland, for all In addition, a 15-day diet restriction experiment was performed using male studies this organ was evaluated without knowledge of the treatment status of rats to determine which thyroid endpoints were body weight-dependent in individual animals. In addition, to minimize potential staining differences with order to evaluate potential confounding from treatment-related decreases in the PAS stain, all slides were stained simultaneously. body weight or body weight gain. The male rat model was used since it is more Hormonal measurements. Blood was collected from all animals at the sensitive to thyroid perturbations than the female rat, due to the higher time of euthanization. Serum was prepared and stored between –65 and –85°C circulating concentrations of TSH (Capen, 1996). Therefore, when screening until analyzed for serum hormone concentrations. For males, serum T, E2, RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 57

DHT, LH, FSH, PRL, TSH, T3, and T4 concentrations were measured using In Vivo Time-Course Study commercially available RIA kits. For the females, serum LH, FSH, PRL, TSH, Study design. The study design consisted of 3 groups of 60 male rats. T , and T concentrations were measured using commercially available RIA 3 4 Male rats were chosen because they are generally more sensitive than female kits. rats to thyroid toxicants (Capen, 1996, 1997). Within each group of 60 rats, 45 Uterine progesterone (PR) and estrogen (ER) receptor concentrations were animals were designated for biochemical/hormonal/histopathological evalua- quantitated as previously described (O’Connor et al., 1996). tion and 15 rats were designated for cell proliferation evaluation. Rats were evaluated for thyroid effects at 3 time points: after 1 week, 2 weeks, and 4 15-Day Dietary Restriction Experiment weeks of dosing. At each time point, 20 rats from each treatment group were The study design and procedures for pathological evaluations and hormonal euthanized and evaluated for thyroid hormone concentrations, hepatic measurements for the dietary restriction experiment were the same as those UDP-GT activity, thyroid follicular cell proliferation, or thyroid gland histo- used in the in vivo male battery. In addition, microsomal preparations were pathology. The following dosages were used in the in vivo thyroid time-course prepared for the evaluation of UDP-GT and 5Ј-deiodinase activities. The experiment: 100 mg/kg/day PB and 1 mg/kg/day PTU. These dosages were procedure for determination of UDP-GT activity was as given below for the in selected based on the results of the Tier I battery (herein reported) to achieve vivo time course study. the maximal endocrine effect and not exceed an MTD after 4 weeks of Microsomal preparations. At necropsy for the dietary restriction experi- treatment. ment, a section of the livers from the rats designated for biochemical analyses Cell proliferation. At each time point, five rats from each group were (5/group) were removed and hepatic microsomes prepared for biochemical designated for thyroid follicular cell proliferation analyses. Four days prior to evaluation. A portion of the liver was homogenized (1 g tissue/8 ml buffer) in euthanization, rats were anesthetized using isoflurane and implanted (subcu- buffer containing 50 mM Tris–HCl, 50 mM Trizma-base, 0.25 M sucrose, and taneously) with Alzet osmotic pumps loaded with 20 mg/ml BrdU dissolved in 5.4 mM EDTA, pH 7.4. The homogenates were centrifuged at 15,000 ϫ g for 0.5 N sodium bicarbonate buffer. At each time point, rats were euthanized and 15 min at 4°C. The resulting supernatants were removed and centrifuged at the thyroid glands were placed into formalin fixative for at least 48 h, routinely 100,000 ϫ g for 70 min at 4°C; these pellets contained the microsomal processed, paraffin embedded, sectioned, and stained for immunohistochemi- fractions. The microsomal pellets were resuspended in the homogenization cal analysis of BrdU incorporation into DNA, or evaluated microscopically. buffer at a protein concentration of 10–20 mg/ml, aliquoted, and stored The duodenum was examined as a positive control to verify proper immuno- between –65°C and –85°C until analyzed for UDP-GT and 5Ј-deiodinase detection of BrdU. activities. The protein content of the microsomes was measured before and Pathological evaluations. At each time point, rats were injected with the after analyses by the Biorad method (Bradford, 1976). Final calculations were final dose of test compound approximately two h prior to sacrifice. Between based on the post-assay protein determination. 0700 and 1000 h, rats were anesthetized using CO2 and euthanized by exsan- 5؅-Deiodinase activity measurements. In the dietary restriction experi- guination. Blood was collected from the descending vena cava and serum ment, 5Ј-deiodinase activity was determined using modifications of the meth- prepared for hormonal analyses. The livers were removed and weighed. The ods of Pazos-Moura (1991) and Leonard and Rosenberg (1980). Briefly, 50 ␮l thyroid glands and surrounding tissue were removed and placed into formalin of reaction mixture [0.1 M potassium phosphate, 1 mM EDTA, 0.5 mM DTT, fixative for at least 48 h prior to trimming and weighing. Following fixation, 125 14.5 ␮l[ I]rT3 (ϳ4.9 ␮Ci, 0.05 nmol)] was preincubated at 37°C for final dissection was performed under a dissecting microscope by one individ- approximately 1 min. Fifty ␮l of microsomes, diluted with liver homogeniza- ual, in order to reduce the variability of the dissection procedure and hence, tion buffer (50 mM Tris–HCl, 50 mM Trizma-base, 0.25 M sucrose, and 5.4 reduce the variability of the thyroid weights. Organ weights were calculated mM EDTA, pH 7.4) to achieve a final protein concentration of 0.5 mg/ml, was relative to body weight. The formalin-fixed thyroid glands were examined added to the reaction mixture. The tubes were vortexed and incubated for 10 microscopically. min at 37°C. The reaction was stopped by the addition of 33 ␮l of ice-cold UDP-Glucuronyltransferase activity measurements. At each time point, BSA-PTU solution (4% BSA, 5 mM PTU) and 133 ␮l of 20% TCA, and the a section of the liver from the rats designated for biochemical analyses tubes were placed on ice until centrifugation. A time zero tube for each sample (5/group) was removed and hepatic microsomes prepared as described previ- was run concurrently by adding 50 ␮l of reaction mixture, 33 ␮l of ice-cold ously. UDP-GT activity was determined spectrophotometrically using a mod- BSA-PTU solution, and 133 ␮l of 20% TCA, and placing the tubes immedi- ification of the method of Bock and co-workers (1983). Briefly, 15 ␮lof ately on ice. A sample blank tube was prepared by adding 50 ␮l of liver p-nitrophenol (66.7 mM) was added to 460 ␮l of microsomes, which had been homogenization buffer, 33 ␮l of ice-cold BSA-PTU solution, and 133 ␮lof resuspended with assay buffer (66 mM Tris–HCl, 10 mM magnesium chloride, 20% TCA, and placing the tube immediately on ice. Total counts per tube were 0.05% Brij 58, pH 7.5) to achieve a final protein concentration of 0.5–1.0 determined by counting a 100 ␮l aliquot of the reaction mixture. All tubes were mg/ml. The tubes were preincubated at 37°C for 2 min prior to the addition of centrifuged for 3 min at 12,000 ϫ g, 4°C. Two-hundred microliters of the 25 ␮l UDPGA (200 mM) to start the reaction. The tubes were vortexed and resulting supernatant (75% of the reaction volume) was applied to a Poly-Prep incubated for 10 min at 37°C. A reaction blank tube for each sample was run column equilibrated with 10% acetic acid, and eluted with 1.7 ml of 10% acetic concurently by substituting the UDPGA with assay buffer. The reaction was acid. The resulting eluate was measured on a ␥-counter to determine 5Ј- stopped by the addition of 0.5 ml ice cold methanol and the tubes placed on ice deiodinase activity. The rate of 5Ј-deiodinase activity (nmoles [125I]rT deio- 3 until centrifugation. All tubes were centrifuged for 10 min at 2000 ϫ g at 4°C. dinated/h/mg protein) was calculated using the following formula: Three hundred microliters of the supernatant was combined with 2.7 ml 0.1 N sodium hydroxide and the absorbance at 405 nm was measured. The rate of 125 2͑CPM 10 min incubation Ϫ CPM time zero͒͑0.05 nmol ͓ I͔rT3͒ UDP-GT activity (nmol/min/mg protein) was calculated using the following ͑0.75͒͑CPM Ϫ total count͒͑mg protein͒͑incubation time of 0.167 hr͒ formula where the extinction coefficient (EC) is 18.1.

The formula was derived on the basis of the following assumptions: (1) that 75% of the final assay mixture was applied to the column, (2) that the protein ͑⌬ Abs405 blank – Abs405 over 10 min incubation) concentration used for the calculation was based on mg protein applied to the ϫ(dilution factor of 20)(1000 nmol) column (i.e., 37.5 ␮l of the 50 ␮l volume), and (3) that the specific gravity of (EC)(protein in mg/ml)(incubation time of 10 min)(1 ␮M) the iodide released was one-half that of the iodothyronines, because rT3 is randomly labeled with 125I in the equivalent 3Ј and 5Ј positions of the phenolic Hormonal measurements. Blood was collected from all animals at the ring. time of sacrifice. Serum was prepared and stored between –65°C and –85°C 58 O’CONNOR ET AL.

TABLE 1 In Vivo Female Battery: Effect of Phenobarbital and Propylthiouracil on Final Body Weight and Uterine Parameters

Estrus Stromal cell Dosage Final body (g) Liver (% body wt) Uterus (g) Uterine ﬂuida conversiona proliferation Epithelial cell heighta

Phenobarbital 0 223 Ϯ 3b 4.6 Ϯ 0.1 0.089 Ϯ 0.005 0/14 0/14 2.3 Ϯ 0.1 18.6 Ϯ 0.6 5 223 Ϯ 4 4.7 Ϯ 0.1 0.092 Ϯ 0.004 0/14 0/14 2.5 Ϯ 0.2 19.5 Ϯ 0.8 25 220 Ϯ 4 4.9 Ϯ 0.1* 0.088 Ϯ 0.004 0/14 0/14 4.1 Ϯ 0.5 20.2 Ϯ 0.9 50 219 Ϯ 4c 5.1 Ϯ 0.1*c 0.090 Ϯ 0.005c 0/13c 0/13c 3.5 Ϯ 0.7 19.8 Ϯ 0.5 100 212 Ϯ 4 5.5 Ϯ 0.1* 0.094 Ϯ 0.005 0/14 0/14 3.2 Ϯ 0.6 21.6 Ϯ 1.1** Propylthiouracil 0 222 Ϯ 4 4.6 Ϯ 0.1 0.076 Ϯ 0.003 0/14 0/14 1.4 Ϯ 0.3 18.9 Ϯ 0.9 0.025 224 Ϯ 4 4.6 Ϯ 0.1 0.091 Ϯ 0.004* 0/14 0/14 1.2 Ϯ 0.3 18.1 Ϯ 0.7 0.25 218 Ϯ 3 4.5 Ϯ 0.0 0.088 Ϯ 0.003* 0/14 0/14 2.4 Ϯ 0.2 21.5 Ϯ 1.4 1 221 Ϯ 3 4.6 Ϯ 0.1 0.079 Ϯ 0.003 0/14 0/14 1.5 Ϯ 0.4 19.4 Ϯ 0.5 10 223 Ϯ 5 4.7 Ϯ 0.1 0.082 Ϯ 0.004 0/14 0/14 1.3 Ϯ 0.3 20.0 Ϯ 0.8

Note. Statistical methods: One-way analysis of variance (ANOVA) and Dunnett’s tests were performed on body weight and organ weight data. Fisher’s Exact test was performed on the uterine fluid and estrus conversion data. Jonckheere’s trend test was performed on the uterine stromal cell proliferation and epithelial cell height data. Dosage, mg/kg/day. Stromal cell proliferation given as % labeled cells; epithelial cell height, ␮m. a Incidence. b Mean Ϯ standard error. c One animal excluded due to the presence of ovarian tissue. * Significantly different (p Ͻ 0.05) from control by Dunnett’s Test. ** Significantly different (p Ͻ 0.05) from control by Jonckheere’s test for trend.

until analyzed for serum hormone concentrations. Serum TSH, T3,T4, and rT3 Statistical Analyses concentrations were measured by commercially available RIA kits. Mean final body weights and organ weights were analyzed by a one-way analysis of variance (ANOVA). When the corresponding F test for differences In Vitro Yeast Studies among test group means was significant, pairwise comparisons between test and control groups were made with Dunnett’s test. Bartlett’s test for homo- Binding of PB and PTU to the human estrogen (ER), androgen (AR), and geneity of variances was performed and, when significant (p Ͻ 0.005), was progesterone (PR) receptor was examined using a yeast reporter gene system followed by nonparametric procedures (Dunn’s test). Serum hormone concen- as previously described (O’Connor et al., 1998b). Receptor-specific com- trations, progesterone and estrogen receptor concentrations, cell proliferation pounds were previously used to validate this system (O’Connor et al., 1998b); indices, uterine morphometry measurements, and UDP-GT and 5Ј-deiodinase these included E2, DHT, and progesterone as the primary ligand for the three activities were analyzed using Jonckheere’s test for trend. If a significant receptor systems, respectively. Briefly, yeast (Saccharomyces cerevisiae) dose-response trend was detected, data from the top-dose group were excluded transformants from single plated colonies were grown overnight at 30°C with and the test repeated until no significant trend was detected. Uterine fluid orbital shaking at 300 rpm in a selective medium containing yeast nitrogen imbibition and estrus conversion data were analyzed by Fisher’s Exact test. base without amino acids plus ammonium sulfate and dextrose. Except for Bartlett’s test, all other significance was judged at p Ͻ 0.05.

Assays were started when the cells reached an OD600nm greater than 1.00.

One hundred microliters of cell suspension, diluted to achieve an OD600nm of 1.00, was placed into each well of a 96-well plate and 100 ␮l selective medium was added. Two ␮l of each test compound, prepared in methanol, was then RESULTS added to each well to develop a dose-response assay in triplicate. These cultures were incubated for3hat30°C with rocking. In Vivo Female Battery

After incubation, cells were read at OD600nm to evaluate the effect of the compounds on cell growth before adding 100 ␮l of assay buffer (60 mM Final body and organ weights, uterine fluid imbibition, and Na2HPO4,40mMNaH2PO4, 10 mM KCl, 1 mM MgSO4, 2 mg/ml ONPG, 50 estrus conversion (Table 1). Body weights were unchanged mM 2-mercaptoethanol, 0.1% SDS, and 200 U/␮l oxalyticase) to each well to in rats treated with PB or PTU. Relative liver weights were assay for ␤-galactosidase induction. The change in concentration of o-nitrophenol, the yellow product that results from ␤-galactosidase activity cleaving significantly increased in rats treated with PB, with the greatest the galactopyranoside, was determined by reading each well at OD410nm after a increase at the highest dosage (120% of control), and were period of one h. Non-specific colorimetric distortions (i.e., debris) were eval- unchanged in rats treated with PTU. Absolute uterine weight uated by also reading each well at OD570nm. (PB only), uterine fluid imbibition, and estrus conversion were For competitive inhibition testing of a compound, approximately one-half unchanged in rats treated with PB or PTU. PTU treatment maximal concentration of E2 for the YTS containing the ER, or DHT for the YTS containing the AR was included in each well of the dose-response for caused a slight, but statistically significant, increase in absolute each test compound. The samples were then treated as above except that the uterine weight at 0.025 and 0.25 mg/kg/day (120 and 116% of percent inhibition was calculated for each test compound. control, respectively). RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 59

TABLE 2 In Vivo Female Battery: Effect of Phenobarbital and Propylthiouracil on Reproductive Hormone Concentrations

Dosage (mg/kg/day) Prolactin (ng/ml) Follicle stimulating hormone (ng/ml) Luteinizing hormone (ng/ml)

Phenobarbital 0 7.3 Ϯ 0.8a 41.7 Ϯ 1.7 29.9 Ϯ 2.4 5 5.2 Ϯ 0.7 37.8 Ϯ 1.7 27.5 Ϯ 1.9 25 5.4 Ϯ 1.1 39.2 Ϯ 1.3 26.5 Ϯ 2.6 50 5.2 Ϯ 0.9 43.5 Ϯ 1.8 25.5 Ϯ 1.6 100 10.1 Ϯ 2.2 52.3 Ϯ 1.8* 24.5 Ϯ 2.2 Propylthiouracil 0 6.7 Ϯ 1.0 41.8 Ϯ 1.1 27.8 Ϯ 1.8 0.025 10.6 Ϯ 1.6 47.2 Ϯ 2.6 27.4 Ϯ 1.8 0.25 10.7 Ϯ 1.3 44.5 Ϯ 1.8 29.2 Ϯ 1.9 1 9.0 Ϯ 1.3 43.1 Ϯ 1.7 25.1 Ϯ 2.3 10 13.1 Ϯ 2.0* 49.6 Ϯ 2.6 26.7 Ϯ 1.6

a Mean Ϯ standard error. * Signiﬁcantly different (p Ͻ 0.05) from control by Jonckheere’s test for trend.

Uterine stromal cell proliferation and epithelial cell height Histopathology. The thyroid glands were examined micro- (Table 1). A statistically significant increase in uterine epi- scopically to identify compounds that target the thyroid gland. thelial cell height was observed at the highest dosage in rats No microscopic changes were observed in the thyroid glands treated with PB (116% of control), and was unchanged in rats from females treated with PB for 5 days. In contrast, PTU treated with PTU. Uterine stromal cell proliferation was un- administration was associated with thyroid gland hypertrophy changed in rats treated with either PB or PTU. and hyperplasia and depletion of colloid at the highest does Uterine receptor content (data not shown). Uterine cyto- tested. solic ER and PR concentrations were analyzed using a single saturating concentration of radio-labeled ligand. PB caused a In Vivo Male Battery statistically significant decrease in ER concentrations, with the Final body and organ weights (Table 4). Mean final body greatest decrease at 50 mg/kg/day (32% of control), but did not weights were significantly decreased at the highest dosage in affect PR concentrations. Uterine ER and PR concentrations males treated with PTU (92% of control), and were unchanged were unaffected by PTU treatment at dosages as high as 10 in rats treated with PB. Relative liver weights were signifi- mg/kg/day. cantly increased in males treated with PB, with the greatest Reproductive hormone concentrations (Table 2). Serum was collected for hormonal analyses approximately 24 h after the last administered dose. A statistically significant increase in TABLE 3 serum FSH concentrations was observed at the highest dosage In Vivo Female Battery: Effect of Phenobarbital and in rats treated with PB (125% of control). Serum LH and PRL Propylthiouracil on Thyroid Hormone Concentrations concentrations were not affected by treatment with PB. A statistically significant increase in serum PRL concentrations Dosage (mg/kg/day) TSH (ng/ml) T3 (ng/dl) T4 (␮g/dl) was observed at the highest dosage in rats treated with PTU (196% of control). Serum LH and FSH concentrations were Phenobarbital 0 13.0 Ϯ 0.4a 79.1 Ϯ 3.5 3.0 Ϯ 0.1 unaffected by PTU treatment at dosages as high as 10 mg/kg/ 5 12.8 Ϯ 0.4 70.7 Ϯ 2.0 3.0 Ϯ 0.2 day. 25 14.0 Ϯ 0.6 66.5 Ϯ 2.9* 2.2 Ϯ 0.1* Thyroid hormone concentrations (Table 3). Serum was 50 13.5 Ϯ 0.6 64.6 Ϯ 2.7* 1.7 Ϯ 0.1* collected for hormonal analyses approximately 24 h after the 100 14.5 Ϯ 1.1 58.9 Ϯ 2.7* 1.4 Ϯ 0.2* Propylthiouracil last administered dose. PTU caused a statistically significant 0 11.6 Ϯ 0.5 70.2 Ϯ 3.4 3.1 Ϯ 0.2 increase in serum TSH concentrations at the highest dosage 0.025 10.8 Ϯ 0.7 64.2 Ϯ 3.2 2.9 Ϯ 0.2 (205% of control). A statistically significant decrease in serum 0.25 10.2 Ϯ 0.4 62.7 Ϯ 3.6 2.7 Ϯ 0.2 1 12.3 Ϯ 0.9 60.6 Ϯ 2.8 2.8 Ϯ 0.3 T3 and T4 concentrations was observed in rats treated with PB and PTU, with the greatest decrease at the highest dosages 10 23.8 Ϯ 0.9* 50.5 Ϯ 2.8* 2.0 Ϯ 0.1* (74% and 72% of control, and 47% and 65% of control, for T3 a Mean Ϯ standard error. and T4 respectively). Serum TSH concentrations were not * Significantly different (p Ͻ 0.05) from control by Jonckheere’s test for affected by treatment with PB. trend. 60 O’CONNOR ET AL.

TABLE 4 In Vivo Male Battery: Effect of Phenobarbital and Propylthiouracil on Final Body Weights and Organ Weights

% Body wt Final body wt Seminal Dosage (g) (% of control)Liver (% body wt) Testes (g) Epididymides (g) ASG unit vesicles Prostate

Phenobarbital 0 413 Ϯ 6a 100 3.9 Ϯ 0.1 3.3 Ϯ 0.0 1.22 Ϯ 0.03 0.596 Ϯ 0.023 0.393 Ϯ 0.016 0.189 Ϯ 0.010 5 415 Ϯ 7 100 3.9 Ϯ 0.1 3.2 Ϯ 0.1 1.23 Ϯ 0.02 0.584 Ϯ 0.019 0.365 Ϯ 0.024 0.203 Ϯ 0.009 25 415 Ϯ 5 100 4.3 Ϯ 0.1* 3.0 Ϯ 0.1 1.17 Ϯ 0.04 0.609 Ϯ 0.018 0.397 Ϯ 0.021 0.203 Ϯ 0.011 50 416 Ϯ 6 101 4.6 Ϯ 0.1* 3.2 Ϯ 0.1 1.26 Ϯ 0.02 0.602 Ϯ 0.026 0.386 Ϯ 0.027 0.199 Ϯ 0.009 100 399 Ϯ 7 97 4.9 Ϯ 0.1* 3.2 Ϯ 0.1 1.25 Ϯ 0.03 0.613 Ϯ 0.023 0.398 Ϯ 0.028 0.199 Ϯ 0.012 Propylthiouracil 0 421 Ϯ 7 100 4.0 Ϯ 0.1 3.2 Ϯ 0.0 1.21 Ϯ 0.02 0.556 Ϯ 0.020 0.369 Ϯ 0.020 0.173 Ϯ 0.010 0.025 418 Ϯ 8 99 4.0 Ϯ 0.1 3.3 Ϯ 0.1 1.17 Ϯ 0.03 0.568 Ϯ 0.013 0.383 Ϯ 0.013 0.178 Ϯ 0.010 0.25 416 Ϯ 6 99 4.0 Ϯ 0.1 3.2 Ϯ 0.1 1.19 Ϯ 0.03 0.553 Ϯ 0.018 0.357 Ϯ 0.020 0.184 Ϯ 0.009 1 404 Ϯ 4 96 3.9 Ϯ 0.1 3.1 Ϯ 0.1 1.17 Ϯ 0.04 0.598 Ϯ 0.020 0.409 Ϯ 0.021 0.177 Ϯ 0.007 10 387 Ϯ 6* 92 3.5 Ϯ 0.1* 3.2 Ϯ 0.1 1.19 Ϯ 0.03 0.604 Ϯ 0.021 0.414 Ϯ 0.020 0.174 Ϯ 0.010

Note. Dosage in mg/kg/day. ASG, accessory sex gland. a Mean Ϯ standard error. * Significantly different (p Ͻ 0.05) from control by Dunnett’s Test. increase at the highest dosage (126% of control). They were in mitotic figures were present in most animals in the 1.0 and significantly decreased in rats treated with PTU, with the 10 mg/kg/day groups. At the 0.025 mg/kg-dose level, equivo- greatest decrease at the highest dosage (88% of control). Tes- cal depletion of colloid was present in 3/14 animals. tes, epididymis, ASG, and the individual component weights of Reproductive hormone concentrations (Table 5). PB ad- the ASG (prostate and seminal vesicles) were unaffected by PB ministration caused a statistically significant increase in serum or PTU treatment at the dosages evaluated. E2 concentrations with the greatest increase at the highest Histopathology. There were no compound-related gross dosage (137% of control). Serum PRL, FSH, and LH concen- lesions and microscopic changes observed in any of the repro- trations were significantly decreased in a dose-dependent man- ductive organs of the PB- or PTU-treated rats. In all PB-treated ner, with the greatest decreases at the highest dosage (42, 80, groups, the primary microscopic change of the thyroid gland and 65% of control, respectively). Although not statistically was pale staining and/or depletion of colloid (Fig. 4). Colloid significant, serum T concentrations were numerically de- changes were somewhat more apparent with PAS staining, as creased at all dosages with the greatest decrease at 50 mg/kg/ the staining intensity was significantly less in affected thyroids day (57% of control). Serum DHT concentrations were not compared to controls (Fig. 4). Colloid depletion was often affected by PB treatment at dosages as high as 100 mg/kg/day. associated with increased numbers of small diameter follicles. PTU administration caused a statistically significant de- Depletion of colloid was generally minimal to mild across crease in serum T concentrations with the greatest decrease at PB-treated groups. Interestingly, colloid depletion was slightly the highest dosage (58% of control). Serum PRL concentra- more severe in some animals in the 50 mg/kg group than in tions were significantly increased in the 1 and 10 mg/kg/day other treated groups, including the 100 mg/kg group. Although groups, with the greatest increase at 1 mg/kg/day (151% of equivocal evidence of microscopic hypertrophy of follicular control). Serum FSH concentrations were significantly in- cells was present in some PB-treated male rats, the variability creased with the greatest increase at the highest dosage (125% in follicular cell size in control male rats precluded a definitive of control). Serum E2, DHT, and LH concentrations were not diagnosis of hypertrophy. affected by PTU treatment at dosages as high as 10 mg/kg/day. Grossly, 3/15 rats from the 10 mg/kg/day PTU treatment group had enlarged thyroid glands. Microscopically, diffuse Thyroid hormone concentrations (Table 6). Thyroid hor- hypertrophy and hyperplasia of thyroid follicular cells were mone analyses were performed in order to evaluate the Tier I present in groups dosed with 0.25 mg/kg/day and above. In male battery for its ability to detect compounds that alter affected thyroid glands, follicular epithelial cells were cuboidal thyroid hormone homeostasis. Serum TSH, T3, and T4 concen- to short columnar with enlarged nuclei, which crowded, and in trations were measured from serum prepared 2 h after the last some follicles, piled up along the basement membrane zone. administered dose. PB and PTU caused statistically significant Follicular lumens were decreased in diameter and were devoid increases in serum TSH concentrations, with the greatest in- of colloid or contained faintly-staining colloid. Slight increases crease at 50 mg/kg/day and 10 mg/kg/day, respectively (151% RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 61

TABLE 5 In Vivo Male Battery: Effect of Phenobarbital and Propylthiouracil on Reproductive Hormone Concentrations

Testosterone Estradiol Dihydrotestosterone Prolactin FSH Luteinizing hormone Dosage (ng/ml) (pg/ml) (pg/ml) (ng/ml) (ng/ml) (ng/ml)

Phenobarbital 0 3.0 Ϯ 0.6a 5.1 Ϯ 0.7 104.6 Ϯ 20.2 12.1 Ϯ 1.5 16.1 Ϯ 0.9 4.9 Ϯ 0.4 5 1.9 Ϯ 0.3 4.8 Ϯ 0.7 149.4 Ϯ 19.4 10.9 Ϯ 1.6 14.2 Ϯ 0.7 4.3 Ϯ 0.3 25 2.1 Ϯ 0.4 4.7 Ϯ 0.6 147.8 Ϯ 24.8 10.4 Ϯ 1.0 13.9 Ϯ 0.6* 3.8 Ϯ 0.3 50 1.7 Ϯ 0.3 6.5 Ϯ 0.6 89.5 Ϯ 17.0 8.7 Ϯ 1.1 13.4 Ϯ 0.5* 3.6 Ϯ 0.3* 100 1.9 Ϯ 0.4 7.0 Ϯ 0.7* 91.9 Ϯ 14.0 5.1 Ϯ 0.7* 12.8 Ϯ 0.8* 3.2 Ϯ 0.3* Propylthiouracil 0 3.3 Ϯ 0.5 10.8 Ϯ 1.2 191.4 Ϯ 27.6 12.8 Ϯ 1.9 16.1 Ϯ 0.7 4.5 Ϯ 0.3 0.025 2.7 Ϯ 0.5 9.3 Ϯ 1.1 190.5 Ϯ 33.7 14.4 Ϯ 1.8 14.6 Ϯ 0.6 4.1 Ϯ 0.2 0.25 2.2 Ϯ 0.3 8.3 Ϯ 1.2 142.0 Ϯ 19.6 12.7 Ϯ 1.7 16.6 Ϯ 1.4 4.5 Ϯ 0.3 1 2.4 Ϯ 0.7* 9.5 Ϯ 1.3 153.0 Ϯ 32.5 19.3 Ϯ 2.3* 17.4 Ϯ 0.8 4.6 Ϯ 0.2 10 1.9 Ϯ 0.5* 8.6 Ϯ 1.0 152.4 Ϯ 27.0 17.5 Ϯ 1.5* 20.1 Ϯ 1.0* 4.7 Ϯ 0.3

Note. Dosage in mg/kg/day. FSH, follicle stimulating hormone. a Mean Ϯ standard error. * Significantly different (p Ͻ 0.05) from control by Jonckheere’s test for trend. and 312% of control, respectively). PB and PTU caused sta- of the ad libitum control at 22, 19, 16, and 13 grams of tistically significant decreases in serum T3 (70% and 8% of feed/day, respectively. Relative liver weights were decreased control, respectively) and T4 concentrations (32% and 0% of in a dose-dependent manner, and were significantly decreased control, respectively) with the greatest decreases at the highest in all treatment groups. The greatest decrease (82% of control) dosages. occurred at a dietary restriction level that achieved a final body weight of 74% of the ad libitum control. Absolute thyroid Dietary Restriction Experiment (Table 7) weight was significantly decreased in all treatment groups, Daily food consumption levels for the dietary restriction with the same magnitude change at dietary restriction level that experiment were targeted to achieve decreases in body weight achieved a final body weight of 85% of the ad libitum control to 95, 90, 85, and 80% of the ad libitum control. Final body (76% of control). Relative thyroid weights were unaffected by weights were significantly decreased at all levels of dietary dietary restriction, suggesting that relative thyroid weights are restriction, achieving final body weights of 90, 85, 79, and 74% a more appropriate endpoint than absolute thyroid weights for detecting compound-related effects on thyroid weight when body-weight effects are observed. Serum concentrations of TABLE 6 TSH, T3, and T4 were decreased by dietary restriction. Serum In vivo Male Battery: Effect of Phenobarbital and TSH concentrations were numerically decreased (considered Propylthiouracil on Thyroid Hormone Concentrations biologically significant) when final body weights were 79% of the ad libitum control or less, and were significantly decreased Dosage (mg/kg/day) TSH (ng/ml) T3 (ng/dl) T4 (␮g/dl) (62% of control) when final body weights were 74% of the ad

Phenobarbital libitum control. Serum T3 and T4 concentrations were signifi- a 0 18.2 Ϯ 1.2 74.1 Ϯ 3.4 3.8 Ϯ 0.2 cantly decreased when final body weights were 85% of the ad 5 19.7 Ϯ 1.4 70.2 Ϯ 4.1 3.4 Ϯ 0.2 libitum control or less. The greatest decrease at dietary restric- 25 19.7 Ϯ 2.0* 65.4 Ϯ 3.2 2.8 Ϯ 0.1* 50 27.5 Ϯ 2.2* 67.5 Ϯ 3.2 2.3 Ϯ 0.1* tion level achieved a final body weight of 74% of the ad libitum 100 24.9 Ϯ 1.8* 52.2 Ϯ 2.9* 1.2 Ϯ 0.1* control (75 and 72% of control, respectively). UDP-GT activity Propylthiouracil was significantly decreased when final body weights were 79% 0 14.6 Ϯ 0.9 62.1 Ϯ 2.7 2.7 Ϯ 0.2 of the ad libitum control or less. The greatest decrease at 0.025 16.1 Ϯ 1.7 49.1 Ϯ 2.2* 3.0 Ϯ 0.2 0.25 20.7 Ϯ 2.0* 43.7 Ϯ 2.5* 1.1 Ϯ 0.1* dietary restriction level achieved a final body weight of 79% of 1 41.0 Ϯ 2.4* 9.1 Ϯ 1.6* 0.0 Ϯ 0.0* the ad libitum control (85% of control). The decrease in 10 45.5 Ϯ 2.9* 5.0 Ϯ 0.9* 0.0 Ϯ 0.0* UDP-GT activity correlated with the decreases in relative liver

weight. Serum rT3 concentrations and hepatic 5Ј-deiodinase Note. TSH, thyroid stimulating hormone. a activity were unaffected by dietary restriction at the levels Mean Ϯ standard error. * Signiﬁcantly different (p Ͻ 0.05) from control by Jonckheere’s test for examined. There were no microscopic changes of the thyroid trend. gland observed at any of the dietary restriction levels. 62

TABLE 7 In Vivo Male Battery: Effect of 15 Days of Dietary Restriction on Thyroid Endpoints in Male Rats

Final body wt Thyroid Glucuronyl- 5Ј-Deiodinase

Feed/day (% of Liver Reverse T3 transferase activity activity

(g) (g) control)(% body w) (g) (% body w) TSH (ng/ml) T3 (ng/dl) T4 (␮g/dl) (ng/ml) (nmol/mg/min) (nmol/mg/hr) AL. ET O’CONNOR ad libituma 414 Ϯ 6b,c 100 3.9 Ϯ 0.1 0.025 Ϯ 0.001 0.006 Ϯ 0.0003 17.3 Ϯ 1.3 80.7 Ϯ 4.0 4.3 Ϯ 0.2 0.073 Ϯ 0.004 34.4 Ϯ 1.1 8.9 Ϯ 0.6 22 373 Ϯ 4* 90 3.6 Ϯ 0.0* 0.021 Ϯ 0.001* 0.006 Ϯ 0.0003 17.0 Ϯ 1.8 79.9 Ϯ 3.6 4.0 Ϯ 0.2 0.091 Ϯ 0.009 32.9 Ϯ 0.5 7.0 Ϯ 0.9 19 351 Ϯ 3* 85 3.4 Ϯ 0.1* 0.019 Ϯ 0.001* 0.006 Ϯ 0.0003 16.7 Ϯ 1.5 68.1 Ϯ 3.7* 3.6 Ϯ 0.2* 0.086 Ϯ 0.009 32.2 Ϯ 1.5 8.9 Ϯ 0.4 16 328 Ϯ 3* 79 3.2 Ϯ 0.0* 0.019 Ϯ 0.001* 0.005 Ϯ 0.0003 14.1 Ϯ 1.1 70.5 Ϯ 3.7* 3.2 Ϯ 0.2* 0.071 Ϯ 0.012 29.2 Ϯ 1.9** 8.1 Ϯ 0.2 13 307 Ϯ 2* 74 3.2 Ϯ 0.0* 0.019 Ϯ 0.001* 0.006 Ϯ 0.0003 10.8 Ϯ 1.5** 60.8 Ϯ 2.8* 3.1 Ϯ 0.2* 0.074 Ϯ 0.009 30.4 Ϯ 2.4** 8.0 Ϯ 0.2

Note. Statistical Methods: One-Way Analysis of Variance and Dunnett’s tests were performed on body weight and organ weight data and Jonckheere’s trend test was performed on the hormonal, UDP-glucuronyltransferase, and 5Ј-deiodinase data. a Ad libitum rats consumed 25.8 g of feed/day. b Mean Ϯ standard error. c The body weights at the start of the experiment were 325.5, 325.7, 321.3, 322.3, and 326.3 g for the ad libitum control, 22, 19, 16, and 13 g/day treatment groups, respectively. * Signiﬁcantly different (p Ͻ 0.05) from control by Dunnett’s test. ** Signiﬁcantly different (p Ͻ 0.05) from control by Jonckheere’s test for trend. RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 63

In Vivo Time-Course Study

Final body and organ weights (Fig. 1). No statistically significant changes in mean final body weights were observed at the 1-week or 2-week time points for the PB- or PTU treated male rats. At the 4-week time point, the mean final body weights of the PB- and PTU-treated rats were significantly decreased to 92 and 87% of control, respectively. Relative liver weights were significantly increased in rats treated with PB at the 1-, 2-, and 4-week time points (129, 128, and 143% of control, respectively). Liver weights were significantly decreased in rats treated with PTU at the 4-week time point (93% of control). Relative thyroid weights were significantly increased at the 2-week and 4-week time points for the PB- treated males (140 and 117% of control, respectively). Thyroid weights were significantly increased at all 3 time points for the PTU-treated rats (133, 240, and 250% of control, respectively). Thyroid hormone concentrations (Fig. 2). In PB-treated male rats, serum TSH concentrations were significantly increased at the 4-week time point (197% of control), and unchanged at the 1-week and 2-week time points. Serum T3 concentrations were significantly decreased at the 1-week and 2-week time points (68 and 81% of control, respectively), and were unchanged at the 4-week time point. Serum T4 concentrations were significantly decreased at all time points (48, 56, and 59% of control at the 1-week, 2-week and 4-week time points, respectively). Serum rT3 concentrations were unaffected by PB treatment at all three time points evaluated. In PTU-treated male rats, serum TSH concentrations were significantly increased at all 3 time points (219, 267, and 426% of control at the 1-week, 2-week and 4-week time points, respectively). At all 3 time points, serum T3 concentrations were significantly decreased (29, 25, and 48% of control at the FIG. 1. The effect of phenobarbital (PB) and propylthiouracil (PTU) 1-week, 2-week and 4-week time points, respectively). Simi- treatment on body weight (A), relative liver weight (B), and relative thyroid weight (C). Groups of 15 male rats were dosed via intraperitoneal injection larly, serum T4 concentrations were significantly decreased at all time points (11, 0, and 14% of control at the 1-week, with 100 mg/kg/day PB or 1 mg/kg/day PTU daily. Body and organ weights were measured after 1 week, 2 weeks, or 4 weeks of treatment. Data are 2-week and 4-week time points, respectively). Serum rT3 con- expressed as mean Ϯ standard error. The numbers within the columns repre- centrations were significantly increased at the 4-week time sent the percentage of the control value. Asterisks denote statistical signifi- point (238% of control), and unchanged at the 1-week or cance from the control group (Dunnett’s test; p Ͻ 0.05). 2-week time points.

UDP-glucuronyltransferase activity (Fig. 3a). In PB- Histopathology. There were no compound-related gross treated male rats, hepatic UDP-GT activity was signiﬁcantly lesions observed in the thyroid glands of the PB-treated ani- increased at all time points (281, 252, and 307% of control at mals. Microscopically, colloid depletion, similar to that noted the 1-, 2-, and 4-week time points, respectively). UDP-GT in the in vivo male battery, was present in most PB-treated activity was unaffected by PTU treatment at all 3 time points animals at the 2- and 4-week time points. Slight colloid deple- evaluated. tion was present in 2/15 PB-treated male rats at the 1-week Thyroid follicular cell proliferation (Fig. 3b). In PTU- time point, but a similar change was also noted in 2 controls treated male rats, thyroid follicular cell proliferation was sig- (one control rat each at the 1- and 2-week time points). niﬁcantly increased at all time points (822, 264, and 209% of For the PTU-treated rats, 4/15 rats had grossly enlarged control at the 1-, 2-, and 4-week time points, respectively). In thyroid glands at the 4-week time point. Microscopic changes PB-treated rats, thyroid follicular cell proliferation was signif- similar to those observed in the in vivo male battery (hyper- icantly increased at the 2-week time point (247% of control) trophy/hyperplasia of follicular cells and colloid depletion) and was unchanged at the 1- and 4-week time points. were present at all time points in PTU-treated male rats. The 64 O’CONNOR ET AL.

FIG. 2. The effect of phenobarbital (PB) and propylthiouracil (PTU) treatment on serum hormone levels. Groups of 15 male rats were dosed via intraperitoneal injection with 100 mg/kg/day PB or 1 mg/kg/day PTU daily. Serum concentrations of TSH (A), T3 (B), T4 (C), or rT3 (D) were measured after 1, 2, or 4 weeks of treatment. Data are expressed as mean Ϯ standard error. The numbers within the columns represent the percentage of the control value. Asterisks denote statistical signiﬁcance from the control group (Jonckheere’s test; p Ͻ 0.05). severity of microscopic changes in the thyroid gland were calculated as a percent of the maximal response. No com- slightly greater at the 2- and 4-week time points compared to pound-related effects on growth were detected in our 3-h assay those observed at the 1-week time point. with any of the test compounds. Neither PB nor PTU had activity in the ER, AR, or PR when evaluated in the direct YTS In Vitro YTS or the YTS competition assays (data not shown). In the YTS, yeast containing constructs that expressed one of the human ER, AR, or PR genes were used to evaluate PB DISCUSSION and PTU for their ability to bind to the receptor and to activate a response element driving an inducible reporter (␤-galactosi- In this report, we have summarized the results of our ongo- dase). During the testing of each compound, a positive control ing Tier I screening battery for 2 well-characterized thyroid (E2 for ER, DHT for AR, and progesterone for PR) was toxicants, PB and PTU. We have previously described the included in each experiment. These positive controls served as results for 6 compounds with distinct endocrine activities using the reference points for each day’s maximum response when this Tier I battery (O’Connor et al., 1998a,b), which brings the comparing test substances across assays, and all results were total number of compounds evaluated to 8. The 2 main goals of RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 65

creted by the pituitary, is particularly important in this feedback mechanism. It induces the thyroid to synthesize thyroid

hormone T4, which is then 5Ј-mono-deiodinated to the more

biologically active T3 (Capen, 1997; Davies, 1993), or by inner

ring deiodination to form rT3, which has no biological function (DeGroot, 1995). The rate of TSH release is controlled by the amount of thyrotropin-releasing hormone (TRH) released by the hypothalamus, as well as by the circulating concentrations

of T3 and T4 (Capen, 1997; DeGroot, 1995). Generally, reduc-

tions in circulating T3 and T4 concentrations will trigger the pituitary to secrete TSH, which in turn results in increased

synthesis of T3 and T4 by the thyroid gland (Capen, 1996, 1997; DeGroot, 1995). The thyroid hormones are eliminated from the body primarily by conjugation reactions in the liver.

T4 is conjugated with glucuronic acid in a reaction catalyzed by thyroxine-UDP-GT (Barter and Klaassen, 1992; Capen, 1996).

T3 is conjugated with sulfate in a reaction catalyzed by phenol sulfotransferase (Capen, 1996). The conjugated products are then excreted in the bile (Hill et al., 1989). Perturbations of thyroid hormone homeostasis can occur through several mechanisms including direct action on the thyroid gland through inhibition of synthesis (e.g., thio- namides, aniline derivatives, substituted phenols), or release of thyroid hormones (e.g., excess iodine, lithium) (Capen, 1996, 1997). However, a wide variety of chemicals and drugs such as PB, spironolactone, chlorinated hydrocarbons, calcium channel blockers, and polychlorinated biphenyls have been shown to FIG. 3. The effect of phenobarbital (PB) and propylthiouracil (PTU) induce hepatic microsomal enzymes, or inhibit 5Ј-deiodinase treatment on hepatic UDP-glucuronyltransferase activity (A) and thyroid fol- (e.g., erythrosine), both of which result in a reduction in the licular cell proliferation (B). Groups of 5 male rats were dosed via intraperi- circulating concentrations of the thyroid hormones (Capen, toneal injection with 100 mg/kg/day PB or 1 mg/kg/day PTU daily. Hepatic microsomes were prepared and analyzed for UDP-glucuronyltransferase ac- 1996, 1997; Masaki et al., 1984; McClain et al., 1989). In tivity after 1, 2, or 4 weeks of treatment. Thyroid follicular cell proliferation general, these agents are less potent in altering thyroid econ- was measured in rats implanted with osmotic minipumps loaded with 20 omy than agents that directly target the thyroid Therefore, a mg/ml BrdU (approximately 4-days of exposure). Data are expressed as comprehensive screen for detecting thyroid toxicants should be mean Ϯ standard error. The numbers within the columns represent the per- able to detect a wide variety of mechanisms of thyroid hor- centage of the control value. Asterisks denote statistical signiﬁcance from the control group (Jonckheere’s test; p Ͻ 0.05). mone disruption.

Detection of PB and PTU in the Tier I Battery this validation are to test the hypothesis that distinct “fingerprints” of endocrine activities can be identified for each EAC, For the proposed Tier I screening battery, the in vivo male and to determine which endpoints should be included in a final battery was the model used for identifying thyroid toxicants Tier I-type screen. In order to accomplish these goals, the due to the greater sensitivity of male rats to thyroid effects (the responses observed for each of the positive controls were in vivo female battery and the in vitro YTS are not designed to compared to the expected responses based on the published detect thyroid toxicants). literature. The effect of diet restriction for 15 days on several PB, a thyroid hormone excretion enhancer, was used to thyroid endpoints in male rats has also been examined to help characterize the ability of the proposed Tier I battery to detect identify possible confounding of results due to decreased body thyroid toxicants that act via enhanced metabolism of thyroid weight. Finally, we have examined several thyroid endpoints hormones. PTU, a thyroid hormone synthesis inhibitor, was for their utility for detecting thyroid modulating effects after 1, used to characterize the ability of the proposed Tier I male 2, or 4 weeks of treatment using PB or PTU. battery to directly identify (i.e., through disruption of thyroid

Homeostasis of thyroid hormone (T3 and T4) synthesis and hormone synthesis or release) thyroid toxicants. As expected, secretion is controlled by a sensitive feedback mechanism, PB administration produced increases in relative liver weight, which involves the hypothalamus, the pituitary, and the thyroid as well as changes in thyroid hormones (increased TSH and gland (reviewed in Capen, 1997; DeGroot, 1995). TSH, se- decreased serum T3 and T4). Deﬁnitive microscopic changes 66 O’CONNOR ET AL.

FIG. 4. Thyroid glands from a control rat stained with hematoxylin and eosin (H and E) (a) or periodic acid-Schiff (PAS) (c), and a rat administered 5 mg/kg/day phenobarbital (PB) for 14 days and stained with H and E (b) or PAS (d). With H and E staining, the colloid of the PB-treated rat is slightly more pallid than the control. Differences in colloid staining are more apparent with the PAS stain, as the intensity of PAS staining is markedly decreased in the PB-treated rat when compared to the control. Bar ϭ 50 ␮m. RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 67 were limited to effects on colloid. The increase in TSH is a In Vivo Time Course Study secondary response following PB administration, and is due to One major factor in developing a screen for detecting thy- the enhanced clearance of serum T3 and T4 via increased biliary excretion as a result of enhanced UDP-GT activity roid toxicants is determining the optimum exposure duration. (McClain et al., 1989). The effects of PTU on thyroid param- In order for a screen to be useful it must be comprehensive, but eters was as expected from previous reports (Capen, 1996, it must also be cost-effective and of short duration. Hence, a 1997) and included a 3-fold increase in serum TSH coupled screen should be no more than 4-weeks in duration. To define which endpoints of thyroid toxicity are useful in a screen, and with decreases in T3 and T4 and thyroid gland hypertrophy/ hyperplasia. to help define the optimal duration of exposure for detecting Overall, the endpoint profiles obtained for PB and PTU were thyroid-modulating effects, we evaluated serum hormone con- characteristic for their mechanisms of action. The hormonal centrations (TSH, T3,T4, and rT3), hepatic UDP-GT activity, pattern in the male battery was consistent with known thyroid liver and thyroid weights, thyroid follicular cell proliferation, and histopathology of the thyroid gland, following 1, 2, or 4 toxicants, namely increased TSH and decreased T3 and T4 (Capen, 1997; Davies, 1993; Hill et al., 1989). The lack of weeks of treatment of male rats with PB and PTU. The goal of clear evidence of thyroid hypertrophy or hyperplasia after PB the thyroid-time course study was to determine the optimum administration is probably due to the lower thyrotrophic po- study duration for detecting thyroid toxicants, and to identify tency of PB compared to PTU (Capen, 1997). The potency which endpoints should be included in a Tier I-type screen for difference between PB and PTU are also underscored by the thyroid toxicants. magnitude of the hormonal alterations, where PTU causes Organ weight measurements are useful endpoints for detect- greater changes in thyroid hormone concentrations than PB. ing thyroid toxicants. For example, increases in relative liver No effects would be expected in the in vivo female battery or weight are typical for hepatic-enzyme inducers such as PB that in the YTS. The male rat is more sensitive to thyroid pertur- cause thyroid gland tumors through a secondary mechanism of bations than the female rat due to the higher circulating con- peripheral metabolism of thyroid hormones (McClain et al., centrations of TSH (Capen, 1996). Therefore, when screening 1989). In the thyroid-time course experiment, PB treatment for thyroid-modulating activity, the male rat is a more appro- increased relative liver weight to a similar magnitude at all 3 priate model than the female. Hence, a female rat model should time points. In contrast, PTU, a direct thyroid toxicant, did not not be used to detect thyroid toxicants due to its lower sensi- increase relative liver weight. Therefore, liver weight measure- tivity, although very potent thyroid toxicants (i.e., PTU) may ments, coupled with other thyroid endpoints, can be useful for produce detectable changes. Interestingly, the female rat has determining mechanisms of action of potential thyroid toxi- been suggested as the model system in the final EDSTAC cants. For a thyroid screen, changes in relative liver weight document (EDSTAC, 1998). may be used as a trigger for measuring hepatic enzyme activities such as UDP-GT and 5Ј-deiodinase, assuming other findings corroborate a possible thyroid effect. 15-Day Dietary Restriction Experiment Relative thyroid-gland weight measurements were more sen- Results of the 15-day diet restriction experiment demon- sitive than predicted based on our previous experience. Thy- strate that caution is warranted when interpreting thyroid end- roid gland weight was increased in a time-dependent manner points in studies where body weight decrements are observed. by PTU with similar increases at the 2- and 4-week time points In the dietary restriction experiment, serum hormone concen- (224 and 229% of control, respectively). Similar results were trations, UDP-GT activity, and relative liver weights were observed following PB administration where thyroid gland affected by dietary restriction at levels that resulted in final weights were 119 and 117% of control at the 2- and 4-week body weight decrements of 85% of ad libitum control or time points, respectively. Therefore, thyroid gland weight mea- greater. The slight decrease in UDP-GT activity was attributed surements are a sensitive endpoint for detecting potent and to the decreased relative liver weights that were observed at weak thyroid toxicants, with a 2- or 4-week duration producing those levels of dietary restriction. The data suggest that rela- similar fold changes. The one important factor in the utility of tive, and not absolute, thyroid weight is the more appropriate thyroid weights is the dissection of the thyroid glands prior to endpoint for identifying compound-related effects on thyroid weighing. The procedure employed for this study was to fix the weight when body weight effects are observed. Body weight thyroid glands and surrounding tissue in formalin fixative for at decrements can confound interpretation of compound-related least 48 h prior to dissection. Following fixation, final dissec- effects. However, by targeting a maximum-tolerated dose tion was performed under a dissecting microscope by one (MTD) where body weight decrements are no greater than 10% individual in order to reduce the variability of the dissection of control when screening for thyroid toxicants, no body procedure and hence, reduce the variability of the thyroid weight-dependent changes in thyroid hormone concentrations, weights. The Tier I battery currently undergoing validation thyroid weights, or UDP-GT activity would be expected, and does not presently include thyroid weights as an endpoint. hence, the potential of confounding will be minimized. However, thyroid gland weights should be included in the 68 O’CONNOR ET AL. proposed Tier I battery currently undergoing validation, in activity. However, the decrease in relative liver weight sug- order to help identify thyroid toxicants. gests that the increase in rT3 concentrations may be due to a Thyroid hormone concentrations are routinely used for iden- reduction in the total hepatic 5Ј-deiodinase activity, although tifying compounds that alter thyroid function. With the increas- 5Ј-deiodinase activity was not examined in the time course ing number of commercially available rat hormone assay kits, study. In a future screen, changes in serum rT3 concentrations examining thyroid hormone homeostasis is a quick and easy may be considered as a trigger for measuring hepatic 5Ј- method to identify compounds that alter thyroid function; deiodinase activity. however, these data require cautious interpretation. Many fac- Consistent with the increase in relative liver weight and with tors can affect thyroid hormone concentrations including diet, the mechanism of action of PB (e.g., enhances thyroid hor- stress, age, and circadian rhythm (Capen, 1997; Hill et al., mone excretion through induction of hepatic enzymes), 1989). Furthermore, detection of small changes in thyroid UDP-GT activity was increased at all time points. The greatest hormone concentrations can be confounded due to normal increase was at the 4-week time point (310% of control), variability between animals (Davies, 1993). Measurement of although the magnitude of the increases was similar between serum TSH may be the most important hormonal endpoint for all 3 time points. Consistent with the mechanism of action (i.e., detecting thyroid toxicants, since all known thyroid toxicants direct thyroid toxicant), PTU administration did not affect exert their effects, either primarily or by a secondary mecha- UDP-GT activity at any time point. While hepatic enzyme- nism. The latter would be through a chronic overstimulation of activity measurements may be useful endpoints for identifying the thyroid gland by enhanced release of TSH from the pitu- thyroid toxicants, they would not be recommended as end- itary (Capen, 1997). In the thyroid time course experiment, points for inclusion in a routine screening battery. They could PTU caused a time-dependent increase in serum TSH concen- be, however, additional endpoints that would be triggered trations, with the greatest increase at the 4-week time point based on changes in relative liver weight. Therefore, one could (426% of control). PB administration caused an increase in prepare and store hepatic microsomes for possible use, based serum TSH concentrations at the 4-week time point only on a trigger of changes in relative liver weight. (197% of control). Surprisingly, PB failed to increase serum Thyroid gland follicular cell proliferation is a relatively TSH concentrations at the 2-week time point, as had been sensitive marker for thyroid gland toxicants; however, the observed in the Tier I male battery (Table 6), although the duration of exposure is a critical factor for detecting increases study design through the 2-week time point of the thyroid time in cell proliferation. In the thyroid time-course study, both PTU course study was identical to the Tier I male battery. The and PB caused statistically significant increases in thyroid reason for the lack of an increase in the time course experiment follicular cell proliferation at the 1-week (212 and 822% of appears primarily due to the concurrent control value being control, respectively) and 2-week (247 and 264% of control, higher than normal compared to our historical control data. respectively) time points, but only PTU caused a statistically However, different laboratories have also had mixed results in significant increase at the 4-week time point (209% of control). detecting an increase in serum TSH concentrations after 2 The labeling indices for PTU were time-dependent with the weeks of exposure to PB, where some researchers report an greatest increase at 1 week. The labeling indices for PB were increase (McClain et al., 1989) and others do not (Davies, similar at the 1- and 2-week time points (212 and 247% of 1993). These discrepancies are likely a result of the low po- control, respectively) and had diminished by 4 weeks. The tency of PB, and illustrate the importance of having redun- increase at the 1-week time point, while not statistically sig- dancy of endpoints for detecting weak thyroid toxicants. With nificant (due to the higher error surrounding the control data), the exception of T3 concentrations at the 4-week time point is considered a biologically significant increase. This is con- after PB treatment, serum T3 and T4 concentrations were sistent with previously published data for another thyroid tox- decreased by PTU and PB-treatment at all time points. How- icant, 1-methyl-3-propylimidazole-2-thione (PTI), where a ever, the magnitude of the changes was greatest at the 2-week 1-week exposure caused greater increases in thyroid follicular time point. cell proliferation than a 3-month exposure (Biegel et al., 1995). Since neither PB nor PTU are inducers/inhibitors of 5Ј- Importantly, both PB and PTU would be identified as com- deiodinase, serum rT3 concentrations were not expected to pounds that induce thyroid follicular cell proliferation if a 1- or change in the thyroid time course study; however, rT3 concen- 2-week exposure were used. trations were still included in the battery of endpoints to At the 1997 Duke University Thyroid Workshop (Durham, examine the reliability of rT3 measurements as a potential NC), entitled Screening Methods for Chemicals That Alter endpoint. As expected, PB did not affect serum rT3 concentra- Thyroid Hormone Action, Function and Homeostasis, thyroid tions at any of the time points evaluated. Surprisingly, PTU gland histopathology (i.e., detection of hypertrophy/hyperpla- caused a statistically significant increase in rT3 concentrations sia) was judged to be the most sensitive parameter for the at the 4-week time point (238% of control). The mechanism for detection of compounds that affect thyroid function. However, this increase is unknown. In this instance, clarification of the it was unclear if a study design that used an exposure of less increase should be made by measuring hepatic 5Ј-deiodinase than 4 weeks would produce detectable microscopic changes of RECOMMENDATIONS FOR SCREENING FOR THYROID TOXICANTS 69 the thyroid gland, particularly with weak thyroid toxicants such PTU caused a slight increase in absolute uterine weight at as PB. PB was reported to produce follicular epithelial cell 0.025 and 0.25 mg/kg/day in the in vivo female battery. This hypertrophy/hyperplasia and colloid depletion in male rats increase was considered spurious due to the lack of a dose- following 20 or 70 days treatment (by intraperitoneal injection) response, the small magnitude of change, and the lack of with 100 mg/kg/day PB (Japundzic, 1969). However, this is the concordance with the other endpoints of the in vivo female only published report of such changes after short duration (i.e., battery or the published literature for PTU. In addition, the Ͻ 4 weeks) exposure to PB. Colloid depletion was also ob- absolute uterine weight for the control group was below the served in the present time course study at 2 and 4 weeks (as range of the historical control data for our laboratory (mean well as the in vivo male battery), but hypertrophy was difficult 0.088 g Ϯ 0.007 standard deviation, range 0.079–0.100 g, n ϭ to confirm by qualitative light microscopy. It is noteworthy that 11), while the increases observed after PTU treatment were colloid depletion was detectable in PB-treated rats at doses that within the historical control range for our control values. did not produce changes in thyroid hormones. These results Consistent with the scientific literature, serum PRL concentra- suggest that thyroid histopathology, specifically, changes in tions were elevated after PTU treatment (DeGroot, 1995). This colloid, may indeed be the most sensitive indicator of pertur- increase in PRL is a result of the enhanced release of TRH bation in thyroid homeostasis. Furthermore, PAS staining of from the hypothalamus and is common among potent thyroid thyroids aids in the detection of subtle changes in colloid. toxicants (DeGroot, 1995). In the in vivo male battery, serum However, caution is warranted in interpreting colloid changes PRL concentrations were also increased. Unexpectedly, serum in the absence of clear hypertrophy and hyperplasia of follic- T concentrations were decreased and serum FSH concentra- ular epithelium. Accurate identification of tinctorial changes in tions were increased in the male battery. The mechanism(s) for colloid requires careful attention to tissue processing, section- the changes in serum T and FSH concentrations are unknown ing, and staining techniques. Inconsistencies in these proce- and have not been previously documented. dures between treated and control groups could results in As expected, in the in vitro YTS, neither PB nor PTU were erroneous conclusions. Furthermore, colloid depletion was active in the 3 receptor systems examined or in the YTS noted in control male rats, albeit in low incidences (2/90 competition assays. controls evaluated in these studies). Thus, the biological significance of colloid depletion, absent other microscopic or Conclusions hormonal changes, is questionable. In this report, two model thyroid toxicants, PB and PTU, Effects of PB and PTU on Reproductive Endpoints were used to characterize the ability of a proposed Tier I screening battery to detect thyroid toxicants. In addition, sev- The primary focus of these studies was the assessment of eral thyroid endpoints [serum hormone concentrations (TSH, thyroid endpoints following exposure to a potent or weak T3,T4, and rT3), hepatic UDP-GT activity, thyroid weight, thyroid toxicant. However, a number of changes were also thyroid follicular cell proliferation, and histopathology of the noted in reproductive parameters. thyroid gland] were evaluated after 1, 2, or 4 weeks of treat- In the in vivo female battery, no changes in any of the ment with PB or PTU. The purpose was to determine which estrogenic endpoints were observed following PB administra- endpoints should be included in a Tier I-type screen for de- tion, except for a slight increase in uterine epithelial cell height tecting thyroid toxicants, and to determine the optimal duration and serum FSH concentrations in the highest treatment group. of exposure. The study duration is a critical factor in determin- These increases were unexpected and the mechanism(s) for the ing what endpoints should be included in a screen for thyroid increases are unknown. However, uterine epithelial cell height toxicants. All of the 9 endpoints examined have utility for can be affected by mechanisms other than estrogenicity (e.g., detecting thyroid toxicants depending on the study design. dopamine regulators and glucocorticoids) (O’Connor et al., However, there was no single time point that was optimal for 1996; 1998a). The slight increase in serum FSH concentrations all the endpoints examined. Nonetheless, the 2-week time point was considered a spurious finding, since there was not concor- appears to provide the best compromise, and it offered either dance in the in vivo male battery where serum FSH concen- greater or equivalent sensitivity for the majority of the end- trations were decreased. In addition, the increase in FSH is not points examined. By having redundancy of endpoints, both consistent with the literature for PB (Kaneko, 1980). In the in potent (i.e., PTU, PTI) and weak (i.e., PB) thyroid toxicants vivo male battery, PB produced changes in a number of repro- would be detected using the proposed 15-day Tier I male ductive hormones (decreased PRL, LH, and FSH; and in- battery. A few enhancements to the proposed Tier I male creased E2). The mechanisms for the decreases in PRL, LH, battery would be to add thyroid gland weights, and potentially and FSH are unknown but may be due to the central sedative thyroid follicular cell proliferation. Therefore, a Tier I male action of PB. The modest increases in E2 were judged spuri- battery designed to detect thyroid toxicants should include ous, due to the small magnitude of change coupled with the thyroid weights, thyroid hormone concentrations (TSH, T3, and decreases in serum T, LH, and FSH concentrations. T4), and thyroid histopathology. Thyroid follicular cell prolif- 70 O’CONNOR ET AL. eration could be used as a confirmation of minimal increases in activities: Guidelines for consistent interim terminology and assay condi- thyroid weight, or could be used to confirm equivocal histo- tions. Biochem. Pharmacol. 32, 953–955. pathology findings in the thyroid gland. Changes in relative Bradford, M. M. (1976). A rapid and sensitive method for the quantification of liver weight and thyroid hormone concentrations could be used microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248–254. as a trigger to perform hepatic UDP-GT and/or 5 -deiodinase Ј Capen, C. C. (1996). Toxic responses of the endocrine system. In Casarett and activity measurements to further identify the mechanisms of DoullЈs Toxicology: The Basic Science of Poisons (C. D. Klaassen, Ed.), pp. action of potential thyroid toxicants, assuming other findings 617–640. McGraw-Hill, New York. corroborate a possible thyroid effect. Capen, C. C. (1997). Mechanistic data and risk assessment of selected toxic EDSTAC is currently recommending a 20-day pubertal fe- endpoints of the thyroid gland. Toxicol. Pathol. 25, 39–48. male model for detecting thyroid toxicants (EDSTAC, 1998). Cook, J. C., Kaplan, A. M., Davis, L. G., and O’Connor, J. C. (1997). The female rat model is not as sensitive at detecting thyroid Development of a Tier I screening battery for detecting endocrine active compounds (EACs). Regul. Toxicol. Pharmacol. 26, 60–68. toxicants as a male model (Capen, 1996, 1997); hence, the Davies, D. T. (1993). Assessment of rodent thyroid endocrinology: advantages current 15-day intact male battery would be a better model than and pit-falls. Comp. Hematol. Int. 3, 142–152. the 20-day pubertal female assay for detecting thyroid toxi- Davis, B. (1993). Female Reproductive Toxicology. Academic Press, San cants. In contrast, the Organization for Economic Cooperation Diego. and Development (OECD) is currently proposing a modified DeGroot, L. J. (1995). Endocrinology (M. Besser, H. G. Burger, J. L. Jameson, 28-day guideline study, consisting of both male and female D. L. Loriaux, J. C. Marshall, W. D. Odell, J. T. J. Potts, and A. H. rats, to screen for thyroid toxicants and other EACs. This study Rubenstein, Eds.), W B Saunders, Philadelphia. design would be sufficient for detecting thyroid toxicants for EDSTAC. (1998). Endocrine Disruptor Screening and Testing Advisory Com- most of the endpoints evaluated. However, some endpoints mittee (EDSTAC) Final Report, August 1998. Hill, R. N., Erdreich, L. S., Paynter, O. E., Roberts, P. A., Rosenthal, S. L., and (i.e., thyroid follicular cell proliferation, serum T3 and T4 Wilkinson, C. F. (1989). Thyroid follicular cell carcinogenesis. Fundam. concentrations) would be less sensitive after a 28-day expo- Appl. Toxicol. 12, 629–697. sure, due to compensatory mechanisms that can occur over Japundzic, M. M. (1969). The goitrogenic effect of phenobarbital-Na on the rat time. Therefore, while the 28-day model would be sufficient thyroid. Acta Anat. 74, 88–96. for detecting most thyroid toxicants, the current data suggest Kaneko, S. (1980). Changes in plasma progestin, prolactin, LH, and FSH at that a 2-week time point would be more sensitive, and would luteal activation with phenobarbital anesthesia in the rat. Endocrinology Jpn also provide a quicker, more cost-effective short-term screen- 27, 431–438. ing battery for detecting potential thyroid toxicants. Leonard, J. L., and Rosenburg, I. N. (1980). Iodothyronine 5Ј-deiodinase from rat kidney: Substrate specificity and the 5Јdeiodination of reverse triiodo- thyronine. Endocrinology 107, 1376–1383. ACKNOWLEDGMENTS Masaki, R., Matsuura, S., and Tashiro, Y. (1984). A biochemical and electron The authors would like to acknowledge the invaluable advice of Ms. Ann M. microscopic study of changes in the content of cytochrome P-450 in rat Mason (Chlorine Chemistry Council) and Drs. James A. Barter (PPG Indus- livers after cessation of treatment with phenobarbital, b-naphtoflavone, or tries), Robert E. Chapin (NIEHS), A. Michael Kaplan (DuPont), William R. 3-methylcholanthrene. Cell Struct. Funct. 9, 53–66. Kelce (Monsanto), Ronald R. Miller (Dow Chemical), and Ellen K. Silbergeld McClain, R. M., Levin, A. A., Posch, R., and Downing, J. C. (1989). The effect (University of Maryland at Baltimore). The authors would like to thank Vivian of phenobarbital on the metabolism and excretion of thyroxine in rats. Thompson, Bryan Crossley, Brian Shertz, Sue Snajdr, Christine Glatt, Suzanne Toxicol. Appl. Pharmacol. 99, 216–228. Craven, Denise Janney, and Susan Nicastro for their technical support. Finally, O’Connor, J. C., Cook, J. C., Craven, S. C., VanPelt, C. S., and Obourn, J. D. authors would also like to thank Dr. Donald P. McDonnell (Duke University) (1996). An in vivo battery for identifying endocrine modulators that are for his technical expertise and generosity in providing the yeast strains for the estrogenic or dopamine regulators. Fundam. Appl. Toxicol. 33, 182–195. YTS and Dr. Kevin W. Gaido (CIIT) for his suggestions throughout this O’Connor, J. C., Cook, J. C., Slone, T. W., Makovec, G. T., Frame, S. R., and project. Davis, L. G. (1998a). An ongoing validation of a Tier I screening battery for detecting endocrine-active compounds (EACs). Toxicol. Sci. 46, 45–60. REFERENCES O’Connor, J. C., Frame, S. R., Biegel, L. B., Cook, J. C., and Davis, L. G. (1998b). Sensitivity of a Tier I screening battery compared to an in utero Barter, R., and Klaassen, C. (1992). UDP-glucuonosyltransferase inducers exposure for detecting the estrogen receptor agonist 17b-estradiol. Toxicol. reduce thyroid hormone levels in rats by an extrathyroidal mechanism. Sci. 44, 169–184. Toxicol. Appl. Pharmacol. 111, 36–42. Paynter, O. E., Burin, G. J., Jeager, R. B., and Gregorio, C. A. (1988). Biegel, L. B., Cook, J. C., O’Connor, J. C., Aschiero, M., Arduengo, A. J., 3rd, Goitrogens and thyroid follicular cell neoplasia. Evidence for a threshold and Slone, T. W. (1995). Subchronic toxicity study in rats with 1-methyl- process. Regul. Toxicol. Pharmacol. 8, 102–-119. 3-propylimidazole-2-thione (PTI): Effects on the thyroid. Fundam. Appl. Pazos-Moura, C. C., Moura, E. G., Dorris, M. L., Rehnmark, S., Melendez, L., Pharmacol. 27, 185–194. Silva, J. E., and Taurog, A. (1991). Effect of iodine deficiency and cold Bock, K. W., Burchell, B., Dutton, G. J., Hanninen, O., Mulder, G. J., Owens, exposure on thyroxine 5Ј-deiodinase activity in various rat tissues. Am. J. I. S., Siest, G., and Tephly, T. R. (1983). UDP-glucuronosyltransferase Physiol. 260, E175–182.