Final Report: In Vitro to In Vivo Screening of Thyroid Hormone Receptor Disrupting Chemicals

EPA Grant Number: R835164
Title: In Vitro to In Vivo Screening of Thyroid Hormone Receptor Disrupting Chemicals
Investigators: Furlow, David , Murk, Albertinka J.
Institution: University of California - Davis , Wageningen University & Research Centre
EPA Project Officer: Klieforth, Barbara I
Project Period: March 1, 2012 through February 29, 2016
Project Amount: $649,345
RFA: Developing High-Throughput Assays for Predictive Modeling of Reproductive and Developmental Toxicity Modulated Through the Endocrine System or Pertinent Pathways in Humans and Species Relevant to Ecological Risk Assessment (2011) RFA Text |  Recipients Lists
Research Category: Computational Toxicology , Endocrine Disruptors , Health , Ecosystems , Safer Chemicals


Objective 1: Validate results from model compound and high throughput screening in the GH3.TRE-LUC cell line.

  1. Validate positive results from candidate model compound and high throughput screening approaches against a battery of endogenous TH target genes in GH3 cells.
  2. Further characterize the thyroid hormone signaling pathway in GH3.TRE-LUC cells.

Objective 2: Screen potential thyroid hormone receptor disrupting chemicals in wild-type and TRE-Luciferase transgenic tadpoles.

  1. Analyze potential thyroid hormone receptor disrupting chemicals, agonists, and antagonists in wild-type premetamorphic tadpoles and during natural metamorphosis using molecular and morphological endpoints.
  2. Analyze thyroid hormone receptor disrupting chemicals, agonists, and antagonists in TRE-Luc transgenicXenopus laevis tadpoles for tissue-specific reporter gene responsesin vivo

Summary/Accomplishments (Outputs/Outcomes):

Major accomplishments during the period of study:

  1. We completed analysis of pilot quantitative high throughput screening of GH3.TRE.LUC cells for thyroid hormone active compounds completed by our collaborators and the National Center for Advancing Translational Sciences (NCATS) and U.S. EPA. Those results were published in Freitas, et al. Identification of thyroid hormone receptor active compounds using a quantitative high-throughput screening platform. Curr Chem Genomics Transl Med. 2014;8:36-46. The results demonstrated that the cell line performed well in qHTS and provided some initial lead compounds to validate.
  2. We completed a screen for a comprehensive set of endogenous thyroid hormone target genes and associated signaling pathway constituents (e.g., TRs, RXRs, coactivators, corepressors, deiodinases, and transporters). Further, pharmacological assays established that our GH3 cells are primarily responding to T3 and other direct agonists via TRb as opposed to TRa.
  3. GH3.TRE-LUC cells were subsequently screened in qHTS with the Tox21 library of approximately 10,000 compounds. A relatively small number of agonists were detected (and several were known natural or synthetic TR ligands). A large number (about 18% of the chemicals tested) were scored as active antagonists; while not significantly decreasing cell viability over concentrations where they significantly inhibited T3 action, a large number of these chemicals were known to be cytotoxic in other assays and often used as chemotherapeutic agents.
  4. We validated the predominant class of active agonist compounds from all HTS screening as RXR active compounds. This finding was surprising because in many cell types (often using overexpressed receptors in assays), the RXR was thought to be a “silent” partner in TR signaling despite being part of the active heterodimeric complex at target genes. Natural, synthetic (including pharmaceuticals), and environmental RXR ligands such as tributyltin and triphenyltin were tested in cells and in vivo with dramatic and unanticipated effects on TR signaling. The results were published in two recent articles: Mengeling BJ, Furlow JD. Pituitary specific retinoid-X receptor ligand interactions with thyroid hormone receptor signaling revealed by high throughput reporter and endogenous gene responses. Toxicol In Vitro. 2015;29:1609-1618 and Mengeling BJ, Murk AJ, Furlow JD. Trialkyltin rexinoid-X receptor agonists selectively potentiate thyroid hormone induced programs of Xenopus laevis metamorphosis. Endocrinology. 2016;157:2712-2723. The latter was a feature cover article.
  5. We developed and validated an induced metamorphosis assay in 1 week old tadpoles that is adaptable to medium throughput screening of potential positive compounds from HTS assays. The assay employs morphological (gill and tail loss, brain remodeling), cytological (whole mount cell death, cell proliferation), endogenous gene expression, and a TH responsive transgenic reporter gene endpoints. A manuscript is in preparation and an example of how the assay was used to characterize organotin–TR interactions is found in Mengeling BJ, Murk AJ  Furlow JD (2016). In addition, the assay was used to validate a new and purer route of synthesis of a novel TR antagonist, NH-3, that we also used to characterize the GH3.TRE-LUC assay described here: Singh L, Pressly B, Mengeling BJ, Fettinger JC, Furlow JD, Lein PJ, Wulff H, Singh V. Chasing the elusive benzofuran impurity of the THR antagonist NH-3: synthesis, isotope labeling, and biological activity. J Org Chem. 2016;81:1870-1876.
  6. Using this precocious metamorphosis assay, we observed a robust potentiation of TR signaling on multiple parameters—morphological changes through transgenic reporter gene activity—when RXR ligands, including organotins, are added in combination with T3. Such interactions open a new area of high concern for RXR interacting chemicals because they interact not only with TRs but a wide range of nuclear receptors that play critical roles in vertebrate (and invertebrate) development, reproduction, and metabolism.
  7. The remarkable effect of RXR ligands on TR signaling uncovered through this grant has led to a series of exciting new collaborations and funding to continue the work. We are working with Robert Tanguay at Oregon State University on RXR ligands and TR signaling in zebrafish, in a collaboration that started as a result of this funding. We also are continuing our work with Dr. Tinka Murk, co-PI on this grant, on RXR ligand effects on settlement and metamorphosis in sea urchins and tunicates, and colleagues at UC Davis on rodent models.  
  8. Lastly, this work led directly to funding from the NIH (NIEHS) to continue our studies on RXR action in Xenopus, delving more deeply into organotin and synthetic rexinoid potentiation of TR signaling in vivo. This includes new CRISPR/Cas9 mediated genome editing technology we have adapted for this system. We look forward to continued interactions with U.S. EPA scientists in this important new area of endocrine disruption and environmental toxicology that has been so far underappreciated until these studies and others.


Objective 1: Validate results from model compound and high throughput screening in the GH3.TRE-LUC cell line.

  1. Characterization of the thyroid hormone induced gene expression program in GH3 cells.

    Given the emerging importance of GH3 cells as an in vitro model for TR action, and their increasing use as a screening system for TH disruption, a fuller understanding of the T3 induced gene expression responses in these cells is important. Therefore, we conducted a microarray analysis to delineate the T3 induced transcriptome in these cells. Our microarray analysis revealed a number of important characteristics regarding the responsiveness of these cells to T3 (Table 1). First, the array detected expression of both TR isotypes, and GH3 cells have been shown to express TRa1 and the non-ligand binding TRa2 splicing isoform, and two TRb splicing isoforms (TRb1 and TRb2) (Ball, et al., 1997). In addition to the TRs, the array indicated that these cells express two RXR isotypes (b and g) (Haugen, et al., 1997), at least one nuclear receptor co-activator, which is T3 inducible (SRC-1), and two nuclear receptor co-repressors (NCoR and SMRT), again in agreement with previous published findings in GH3 cells (Misiti, et al., 1998). Type I deiodinase (Dio1) expression was both detected in controls and strongly induced by T3, with lower expression of the Type II deiodinase (Dio2). The T3- and T4-inactivating Type III deiodinase (Dio3) was not detected in these experiments, although a probe was included on the array. Expression of the TRs, RXRs (and associated coregulators), the low level of Dio3 expression, and abundant heterodimeric LAT1/4F2hc TH transporter expression, are all consistent with high sensitivity of these cells to T3.

    In terms of the T3 induced gene expression program in these cells, in general, up-regulated genes mostly fell into two general categories: genes often induced by various extracellular signals, including hormones, and genes involved in angiogenesis or as a response to hypoxia. In the former group, there were several genes identified that have been identified as TH responsive genes in several different contexts; these genes may be considered a common “core” that when taken together, reflect the presence of T3 and a functional TR. For example, the aforementioned Dio1 gene, the B-cell leukemia/lymphoma 3 (Bcl3), sonic hedgehog (Shh), multidrug resistance (Abcb1b) (Kurose, et al., 2008), the corepressor hairless (hr) (Thompson and Potter, 2000), and Klf9/BTEB genes (Denver and Williamson, 2009) (Table 2) have all been identified as TH inducible in other cell types and contain functional TREs, adding additional confidence that the array detected bona fide T3 responsive genes.

  2. Characterization of the thyroid hormone isotype specificity of GH3.TRE-LUC cells.

    We also demonstrated that the reporter gene assay developed in GH3.TRE-LUC cells is an excellent surrogate for specific endogenous gene responses to T3, isotype thyromimetic drugs. The reporter assay provides greater speed and more specificity than the previously developed T-screen that relies on an indirect detection of cell proliferation. Since induction or inhibition of cell proliferation could occur via multiple pathways unrelated directly to TR activity, development of more specific assays for TR function, such as the reporter system described here, that closely mimics endogenous gene behaviors, is highly desirable. In comparing T screen proliferation results to the reporter and endogenous gene (GH) assays, we found close concordance among the assays, including the same rank order of potency of T3 vs. two thyromimetic isotype selective compounds, GC-1 and CO23 (T≥ GC-1>>CO-23). GC-1 is highly selective for TRb (Furlow, et al., 2004) whereas CO23 is highly selective for TRa  (Ocasio and Scanlan, 2006). Thus, for these compounds, the reporter gene assay is strongly predictive of both a phenotypic response (proliferation) and an endogenous gene response, but in a much more convenient format that is adaptable for high throughput assay development.

  3. Analysis of positive hits from HTS screening of a series of chemical libraries.

    To adapt the use of GH3.TRE-Luc reporter gene cell line for a quantitative high-throughput screening (qHTS) platform, we miniaturized the reporter gene assay to a 1536-well plate format. One-thousand two-hundred eighty chemicals from the Library of Pharmacologically Active Compounds (LOPAC) and the National Toxicology Program (NTP) 1408 compound collection were analyzed to identify potential thyroid hormone receptor (TR) agonists and antagonists, with the help of Menhang Xia’s group at the National Center for Advanced Translational Sciences and with the collaborative efforts of Dr. Keith Houck of the U.S. EPA (Freitas, et al., 2014). Of the 2,688 compounds tested, eight scored as potential TR agonists when the positive hit cut-off was defined at  ≥ 10% efficacy, relative to maximal triiodothyronine (T3) induction, and with only one of those compounds reaching ≥ 20% efficacy. Five potential TR antagonists were identified. None of the inactive compounds were structurally related to T3, nor had been reported elsewhere to be thyroid hormone disruptors, so false negatives were not detected. None of the low potency (> 100 µM) TR agonists resembled T3 or T4, thus these may not bind directly in the ligand-binding pocket of the receptor. For TR agonists, in the qHTS, a hit cut-off of ≥ 20% efficacy at 100 μM may avoid identification of positives with low or no physiological relevance. Therefore, we concluded that the miniaturized GH3.TRE-Luc assay offers a promising addition to the in vitro test battery for endocrine disruption, and given the low percentage of compounds testing positive, its high-throughput nature is an important advantage for future toxicological screening (Freitas, et al., 2014).

    Again, in collaboration with the Xia group at NCATS and Dr. Houck at the EPA, the larger Tox21 library of approximately 9,000 unique chemicals was run against the GH3.TRE-LUC cells in quantitative HTS. The results were accessible to us via the iCSS ToxCast dashboard ( In subsequent analysis, Dr. Xia’s group applied a more stringent set of filters and provided early access to their analysis via As in the pilot studies with the smaller LOPAC and NTS libraries, the agonist mode of the screen was run without any T3 present. For each dose response curve, T3 was used as a positive control. In order for a chemical to be considered a positive hit, it had to increase luciferase levels to at least 20% maximal T3 activation reached by 10 μM of the compound. Out of the larger number of chemicals screened, only 22 were considered active agonists by this criteria. The top 14 compounds that scored positive on the agonist assay with an AC50 of < 10 µM are found in Table 3.

    The agonist mode assay revealed that seven of the active agonists belong to the known natural and synthetic thyroid hormone class. Each of the natural and synthetic THs activated the Luc gene assay to at least 100% maximal T3 activity, meaning they are considered full agonists. However, the concentration at which each agonist reached maximal T3 activation varied. Seven of the top eight most potent chemicals (with the exception of 13-cis retinoic acid) belong to the natural and synthetic thyroid hormone class. Surprisingly, four natural and synthetic retinoids (all trans retinoic acid, 13-cis retinoic acid, 9-cis retinoic acid, and acitretin) also scored positive in the agonist mode assay as well. Retinoids also had been detected in the pilot screens, so it was not a completely unexpected result. For example, ATRA activated the Luc gene assay to 31.6% maximal T3 activation. Luc activity peaked at a concentration of 1.65 μM, after which a sharp decrease in activity was observed that coincided with general cytotoxicity.

    The antagonist mode of the screen was run in the presence of 1nM T3. In order for a chemical to be considered a positive hit, it had to inhibit the T3 induced response by at least 20% by 10 μM. In addition to the antagonist assay, a viability assay also was run to ensure that cytotoxicity was not contributing to the decrease in luciferase levels. Out of the roughly 10,000 chemicals screened, there were initially 1,800 positive antagonists, a surprisingly high number. In a recent re-evaluation of a wide range of Tox21 supported screens, Hsieh, et al. (2015) sought to reduce potential artifacts by excluding known cytotoxic agents or known inhibitors of luciferase activity, among other critieria. For example, general transcription inhibitors like Actinomycin D and translation inhibitors like cycloheximide are filtered out. Such analysis with additional curve fitting further reduced the potential antagonist list to about 300 chemicals of interest.  We further limited our analysis to only include compounds that met the following three requirements: (1) decreased T3 induced luciferase activity by 20% or more, (2) an AC50 of less than 10 μM, and (3) a viability AC50 that is at least two times greater than the luciferase AC50. Using this restricted approach, we were able to narrow down the list of active antagonists to 25 high priority compounds for further testing. This list of active antagonists can be found in Table 4. Of particular interest to us was the group of related compounds of the strobilurin class of fungicides as well as a set of related tyrosine kinase inhibitors known to inhibit TH uptake into cells (Braun, et al., 2012) (pazopanib is the most potent in this analysis).

  4. Further investigation of the retinoid induction of the reporter gene in GH3.TRE-LUC cells.

    As noted, one major surprise from the high-throughput screening experiments using these cells was that several retinoid/rexinoid compounds induced Luc activity, although the retinoic acid receptor (RAR) specific synthetic agonist TTNPB did not. We more fully characterized the retinoid response and showed that even in the case of all trans retinoic acid (ATRA) and 13-cis retinoic acid (13cRA), the retinoids appear to be functioning through retinoid –X receptors (RXRs) (Mengeling and Furlow, 2015), which are heterodimer partners with TRs, as RXR antagonists abrogated the retinoid-induced Luc activation. We investigated whether this retinoid response also induced endogenous TR target genes in the GH3.TRE-Luc cells using the genes identified in the previous objective, and we showed that Luc activity correlated extremely well with TR target gene activation. In addition, synthetic RXR-specific agonists significantly activated all the tested TR target genes, but interestingly, the extent compared to the maximal T3-induction the retinoids/rexinoids could reach varied on a gene by gene basis, mostly through differences in the fold activation by T3 (Mengeling and Furlow, 2015). The organotin compounds tributyltin and triphenyltin, chemicals of high concern for invertebrates and vertebrates as well as known RXR interacting chemicals, followed the same pattern as the natural and synthetic retinoids in GH3.TRE-LUC cells (Mengeling, et al., 2016). Because they induced the reporter gene like the other retinoids and synthetic RXR ligands but generally did not reach 20% of the T3 response, they were scored as negative hits on HTS. However, when we checked individually in our hands, the chemicals often reached over 20% activation at low concentrations (1 nM) prior to concentrations that led to general toxicity. The non RXR binding organotin, trimethyltin, did not show any response prior to general toxicity. Given their low dose effects, their putative RXR mediated mode of action that is novel for TR signaling, and their presence in environmental samples and in human blood samples, we further investigated the organotins in vivo.

  5. Prioritization of chemicals for in vivo testing.

    Given the lower throughput nature of the in vivo assays, we applied the following criteria for prioritization of positive hits from the GH3.TRE-LUC assay.  As far as the small number of agonists, other than using T4 and T3 as positive controls, our focus was on the retinoid story with comparisons to marginally active yet known RXR ligands such as the organotins. Given the large number of antagonists, we did not test such chemicals if they also repressed Luc- acivity in the absence of T3, implying a general inhibition of luciferase activity or gene expression rather than specifically antagonizing T3 activity per se. These included 5-Fluoruracil, 1-acetyl-2-phenylhydrazine and tranilast, among many others we tested. We also did not have access to the Tox21 library of compounds (an MTA is in place, but they were not ready to be shipped as of this writing); therefore, we relied exclusively on commercial sources in all experiments. Chemicals that were only available in small amounts and/or were prohibitively expensive were initially excluded. In general, we observed very similar results in our hands compared to the reported HTS results; chemicals that did not satisfy that simple criterion also were not screened further.

Objective 2. Assaying positive hits from HTS in vivo: Xenopus laevis metamorphosis

  1. Development of an early stage metamorphosis assay using transgenic Xenopus laevis.

    One of the most dramatic effects of any hormone in nature can be observed in amphibian metamorphosis, a process completely dependent on THs that are structurally identical from frog to man (Furlow and Neff, 2006). Most work on the molecular underpinnings of metamorphosis has been conducted using the frog Xenopus laevis, a system that has proven to be a particularly useful biological assay for identifying and characterizing novel thyroid hormone receptor agonists and antagonists, and more recently for screening environmental thyroid hormone disrupting chemicals. During metamorphosis, de novo adult tissue growth and larval tissue death during metamorphosis is coordinated by rising levels of TH (Furlow and Neff, 2006). TH also induces remodeling in tissues that undergo both larval cell death and adult cell proliferation and differentiation. These morphological changes are accompanied by extensive tissue-specific gene expression changes that have been extensively characterized. Like their mammalian and avian counterparts, Xenopus laevis has two highly conserved receptor isoforms, xTRa and xTRb, encoded by two separate genes, and together with their heterodimer partner RXR that also is well conserved in Xenopus, they modulate target gene transcription. Both genes are also autoinduced by TH, particularly TRb (Furlow and Neff, 2006).

    There are multiple advantages of the Xenopus system for studying basic developmental endocrinology, as well as its application in environmental toxicology. For example, large numbers of embryos can be obtained year round, and they develop competence to respond to exogenously added TH and synthetic analogs shortly after hatching. Given the conservation of the TH signaling pathway from frog to man, and the exquisite specificity of metamorphosis on TH, the EPA and OECD proposed the development of the Amphibian Metamorphosis Assay as a Tier 1 battery component for endocrine disrupting chemicals (; However, the assay is time consuming, uses a large number of animals and large water volumes (and compounds) with a specialized flow through water supply not routinely available in most laboratories, it relies heavily on limb development as the primary endpoint for developmental staging along with thyroid histology, and it does not provide direct mode of action information.

    Therefore, we chose to test whether the cell-based screening results are relevant in a convenient and specific in vivo assay for TH action using induced metamorphosis of 1 week old, stage 48 Xenopus laevis tadpoles. The tadpoles are quite small and, therefore, small static volumes of rearing solutions can be used. We compared a set of natural TR agonists (T3 and T4), a synthetic antagonist (NH-3) (Lim, et al., 2002), and a set of potential TH disrupting chemicals (TBBPA and BDE-47) to develop the assay. Compounds were tested for specific morphological changes representing reproducible and quantifiable endpoints of TH action, and conveniently measured gill loss, lower jaw remodeling, brain (especially tectum) expansion, and tail resorption were observed. We augmented gross morphology measurements with a whole mount immunocytochemical assay to quantify proliferating cells (phosphohistone H3) and dying cells (caspase – 3 activation) in a three-dimensional assay for a known cellular response to TH in vivo. We further demonstrated that endogenous strongly induced TH responsive genes that can be used as pathway specific biomarkers. Lastly, we developed a transgenic, TH response element driven reporter gene that is a convenient substitute for gene expression responses to suspected TR interacting compounds. The reporter gene cosegregates with another transgene driving GFP expression in the lens of the eye resulting in a convenient screen for animals outbred in matings with wild type animals. The reporter gene is highly specific, is highly inducible from a low baseline, and is activated in a sensitive dose responsive manner by T3 (and inhibited by the antagonist NH-3) added to the rearing water. Overall, there was an excellent concordance among multiple assays for the tested chemicals: T3 and T4 as agonists, NH-3 as a dose responsive antagonist, and TBBPA as a lower potency but still highly significant antagonist across multiple endpoints. Only BDE-47 did not show a specific response, and may need additional metabolic activation to be functional as an agonist. TBBPA is only weakly active in GH3 cells, so the one week induced metamorphosis assay may be more sensitive for some compounds while at the same time revealing potentially important cell type selective actions.

  2. In vivo assays for HTS hits from GH3 cells: focus on RXR ligands

    We became very interested in the recently revealed RXR agonist effects on thyroid hormone (TH) signaling in GH3.TRE-Luc cells. We turned to a simple, robust, and specific in vivo model system of TH action described above: metamorphosis of Xenopus laevis, the African clawed frog. Using the precocious metamorphosis assay, we found that synthetic RXR ligands (bexarotene) as well as the organotins TBT and TPT, but not TMT (a non RXR interacting organotin), greatly potentiated the effect of TH treatment on resorption phenotypes of the tail, which is lost at metamorphosis, and in the head, which undergoes extensive remodeling including gill loss (Mengeling et al. 2016). Consistent with these responses, TH-induced caspase-3 activation in the tail was enhanced by co-treatment with TBT and bexarotene. Induction of transgenic reporter and endogenous genes related to cartilage remodeling were not induced by TBT or bexarotene alone, but TH induction was significantly potentiated by TBT and bexarotene. However, induction of other thyroid hormone receptor target genes such as TRβ and deiodinase 3 by TH were not affected by TBT or bexarotene co-treatment.

    This level of synergy between RXR ligands and TR action is fairly unprecedented and challenges dogma, yet was indicated to a degree by the GH3.TRE-LUC HTS results. The involvement of RXR was further supported by parallel studies that showed two different RXR antagonists inhibited TR action in GH3.TRE-LUC cells (Mengeling and Furlow, 2015) and in tadpoles. The biggest difference between the GH3. TRE-LUC cells and tadpoles is the lack of apparent affect of the RXR ligands, including the organotins, alone on TH dependent responses which was how the chemicals were flagged for concern in the HTS results in the first place. Hence, the GH3. TRE-LUC cells detected an important interaction that played out in vivo, nevertheless with a different feature in the absence of T3 that remains to be explored.

  3. Additional chemicals tested in vivo.

    While most of our focus necessarily centered on the surprising and important RXR story, we did explore additional compounds indicated by the qHTS results in GH3.TRE-LUC cells. Besides TBBPA, which already has been discussed, three are highlighted here. First, we tested Equilin which was a potent agonist in GH3, and ordinarily an estrogen. Notable, more potent estrogens do not activate the system. TRE-LUC cells, as well as two putative antagonists, Pyraclostrobin (one of the strobilurins) and SB 205384 that showed significant inhibition of T3 signaling in GH3.TRE-LUC cells with little effect on cell viability or basal luciferase activity. To date, Equilin has not shown any characteristic activity in the absence of T3 in tadpoles, and may be following the RXR ligand example of requiring T3 to be present to exert its full effects in tadpoles. Likewise, SB 205384 has not proven to be an antagonist of T3 action in tadpoles. In both cases, we do not know the extent of uptake into tadpoles that is a limiting feature of the assay. By contrast, pyraclostrobin has shown promising albeit somewhat complicated results to date. It appears to inhibit T3 induced expansion of the tectum but potentiates T3 action in the resorption of the gills. Follow up studies with the transgenic reporter gene and our battery of cytological assays currently are being performed.

Table 1.

Gene symbol Definition T3 FOLD CHANGE
Known direct TH response genes
LOC686871 PREDICTED: similar to B-cell leukemia/lymphoma 3 (Bcl3) 11.8
Dio1 Deiodinase, iodothyronine, type I 7.9
Shh Sonic hedgehog 5.4
Abcb1b ATP-binding cassette, sub-family B (MDR/TAP), member 1B 5.3
Hr Hairless homolog 4.8
Acta1 Actin, alpha 1, skeletal muscle 4.1
Genes linked to hormonal/extracellular stimuli
Vdr Vitamin D (1,25- dihydroxyvitamin D3) receptor 4.0
Irak2 Interleukin-1 receptor-associated kinase 2 4.5
Sstr2 Somatostatin receptor 2 4.5
Rara Retinoic acid receptor alpha 1 2.4
Bmp6 Bone morphogenetic protein 6 2.3
Esr1 Estrogen receptor 1 2.0
Genes linked to angiogenesis
Epas1 Endothelial PAS domain protein 1 (HIF2a) 5.2
Pgf Placental growth factor 5.1
Nrp1 Neuropilin 1 3.3
Hmox1 Heme oxygenase (decycling) 1 3.2
Efna1 Ephrin A1 3.0
Prok2 Prokineticin 2 2.7
Angpt1 Angiopoietin 1 2.6

Table 2.

Gene symbol Definition Average signal,
Average signal,
Thra Thyroid hormone receptor alpha 1356.8 1164.7
Thrb Thyroid hormone receptor beta 2448.1 1562.9
Rxra Retinoid-X receptor alpha ND ND
Rxrb Retinoid-X receptor beta 1465.0 2028.9
Rxrg Retinoid-X receptor gamma 245.2 203.7
Ncoa1 Nuclear receptor coactivator 1 (SRC-1) 2021.3 4243.0
Ncoa2 Nuclear receptor coactivator 2 (SRC-2, Tif2) ND ND
Ncoa3 Nuclear receptor coactivator 3 (SRC-3, ACTR) ND ND
Ncor1 Nuclear receptor corepressor 1 (NCoR) 602.2 723.1
Ncor2 Nuclear receptor corepressor 2 (SMRT) 527.0 655.8
Dio1 Deiodinase, Type I 238.2 1872.7
Dio2 Deiodinase, Type II 56.3 65.4
Dio3 Deiodinase, Type III ND ND
SLC16A2 Solute carrier family 16, member 2 (MCT8) ND ND
SLCO1C1 Solute carrier organic anion transporter family, member 1C1 (OATP1c1) ND ND
SLC3A2 Solute carrier family 3, member 2 (4F2hc) 8997.8 7666.7
SLC7A5 Solute carrier family 7, member 5 (LAT1) 2737.5 1606.2
Rpl8 Ribosomal protein L8 32692.9 34535.4
Gapdh Glyceraldehyde 3-phosphate dehydrogenase 3580.6 3878.3

Table 3.

Agonists Tox21 screen
Chemical name AC50 (μM) Notes
3, 3’, 5- Triiodo-1-thyronine sodium salt 0.00029 TR agonist
Tetrac 0.00336 TR agonist
Tiratricol 0.00526 TR agonist
CP-634284 0.0055 TR agonist
3, 5, 3’- Triiodothyronine 0.0101 TR agonist
Levothyroxine 0.011 TR agonist
13-cis Retinoic Acid 0.16 Retinoid
3, 3’, 5’- Triiodo-L-thyronine 0.16 TR agonist
Anilazine 0.23 Triazine
All Trans Retinoic Acid 0.24 Retinoid
9-cis Retinoic Acid 1.53 Retinoid
Equilin 1.93 Estrogen
Acitretin 3.35 Retinoid
Toltrazuril sulfone 8.41 Triazinedione

Table 4.

Antagonists Tox21 screen
Chemical name AC50 (μM) Viability AC50 (μM) Notes
Pyraclostrobin 0.083 44.9 Strobilurin
Pyridaben 0.153 31.8  
Picoxystrobin 0.243 .830 Strobilurin
1-Ethylpyridinium bromide 0.630 31.6  
Fluoxastrobin 0.659 25.3 Strobilurin
Floxuridine .810 9.4  
Phanquinone 1.81 29.8  
Proflavin hydrochloride 2.344 25.3  
9-Aminoacridine 2.40 28.7  
Topotecan hydrochloride 2.56 27.6  
Proguanil hydrochloride 2.75 9.02  
Kresoxim-methyl 3.06 8.31 Strobilurin
Pazopanib 3.62 9.44 TH uptake inhibitor
5HPP-33 4.67 17.2  
Proflavin hemisulfate 4.92 26.6  
Dimethyldithiocarbamic acid, dimethylamine 5.09 23.2  
Diclazuril 5.31 13.3  
Picodralazine 5.31 35.5  
Dinoprostone 5.31 16.8  
Acriflavine hydrochloride 6.68 26.6  
Nolatrexed dihydrochloride 6.94 18.1  
5,6-Benzoflavone 7.41 77.0 AhR agonist
Indoprofen 7.79 21.1  
Chlorfluazuron 7.96 35.5  
Guggulsterones E 8.28 76.7  


Out of approximately 10,000 chemicals ultimately screened against GH3.TRE-LUC cells, only a small number (about 30 or so) were considered active partial agonists, and none of the chemicals outside the known natural or synthetic ligands for the TR were full agonists. A relatively large number of chemicals (ranging in some estimates as high as over 10-20% of the tested chemicals) were considered active antagonists. The potent agonist activity of the natural and synthetic thyroid hormones helped to validate the GH3 based luciferase assay as suitable system to detect chemicals that act as agonists or antagonists of this signaling pathway. For the chemicals scoring as partial agonists, we suspect that the significant number of natural and synthetic retinoids we detected are acting through the TR heterodimeric partner RXR as a novel means of inhibition of TR activity worthy of considerable attention, but overall the number of chemicals capable of activating the complex, especially directly interacting with TR, is quite small. For antagonists, several interesting chemicals were found, including clusters of related chemicals that may inhibit via a common function. Interesting possibilities for inhibiting TH action in cells have been potentially revealed that warrant further study, including RXR antagonism, inhibition of uptake into cells, and increased TH inactivation. However, it appears that the vast majority of the original long list of chemicals scored as antagonists also inhibit the luciferase reporter activity alone, despite not being previously scored as cytotoxic in the range of concentrations tested. Many of these were steroid hormones, while another large class consisted of chemotherapeutic agents. The specific or general inhibitory mechanisms remain unclear, but chemicals showing this pattern were of lower priority for further testing.

Our in vivo assays so far have revealed important but arguably mixed results for predictive toxicology in that some chemicals that are not considered very active in cells (TBBPA) appear to be clearly active in vivo, whereas others that are potent antagonists in cells (Pyraclostrobin) show a complicated tissue specific set of responses in vivo, a situation not unlike is commonly observed with many designed NR ligands that turn out to be mixed agonists/antagonists. The retinoid story is clearly the most promising to come out of these studies, and has garnered good attention in the field and additional funding to explore further. Yet in this case, a simple extrapolation from the GH3.TRE-LUC cells was not quite possible: a strong and positive interaction was predicted, certainly, but not entirely in the same pattern in vivo with regard to effects in the presence and absence of T3. That said, a potentially rich source of additional chemicals to test may be revealed from examining existing RXR based HTS assays, and the design of RXR antagonist screens.

Furthermore, careful characterization of the latest HTS results and refined analyses reveal chemicals that may act via multiple mechanisms besides interacting with RXR and TRs directly. We will continue to test recently uncovered candidates like the strobilurin class of compounds, equilin in the presence and absence of T3, pazopanib and its related family of compounds that may inhibit TH uptake, and 5,6-benzoflavone and other AhR ligands that may affect TH metabolism using our in vivo using amphibian metamorphosis assays. These studies will be carefully designed with response curves to limit general toxicity, and examination of multiple tissues and developmental time points. The newly developed TRE-luc transgenic Xenopus lines developed here remain a powerful tool to examine interactions of these and other chemicals affecting TR signaling in vivo.


Ball, S.G., Ikeda, M., Chin, W.W., 1997. Deletion of the thyroid hormone beta1 receptor increases basal and triiodothyronine-induced growth hormone messenger ribonucleic acid in GH3 cells. Endocrinology 138, 3125-3132.

Braun D., Kim T.D., le Coutre P., Kohrle J., Hershman J.M., Schweizer U., 2012 Tyrosine kinase inhibitors noncompetitively inhibit MCT8-mediated iodothyronine transport. Journal of Clinical Endocrinology and Metabolism  97 E100-E105.

Denver, R.J., Williamson, K.E., 2009. Identification of a thyroid hormone response element in the mouse Kruppel-like factor 9 gene to explain its postnatal expression in the brain. Endocrinology 150, 3935-3943.

Freitas, J., Miller, N., Mengeling, B.J., Xia, M., Huang, R., Houck, K., Rietjens, I.M., Furlow, J.D., Murk, A.J., 2014. Identification of thyroid hormone receptor active compounds using a quantitative high-throughput screening platform. Curr Chem Genom Transl Med 8, 36-46.

Furlow, J.D., Neff, E.S., 2006. A developmental switch induced by thyroid hormone: Xenopus laevis metamorphosis. Trends Endocrinol Metab 17, 40-47.

Furlow, J.D., Yang, H.Y., Hsu, M., Lim, W., Ermio, D.J., Chiellini, G., Scanlan, T.S. 2004. Induction of larval tissue resorption in Xenopus laevis tadpoles by the thyroid hormone receptor agonist GC-1. J Biol Chem 279:26555-26562.

Haugen, B.R., Brown, N.S., Wood, W.M., Gordon, D.F., Ridgway, E.C. 1997. The thyrotrope-restricted isoform of the retinoid-X receptor-gamma1 mediates 9-cis-retinoic acid suppression of thyrotropin-beta promoter activity. Mol Endocrinol 11:481-489.

Hsieh, J.-H., Sedykh, A., Huang, R., Xia, M., Tice, R.R. 2015. A data analysis pipeline accounting for artifacts in Tox21 quantitative high-throughput screening assays. Journal of Biomolecular Screening 20:887-897.

Kurose, K., Saeki, M., Tohkin, M., Hasegawa, R. 2008. Thyroid hormone receptor mediates human MDR1 gene expression-Identification of the response region essential for gene expression. Arch Biochem Biophys 474;82-90.

Lim, W., Nguyen, N.H., Yang, H.Y., Scanlan, T.S., Furlow, J.D. 2002. A thyroid hormone antagonist that inhibits thyroid hormone action in vivo. J Biol Chem 277;35664-35670.

Mengeling, B.J., Furlow, J.D. 2015. Pituitary specific retinoid-X receptor ligand interactions with thyroid hormone receptor signaling revealed by high throughput reporter and endogenous gene responses. Toxicology In Vitro 29;1609-1618.

Mengeling B.J., Murk A.J., Furlow, J.D. 2016. Trialkyltin rexinoid-X receptor agonists selectively potentiate thyroid hormone induced programs of Xenopus laevis metamorphosis. Endocrinology 157:2712-2723.

Misiti, S., Schomburg, L., Yen, P.M., Chin, W.W., 1998. Expression and hormonal regulation of coactivator and corepressor genes. Endocrinology 139;2493-2500.

Ocasio, C.A., Scanlan, T.S. 2006. Design and characterization of a thyroid hormone receptor alpha (TRalpha)-specific agonist. ACS Chem Biol 1;585-593.

Thompson, C.C., Potter, G.B. 2000. Thyroid hormone action in neural development. Cereb Cortex 10;939-945.

Journal Articles on this Report : 5 Displayed | Download in RIS Format

Other project views: All 10 publications 5 publications in selected types All 5 journal articles
Type Citation Project Document Sources
Journal Article Freitas J, Miller N, Mengeling BJ, Xia M, Huang R, Houck K, Rietjens IM, Furlow JD, Murk AJ. Identification of thyroid hormone receptor active compounds using a quantitative high-throughput screening platform. Current Chemical Genomics and Translational Medicine 2014;8:36-46. R835164 (2012)
R835164 (Final)
  • Full-text from PubMed
  • Abstract from PubMed
  • Associated PubMed link
  • Full-text: Bentham Open-Full Text HTML
  • Abstract: Bentham Open-Abstract
  • Other: Bentham Open-Full Text PDF
  • Journal Article Mengeling BJ, Furlow JD. Pituitary specific retinoid-X receptor ligand interactions with thyroid hormone receptor signaling revealed by high throughput reporter and endogenous gene responses. Toxicology in Vitro 2015;29(7):1609-1618. R835164 (2012)
    R835164 (Final)
  • Abstract from PubMed
  • Full-text: ScienceDirect-Full Text HTML
  • Abstract: ScienceDirect-Abstract
  • Other: ScienceDirect-Full Text PDF
  • Journal Article Mengeling BJ, Murk AJ, Furlow JD. Trialkyltin rexinoid-X receptor agonists selectively potentiate thyroid hormone induced programs of Xenopus laevis metamorphosis. Endocrinology 2016;157(7):2712-2723. R835164 (Final)
  • Abstract from PubMed
  • Full-text: Oxford-Full Text HTML
  • Other: Oxford-Full Text PDF
  • Journal Article Mengelinga BJ, Wei Y, Dobrawa LN, Streekstra M, Louisse J, Singh V, Singh L, Lein PJ, Wulff H, Murk AJ, Furlow JD. A multi-tiered, in vivo, quantitative assay suite for environmental disruptors of thyroid hormone signaling. Aquatic Toxicology 2017;190:1-10. R835164 (Final)
    R835550 (Final)
  • Abstract from PubMed
  • Full-text: ScienceDirect-Full Text HTML
  • Abstract: ScienceDirect-Abstract
  • Other: ScienceDirect-Full Text PDF
  • Journal Article Singh L, Pressly B, Mengeling BJ, Fettinger JC, Furlow JD, Lein PJ, Wulff H, Singh V. Chasing the elusive benzofuran impurity of the THR antagonist NH-3: synthesis, isotope labeling, and biological activity. Journal of Organic Chemistry 2016;81(5):1870-1876. R835164 (Final)
    R835550 (Final)
  • Abstract from PubMed
  • Full-text: eScholarship-Full Text PDF
  • Abstract: ACS-Abstract
  • Other: ACS-Full Text PDF
  • Supplemental Keywords:

    thyroid hormone receptor, retinoid-X receptor, reporter gene assay, transgenic animals, high throughput assays, retinoids, organotins, agonist, antagonist, gene expression, endocrine disruption;

    Progress and Final Reports:

    Original Abstract
  • 2012 Progress Report
  • 2013 Progress Report
  • 2014 Progress Report