Evaluation of potential sodium-iodide symporter (NIS) inhibitors using a secondary Fischer rat thyroid follicular cell (FRTL-5) radioactive iodide uptake (RAIU) assay

Abstract

The Fischer rat thyroid follicular cell line (FRTL-5) endogenously expresses the sodium-iodide symporter (NIS) and has been used to identify environmental chemicals that perturb thyroid hormone homeostasis by disruption of NIS-mediated iodide uptake. Previously, a high-throughput radioactive iodide uptake (RAIU) screening assay incorporating the hNIS-HEK293T-EPA cell line was used to identify potential human NIS (hNIS) inhibitors in 1028 ToxCast Phase I (ph1_v2) and Phase II chemicals. In this study, the FRTL-5 cell line was evaluated and applied as a secondary RAIU assay coupled with cell viability assays to further prioritize highly active NIS inhibitors from the earlier screening. Assay validation with 10 reference chemicals and performance assessment by chemical controls suggest the FRTL-5 based assays are robust and highly reproducible. Top-ranked chemicals from the ToxCast screening were then evaluated in both FRTL-5 and hNIS RAIU assays using newly sourced chemicals to strengthen the testing paradigm and to enable a rat vs. human species comparison. Eighteen of 29 test chemicals showed less than 1 order of magnitude difference in IC₅₀ values between the two assays. Notably, two common perfluorinated compounds, perfluorooctanesulfonic acid (PFOS) and perfluorohexane sulfonate (PFHxS), demonstrated strong NIS inhibitory activity [IC₅₀ −6.45 (PFOS) and −5.70 (PFHxS) logM in FRTL-5 RAIU assay]. In addition, several chemicals including etoxazole, methoxyfenozide, oxyfluorfen, triclocarban, mepanipyrim, and niclosamide also exhibited NIS inhibition with minimal cytotoxicity in both assays and are proposed for additional testing using short-term in vivo assays to characterize effects on thyroid hormone synthesis.

Introduction

There is growing global concern that exposure to chemicals in the environment may be contributing to the observed rise in thyroid disease (Davies et al. 2015) and altered thyroid hormone homeostasis in the human population (Ferrari et al. 2017; Ghassabian and Trasande 2018). Thyroid hormones are important for a number of normal physiological processes such as regulation of metabolic and cardiac function, as well as the development of the brain and sympathetic nervous system (Murk et al. 2013; Yen 2001). Disruption of the thyroid hormone signaling pathways can lead to serious health effects as documented in patients with thyroid disease (Haddow et al. 1999; Taylor et al. 2013) and during pregnancy where suppression of maternal thyroid hormone levels has been implicated in adverse effects on fetal neurodevelopment and cognitive deficits in the offspring (Korevaar et al. 2017; Moog et al. 2017; Zimmermann and Boelaert 2015). Experimental studies have identified a diverse group of environmental chemicals that can perturb thyroid hormone homeostasis through multiple mechanisms of action, including effects on hormone synthesis, secretion, transport and metabolism (Brucker-Davis 1998; Capen and Martin 1989; Crofton 2008; Ghassabian and Trasande 2018), resulting in effects on circulating thyroid hormones, brain development and other physiological processes in laboratory animals (Escobar et al. 2004; Gilbert et al. 2013; Gilbert and Zoeller 2010; Hurley 1998; Köhrle 2008; O’Shaughnessy et al. 2018). However, progress in the identification of chemicals with potential to disrupt thyroid hormone homeostasis and function has been limited by the lack of available in vitro assays/tools to evaluate effects on the array of molecular targets within the thyroid signaling pathway. This has recently stimulated global initiatives to identify and prioritize the development/validation of new computational and in vitro screening assays to evaluate larger libraries of chemicals for potential thyroid-disrupting activity (Murk et al. 2013; Noyes et al. 2019; OECD 2017; Thomas et al. 2019; Zuang et al. 2018).

The U.S. Environmental Protection Agency’s Endocrine Disruptor Screening Program (EDSP) has been working toward using more high-throughput and computational toxicology methods to screen and prioritize chemicals for thyroid hormone disruption, including targeted assays that evaluate molecular components integral to thyroid hormone synthesis (e.g., sodium-iodide symporter and thyroid peroxidase) and metabolism (e.g., deiodinases) (Hallinger et al. 2017; Hornung et al. 2018; Hornung et al. 2015; Olker et al. 2018a; Olker et al. 2018b; Paul Friedman et al. 2016; Paul et al. 2014). Our laboratory recently developed a screening approach to detect chemicals that inhibit the sodium-iodide symporter (NIS) (Hallinger et al. 2017), a transmembrane glycoprotein that mediates the active uptake of extracellular iodide into thyroid follicular cells. The NIS-mediated transport of iodide is the first step in the biosynthesis of thyroid hormones, triiodothyronine (T3) and thyroxine (T4), and represents a key molecular target within the thyroid signaling pathway for environmental chemicals to disrupt (Carrasco 1993; De Groef et al. 2006). To develop the screening approach, we adapted a radioactive iodide uptake (RAIU) assay previously used to screen pharmaceutical chemicals for NIS inhibitory activity (Lecat-Guillet et al. 2007), and optimized the assay using a stable human NIS-expressing HEK293T cell line (hNIS-HEK293T-EPA) (Hallinger et al. 2017). The hNIS RAIU assay was validated using well-characterized reference chemicals, quality controls (QC), and a small set of test chemicals (Hallinger et al. 2017). This assay was used for high-throughput screening of the ToxCast Phase I (ph1_v2) and II chemical libraries whereby 1,028 unique chemicals were initially tested in a tiered format using a single-concentration RAIU assay. Those chemicals that significantly inhibited NIS were further tested in parallel, multi-concentration RAIU and cell viability assays (Wang et al. 2019; Wang et al. 2018). A novel chemical potency ranking approach incorporating multi-concentration RAIU and cell viability responses was also developed to provide each test compound a single score that was normalized relative to the known NIS inhibitor, sodium perchlorate. Using this ranking system, we identified a group of chemicals including insecticides, fungicides, perfluoroalkyl substances, herbicides, and antimicrobials that had the greatest potential to inhibit hNIS activity.

The aim of the current study was to further examine the NIS inhibitory activity for the highest ranked chemicals identified in our previous screening of ToxCast chemical libraries (Wang et al. 2019; Wang et al. 2018) using a secondary RAIU assay. The Fischer rat thyroid follicular (FRTL-5) cell line expresses endogenous NIS and has been shown to maintain biochemical and morphological characteristics of primary thyroid follicular epithelial cells when maintained in custom medium (Ambesi-Impiombato et al. 1980; Weiss et al. 1984). In addition, these cells have been used extensively to study the process of iodide transport (Kaminsky et al. 1991; Lecat-Guillet et al. 2008; Waltz et al. 2010), to screen pharmaceuticals and other environmental chemicals for NIS inhibition in several different assay formats (Di Bernardo et al. 2011; Radovic et al. 2005; Waltz et al. 2010; Wu et al. 2016), and to identify alterations in the expression of genes related to iodide uptake and synthesis of thyroid hormones (Lee et al. 2018; Lee et al. 2017; Xiong et al. 2018). The use of these cells in a secondary RAIU assay provided a method to examine NIS activity in a well characterized, biologically relevant, thyroid follicular cell model for comparing the concordance with data derived from the hNIS-HEK293T-EPA cell line.

Results from this study confirm our previous findings and enhance the weight of evidence for a group of organic environmental chemicals as highly active NIS inhibitors. Top ranked chemicals selected from earlier screening were tested in both the FRTL-5 and hNIS-HEK293T-EPA RAIU assays to enable a direct comparison of rat and human NIS-mediated iodide uptake. New chemicals were procured for this study to provide a testing paradigm that was independent of previous screening. In addition, we evaluated the effects of these chemicals on cell viability using several methods to characterize and interpret the ability of each chemical to directly or indirectly compromise NIS function. This study further strengthens the current knowledge of environmental chemicals that can potentially disrupt thyroid homeostasis through the inhibition of NIS, and will support the development of NIS inhibition test guidelines for the Organisation for Economic Co-operation and Development (OECD).

Materials and Methods

Materials

Bovine thyroid stimulating hormone (bTSH) was obtained from Dr. Parlow at the National Hormone and Peptide Program (NHPP; Torrance, CA). Gibco fetal bovine serum (FBS) and L-glutamine were from ThermoFisher Scientific (Waltham, MA). Tissue-culture flasks (T-75) and 96-well polystyrene plates were from Corning (Corning, NY). CellTiter-Glo kits were purchased from Promega (Madison, WI). Carrier-free ¹²⁵I (Catalog# NEZ033) was purchased from PerkinElmer (Waltham, MA). All other reagents were obtained from Millipore-Sigma (St. Louis, MO) unless otherwise noted.

Cell culture

Fischer rat thyroid cells (FRTL-5) were obtained from the European Collection of Authenticated Cell Culture (ECACC No. 91030711) distributed by Sigma-Aldrich. Cells were maintained in Coon’s modified F-12 medium supplemented with 5% FBS, 2 mM L-glutamine, 10 mU/mL bTSH, 10 μg/mL insulin, 10 nM hydrocortisone, 5 μg/mL transferrin, 10 ng/mL gly-his-lys, 10 ng/mL somatostatin, and penicillin-streptomycin at 37°C and 5% CO₂. Cells of passage <20 were plated in clear, flat-bottom, tissue culture treated 96-well plates at a density of 3.0 × 10⁴ cells/well in 200μL and incubated for 48 hours prior to RAIU or cytotoxicity assays.

hNIS-HEK293T-EPA (hNIS) cells (<25 passages) were grown and maintained in Gibco Dulbecco’s Modified Eagle Medium (DMEM) (ThermoFisher Scientific, Waltham, MA) as previously described (Hallinger et al., 2017). Cells were seeded at a density of 4.0 × 10⁴ cells/well in 200μL medium and incubated at 37°C and 5% CO₂ for 40 hours prior to the RAIU assay. To ensure cell adherence, all plates were coated with high molecular weight (MW: 150,000 to 300,000) 0.01% poly-L-lysine and rinsed twice with sterile deionized H₂O prior to seeding. To maintain cell adherence, gentle aspiration and dispense cycles on the plate washer (Tecan, Männedorf, Switzerland) were used with hNIS cells.

Control and test chemicals

Chemical names, CAS numbers, sources, and purities are listed (Table 1). All chemicals were solubilized in DMSO at 10 mM stock concentration. Chemicals included 10 reference chemicals (monovalent and divalent anions; Weiss et al., 1984; Lecat-Guillet et al., 2007), 5 QC chemicals (Hallinger et al., 2017), and 29 test chemicals (TC). QC chemicals included sodium perchlorate (NaClO₄; RAIU assay positive control), sodium nitrate (NaNO₃; RAIU assay EC₈₀ control), sodium thiocyanate (NaSCN; RAIU assay EC₂₀ control), 2,4-dichlorophenoxyacetic acid (2,4-D; RAIU and cell viability negative controls), and 2,3-dichloro-1.4-napthoquinone (DCNQ; cell viability assay positive control). TCs included the top-ranked NIS inhibitors identified from Tox Cast Phase I (ph1_v2) and Phase II chemical libraries (Wang et al., 2018; Wang et al., 2019). Additional chemicals tested to compare results in FRTL-5 and hNIS RAIU assays included thyroid hormone disruptors, ATPase inhibitors, and perfluorinated chemicals. All test chemicals were assayed in multi-concentration format (0.001 μM-100 μM), with each of six concentrations tested in duplicate wells/plate, for a minimum of 3 experimental bioreplicates (i.e., using cells from independent passages). A detailed assay plate map with reference, QC, and TC location and concentrations is provided (Fig. S1).

Table 1.

List of chemicals used in the study

Reference chemicals	Abbreviation	CAS #	Source	Purity (%)	Purpose
Sodium tetrafluoroborate	NaBF4	13755–29-8	Sigma-Aldrich	98	Validation
Potassium hexafluorophosphate	KPF6	17084–13-8	Sigma-Aldrich	98	Validation
Potassium perchlorate	KClO4	7778–74-7	Sigma-Aldrich	≥99	Validation
Sodium bromide	NaBr	7647–15-6	Sigma-Aldrich	≥99.0	Validation
Sodium carbonate	Na2CO3	497–19-8	Sigma-Aldrich	99.99	Validation
Sodium fluoride	NaF	7681–49-4	Sigma-Aldrich	≥99	Validation
Sodium sulfite	Na2SO3	7757–83-7	Sigma-Aldrich	≥98.0	Validation
Quality Control (QC) chemicals	Abbreviation	CAS #	Source	Purity (%)	Purpose
2,4-Dichlorophenoxyacetic acid	2,4-D	94–75-7	Sigma-Aldrich	98	Negative (RAIU; CTG)
2,3-dichloro-1,4-naphthoquinone	DCNQ	117–80-6	Sigma-Aldrich	98	Positive (CTG)
Sodium thiocyanate a	NaSCN	540–72-7	Sigma-Aldrich	98–102	Positive (RAIU)
Sodium nitrate a	NaNO3	7631–99-4	Sigma-Aldrich	≥99.0	Positive (RAIU)
Sodium perchlorate a	NaClO4	7601–89-0	Sigma-Aldrich	≥98.0	Positive (RAIU)
Test chemicals b	Abbreviation	CAS #	Source	Purity (%)	Purpose
Etoxazole		153233–91-1	Sigma-Aldrich	99.4	Ph1
Niclosamide		50–65-7	Sigma-Aldrich	≥ 98.0	Ph1
3-Iodo-2-propynyl-N-butylcarbamate		55406–53-6	Sigma-Aldrich	≥ 98.5	Ph1
Rotenone		83–79-4	Sigma-Aldrich	> 95	Ph1
Cyprodinil		121552–61-2	Santa Cruz	97	Ph1
Methoxyfenozide		161050–58-4	Sigma-Aldrich	99.8	Ph1
Oxyfluorfen		42874–03-3	Sigma-Aldrich	99.9	Ph1
Triphenyltin hydroxide		76–87-9	Sigma-Aldrich	na	Ph1
Pyridaben		96489–71-3	Sigma-Aldrich	na	Ph1
Captan		133–06-2	Sigma-Aldrich	≥ 98.0	Ph1
Fluroxypyr-meptyl		81406–37-3	Sigma-Aldrich	≥ 98.0	Ph1
Fipronil		120068–37-3	Sigma-Aldrich	≥ 95.0	Ph1
2-(thiocyanomethylthio) benzothiazole	TCMTB	21564–17-0	Chem Service	99.5	Ph1
Perfluorooctane sulfonic acid (free acid)	PFOS	1763–23-1	Evotec	98	Ph1, Ph2
Perfluorooctane sulfonic acid, potassium salt	PFOS-K	2795–39-3	Sigma-Aldrich	≥ 98.0	Ph2
Perfluorohexane sulfonate, potassium	PFHxS-K	3871–99-6	Sigma-Aldrich	≥ 98.0	Ph2
Phenolphthalein		77–09-8	Sigma-Aldrich	98–102	Ph2
Tributyltin chloride	TBTC	1461–22-9	Sigma-Aldrich	96	Ph2
Benz(a)anthracene		56–55-3	Sigma-Aldrich	99	Ph2
Triclocarban		101–20-2	Sigma-Aldrich	99	Ph2
Mepanipyrim		110235–47-7	Sigma-Aldrich	99.9	Ph2
4-chloro-1,2-diaminobenzene		95–83-0	Sigma-Aldrich	97	Ph2
Additional chemicals c	Abbreviation	CAS #	Source	Purity (%)	Purpose
Digitoxin		71–63-6	MCE	95.33	ATPase Inhibitor
Ouabain, octahydrate		11018–89-6	Calbiochem	98.1	ATPase Inhibitor
Digoxin		20830–75-5	MCE	98.46	ATPase Inhibitor
Perfluorooctanonic acid, ammonium salt	PFOA-ammonium	3825–26-1	Sigma-Aldrich	≥ 98.0	PFAS
Perfluorooctanonic acid (free acid)	PFOA	335–67-1	Sigma-Aldrich	≥ 98.0	PFAS
6-Propyl-2-thiouracil	PTU	51–52-5	Sigma-Aldrich	99.2	TPO Inhibitor
2-Mercapto-1-methylimidazole	MMI	60–56-0	Sigma-Aldrich	≥ 99	TPO Inhibitor

^aUsed as a Reference chemical and selected as a QC chemical for each plate.

^bSelected from Table 1 (Wang et al., 2018) and Table 1 (Wang et al., 2019).

^cNew chemicals selected for testing in FRTL-5 and hNIS RAIU and cell viability assays.

na: not available

Abbreviations: CTG: CellTiter-Glo cell viability assay, RAIU: radioactive iodide uptake assay, TPO: thyroid peroxidase.

Iodide uptake assay (FRTL-5 and hNIS)

RAIU assays were conducted as previously described in our laboratory (Hallinger at al., 2017) with minor modifications. Carrier-free ¹²⁵I was diluted with uptake buffer (Hanks’ balanced salt solution supplemented with 10 mM Hepes, pH 7.4). Following the incubation of FRTL-5 (48 h) and hNIS-HEK-293T-EPA (40 h), cells were washed twice with 200 μL of uptake buffer using a 96-well Tecan HydroSpeed plate washer. Uptake buffer (89 μL) was then added back to each well. Chemicals (1 μL at 100X) were added to each well, immediately followed by the addition of 10 μL ¹²⁵I (i.e., 0.1 μCi in a final volume of 100 μL/ well). Cells were incubated at room temperature for 2 h and then washed twice with 200 μL ice cold uptake buffer. Radioactive iodide was released from the cells by lysing with 100 μL of 0.1 N NaOH for 10 minutes. The entire volume in each well was transferred into separate 12×75mm tubes and the ¹²⁵I quantified using a Wizard gamma counter (PerkinElmer, Shelton, CT). RAIU was measured in counts per minute (CPM).

CellTiter-Glo cell viability assay (FRTL-5 and hNIS)

CellTiter-Glo cell viability assays were conducted in parallel with the RAIU assays to identify chemicals that caused a decrease in RAIU due to cytotoxicity. Using this method, the number of viable cells in each well was determined based on the quantification of ATP present, an indication of metabolically active cells. FRTL-5 and hNIS-HEK293T-EPA cells were seeded in white, 96-well plates at the same concentrations as the RAIU assays. Following the incubation periods (48 h (FRTL-5) and 40 h (hNIS)), plates were washed twice with 200 μL uptake buffer utilizing a Tecan HydroSpeed plate washer. Uptake buffer (99 μL) was then added back to each well and 1 μL of each chemical was added to duplicate wells to achieve a final volume of 100 μL/well. Following a 2 h incubation period at room temperature, 100 μL of CellTiter-Glo reagent was dispensed into each well. Plates were placed on orbital shaker for 5 minutes at 700 rpm, followed by 10 minutes of incubation at room temperature. Plates were immediately scanned on a FLUOstar Omega microplate reader (BMG Labtech, Cary, NC) where luminescent signal was quantified as relative light units (RLU) indicating relative ATP concentration.

Propidium iodide cytotoxicity assay (FRTL-5)

An orthogonal propidium iodide (PI) cytotoxicity assay was conducted as described for 29 test chemicals (Table 1) for the FRTL-5 RAIU assay to assess membrane integrity. PI is excluded from viable cells but can penetrate the membrane of dying or dead cells. For imaging compatibility, FRTL-5 cells were seeded in 96-well black-walled, TC-treated, optical bottom microplates (ThermoFisher Scientific, Waltham, MA) pre-coated with poly-L-lysine (MW 70,000–150,000) (0.1 mg/ml) (Millipore Sigma, Burlington, MA). Each plate contained 18 DMSO (1% v/v) solvent control wells and 18 digitonin (24 μM) positive control wells. Following test chemical treatment for 2 h, cells were washed twice with uptake buffer, stained for 20 minutes with PI (1 μg/ml) in uptake buffer, and washed two additional times using a ViaFlo 96/384 semi-automated liquid handler (Integra Biosciences, Hudson, NH). Cells were fixed with 4% paraformaldehyde in phosphate buffered saline (PBS), containing a Hoechst 33342 (5 μg/ml) (ThermoFisher Scientific, Waltham, MA) nuclear counterstain, for 20 minutes, washed twice with PBS, and sealed in PBS using MicroAMP optical adhesive film (ThermoFisher Scientific, Waltham, MA). Four experimental bioreplicates (n=4) were performed for the chemical set.

High content image analysis was performed to determine total cell counts and percent PI responders for each test well. Images were acquired on a Perkin Elmer Opera Phenix (PerkinElmer, Waltham, MA) using a standard 10X, 0.3 NA air objective and 4.4 mega pixels CMOS camera. Independent channels for Hoechst 33342 (ex/405 nM; em/435–480 nM) and PI (ex/561 nM; em/570–630 nM) were collected across 12 fields per well. Image analysis was performed in the instrument-based Harmony software using Hoechst-stained nuclei as primary channel objects. Cell counts were determined by the total object counts per well. The percent PI responders were quantified at the single-cell level using a plate-based PI intensity threshold set at three standard deviations from the mean PI intensity for all DMSO negative control wells. Intra-assay precision and Z’ values (Zhang et al. 1999) were monitored across all plates to evaluate assay performance.

Data analysis

Iodide uptake and ATP cell viability data were normalized to the median DMSO activity of each assay plate and expressed as percent of control activity (% control). The results presented are from a minimum of 3 experimental bioreplicates. A significant threshold of 20% inhibition was chosen in this study for the RAIU and cytotoxicity assays based on our previous NIS inhibitor screening studies (Hallinger et al. 2017; Wang et al. 2019; Wang et al. 2018). Cell count data were represented as percent of the mean DMSO control counts on each assay plate. The PI cytotoxicity assay results were reported as percentage of positive PI responders in each well. The PI results, when plotted with other results, were inversely converted to downtrend to match other inhibitory effects results for easier visualization and interpretation. Dose-response curves were fit using the 4-parameter logistic model in GraphPad Prism 7.0 (La Jolla, CA), with the top and bottom asymptotes constrained to 100 and 0, respectively. The IC₅₀ obtained from the model fit represents the absolute IC50, i.e., the concentration where 50% inhibition occurs.

Z’ values, coefficients of variation (CV) of DMSO, and IC₅₀ of positive controls were calculated by each 96-well assay plate. CV of DMSO was calculated as CV = SD_DMSO/μ_DMSO, where SD_DMSO and μ_DMSO are the standard deviation and mean of the raw response value for DMSO control wells, respectively. Z’ values (Zhang et al. 1999) were calculated as: $Z^{\prime }=1-\frac{3\left({\sigma }_{pos}+{\sigma }_{DMSO}\right)}{\left|{\text{μ}}_{pos}-{\text{μ}}_{DMSO}\right|}$, where σ_pos and μ_pos are the standard deviation and mean of normalized positive control (NaClO₄ for RAIU assay, DCNQ for CellTiter-Glo cytotoxicity assay, Digitonin for PI cytotoxicity assay) values at the highest 100 μM concentration. σ_DMSO and μ_DMSO are the standard deviation and mean of normalized DMSO control values.

Results and Discussion

FRTL-5 RAIU assay validation

The predictivity of the FRTL-5 cell-based RAIU assay for NIS inhibition was first assessed by testing 10 reference inorganic chemicals that are known NIS inhibitors or inactives (Dong et al. 2019; Hallinger et al. 2017; Lecat-Guillet et al. 2007; Weiss et al. 1984). IC₅₀ values for the 10 reference chemicals are listed in Table 2. The FRTL-5 RAIU assay produced IC₅₀ values with small variations that were comparable to our previous study using the hNIS cell line assay (Hallinger et al. 2017), as well as other reports using the FRTL-5 cell lines (Waltz et al. 2010). For example, the IC₅₀ for sodium perchlorate, a commonly used reference NIS inhibitor, was reported as −6.76 (logM) in the FRTL-5 RAIU assay, a comparable value to the IC₅₀ of −6.06 (logM) reported in the hNIS RAIU assay from our previous study (Hallinger et al. 2017).

Table 2.

IC₅₀ values for reference chemicals tested in the FRTL-5 RAIU assay

Reference Chemicalsa	FRTL-5 IC50 (logM)
Potassium hexafluorophosphate (KPF6)	−8.00 ± 0.11
Potassium perchlorate (KClO4)	−6.77 ± 0.12
Sodium perchlorate (NaClO4)	−6.76 ± 0.09
Sodium tetrafluoroborate (NaBF4)	−6.12 ± 0.06
Sodium thiocyanate (NaSCN)	−4.90 ± 0.07
Sodium nitrate (NaNO3)	−3.00 ± 0.05
Sodium bromide (NaBr)	−1.75 ± 0.05
Sodium fluoride (NaF)	Inactiveb
Sodium carbonate (Na2CO3)	Inactive
Sodium sulfite (Na2SO3)	Inactive

Results from 3 experimental bioreplicates with mean ± SD.

^aConcentration range: −9 to −4 logM (KPF₆, KClO₄, NaClO₄, NaBF₄, NaSCN, Na₂CO₃, Na₂SO₃); −6 to −1 logM (NaNO₃); −4 to −0.3 logM (NaBr)

^bChemical inactive up to the limit of cytotoxicity at 1×10⁻²M (−2 logM).

FRTL-5 and hNIS RAIU assay performance

The performance of the FRTL-5 RAIU assay was monitored per assay plate throughout the entire study using several indicators including: CV of DMSO control activity, Z’ value, and responses of positive and negative control chemicals. Across all the FRTL-5 RAIU assay plates tested in this study, the variability in iodide uptake was low in DMSO controls with mean CV of 5.18 ± 2.68. The Z’ value, a commonly used measure of assay quality, averaged 0.84 ± 0.08, indicating suitable dynamic range and acceptable variability. The IC₅₀ of the positive control NaClO₄ averaged −6.83(logM) with a small standard deviation of 0.17, a value that is consistent with previously reported data (Hallinger et al., 2016; Wang et al., 2018, 2019). The hNIS RAIU assay also demonstrated good performance comparable to our previous high-throughput screening studies with mean DMSO CV of 4.52 ± 2.1 and mean Z’ of 0.86 ± 0.06. Overall, summary statistics of assay performance for the FRTL-5 and hNIS RAIU assays indicated good reproducibility and reliability.

Evaluation of test chemicals in FRTL-5 and hNIS assays

A total of 29 test chemicals were evaluated in both the FRTL-5 and hNIS based assays. The IC₅₀ obtained from the respective RAIU and cytotoxicity assays are presented in Table 3. Dose-response curves for test chemicals are provided in SI file 1. The hNIS RAIU responses in the current study demonstrated strong concordance (Table 3) with hNIS from the previous screening based on IC₅₀ for RAIU and cell viability assays reported by Wang et al., (2018, 2019). However, several chemicals in the current hNIS assay including cyprodinil, captan, fluroxypyr-methyl, and phenolphthalein, were more sensitive to cytotoxicity as compared to our prior screening (Wang et al., 2019).

Table 3.

IC₅₀ (logM) values for all test chemicals evaluated in FRTL-5 and hNIS cell lines

Test chemicals	CAS #	IC50 (logM)
		FRTL5 Cytotox	FRTL5 RAIU	hNIS Cytotox	hNIS RAIU	hNIS RAIU (absEC50 from Wang et al. 2018, 2019)
Etoxazole	153233–91-1	a	−5.50	a	−5.87	−5.88
Niclosamide	50–65-7	−4.07	−6.04	−3.03	−7.31	−6.79
3-Iodo-2-propynyl-N-butylcarbamate b	55406–53-6	−4.36	−5.03	a	−5.51	−5.09
Rotenone	83–79-4	-	−4.09	a	−6.26	−6.22
Cyprodinil	121552–61-2	−4.08	−5.55	−4.22	−4.77	−4.43
Methoxyfenozide	161050–58-4	-	−5.66	-	−5.55	−4.72
Oxyfluorfen	42874–03-3	-	−5.66	-	−5.72	−4.39
Triphenyltin hydroxide	76–87-9	−5.40	−6.08	−3.62	−5.93	−5.39
Pyridaben	96489–71-3	−4.39	−4.10	a	−6.35	−7.49
Captan	133–06-2	−5.12	−5.15	−4.64	−5.06	−4.50
Fluroxypyr-meptyl	81406–37-3	−4.60	−4.82	−4.39	−5.39	−4.74
Fipronil	120068–37-3	−4.71	−4.83	−4.37	−5.12	−4.63
2-(thiocyanomethylthio) benzothiazole	21564–17-0	−5.01	−5.65	−4.48	−5.33	−4.37
PFOS (free acid)	1763–23-1	a	−6.45	a	−5.87	−4.74
PFOS-K	2795–39-3	a	−6.08	a	−5.33	−4.70
PFHxS-K	3871–99-6	a	−5.70	-	−5.42	−4.77
Phenolphthalein	77–09-8	−4.51	−4.80	−4.39	−6.42	−6.13
Tributyltin chloride	1461–22-9	−5.88	−6.48	−5.55	−7.18	−5.60
Benz(a)anthracene	56–55-3	-	−3.97	-	a	−4.54
Triclocarban	101–20-2	−4.01	−6.13	a	−5.25	−5.11
Mepanipyrim	110235–47-7	-	−5.21	-	−4.83	−4.27
4-chloro-1,2-diaminobenzene	95–83-0	a	−5.51	-	a	−4.77
Digitoxin	71–63-6	-	−3.13	-	−6.74
Ouabain, octahydrate	11018–89-6	-	−3.17	-	−6.57
Digoxin	20830–75-5	-	−3.23	-	−6.20
PFOA-ammonium	3825–26-1	a	−4.52	-	−4.15
PFOA (free acid)	335–67-1	-	−4.60	-	−4.12
6-Propyl-2-thiouracil (PTU)	51–52-5	-	-	-	-
2-Mercapto-1-methylimidazole (MMI)	60–56-0	-	-	-	-

Results from at least 3 experiments

^-:inactive (inhibition <20% threshold)

^a:active inhibition with no reportable IC₅₀ (inhibition at maximum concentration >20% but < 50%)

^b:Confirmed interference by iodide content in the chemical (Hornung et. al., 2018; Wang et al., 2018)

When comparing the RAIU IC₅₀ of test chemicals evaluated by the FRTL-5 and the current hNIS assays (Fig. 1), most chemicals demonstrated similar potencies. Twenty five of the 29 test chemicals produced RAIU IC₅₀ values in both assays (i.e., inhibition at the maximum concentration was ≥ 50%), and 18 of them showed less than 1 order of magnitude difference in IC₅₀ between the two RAIU assays, indicating good agreement between the two assays. Seven chemicals including niclosamide, phenolphthalein, digoxin, digitoxin, ouabain, rotenone, and pyridaben showed differences of greater than 1 order of magnitude in RAIU IC₅₀ for the two assays. While inhibition of NIS activity was observed for both 4-chloro-1,2-diaminobenzene and benz(a)anthracene, no IC₅₀ was reported for the hNIS RAIU assay as the inhibition at the maximum concentration was < 50%. The two well-known thyroid peroxidase (TPO) inhibitors, PTU and MMI, demonstrated no inhibitory activity (<20% inhibition) in either the FRTL-5 or hNIS RAIU assays (Table 3 and SI file 1).

Figure 1.

Comparison of RAIU IC₅₀ (logM) responses between FRTL-5 and hNIS assays for test chemicals. Dots and triangles represent the IC₅₀ values, with the difference between two assays indicated by horizontal lines. Missing triangle symbols for 4-chloro-1,2-diaminobenzene and benz(a)anthracene indicate no reportable IC₅₀ values when tested in the hNIS assay. PTU and MMI were inactive in both assays.

Six test chemicals, etoxazole, niclosamide, methoxyfenozide, oxyfluorfen, mepanipyrim and triclocarban demonstrated relatively potent RAIU inhibition (IC₅₀ < 1×10⁻⁵ M) in both cell lines with moderate to low cytotoxicity (Fig. 2). With the exception of triclocarban, previously identified as a NIS inhibitor using FRTL-5 based colorimetric iodide uptake assay (Wu et al. 2016), these additional 5 chemicals were novel findings in the FRTL-5 assay. Methoxyfenozide, oxyfluorfen and mepanipyrim all exhibited full sigmoidal RAIU response curves with no cytotoxicity in both cell lines. Overall, the consistent RAIU effects of these six chemicals in both hNIS and FRTL-5 cell lines, in addition to their moderate to low cytotoxicity, render these compounds higher priority for further in vivo and mechanistic studies.

Figure 2.

Selected chemicals that exhibit strong RAIU inhibition with moderate to low cytotoxicity (CellTiter-Glo) in both FRTL-5 and hNIS assays. All data are presented as medians with upper and lower ranges for 3 experimental bioreplicates of each assay. The red line indicates 20% inhibition level.

Five common perfluorinated chemicals PFOS, PFOS-K, PFOA, PFOA-ammonium, and PFHxS-K were also evaluated (Figure 3). The FRTL-5 cell line exhibited slightly higher sensitivity than the hNIS cell line for each of these compounds, yielding marginally lower RAIU IC₅₀ values. In addition, all five compounds demonstrated little to no cytotoxicity in both cell lines. The free acid and salt forms for PFOS and PFOA yielded similar responses in both the FRTL-5 and hNIS assays, suggesting the acidity of the PFOS and PFOA had minimal effect on the RAIU assay. PFOA and PFOA-ammonium demonstrated the weakest potencies, with RAIU IC₅₀ in the range of −4.60 to −4.12 (logM). PFOS, PFOS-K, and PFHxS-K all demonstrated full dose-response curves in both cell lines, producing RAIU IC₅₀ values in the range of −5.42 to −6.45 (logM), which were very close to that of sodium perchlorate (−6.76 in FRTL-5). Considering the relatively strong potency of NIS inhibition demonstrated by these five perfluorinated compounds and previous reports of these compounds on thyroid hormone disruption (Lau et al. 2007), further study into the in vivo mechanism is warranted. Another compound, sodium fluoride (NaF), that has been reported as having a positive association with thyroid hormone disruption in several epidemiology studies (Kheradpisheh et al. 2018; Malin et al. 2018) was also tested in this study. NaF demonstrated no NIS inhibition until tested at 1×10⁻² M; a concentration that also produced strong confounding cytotoxicity. This result is consistent with a previous FRTL-5 study that reported NaF as inactive for NIS inhibition (Waltz et al. 2010).

Figure 3.

Dose-response curves for five PFAS chemicals evaluated in FRTL-5 and hNIS assays. All data are presented as medians with upper and lower ranges. The red line indicates 20% inhibition level. Cytotoxicity data were obtained using the CellTiter-Glo assay.

As the normal function of NIS depends on the cross-membrane sodium gradient maintained by the Na⁺/K⁺-ATPase (Carrasco 1993), our previous studies (Hallinger et al. 2017; Wang et al. 2018) demonstrated that the hNIS RAIU assay is sensitive to chemicals that disrupt the function of the Na⁺/K⁺-ATPase through either an indirect inhibition of mitochondria function (e.g., rotenone, pyridaben) (Barrientos and Moraes 1999; Navarro et al. 2010), or direct inhibition of Na⁺/K⁺ ATPase (e.g. ouabain). Therefore, these three chemicals, in addition to two other known ATPase inhibitors (digoxin and digitoxin), were tested in the two RAIU assays (Fig. 3). The three Na⁺/K⁺-ATPase inhibitors (ouabain, digoxin, and digitoxin) demonstrated very large differences in potencies between the two cell lines, with greater than 2 orders of magnitude difference in RAIU IC₅₀. The hNIS cell line produced similar responses to the three Na⁺/K⁺-ATPase inhibitors, with IC₅₀ in the range of −6.20 to −6.74 (logM). In comparison, the FRTL-5 cell line did not show RAIU inhibition unless tested at chemical concentrations greater than 1×10⁻⁴ M (IC₅₀ between −3.13 to −3.23, logM), suggesting the FRTL-5 RAIU assay is much less sensitive to the influence of Na⁺/K⁺-ATPase inhibitors. One reason for this difference between the cell lines in response to Na⁺/K⁺-ATPase inhibitors may be related to the structural difference of Na⁺/K⁺-ATPase protein in the rat compared to human (Chanda et al. 2008). A large difference of responses between the two cell lines was also observed for pyridaben and rotenone, pesticides known to inhibit mitochondria function (Navarro et al. 2010). Both pyridaben and rotenone demonstrate atypical dose-response curves in the hNIS cell line (Figure 4) with significant RAIU inhibition starting at 1×10⁻⁸ M; a finding that is in agreement with our previous screening results (Wang et al., 2018). In contrast, the change in RAIU observed in the FRTL-5 cell line occurred at the same concentration as that of cytotoxicity.

Figure 4.

Dose-response curves of test chemicals for RAIU and cell viability assays that (A) directly inhibit Na⁺/K⁺-ATPase function or (B) inhibit mitochondria ATP production. All data are presented as medians with upper and lower ranges. The red line indicates 20% inhibition level. Cytotoxicity data was obtained from the CellTiter-Glo assay. Note: To detect ATPase inhibition, higher concentrations were tested for FRTL-5 cells (−6 to −3 logM) versus hNIS (−8 to −4 logM).

Chemical cytotoxicity in FRTL-5 assay

Chemical-induced adverse effects on cell viability following the 2 h exposure period were evaluated to filter out potential false positives. As a part of our routine screening, all RAIU assays were conducted in parallel with a Cell Titer-Glo cytotoxicity assay that measures a change in cellular ATP level as an indicator of cell health. An orthogonal propidium iodide (PI) dead-cell viability assay for the FRTL-5 cell line, which uses PI nuclear staining to indicate permeable membranes or compromised cells, was included. Surrogate cell counts (Hoechst-stained nuclei) in each test well were also obtained to quantitate the percentage of PI staining per well and total plate adherence of the cell population. The PI, cell counts, RAIU, and Cell Titer-Glo cytotoxicity data for all FRTL-5 tested chemicals are shown in SI file 2.

For most chemicals, both the PI and cell count data were consistent with the results observed using the CellTiter-Glo assay. This was the case for the chemicals with the strongest RAIU inhibitory responses, such as methoxyfenozide, oxyfluorfen, and the perfluorinated chemicals, where results of the PI and cell count data further confirmed that NIS inhibition was not likely caused by loss of cells or compromised membrane integrity. For other chemicals, the additional viability assay data pointed to mitochondrial toxicity as a mechanism explaining the decrease in ATP levels from the Cell Titer-Glo assay. For example, pyridaben and fipronil both have been reported to inhibit mitochondrial function (Navarro et al. 2010; Tavares et al. 2015), and in our study, each displayed identical concentration-response curves showing significant NIS inhibition and reduction of ATP with no significant PI or cell count changes. Overall, our results support the continued use of the CellTiter-Glo assay for high throughput NIS inhibitor screening to cover a broad range of cellular mechanisms of cytotoxicity.

Conclusion

This study demonstrated the application and integration of the FRTL-5 based secondary RAIU assay as part of the in vitro NIS inhibitor identification workflow. The FRTL-5 assay exhibited excellent reproducibility, reliability, and sensitivity comparable to the hNIS assay in testing the 10 reference chemicals. Evaluation of 29 newly acquired environmental organic compounds that were highly active in our previous ToxCast screening in both the FRTL-5 and hNIS RAIU assays further confirms and provides additional weight of evidence for the inhibitory activity of these compounds to the rat and human NIS. Eighteen of 29 test chemicals showed less than 1 order of magnitude difference in IC₅₀ values between the two assays. Several chemicals including etoxazole, methoxyfenozide, oxyfluorfen, triclocarban, mepanipyrim, and niclosamide displayed relatively potent RAIU inhibition in both the FRTL-5 and hNIS assays with minimal cytotoxicity. Five common PFAS also demonstrated strong RAIU inhibition with nominal cytotoxicity and good concordance between these two assays, with PFOS and PFHxS exhibiting the most potent IC₅₀ similar to that observed for sodium perchlorate. The CellTiter-Glo viability assay, as an integral part of the NIS inhibitor identification workflow, was compared with the PI cytotoxicity assay for all test chemicals using the FRTL-5 cell line and was concluded as sufficient and reliable to gauge various types of chemical cytotoxicity. Overall, our results have demonstrated the advantages of including the FRTL-5 assay into the NIS inhibitor identification workflow by confirming NIS inhibition activity for over 20 of the highest ranked organic environmental chemicals. These results further refine the prioritization of chemicals for additional testing such as short-term in vivo assays to characterize the impact on thyroid hormone synthesis. The identified chemical activity will also provide critical information to facilitate structure-activity investigation for NIS inhibition, as well as the development of adverse outcome pathways (AOP) and OECD test guideline.

Supplementary Material

Sup1

SI file1 1: Dose-response curves of test chemicals evaluated in FRTL-5 and hNIS assays

Sup2

SI file1 2: Dose-response of propidium iodide cytotoxicity and cell counts in FRTL5