Cytotoxicity evaluation and mechanism of endocrine-disrupting chemicals by the embryoid body test

Abstract

Endocrine-disrupting chemicals (EDCs) are a structurally diverse class of synthetic and natural compounds. EDCs can cause non-communicable diseases such as obesity, type 2 diabetes, thyroid disorders, neurodevelopmental disease, hormone-dependent cancers, and reproductive disorders. The embryoid body test (EBT) is a developmental toxicity test method that determines the size of embryoid bodies (EBs) and the viability of mouse embryonic stem cells (mESCs) and fibroblasts (3T3 cells). The present study used the EBT to perform cytotoxicity evaluations of 10 EDCs and assessed the mechanistic relationship between endoplasmic reticulum (ER) stress and cytotoxicity. According to the statistical analysis and prediction model results, methylparaben, butylparaben, propylparaben, ethylparaben, triclosan, octylphenol, methoxychlor, bisphenol A, and diethylstilbestrol were classified as cytotoxic, but trichloroacetic acid was non-toxic. Classification accuracy was 90%. The mechanistic study showed that the cytotoxicities of butylparaben, propylparaben, octylphenol, and triclosan were induced by ER stress. The mRNA expressions of BiP, CHOP, and ATF4 were significantly higher following treatments with four EDCs compared to those after the control treatment. Compared to the control treatment, the mRNA levels of XBP1u and XBP1s increased significantly after butylparaben and propylparaben treatments, but did not increase with octylphenol and triclosan treatments. These results indicate that the EBT can be applied as an alternative toxicity test when evaluating the cytotoxicity of EDCs.

Introduction

Toxicity tests are necessary for assessing the safety of or risk to an organism exposed to various chemicals. The embryoid body test (EBT) is a developmental toxicity test method that allows measurement of the size of embryoid bodies (EBs) and determination of the viability of mouse embryonic stem cells (mESCs) and fibroblasts (3T3 cells). The EBT method of determining embryonic body size is distinct from that of the embryonic stem cell test (EST), which assesses the beating behavior of cardiomyocytes. The EBT is important because embryoid bodies have three germ layers in the early stages of development and the toxic effects on all three layers can be determined. Although the EST requires about 10 days or more to determine toxicity, the EBT can be performed in only 4 days [1–5]. By applying the EBT, Hong et al. [2] observed that cell damage caused by toxicity in mouse EB is associated with a reduction in EB size through necrotic cell death and damage to differentiation by apoptosis. Also, that report confirmed that the mouse EB can be used as a screening tool to assess the toxicity characteristics of pharmacological drugs. Kang et al. [3] assessed the developmental toxicity of 21 chemicals and showed 90.5% accuracy of toxicity predictions. Lee et al. [4, 5] updated a toxicity prediction model using toxicity classification results for 129 chemicals and analyzed the second-phase study results by applying a newly updated prediction model.

Endocrine-disrupting chemicals (EDCs) are a structurally diverse class of synthetic and natural compounds. EDCs interfere with the endocrine system by mimicking hormones, blocking their effects, or interfering with their synthesis or excretion [6, 7]. Such disruptions can result in non-communicable diseases such as obesity, type 2 diabetes, thyroid disorders, neurodevelopmental disease, hormone-dependent cancers, and reproductive disorders [8]. EDCs act through receptor-mediated and non-receptor-mediated mechanisms to regulate different components of the endocrine system. EDCs are typically identified as antagonists or agonists of endogenous hormone receptors. However, EDCs be able to act through multiple mechanisms that affect hormone synthesis, transport, and metabolism [9]. Therefore, the present study used the EBT to evaluate the cytotoxicity of EDCs and assessed the mechanistic relationship between the endoplasmic reticulum (ER) stress and cytotoxicity.

Materials and methods

Endocrine-disrupting chemicals

All chemicals were purchased from Sigma-Aldrich (St. Louis, MO, USA) and were prepared in dimethyl sulfoxide (DMSO): methylparaben, butylparaben, propylparaben, ethylparaben, triclosan, octylphenol, methoxychlor, bisphenol A, diethylstilbestrol, and trichloroacetic acid.

Culture for ES-E14TG2a and 3T3 cells

The mouse ESCs (ES-E14TG2a) and 3T3 cells (Clone A31) were purchased from the American Type Culture Collection (ATCC; Manassas, VA, USA). The mESCs were cultured on mitomycin C-treated mouse embryonic fibroblasts (mEFs) at 37 °C in a 5% CO₂ humidified incubator. The culture medium consisted of DMEM/F-12 (Gibco-BRL, Logan, UT, USA) supplemented with 10% heat-inactivated certified fetal bovine serum (FBS; Gibco-BRL), non-essential amino acids (NEAA, 1X; Gibco-BRL), 2-mercaptoethanol (10^–4 M), penicillin (100 U/mL), streptomycin (100 µg/mL), and mouse leukemia inhibitory factor (mLIF, 10 ng/mL; Millipore, Darmstadt, Germany). The 3T3 cells were cultured in DMEM (Gibco-BRL) supplemented with 10% heat-inactivated FBS, penicillin (100 U/mL), and streptomycin (100 µg/mL).

Three measurement endpoints

Cells in an undifferentiated condition (mES) were treated for four days with EDCs. Cell survival rates were determined by CCK assay and the IC₅₀ (half-maximal inhibitory concentration) values were calculated. Differentiated 3T3 cells were treated for four days with EDCs. The cell survival rates were determined by CCK assay and the IC₅₀ values were calculated. Cells in the EB-forming stage were treated for three days with various concentrations of EDCs, after which the sizes of EBs were measured and the ID₅₀ (half inhibition concentration for embryoid bodies area) values were calculated.

Cell viability and embryoid bodies area measurement

The IC₅₀ values of mES cells and 3T3 cells were measured by CCK assay. 50 µL of cell suspension (500 cells/well) was plated into wells of a 96-well plate. Two hours after cell seeding, 150 µL of culture medium was added to each well with 1.333 times the final concentration of EDCs. The treated cells were grown in the culture medium for 4 days. After 4 days, the medium of the chemical-treated plate was removed and washed twice with PBS. Medium with CCK solution (Dojindo Laboratories, Tokyo, Japan) added, and cells were incubated for 1 h. Absorbance is measured at 450 nm by using Epoch Microplate Spectrophotometer (BioTek Instruments, Winooski, VT, USA). The IC₅₀ values were calculated using Excel and GraphPad Prism 6.01 (GraphPad Prism Software, San Diego, CA, USA). The percentage of cell viability was determined from the absorbance of each well relative to the absorbance of the control group (DMSO 0.1—0.5% or media) which was set at 100%.

For the formation of EBs, 84 drops with 20 µL of cell suspension per drop (800 cells) were plated on the lid of a 90 mm Petri dish (SPL, Pocheon, South Korea) and cultured as hanging drops with/without chemicals for three days after turning over the lids. d-PBS (7 mL) was added to the bottom dish (prevent drying of the drops). Three days after hang-in drop, the formed EB was transferred to a non-coated Petri dish with a medium. The EB of photographic image files has been converted to black using Image J software (NIH, Bethesda, MD, USA) (8-bit type convert: Image → Adjust → Threshold). After obtaining the number of black pixels, the EB area was calculated to divide the number of black pixels by the number of EBs. The average EB area obtained using excel was calculated as a percentage relative to the mean value of the control group (DMSO 0.1—0.5%) set at 100%. The obtained percentage value was analyzed through GraphPad Prism Software to calculate an ID₅₀.

Statistical analysis and prediction model (PM)

Concentration–response curves were generated for each test chemical and a one-site fit to a three-parameter logistic function was obtained by using GraphPad Prism (v. 6.01). The IC₅₀ E14, IC₅₀ 3T3, and ID₅₀ EB values were used to classify the compounds as non-toxic or toxic. A linear discriminant analysis (LDA) was used to identify endpoints via stepwise discriminant analysis (SPSS) [10]. LDA is a method of classifying given predictive variables into a class with the largest posterior probability. Discriminant analysis approaches assume that the conditional probability density functions are both normally distributed with mean and covariance parameters, respectively. The LDA classifier assumes that the observations in the kth class are drawn from a multivariate Gaussian distribution N which is a class-specific mean vector and is a covariance matrix that is common to all kth classes. By plugging the density function for the kth class, into Bayes’ theorem and performing a little bit of algebra, discriminant function,

$$\begin{array}{c}discriminantfunction=-0.1139733\times {log}_{10}I{C}_{50}(ESC)\\ -0.2303571\times {log}_{10}I{C}_{50}(3T3)-0.6726328\times {log}_{10}I{D}_{50}(EB)+2.723601\\ \end{array}$$

It was assumed that the observations in the toxic (or non-toxic) were drawn from a multivariate Gaussian distribution, where ${\mu }_k$ is a class-specific mean vector and $\mathrm{\Sigma }$ is a common variance–covariance.

For predicting, it is calculated the discriminant function, and then classified using the allocation rule; if the calculated discriminant function value is $\le -0.667$, it is classified as non-toxic, otherwise, it is classified as toxic.

Model performance was evaluated by calculating the accuracy for predicting outcomes.

$$\text{accuracy}\left(\text{\%}\right)=\frac{TP+TN}{TP+FN+TN+FP}\times 100$$

		Predict
		Non-toxic	Toxic
Original	Non-toxic	True negative; TN	False positive; FP
	Toxic	False negative; FN	True positive; TP

True Negative (TN) refers to the accurate classification of non-toxic chemicals as non-toxic chemicals. False Negative (FN) refers to the misclassification of non-toxic chemicals as toxic chemicals. True Positive (TP) refers to the accurate classification of toxic chemicals as toxic chemicals. False Positive (FP) refers to the misclassification of toxic chemicals as non-toxic chemicals.

Total RNA extraction and real-time PCR

Total RNA was extracted from the entire brain of each mice using Trizol reagents (Ambion, Austin, TX, USA) according to the manufacturer’s protocols. cDNA synthesis was performed as previously described [11]. Qantitative real-time PCR analysis was conducted using a QuantStudio 3 (Applied Biosystems, Foster City, CA). GAPDH served as an internal vehicle. Primer sequences were as follows: BiP, forward CTATTCCTGCGTCGGTGTGT, reverse GCAAGAACTTGATGTCCTGCT; CHOP, forward CCTGAGGAGAGAGTGTTCCAG, reverse CTCCTGCAGATCCTCATACCA; ATF4, forward GAGCTTCCTGAACAGCGAAGTG, reverse TGGCCACCTCCAGATAGTCATC; XBP-1u, forward TGAGAACCAGGAGTTAAGAACACGC, reverse CACATAGTCTGAGTGCTG-CGG; XBP-1 s, forward TGAGAACCAGGAGTTAAGAACACGC, reverse CCTGCACCTGCTGC-GGAC; glyceraldehyde-3-phosphate dehydrogenase (GAPDH), forward TGGAAAGCTGTGGCGT-GAT, reverse TGCTTCACCACCTTCTTGAT.

Statistical analysis

All statistical analyses were conducted by using ANOVA followed by Tukey’s test for multiple comparisons. The analysis was performed using the Prism Graph Pad v. 6.01 (Graph Pad Software, San Diego, CA, USA). The values are presented as means ± SEM of at least three separate experiments, in which case a representative result is depicted in the figures. P values < 0.05 were considered statistically significant.

Results

Initially, we performed cytotoxicity evaluations of 10 EDCs by using the EBT. Characterizations of 10 EDCs (methylparaben, butylparaben, propylparaben, ethylparaben, triclosan, octylphenol, methoxychlor, bisphenol A, diethylstilbestrol, and trichloroacetic acid) are presented in Table 1. For testing, EBs were exposed for four days to various concentrations of the 10 EDCs (Fig. 1). The EB size was markedly decreased after exposure to the following EDC concentrations: 3 × 10^–1 mg/mL of methylparaben, butylparaben, propylparaben, and ethylparaben; 3 × 10^–2 mg/mL of triclosan and octylphenol; 3 × 10 mg/mL of trichloroacetic acid; 7.5 × 10^–2 mg/mL of methoxychlor; 1 × 10^–1 mg/mL of diethylstilbestrol; however, bisphenol A did not significantly affect EB size (Fig. 1). Following treatment with nine of the EDCs, the treated EBs appeared uneven and shrunken, indicating that the observed EB size reduction reflects the developmental toxic effects of the EDC.

Table 1.

Information on 10 EDCs including chemical name, molecular weight, in vivo class, and uses

No	Chemicals	MW	In vivo class (Reference)	Uses
1	Methylparaben	152.15	Toxicity (Brand, Petric)	Preservative & antiseptics
2	Butylparaben	194.23	Toxicity Petric
3	Propylparaben	180.2	Toxicity Brand, Petric
4	Ethylparaben	166.17	Toxicity Brand, Petric
5	Triclosan	289.54	Toxicity Yueh
6	Octylphenol	206.32	Toxicity (Bøgh)	Intermediate in the production of phenolic resins
7	Trichloroacetic acid	163.39	Toxicity (EPA, 2011)	Cosmetic treatments, e.g. chemical peels and tattoo removal
8	Methoxychlor	345.65	Toxicity	Pesticides
9	Bisphenol A	364.43	Toxicity	Plastics
10	Diethylstilbestrol	268.35	Toxicity	Nonsteroidal estrogen medication

Fig. 1.

Reduction in EB size according to the toxicity of EDCs. The EBs were formed for four days via the hanging drop method with various concentrations of EDCs. The scale bar means 100 μm

To examine further the cytotoxicity of the 10 EDCs, undifferentiated mESCs (ES-E14TG2a), 3T3 cells, and EBs underwent EBTs after being treated with various concentrations of the EDCs. Three half-maximal inhibitory concentrations (IC₅₀ E14, IC₅₀ 3T3, and ID₅₀ EB) were obtained from the sigmoidal curves, indicating the EDCs had dose-dependent cytotoxic effects (Fig. 2, Table 2). Based on the statistical analysis and prediction model results, nine of the EDCs (methylparaben, butylparaben, propylparaben, ethylparaben, triclosan, octylphenol, methoxychlor, bisphenol A, and diethylstilbestrol) were classified as toxic, and one, trichloroacetic acid, was classified as non-toxic (Table 3). The classification accuracy of the predictive model was 90%.

Fig. 2.

Concentration–response curves for EDCs. Cell survival of undifferentiated ESCs (A) and 3T3 cells (B) and EB area (C) curves were obtained with various EDCs in a culture medium after chemical exposures. The endpoint “viability and EB size” are based on the reduction of the substance resazurin. The coefficient of determination (R²) was indicated in each graph. The values represent the average ± SEM of three separate experiments

Table 2.

Summary of three endpoints values

No	Chemicals	IC50 E14 [mg/mL]	IC50 3T3 [mg/mL]	IC50 EB [mg/mL]
1	Methylparaben	0.08052	0.04901	0.2395
2	Butylparaben	0.01204	0.004106	0.03442
3	Propylparaben	0.02070	0.005444	0.03586
4	Ethylparaben	0.04369	0.01824	0.1076
5	Triclosan	0.01034	0.007082	0.01648
6	Octylphenol	0.005101	0.007801	0.01696
7	Trichloroacetic acid	0.6842	0.3940	0.4041
8	Methoxychlor	0.02086	0.03913	0.05988
9	Bisphenol A	0.01493	0.01549	0.06003
10	Diethylstilbestrol	0.006221	0.009183	0.002299

Table 3.

Evaluation classifications by the predictive model

No	Chemicals	In vivo class	Predictive model class
1	Methylparaben	Toxicity	Toxicity
2	Butylparaben	Toxicity	Toxicity
3	Propylparaben	Toxicity	Toxicity
4	Ethylparaben	Toxicity	Toxicity
5	Triclosan	Toxicity	Toxicity
6	Octylphenol	Toxicity	Toxicity
7	Trichloroacetic acid	Toxicity	Non-toxicity
8	Methoxychlor	Toxicity	Toxicity
9	Bisphenol A	Toxicity	Toxicity
10	Diethylstilbestrol	Toxicity	Toxicity

Next, we investigated the cytotoxic mechanism associated with the effects of ER stress on four EDCs (butylparaben, propylparaben, octylphenol, and triclosan). The assessed ER stress markers were binding immunoglobulin protein (BiP), C/EBP homologous (CHOP), activating transcription factor 4 (ATF4), X-box-binding protein 1 unspliced (XBP-1u), and X-box-binding protein 1 spliced (XBP-1 s). Compared to the control levels, the mRNA expression levels of BiP, CHOP, and ATF4 increased significantly following treatment by each of the four EDCs. The mRNA levels of XBP-1u and XBP-1 s increased significantly from the control level with butylparaben and propylparaben treatments, but did not change significantly with octylphenol and triclosan treatments (Fig. 3). Together, these results can be applied to the use of EBT as an alternative toxicity test to evaluate the cytotoxicity of other EDCs.

Fig. 3.

mRNA levels of ER stress-related genes were assessed in EBs lysates by real-time PCR. mRNA levels were normalized to the GAPDH level. Data shown are the means ± SEMs and were analyzed by one-way ANOVAs with Bonferroni’s correction. *p < 0.05, **p < 0.01, ***p < 0.001 versus control

Discussion

The predictive capacity of the EBT to estimate the cytotoxic effects of 10 EDCs was assessed, and the toxicity of those EDCs was evaluated (Tables 2, 3). The EDCs used in this study were substances to which human and animal life are often exposed. Parabens are a group of chemicals that are commonly used as preservative ingredients in cosmetic products, personal hygiene products, food products, and pharmaceuticals [12, 13]. Triclosan is an antimicrobial widely used for personal hygiene and household products such as mouthwash, detergents, soaps, deodorants, toothpaste, and disinfecting lotions [14, 15]. Octylphenol is used in rubber, pesticides, and paints [16], methoxychlor is used as an insecticide [17], bisphenol A is used to make polycarbonate plastic [18], and diethylstilbestrol is used as a nonsteroidal estrogen medication [19]. Thus, EDCs are widely used in products related to the daily life of humans and animals, and they are well known for their capacity to disturb the endocrine system [20, 21]. The EBT results showed that 9 EDCs (methylparaben, butylparaben, propylparaben, ethylparaben, triclosan, octylphenol, methoxychlor, bisphenol A, and diethylstilbestrol), which are toxic in vivo, were accurately evaluated as toxic, whereas trichloroacetic acid, which is toxic in vivo, was assessed to be non-toxic.

Trichloroacetic acid is used in various medicinal products and organic chemicals and is used as a soil sterilizer. Industrially, trichloroacetic acid-based chemicals are used as solvents in the plastics industry and as etching and pickling agents for the surface treatment of metals. Trichloroacetic acid is also used in medical applications as an antiseptic for albumin detection and as a skin peeling agent [22, 23]. Trichloroacetic acid at concentrations of 15–35% has been used in skin peeling treatments to treat conditions such as actinic damage, scars, wrinkles, and abnormal skin pigmentation. Concentrations that are 45% or more are associated with an increased risk of scarring [24, 25]. Animal studies show that trichloroacetic acid-induced skin damage is characterized by an early inflammatory response, epidermal loss, and collagen degeneration [26]. Adverse side effects in patients receiving the trichloroacetic acid-based skin peel procedure have shown infection, transient hyperpigmentation, persistent erythema, acne or cyst formation, keratoacanthomas, and fine crusting [24, 25]. However, in vitro model results may not reflect toxicity accurately because they do not assess in vivo metabolic activities. Also, when trichloroacetic acid is administered to dogs, mice, rats, or humans, it may be toxic or non-toxic depending on the concentration administered and the duration in the body. In this study, trichloroacetic acid was classified as non-toxic based on the EBT-based cytotoxicity evaluation.

Recently, various reports have suggested that the biological mechanism related to ER stress is upregulated by EDCs exposure. The mRNA and protein levels associated with ER stress undergo abnormal changes in various species, including mice and rats (27–29). The present study indicates the possibility that monitoring the expression of an ER stress gene could be useful in the assessment and analysis of EDC cytotoxicity in the laboratory. Furthermore, it is crucial to consider the relationship between EBT results and ER stress upon EDC exposure. Overall, the results of the EBT for EDC cytotoxicity evaluation indicated that stem cells could be useful when evaluating the toxicity of various chemicals, including EDCs, in industrial situations. However, the EDCs in this study can produce abnormal disruptions during early embryonic development, which is a critical problem. The EBT can be as good an indicator of EDC toxicity as those of other in vitro or in vivo sources such as OECD guidelines and alternative developmental toxicity tests. Finally, the use of stem cell-based tests to evaluate EDC toxicity needs further study using numerous EDCs in order to determine their biological and chemical significance.

This study demonstrates that the EBT may be used to evaluate the cytotoxicity of EDCs. Increased expression levels of ER stress-related genes were observed following EDC treatment. Our results suggest that the EBT can be used to evaluate the cytotoxicity of EDCs. Further study will focus on elucidating the mechanism through which EDCs increase the expression levels of ER stress-related genes.

Author contributions

E-MJ, Y-MY and E-BJ conceived the study, designed, performed and analyzed the experiments, and wrote the paper. J-HL performed the experiments. E-BJ supervised the study.

Funding

This research was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2021R1A2C2093275) and supported by Global Research and Development Center (GRDC) Program through the NRF funded by the Ministry of Education, Science and Technology (2017K1A4A3014959).

Declarations

Conflict of interest

The authors have no conflict of interest to disclose.