Cardiac Development in the Presence of Cadmium: An in Vitro Study Using Human Embryonic Stem Cells and Cardiac Organoids

Abstract

Background:

Exposure to cadmium (Cd) is associated with cardiovascular diseases. Maternal Cd exposure is a significant risk factor for congenital heart disease. However, mechanisms of Cd on developmental cardiotoxicity are not well defined.

Objectives:

We evaluated the effects of Cd on the different stages (mesoderm, cardiac induction, cardiac function) of cardiac development using an early embryo development in vitro model and two- or three-dimensional (2- or 3D) cardiomyocyte and cardiac organoid formation models mimicking early cardiac development.

Methods:

Embryonic stem cells (ESCs) form 3D aggregates, called embryoid bodies, that recapitulate events involved with early embryogenesis (e.g., germ layer formation). This model was used for early germ layer formation and signaling pathway identification. The 2D cardiomyocyte differentiation from the $NKX\mathit{2}-{\mathit{5}}^{eGFP/w}$ human ESCs model was used to explore the effects of Cd exposure on cardiomyocyte formation and to model mesoderm differentiation and cardiac induction, allowing us to explore different developmental windows of Cd toxicity. The 3D cardiac organoid model was used in evaluating the effects of Cd exposure on contractility and cardiac development.

Results:

Cd ($0.6\mathrm{\mu }\mathrm{M}$; $110\text{\hspace{0.17em}ppb}$) lowered the differentiation of embryoid bodies to mesoderm via suppression of $\text{Wnt}/\mathrm{\beta }\text{-catenin}$–signaling pathways. During early mesoderm induction, the mesoderm-associated transcription factors MESP1 and EOMES showed a transient up-regulation, which decreased later in the cardiac induction stage. Cd ($0.15\mathrm{\mu }\mathrm{M}$) lowered mesoderm formation and cardiac induction through suppression of the transcription factors and mesoderm marker genes HAND1, SNAI2, HOPX, and the cardiac-specific genes NKX2-5, GATA4, troponin T, and $\text{alpha}\text{-actinin}$. In addition, Cd-induced histone modifications for both gene activation (H3K4me3) and repression (H3K27me3), which play vital roles in regulating mesoderm commitment markers. The effects of Cd inhibition on cardiomyocyte differentiation were confirmed in 3D cardiac organoids.

Discussion:

In conclusion, using a human ESC-derived 2D/3D in vitro differentiation model system and cardiac organoids, we demonstrated that low-dose Cd suppressed mesoderm formation through mesoderm gene histone modification, thus inhibiting cardiomyocyte differentiation and cardiac induction. The studies provide valuable insights into cellular events and molecular mechanisms associated with Cd-induced congenital heart disease. https://doi.org/10.1289/EHP11208

Introduction

Congenital heart disease (CHD) occurs in $\sim 1\%$ of newborns in the United States.^1,2 This disease can arise from many distinct etiologies that elicit diverse cellular and molecular responses during cardiac development.³ Many chemicals remain untested for their potential developmental toxicological effects. Overwhelming evidence indicates that environmental factors play a role in the development and progression of CHD.^4,5 As an alternative to primary cardiomyocytes and myocardial tissue, embryonic stem cell (ESC)–⁶ or induced pluripotent stem cell (iPSC)–derived cardiomyocytes⁷ and three-dimensional (3D) cardiac organoids⁸ have emerged as physiologically relevant models for in vitro cardiotoxicity screening. However, ESC/iPSC-derived cells can only mimic the chemical-associated cardiophysiologic phenotypes, such as positive and negative inotropic effects and prolonged repolarization.⁹ Therefore, these differentiated cardiomyocyte models are not suitable for examining the chemical effects initiated during the early developmental stages of heart development, such as during the formation of cardiomyocytes.

ESCs offer an excellent opportunity to study the mechanisms of developmental toxicity.¹⁰ In vitro cardiomyocyte differentiation models have defined stages of mesoderm induction and cardiac induction, followed by differentiation to cardiomyocytes, providing a convenient tool for exploring the molecular and cellular events involved in developmental cardiotoxicity. The ESC test has been used in combination with quantitative expression analysis of cardiac-specific genes as a model for predicting the inhibition of cardiac differentiation and cytotoxicity.¹¹ 13-cis-Retinoic acid, a skin acne drug, has been shown to disrupt mesoderm formation during cardiac differentiation using human ESCs (hESCs) and iPSCs.¹² In other studies, mouse ESCs were differentiated into cardiomyocytes to investigate the association between common antiepileptic drugs and developmental cardiotoxicity and congenital cardiovascular defects.¹³ Arsenic trioxide exposure altered cardiomyocyte marker gene expression and led to the reduction of cardiomyocyte sarcomeric proteins, as well as reducing beat frequency.¹⁴

NKX2-5 is a homeodomain transcription factor of the homeobox 1 gene (NKX2) that is expressed during early cardiac morphogenesis.¹⁵ Together with GATA-binding protein 4 (GATA4), NKX2-5 acts as a transcriptional activator of natriuretic peptide A (NPPA/ANF), two early specific markers of differentiating myocardium of the developing heart.¹⁶ In addition, NKX2-5 serves as a master regulatory protein during heart development and in the mature heart throughout life.¹⁷ Patients with various congenital heart malformations—such as atrial and ventricular defects, conotruncal abnormalities, and atrioventricular defects—have been found to have mutations in NKX2-5.¹⁸ A hESC line containing an NKX2-5 reporter is an ideal model system for investigating developmental cardiotoxicity. This model allows for real-time monitoring of the differentiation of cardiac cells from hESCs, the purification of hESC-committed progenitor cells (hESC-CPCs) and cardiomyocytes (hESC-CMs) derived from hESCs, and the standardization of differentiation protocols.¹⁹

Cadmium (Cd) exposure can give rise to a variety of developmental defects. These exposures have been associated with coronary heart disease and cardiovascular diseases, including hypertension, atherosclerosis, increased oxidative stress, and endothelial dysfunction in a transgenic mouse study.²⁰ Other mouse models have demonstrated that maternal/prenatal Cd exposure caused differential expression of genes and pathways associated with heart development and hypertension in the hearts of newborns²¹ and altered the expression of genes that affect cardiomyocyte proliferation and differentiation, including GATA4.²² Zebrafish exposed to Cd during development had significantly lower heart rates during development²³ and as adults.²⁴ In a group of Chinese patients, maternal exposure to Cd was a significant risk factor for CHDs in the offspring.²⁵ Another recent epidemiological study suggested that maternal Cd exposure could affect body mass index, hypertension, and atrial fibrillation in newborns and that epigenetic changes (i.e., DNA methylation) may play a role in in these Cd-associated disease outcomes.²⁶ However, there is little direct evidence of cellular and molecular events in Cd-induced CHD in humans owing to the lack of appropriate in vitro models. In the present study, we aimed to investigate the key developmental stage (mesoderm formation and cardiac induction) and molecular target (e.g., epigenetic marker) of Cd exposure on cardiomyocytes during cardiac development using an in vitro 2D cardiomyocyte differentiation quantification model and 3D cardiac organoid contractility model.

Materials and Methods

Cell Maintenance and Differentiation

$\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were maintained in StemFlex medium (A3349401; ThermoFisher Scientific) until 80%–90% confluent for cell passaging. $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were a kind gift from E.G. Stanely and were created through the insertion of sequences encoding enhanced green fluorescent protein (GFP), or eGFP, into the NKX2-5 locus of the hESCs by homologous recombination.¹⁹ Cardiomyocyte differentiation was performed in 96-well plates (353219; Corning) using a monolayer differentiation method with a STEMdiff Cardiomyocyte Differentiation Kit (05010; STEMCELL Technologies Inc.), as described in the protocol. This differentiation to cardiomyocytes in this protocol includes the mesoderm formation and cardiac induction stages. Typically, on day 0 (D0), the $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were seeded on Matrigel (354230; Corning)-coated plates at a density of $3\times {10}^4$ cells per well in $200\mathrm{\mu }\mathrm{L}$ of STEMdiff Cardiomyocyte Differentiation Medium A from the STEMdiff Cardiomyocyte Differentiation Kit and incubated at 37°C for 2 d. On D2, the medium was gently removed, $200\mathrm{\mu }\mathrm{L}$ of STEMdiff Cardiomyocyte Differentiation Medium B was added, and the plates were incubated at 37°C for 2 d. On D4 and again on D6, the medium was gently removed, $200\mathrm{\mu }\mathrm{L}$ of STEMdiff Cardiomyocyte Differentiation Medium C was added, and the plates were incubated at 37°C for 2 d. On D8, the medium was removed, the cells were washed with phosphate-buffered saline (PBS) before cell collection and analysis. The differentiating cells were harvested at the ends of D2 and D8 for analysis. The differentiation protocol and initial cell density applied to all the experiments (Figures 1, 2, 4, and 5). Cadmium chloride (202908; Sigma-Aldrich) stock solution ($100\text{\hspace{0.17em}mM}$) was prepared in distilled water and stored at $–20°\mathrm{C}$ before use. A dose range of $0.001–100\mathrm{\mu }\mathrm{M}$ Cd was applied to the adenosine triphosphate (ATP) cell viability luciferase assay and $\text{NKX}2\text{-}{5}^{\text{eGFP}}$ fluorescence signal quantification (Figure 2). The median effective concentration (${\text{EC}}_{50}$) dose was used on the monolayer cardiomyocyte differentiation Cd-exposure model unless otherwise specified. UNC1999 [$N$-[(1,2-dihydro-6-methyl-2-oxo-4-propyl-3-pyridinyl)methyl]-1-(1-methylethyl)-6-[6-[4-(1-methylethyl)-1-piperazinyl]-3-pyridinyl]-1$H$-indazole-4-carboxamide; an enhancer of zeste homolog 2 (EZH2) and EZH1 dual inhibitor] was added at doses of 2, 10, and $50\text{\hspace{0.17em}nM}$ to cardiomyocyte differentiation medium with or without Cd (Figure 5). Distilled water was used as the vehicle control and diluted in the medium at 1:1,000 for all experiments with Cd exposure.

Figure 1.

Quantification of ${\text{GFP}}^{+}$ cardiomyocytes. $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were seeded at a density of 30,000 cells/well in 96-well plates for cardiomyocyte differentiation. (A–D) ${\text{GFP}}^{+}$ cardiomyocytes were quantified by IncuCyte live cell image analysis by ratio of green object confluence (percentage) to phase object confluence (percentage) and analyzed via one-way ANOVA in Prism (version 8; Graphpad). ($n=9/\text{group}$.) The data are provided in Table S1 and graphed in (D) as $\text{means}+\text{SDs}$. (*$p<0.05$; $N=3$.) (E,F). Cardiomyocyte differentiation ratio was verified by flow cytometry single cell analysis. (G) Gene expression at D0, D2, and D8 was normalized to the mean $2^{-\mathrm{\Delta }\mathrm{\Delta }\text{Ct}}$ value of D0 and analyzed via two-tailed $t$-test. ($n=3$.) The data are provided in Excel Table S1 and graphed in (G) as ${\log }_2\left[\text{mean\hspace{0.17em}}\left(2^{-\mathrm{\Delta }\mathrm{\Delta }\text{Ct}}\right)\right]$. Note: $\mathrm{\Delta }\mathrm{\Delta }\text{Ct}$, delta delta Ct; ANOVA, analysis of variance; Cardio diff, cardiomyocyte differentiation; D, day of exposure; FC, fold change; FITC, fluorescein isothiocyanate; GFP, green fluorescent protein; hESCs, human embryonic stem cells; SD, standard deviation.

Figure 2.

The viability of differentiating cardiomyocyte and ${\text{GFP}}^{+}$ cardiomyocytes were assessed after 8 d of exposure to Cd. (A) ${\text{GFP}}^{+}$ cardiomyocytes exposed to different doses of Cd. (B) Viability was assessed using the CellTiter-Glo assay (Promega) in 96-well plates and ${\text{LC}}_{10}$ was calculated based on the equation $Y=100/(1+{10}^{(\text{LogLC}50-\text{X})\times \text{HillSlope}})$, $\text{HillSlope}=-1.0$ and ${\text{LC}}_{50}$ was calculated under nonlinear regression curve fit and equation under log(inhibitor) vs. response in Prism software (version 8; Graphpad). ($n=4–5/\text{group}$.) The data are provided in Table S2 and graphed in (B) as normalized $\text{means}+\text{SDs}$. (C) ${\text{GFP}}^{+}$ cardiomyocyte area was quantified by IncuCyte live cell image analysis at 8 d and ${\text{EC}}_{50}$ was calculated under nonlinear regression curve fit and equation under log(inhibitor) vs. response in Prism software (version 8; Graphpad). ($n=5/\text{group}$.) The data are provided in Table S3 and graphed in (C) as $\text{means}+\text{SDs}$. (D) $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were exposed during the time periods D0–2 and D2–4 at various Cd doses, and GFP percentage was evaluated at the end of 8 d post-treatment. Data was analyzed by two-way ANOVA with a significance threshold of $p<0.05$ followed by a Dunnett’s multiple comparisons test. The data are provided in Table S4 and graphed in (D) as $\text{means}+\text{SDs}$. (*, ^#$p<0.05$; $N=5$.) Note: ANOVA, analysis of variance; ATP, adenosine triphosphate; Cd, cadmium; Conc., concentration; D, day of exposure; ${\text{EC}}_{50}$, median effective concentration; GFP, green fluorescent protein; hESCs, human embryonic stem cells; ${\text{LC}}_{10}$, lethal concentration for 10%; ${\text{LC}}_{50}$, lethal concentration for 50%; SD, standard deviation.

Figure 4.

RT-PCR gene quantification of differentiating cardiomyocytes at D2 and D8. (A) Confirmation of selected mesoderm transcription factors and marker genes at 2 d. Gene expression in the Cd ($0.15\mathrm{\mu }\mathrm{M}$) exposure group was normalized to that in the water vehicle control group and analyzed via two-tailed $t$-test (*$p<0.05$; $N=3$). The data are provided in Excel Table S5 and graphed in (A) as normalized $\text{means\hspace{0.17em}}+\text{\hspace{0.17em}SDs}$. (B) Expression of cardiomyocyte marker genes was quantified at D8. Gene expression in the Cd ($0.15\mathrm{\mu }\mathrm{M}$)-exposure group was normalized to that of the water vehicle control group and analyzed via two-tailed $t$-test (*$p<0.05$; $N=3$). The data are provided in Excel Table S6 and graphed in (B) as normalized $\text{means\hspace{0.17em}}+\text{\hspace{0.17em}SDs}$. Note: Cd, cadmium; Con, control; D, day of exposure; RT-PCR, real-time polymerase chain reaction; SD, standard deviation.

Figure 5.

Histone protein methylation and cardiomyocyte differentiation in 2D hESC to cardiomyocyte differentiation model with or without UNC1999 and cadmium exposure. (A) Histone protein methylation was quantified by western blot in EBs on D2 of differentiation. ($n=3/\text{group}$.) The data are provided in Excel Table S7 and a representative blot of replicate 1 (R1) is shown (A). (B) Selective inhibitor UNC1999, which targets EZH2 (controls H3k27me3) was added together with cadmium $0.15\mathrm{\mu }\mathrm{M}$ during cardiomyocyte differentiation, and ${\text{GFP}}^{+}$ cardiomyocytes were quantified at D8 (*$p<0.05$; $N=5$). Data was analyzed by two-way ANOVA with a significance threshold of $p<0.05$ followed by a Dunnett’s multiple comparisons test. The data are provided in Table S5 and graphed in (B) as $\text{means\hspace{0.17em}}+\text{\hspace{0.17em}SDs}$. Note: ANOVA, analysis of variance; Cd, cadmium; Ctrl, control; D, day of exposure; EBs, embryoid bodies; EZH2, enhancer of zeste homolog 2; GFP, green fluorescent protein; hESCs, human embryonic stem cells; R1, replicate 1; SD, standard deviation; UNC1999, $N$-[(1,2-dihydro-6-methyl-2-oxo-4-propyl-3-pyridinyl)methyl]-1-(1-methylethyl)-6-[6-[4-(1-methylethyl)-1-piperazinyl]-3-pyridinyl]-1$H$-indazole-4-carboxamide.

Formation of 3D embryoid bodies (EBs) and differentiation to germ layers was performed in 384-well plates (3830; Corning) using Essential 6 Medium (A1516401; ThermoFisher Scientific) with a cell density of 1,000 cells/well for 8 d. A single EB was formed in each well. Cd stock solution was diluted to the doses 0.1, 0.5, 1, 5, 10, and $50\text{\hspace{0.17em}mM}$ in distilled water followed by 1:1,000 dilution in medium for the 96-h exposure and viability assay. The lethal concentration for 10% (${\text{LC}}_{10}$) dose was calculated from the viability assay and used together with a medium change every other day from D1 to D8 of exposure. We used the TaqMan hPSC Scorecard Array ($n=1$ array) and polymerase chain reaction (PCR) to evaluate the pluripotency and expression of mesoderm-associated genes in EBs treated with the ${\text{LC}}_{10}$ dose of Cd ($0.6\mathrm{\mu }\mathrm{M}$). The Scorecard evaluates the expression of genes associated with pluripotency, as well as those specific to each of the three germ layers. These data are shown in Figure 3.

Figure 3.

Viability, differentiation, and pluripotency capability of differentiating EBs exposed to Cd. (A) Images of an hESC at D0 and EBs at D7 with or without Cd ($0.6\mathrm{\mu }\mathrm{M}$) treatment. (B) Viability was assessed using the CellTiter-Glo assay (Promega) after a 96-h exposure. ${\text{LC}}_{50}$ and ${\text{LC}}_{10}$ values were calculated using Probit analysis in JMP (version 13.0; JMP Statistical Discovery LLC). ($n=8–10/\text{group}$.) The data are provided in Excel Table S2 and graphed in (B) as normalized data points. (C) EB differentiation was assessed using the TaqMan hPSC Scorecard Panel following 7 d of exposure to equipotent (${\text{LC}}_{10}$) concentrations. ($n=1$ assay.) The data are provided in Excel Table S3 and graphed in (C) as normalized fold change. (D) Gene expression in EBs exposed to Cd ($0.6\mathrm{\mu }\mathrm{M}$) normalized to that of those exposed to water and analyzed via two-tailed $t$-test. (*$p<0.05$; $N=3$.) The data are provided in Excel Table S4 and graphed in (D) as normalized $\text{means}\pm \text{SEMs}$. Note: ATP, adenosine triphosphate; Cd, cadmium; Con, control; Conc., concentration; Ctrl, water vehicle control; D, day of exposure; EBs, embryoid bodies; fc, fold change; hESCs, human embryonic stem cells; ${\text{LC}}_{10}$, lethal concentration for 10%; ${\text{LC}}_{50}$, lethal concentration for 50%; SEM, standard error of the mean.

Cardiac Organoid Formation

Organoid formation was conducted following the protocol published by the Aguirre group.²⁷ Briefly, $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were dissociated and seeded in round-bottom ultra–low-attachment 96-well plates (4520; Corning) on $-\mathrm{D}2$ at $1\times {10}^4\text{cells}/\text{well}$. After 24 h ($-\mathrm{D}1$), $50\mathrm{\mu }\mathrm{L}$ of medium was carefully removed from each well, and $200\mathrm{\mu }\mathrm{L}$ of fresh Essential 8 Flex medium was added for a final volume of $250\mathrm{\mu }\mathrm{L}/\text{well}$. On D0, $166\mathrm{\mu }\mathrm{L}$ of medium was removed from each well and $166\mathrm{\mu }\mathrm{L}$ of Roswell Park Memorial Institute (RPMI) 1640/B-27, minus insulin medium (11875, A369520; ThermoFisher Scientific) containing CHIR99021 (S1263; Selleck) was added at a final concentration of $4\mathrm{\mu }\mathrm{M}/\text{well}$ along with BMP4 at 0.36 pM ($1.25\text{\hspace{0.17em}ng}/\mathrm{mL}$; 314-BP-010; R&D) and Actin A at 0.08 pM ($1\text{\hspace{0.17em}ng}/\mathrm{mL}$; 338-AC-010; R&D) for 24 h. On D1, $166\mathrm{\mu }\mathrm{L}$ of medium was removed and replaced with fresh RPMI1640/B-27, minus insulin medium. On D2, RPMI1640/B-27, minus insulin medium containing IWR-1 (S7086; Selleckchem) was added for a final concentration of $5\mathrm{\mu }\mathrm{M}$, instead of the Wnt-C59 used in the original protocol. On D4, the medium was switched to RPMI1640/B-27, minus insulin medium. On D6, the medium was changed to RPMI1640/B-27 (17504044; ThermoFisher Scientific). On D7, a second $2\text{-}\mathrm{\mu }\mathrm{M}$ CHIR99021 exposure was conducted for 1 h in RPMI1640/B-27. Subsequently, the medium was changed every 48 h for organoid maintenance. A diluted cadmium chloride solution ($0.001–10\text{\hspace{0.17em}mM}$) in medium was added every 48 h during differentiation for viability and contraction rate analysis, and the noncytotoxic dose of $1\mathrm{\mu }\mathrm{M}$ was used for the quantitative PCR (qPCR). Human heart total RNA was purchased (AM7966; ThermoFisher Scientific) and reversely transcribed to complementary DNA (cDNA) and used as a reference of heart-associated gene expression.

Image Acquisition and Analysis

Cardiomyocytes derived from $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs at the initial density of $3\times {10}^4\text{\hspace{0.17em}cells}/\text{well}$ at different days were allowed to adhere in 96-well flat clear-bottom black-walled polystyrene tissue culture treated microplates (353219; Corning). The plates were transferred to an IncuCyte ZOOM platform (40986; Essen BioScience) that was housed inside an incubator at 37°C/5% carbon dioxide for the duration of the assay. Four images per well from two technical replicates were taken using a $10\times$ objective and then analyzed using the IncuCyte Basic Software. Green channel acquisition time was 450 ms. In phase contrast, a mask was applied to exclude cells from background to perform the cell segmentation analyses. To exclude objects $<50{\mathrm{\mu }\mathrm{m}}^2$, an area filter was applied. The adaptive method of background nonuniformity correction with a threshold adjustment at 0.23 green corrected units was employed to remove green channel background noise. Fluorescence signal was quantified applying a mask (Figure 1B). Edge split was used to split closely spaced objects to obtain a more accurate quantification of the fluorescent objects.

Flow Cytometry

D8 cardiomyocytes derived from $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were blocked with flow cytometry buffer [PBS; 2% FBS, $25\text{\hspace{0.17em}mM}$ 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), $5\text{\hspace{0.17em}mM}$ ethylenediaminetetraacetic acid (EDTA)] for 15 min on ice while protected from light. The cells were then washed three times and pelleted at $\mathrm{5,000}\times g$ for 10 min before being resuspended in flow cytometry buffer. The samples were run on a BD LSR Fortessa flow cytometer (BD Biosciences) and analyzed with FlowJo software (version 10; Tree Star Inc.).

Western Blotting

The $\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were seeded on Matrigel-coated plates (354230; Corning) in STEMdiff Cardiomyocyte Differentiation Medium A and incubated at 37°C for 2 d with or without Cd. The attached cells were manually scraped loose and then lysed in cytoplasmic protein extraction buffer (NE1) with dithiothreitol (DTT) and protease inhibitor cocktail, and nuclear protein was extracted in nuclear extraction buffer (NE2) with DTT and protease inhibitor cocktail using the EpiQuick nuclear extraction kit (OP-0002-1; EPIGENTEK). Total protein content was determined by bicinchoninic acid assay (23225; ThermoFisher Scientific), and $7\mathrm{\mu }\mathrm{g}$ of protein per sample was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis using 4%–12% bis-Tris gels (3450124; Bio-Rad). Proteins were transferred to $0.45\text{-}\mathrm{\mu }\mathrm{M}$ nitrocellulose membranes (1620115; Bio-Rad) and blocked overnight at 4°C in a PBS-based blocking solution [$1\times$ PBS, 0.1% Tween-20 with 3% wt/vol bovine serum albumin (BSA)]. Histone methylation was detected using $\sim 5\mathrm{\mu }\mathrm{g}$ of primary Histone H3, H3K4 me3, and H3K27 me3 antibodies (61800, 39160, and 39157; ACTIVE MOTIF) at room temperature for 1 h. The band was detected by secondary mouse antirabbit IgG antibody (1:2,000) (93702S; Cell Signaling) in blocking buffer with gentle agitation for 1 h at room temperature and chemiluminescent reaction in SuperSignal West Pico PLUS Chemiluminescent Substrate (34580; Thermofisher). Images of bands were taken with the Azure C300 imaging system (Azure Biotech) and quantified with ImageJ software.²⁸

Immunofluorescence

Using a $\mathrm{1,000}\text{-}\mathrm{\mu }\mathrm{L}$ pipette to avoid disruption, the cardiac organoids were transferred to microcentrifuge tubes and fixed in a 4% paraformaldehyde solution at 4°C overnight. This was followed by three washes in PBS and incubation in Clear, Unobstructed Brain/Body Imaging Cocktails and Computational analysis (CUBIC)-L solution for delipidation for 1 d. The samples were then incubated in blocking/permeabilization solution (0.5% BSA in PBS, 0.5% Triton X-100) at 4°C overnight. The cardiac organoids were incubated with primary antibodies diluted 1:200 in the blocking/permeabilization solution [troponin T/TNNT2 (MA5-12960; Invitrogen), Thy-1 cell surface antigen/THY1 (328101; BIOLEGEND), nuclear factor of activate T cells 1/NFATC1 (MA3-024; Invitrogen), platelet and endothelial cell adhesion molecule 1/PECAM1 (MA5-13188; Invitrogen), chicken ovalbumin upstream promoter transcription factor II/COUP-TFII (PP-H7147-00; Perseus Proteomics); and myosin light chain 2/MYL2 (ab79935; Abcam)] for 2 d on a thermal mixer at 300 rpm at 4°C followed by three washes in PBS and incubation with secondary antibodies [goat antirabbit (A11012; Invitrogen), goat antimouse (A25534; Invitrogen), and donkey antirabbit (A31573; Invitrogen)] for 2 d in the dark. The stained cardiac organoids were incubated in CUBIC-R solution for 1 d for clearing before imaging. The images of the stained organoids were taken with a Zeiss LSM 710 inverted confocal microscope (Zeiss).

ATP Cell Viability Luciferase Assay

$\text{NKX}2\text{-}{5}^{\text{eGFP}/\mathrm{w}}$ hESCs were seeded in 96-well plates for cardiac differentiation. At the end of D8 differentiation, cells in the control nontreated group and in the Cd-treatment group were incubated with CellTiter-Glo reagent (G7570; Promega) and mixed on the orbital shaker for 2 min. The outer wells in each 96-well plate contained medium without cells and were also incubated for background luminescence value detection. After a 10-min incubation, the luminescence plates were read in the Biotek Synergy 4 plate reader (Agilent).

Quantitative Polymerase Chain Reaction

RNeasy kits (74004; Qiagen) were used for RNA extraction. An ND-100 spectrophotometer (Nano Drop Technologies) was used to verify RNA quality. The RNA purity was defined by the ratio of absorbance at $260\text{\hspace{0.17em}nm}$ and $280\text{\hspace{0.17em}nm}$. The QuantiTect Reverse Transcription kit (205311; Qiagen) was used to reverse transcribe cDNA from $500\text{\hspace{0.17em}ng}$ of RNA, according to the manufacturer’s instructions. Twenty nanograms of RNA reverse-transcribed cDNA were used as a template in qPCR experiments using specific primers ($500\text{\hspace{0.17em}nM}$) and Fast SYBR Green Master Mix as the detection chemistry (43-856-12; ThermoFisher Scientific). The thermal cycle profile was as follows: 95°C for 5 min, followed by 40 cycles at 95°C for 20 s, 60°C for 20 s, and 72°C for 20 s, and then a final phase of 72°C for 5 min. Every experiment included a melt curve analysis. A Step One Plus platform (Applied Biosystems) was used for all experiments and data were analyzed with the StepOne software. Cycle threshold (Ct) values were determined, and relative mRNA contents were inferred from normalization of the gene of interest expression to that of the housekeeping gene (Delta Ct). Relative expression results were plotted as $2^{-\mathrm{\Delta }\mathrm{\Delta }\text{Ct}}$. Primers were designed in the Primer3 web (version 4.1.0; ELIXIR) and primer sequences are shown in in Excel Table S9.

Statistics

Prism (version 8; Graphpad Software) and JMP (version 13.0; JMP Statistical Discovery LLC) software were used for statistical difference analysis and figure generation. Data were analyzed by one-way analysis of variance (ANOVA) to determine significant differences between the means of independent groups or two-way ANOVA and post hoc testing with correction for comparison to determine significant differences affected by two variables in combination in Prism (GraphPad Software; version 8). Gene expression difference was analyzed by two-tailed $t$-test comparison. Statistical details are in the figure legends and tables. Median lethal concentration (${\text{LC}}_{50}$), ${\text{LC}}_{10}$, ${\text{EC}}_{50}$, and other statistical analyses were performed using Prism (version 8; Graphpad), ${\text{LC}}_{50}$ and ${\text{EC}}_{50}$ were calculated under nonlinear regression curve fit and equation under log(inhibitor) vs. response; ${\text{LC}}_{10}$ was calculated based on equation $Y=100/(1+{10}^{(\text{LogIC}50-\text{X})\times \text{HillSlope}})$, $\text{HillSlope}=-1$ in Prism (version 8; GraphPad). All raw numbers and normalized data as well as $p$-values are provided in the experiment supplemental tables in the Supplemental Material (Table S1–S7 and Excel Table S1–S8, S10).

Results

Optimization and Validation of in Vitro Cardiomyocyte Differentiation Model

To characterize the in vitro cardiomyocyte differentiation, we performed a monolayer-differentiation using NKX 2-5 GFP hESC cell line (NKX2.5-hESC) into cardiomyocytes and quantified the fluorescent signal from D6 to D10 (Figure 1). The NKX 2-5 reporter GFP was expressed in early cardiac mesoderm cells throughout the formation of the left ventricle and atrial chambers. The automated quantification of ${\text{GFP}}^{+}$ cell area to total cell ratio in the IncuCyte indicated that ${\text{GFP}}^{+}$ cells increased from $24.01\%\pm 2.93\%$ on D6 to $48.52\%\pm 6.47\%$ by D8 and increased to $58.41\%\pm 5.43\%$ by D10 (Figure 1D; Table S1). To confirm the reliability of the IncuCyte quantification, the differentiated cells at D8 were also quantified by flow cytometry based on NKX 2-5 reporter GFP expression. The ${\text{GFP}}^{+}$ cells ratio was 51.6%, which was consistent with the ${\text{GFP}}^{+}$ cardiomyocytes ratio by IncuCyte analysis at D8 (Figure 1E,F). Mesoderm-associated transcription factors mesoderm posterior bHLH transcription factor 1 (MESP1) and eomesodermin (EOMES) showed a transient up-regulation in the first 2 d of differentiation, whereas the mesoderm marker genes heart and neural crest derivatives expressed 1 (HAND1), forkhead box A2 (FOXA2), snail family transcriptional repressor 2 (SNAI2), and HOP homeobox (HOPX) were up-regulated on D2 and D8. Cardiac-specific genes NKX2-5, GATA4, and troponin T (TNNT2/cTNT) were all highly up-regulated in differentiating cardiomyocytes at D8 (Figure 1G; Excel Table S1).

Effect of Cd on Cardiomyocyte Differentiation

To examine the effect of Cd exposure on human cardiomyocyte differentiation, differentiating cardiomyocytes were exposed to a dose range of Cd from 0.001 to $100\mathrm{\mu }\mathrm{M}$ and ${\text{GFP}}^{+}\text{-cardiomyocytes}$ were quantified at the end of D8 of differentiation (Figure 2A). The calculated ${\text{LC}}_{50}$ was $14.93\mathrm{\mu }\mathrm{M}$ (Figure 2B; Table S3), whereas the calculated ${\text{EC}}_{50}$ was $0.15\mathrm{\mu }\mathrm{M}$ (Figure 2C; Table S2). To help identify the possible window of inhibitory effects of Cd during cardiomyocyte differentiation, cells were exposed during early mesoderm formation (D0 to D2) or during later cardiac induction and cardiomyocyte differentiation (D2 to D4), and ${\text{GFP}}^{+}$ cardiomyocytes were quantified at the end of D8 differentiation. Cardiomyocytes exposed to 0.1 to $10\mathrm{\mu }\mathrm{M}$ Cd during mesoderm formation (D0–2) had a lower percentages of $\text{GFP}\text{-positive}$ cardiomyocytes compared with control cells, whereas those exposed to Cd during cardiac induction (D2–4) had lower percentages of $\text{GFP}\text{-positive}$ cardiomyocytes only at the $10\text{-}\mathrm{\mu }\mathrm{M}$ exposure (Figure 2D; Table S4).

Effect of Cd on Mesoderm- and Cardiac-Associated Gene Expression and Histone Methylation during Cardiomyocyte Differentiation

To evaluate the mechanism of Cd-induced down-regulation of cardiomyocyte formation during different stages of differentiation (i.e., mesoderm formation, cardiomyocyte differentiation), the 3D EB model and 2D cardiomyocyte differentiation model were used (Figure 3A). The 96-h viability assay determined the ${\text{LC}}_{10}$ of Cd exposure in the 3D EB model to be $0.6\mathrm{\mu }\mathrm{M}$ (Figure 3B; Excel Table S2). We further evaluated mesoderm-related gene expression in the 3D EB model after exposure to Cd at the ${\text{LC}}_{10}$. EB exposed to $0.6\text{-}\mathrm{\mu }\mathrm{M}$ Cd exhibited down-regulation of several mesoderm-related genes (HAND1, HOPX, IL6ST, ODAM, SNAI2, and TM4SF1) compared with control cells, whereas other mesoderm-associated genes (ABCA4, HEY1, PDGFRA, RGS4, and TBX3) were not significantly different after 7 d (Figure 3C; Excel Table S3). The WNT signaling key regulators WNT1, WNT3a, $GSK3\mathrm{\beta }$, and beta-catenin 1 (CTNNb1) were also expressed at a lower level in EB exposed to $0.6\text{-}\mathrm{\mu }\mathrm{M}$ Cd ${\text{LC}}_{10}$ (Figure 3D; Excel Table S4). To confirm the effect of Cd on mesoderm formation, the hESCs were differentiated to mesoderm-stage cells (D0–2) using the 2D model. Mesoderm-associated transcription factors [MESP1, Mix paired-like homeobox (MIXL1), and EOMES] were expressed at a lower level after treatment with $0.15\text{-}\mathrm{\mu }\mathrm{M}$ Cd (${\text{EC}}_{50}$), and mesoderm markers (HAND1, SNAI2, and HOPX) were all were expressed at a lower level at the end of D2 of treatment compared with control cells at the same time point (Figure 4A; Excel Table S5). Cardiac-specific genes NKX2-5, GATA4, cTNT, and $\mathrm{\alpha }\text{-actinin}$ were expressed at a lower level at the end of D8 in cells exposed to $0.15\text{-}\mathrm{\mu }\mathrm{M}$ Cd compared with control cells (Figure 4B; Excel Table S6). Histone methylation was analyzed by western blot to help explain this gene regulation. H3K27 trimethylation (H3K27me3), associated with inactive gene promoters, was apparently more abundant with increasing Cd concentrations ($141.07\%\pm 19.84\%$ in the $0.15\text{-}\mathrm{\mu }\mathrm{M}$ Cd group and $169.75\%\pm 51.28\%$ in the $1.5\text{-}\mathrm{\mu }\mathrm{M}$ Cd group, compared with the non-Cd treatment group after intensity normalization to histone H3), whereas H3K4 trimethylation (H3K4me3), a characteristic of active enhancers, was apparently less abundant in the D2 cardiomyocyte differentiation samples ($85.92\%\pm 5.81\%$ in the $0.15\text{-}\mathrm{\mu }\mathrm{M}$ Cd group and $54.18\%\pm 20.4\%$ in the $1.5\text{-}\mathrm{\mu }\mathrm{M}$ Cd group, compared with the non-Cd treatment group (Figure 5A; Excel Table S7). To confirm the role of H3K27 trimethylation in cardiomyocyte differentiation, the selective inhibitor UNC1999, which targets EZH2 in control of H3K27me3, was added together with Cd ($0.15\mathrm{\mu }\mathrm{M}$) during cardiomyocyte differentiation in 8 d. In cells exposed to Cd, the percentage of ${\text{GFP}}^{+}$ cells in cultures was gradually higher with increasing UNC1999 treatment such that the amount of ${\text{GFP}}^{+}$ cells in those samples exposed to both Cd and $50\text{-nM}$ UNC1999 approximated the amount of ${\text{GFP}}^{+}$ cells in non-Cd– and UNC1999-treatment samples (Figure 5B; Table S5). In hESCs exposed to $0.15\text{-}\mathrm{\mu }\mathrm{M}$ Cd for 2 d, cells additionally treated with UNC1999 exhibited significantly higher expression of HOPX and FOXA2 (two genes that were expressed at a lower level in differentiating hESCs exposed to $0.15\text{-}\mathrm{\mu }\mathrm{M}$ Cd for 2 d; Figure 4A) at doses of 10 and $50\mathrm{\mu }\mathrm{M}$ (but not at $2\mathrm{\mu }\mathrm{M}$) (Figure S1).

Cd Exposure and Cardiac Organoids

The cardiac organoids derived from NKX2.5-hESC in this study start beating and expressing NKX 2-5 GFP at approximately D6 (Figure S2A) and were quantified on D8 (Figure 6). Confocal analysis of the organoids for cardiac markers revealed the presence of all major cardiac lineages, including cardiomyocytes ($\text{TNNT}{2}^{+}$), cardiac fibroblasts ($\text{THY}{1}^{+}$), endocardial cells ($\text{NFATC}{1}^{+}$), and endothelial cells ($\text{PECAM}{1}^{+}$). Both atrial (${\text{COUP-TFII}}^{+}$) and ventricular ($\text{MYL}{2}^{+}$) cardiomyocytes were also present in the cardiac organoids (Figure 6A; Figure S2). Based on a cell viability assay, organoids treated with lower doses of Cd ($0.001–1\mathrm{\mu }\mathrm{M}$) did not exhibit differences in viability compared with control organoids; however, organoids treated with $10\text{-}\mathrm{\mu }\mathrm{M}$ Cd had significantly lower viability (17.8%), suggesting that this dose was cytotoxic (Figure 6B; Table S6). To identify the effect of Cd on the cardiac organoid contraction, the contraction rate was analyzed following Cd treatment for 8 d. Organoids treated with Cd ($1.0\mathrm{\mu }\mathrm{M}$) had a significantly lower contraction rate ($19\pm 2.16\text{\hspace{0.17em}beats}/\text{min}$) compared with the contraction rate in the untreated control organoids ($40\pm 1.71\text{\hspace{0.17em}beats}/\text{min}$) (Figure 6C; Table S7). Gene expression of major mature cardiomyocyte markers in the differentiated cardiac organoids [NKX2.5, TNNT2, myosin heavy chain 7 (MYH7), calcium voltage-gated channel subunit alpha1 C (CACNA1C), and hyperpolarization-activated cyclic nucleotide gated potassium channel 4 (HCN4)] was higher in Cd-treated organoids than water-treated control. Organoids treated with Cd ($0.1\mathrm{\mu }\mathrm{M}$) had lower expression of contractile- and structural-related genes TNNT2 and MYH7 and development and maturation genes NKX2.5 and CACNA1C. Organoids treated with Cd also exhibited lower expression of the arterial endothelium marker T-box transcription factor 20 (TBX20). The cardiac fibroblast marker discoidin domain receptor tyrosine kinase 1 (DDR1) was higher in organoids treated with Cd (Figure 6D; Excel Table S8).

Figure 6.

Cardiac organoid viability and contraction in the presence of Cd exposure during differentiation. (A) Characterization of cardiac organoids by immunostaining with antibodies specific for cardiac fibroblasts (THY1), cardiomyocytes (cTnT1), endocardial cells (NFATC1). (B,C) The viability ($n=6/\text{group}$) and beating frequency of differentiating cardiac organoids ($n=4/\text{group}$) were assessed at D8 of exposure to Cd. Raw data was normalized to mean values of water vehicle control and $\text{means\hspace{0.17em}}+\text{\hspace{0.17em}SDs}$ are plotted in (B,C). *$p<0.05$, relative to water vehicle control (one-way ANOVA with post hoc multiple comparison correction using a Tukey’s multiple comparisons test). The data are provided in Tables S6 and S7 and graphed in (B,C). (D) RT-PCR confirmation of the expression of selected genes associated with cardiomyocytes, endocardial cells, and cardiac fibroblasts 8 d of cadmium exposure in organoids treated with cadmium, cells treated with water, and human heart tissue. Gene expression was normalized to the water vehicle control group and analyzed via two-tailed $t$-test (*$p<0.05$; $N=3$). The data are provided in Excel Table S8 and graphed in (D) as mean $2^{-\mathrm{\Delta }\mathrm{\Delta }\text{Ct}}$ values in the heatmap. Note: $\mathrm{\Delta }\mathrm{\Delta }\text{Ct}$, delta delta Ct; ANOVA, analysis of variance; ATP, adenosine triphosphate; Cd, cadmium; Conc., concentration; Ctrl, control; D, day of exposure; DAPI, 4′,6-diamidino-2-phenylindole; RT-PCR, real time polymerase chain reaction; SD, standard deviation; THY1, Thymus cell antigen-1.

Discussion

Using an efficient 2D/3D cardiomyocyte differentiation model system and a cardiac organoid model, this study established an in vitro toolbox for high-throughput screening of developmental cardiotoxicants and for investigating cardiac function. The self-organized cardiac organoids showed similarities in both morphology and cell-type complexity to human fetal hearts and could, thus, provide some insight into how congenital heart disorders may originate. In addition, the 3D EB model provided a platform for high-throughput screening of environmental developmental cardiotoxicants. Using these model systems, we conclude that Cd exposure inhibited several key factors involved in mesoderm formation and cardiomyocyte differentiation at a noncytotoxic dose, which, in turn, resulted in the inhibition of cardiac organoid contraction. Together, these studies provide valuable insights into Cd-induced CHD and cardiovascular disease.

Transcription factor NKX2-5 is expressed upon cardiac crescent formation similar to how this factor is expressed in cardiomyocytes derived from ESCs in vitro.²⁹ 2D cardiomyocyte differentiation in the hESC-NKX2-5 reporter line used in this study resulted in continuous expression of NKX2-5-eFGP fluorescence after NKX2-5 transcription factor activation (Figure 1A–C). This provides the opportunity for rapid in vitro screening of developmental cardiotoxicants with real-time GFP expression analysis. This 2D cardiomyocyte differentiation model also mimics mesoderm formation and cardio-mesoderm differentiation to cardiomyocytes, thereby providing different developmental windows for exploring possible mechanisms of action of developmental cardiotoxicants. In the 3D cardiac organoid, ${\text{GPF}}^{+}$ cardiomyocytes could be isolated for further quantification, providing a cost-efficient analysis tool for developmental cardiotoxicity research.

According to U.S. Environmental Protection Agency guidance, the reference dose (RfD) for Cd in drinking water is $0.0005\mathrm{mg}/\mathrm{kg}\text{\hspace{0.17em}per day}$ and the RfD for dietary exposure to Cd is $0.001\mathrm{mg}/\text{kg per day}$.³⁰ However, Cd exposure levels could be much higher in mining and industrial settings or in contaminated and polluted land, as in some rice paddies in Japan.³¹ According to the World Health Organization Environmental health criteria 135 report, blood Cd levels in exposed workers in an alkaline battery factory were generally between 5 and $50\mathrm{\mu }\mathrm{g}/\mathrm{L}$, which is $\sim 45–450\text{\hspace{0.17em}nM}$,³² but extreme exposures caused Cd concentrations in the blood as high as $300\mathrm{\mu }\mathrm{g}/\mathrm{L}$ ($\mathrm{2,670}\text{\hspace{0.17em}nM}$).³³ The placenta can act as a partial barrier to fetal Cd exposure, and studies have shown anywhere between 50% and 67% of Cd can cross the placenta.^34,35 Cd concentrations in cord blood have been found to be approximately up to 50% of the concentration in maternal blood.³⁶

Cd induced morphological changes and apoptosis in hESC-derived cardiomyocytes.³⁷ However, the possible toxicity of Cd during the formation of these cardiomyocytes was not identified. In the present study, hESCs were chronically exposed to Cd during their differentiation to cardiomyocytes, as indicated by the expression of NKX2-5-eGFP. The inhibition of developing cardiomyocytes occurred at a much lower dose ($0.15\mathrm{\mu }\mathrm{M}$) than in a previous study using hESC-derived cardiomyocytes ($30\mathrm{\mu }\mathrm{M}$) as an acute model for 24-h viability assay.³⁷ The difference in effective doses between cardiomyocyte viability and cardiomyocyte differentiation suggests that cardiomyocyte differentiation during development could be more sensitive to Cd than fully differentiated cardiomyocytes are. Early developmental exposure to Cd may be a more sensitive stage when compared with later life exposure of differentiated cardiomyocytes.

Cd is a known cardiotoxicant; however, the mechanisms of these toxic effects during development are not well defined. Epidemiologic studies have suggested maternal exposure to Cd may be a significant risk factor for CHDs in offspring.²⁵ Similarly, animal studies have shown Cd exposure can be highly toxic to minnow embryos, with the most pronounced morphological alterations occurring in development of the heart.³⁸ The present study mimics early human heart development by using a 3D EB and germ layer formation model. EBs are formed from the spontaneous differentiation of hESCs and comprise cells from all three germ layers (the ectoderm, mesoderm, and endoderm)³⁹; whereas cardiomyocytes derive from the mesoderm layer. Wnt signaling is required for development of ESC-derived mesoderm, including cardiovascular mesoderm.⁴⁰ Cd exposure inhibited the canonical Wnt signaling via down-regulation of WNT1 and WNT3a gene expression. Wnt signaling regulated the cardiac lineage factor MESP1 through promoter binding.⁴¹ The present study demonstrated that Cd treatment of hESCs inhibited mesoderm gene expression and disrupted mesoderm formation which could, thus, interfere with downstream cardio-mesoderm formation. The 2D cardiomyocyte differentiation model gene expression data confirmed that cardio-mesoderm related genes, such as MESP1 and EOMES, were down-regulated with Cd treatment (Figure 4). The transcription factor EOMES binds directly to regulatory sequences of the Mesp1 gene and is involved in early embryonic patterning and cardiac differentiation.⁴² MESP1 was required for regulating the initiation of the cardiac transcription factor cascade to direct the generation of cardiac mesoderm.⁴³ HAND1 gene expression in mice was detected throughout lateral and extraembryonic mesoderm during heart development and is involved in patterning the left ventricle.⁴⁴ Cardiac progenitor cells residing in the precardiac mesoderm expressed HOPX prior to expressing troponin T, a component of the contractile sarcomere apparatus of myocytes.⁴⁵ SNAI2 is known for its activity in lateral plate mesoderm formation in mice.⁴⁶ FOXA2 contributed to the cardiac mesoderm formation and progenitor population during cardiac differentiation that gives rise primarily to cardiovascular cells of the ventricles.⁴⁷ The results of this study suggest that, altogether, the suppressive effects of Cd on these factors inhibited the formation of mesoderm and cardio-mesoderm during the differentiation of hESCs to cardiomyocytes. This conclusion, made in the 2D cardiomyocyte differentiation model, was further supported in the 3D cardiac organoid formation model in which the contractility was less and the cardiac contractile and maturation related structure gene expression was lower. However, the cardiac fibroblast marker gene DDR1 was more highly expressed with Cd treatment, which might indicate a Cd-associated alteration of the cell composition. The higher dose effect in the 3D model compared with the 2D cardiomyocyte differentiation model could be due to the lower contact to chemical exposure inside the 3D organoid.

Histone 3 lysine 27 (H3K27) trimethylation (H3K27m3) is characteristic of inactive/repressive chromatin, whereas H3K4 trimethylation (H3K4me3) is characteristic of active promoters.⁴⁸ Pluripotent ESCs have been shown to have H3K4me3 and H3K27me3 located together at promoter and enhancer regions of poised key developmental genes.⁴⁹ During cardiac lineage formation, dynamic chromatin modifications, such as the acquisition of H3K4me3 and H3K27me3 by various transcription factors, have been identified.⁴⁹ H3K4me3 modifications at the promoters of mesodermal genes, such as HOPX, were required for active transcription and mesoderm formation,⁵⁰ whereas disruption of the site modification led to malformation of the mesoderm layer during development.⁵¹ In this regard, mesoderm formation is regulated by WNT-intermediated epigenetic modification. It is possible that Cd treatment resulted in up-regulation of H3K27me3, which plays a role in the inhibition of cardiomyocyte differentiation via modification of expression of mesoderm-associated genes, such as FOXA2. FOXA2 is a transcription factor that regulates the differentiation of hESCs to early mesoderm and was bivalently marked with both H3K4me3 and H3K27me3, which are histone modifications for gene activation and repression, respectively.⁵²

Our study has limitations. First, the hESC line used in this study is from a single donor embryo. Using other hESC or iPSC lines with different genetic and sex backgrounds could help determine possible genetic diversity in developmental cardiotoxicity caused by Cd exposure. Second, the cardiac organoids used in this study did not contain every cell type of the heart and lacked some structural similarity to the human heart, which may somewhat limit the extrapolation of the observed effects to in vivo models or to humans. And third, the percentage of Cd taken up by the placenta may be influenced by several factors, including the timing and level of exposure, route of exposure, different environmental settings, or genetic variation. Therefore, the concentrations of Cd used in this study may not be the same as exposure levels for all populations or individuals.

In closing, our findings suggest that Cd exposure inhibits the differentiation of hESCs to cardiomyocytes and may help explain the association of Cd exposure and CHD etiology at the cellular level. Moreover, this hESC differentiation model could provide a platform for further mechanistic investigation of the potential effects of Cd exposure on molecular targets during human heart development. A suggested mechanism involved in these effects of Cd is shown in Figure 7. The difference in sensitivity to Cd exposure during the specific stages of differentiation (i.e., mesoderm induction vs. cardiac induction) from this study suggests a window of sensitivity during mesoderm formation, which could mean a possibly lower risk with later Cd exposure during fetal development. Furthermore, these results indicate that Cd could potentially affect other tissues derived from mesoderm formation. As such, this study outlines a strategy that may be used to help elucidate the adverse outcome pathways of known or possible early life developmental cardiotoxicants. Furthermore, the 2D and 3D model systems developed and used in this study could be a valuable toolbox to use for high-throughput screening of developmental toxicants.

Figure 7.

Schematic illustration of the predicted mechanism of Cd-mediated effect on cardiomyocyte differentiation and cardiac organoid formation from the embryonic stem cells. Note: Cd, cadmium; H3K27me3, histone H3 protein tri-methylation at the 27th lysine residue; H3K4me3, histone H3 protein tri-methylation at the fourth lysine residue.