Coumestrol Induces Oxidative Stress and Impairs Migration and Embryonic Growth

Abstract

Although vegetarianism has grown in popularity especially among pregnant women, the effects of phytoestrogens in placentation lacks understanding. Factors such as cellular oxidative stress and hypoxia and external factors including cigarette smoke, phytoestrogens, and dietary supplements can regulate placental development. The isoflavone phytoestrogen coumestrol was identified in spinach and soy and was found to not cross the fetal-placental barrier. Since coumestrol could be a valuable supplement or potent toxin during pregnancy, we sought to examine its role in trophoblast cell function and placentation in a murine pregnancy. After treating trophoblast cells (HTR8/SVneo) with coumestrol and performing an RNA microarray, we determined 3,079 genes were significantly changed with the top differentially changed pathways related to the oxidative stress response, cell cycle regulation, cell migration, and angiogenesis. Upon treatment with coumestrol, trophoblast cells exhibited reduced migration and proliferation. Additionally, we observed increased reactive oxygen species accumulation with coumestrol administration. We then examined the role of coumestrol within an in vivo pregnancy by treating wildtype pregnant mice with coumestrol or vehicle from day 0 to 12.5 of gestation. Upon euthanasia, fetal and placental weights were significantly decreased in coumestrol treated animals with the placenta exhibiting a proportional decrease with no obvious changes in morphology. Therefore, we conclude that coumestrol impairs trophoblast cell migration and proliferation, causes accumulation of reactive oxygen species, and reduces fetal and placental weights in murine pregnancy.

In Brief:

Healthy development of the placenta is dependent on trophoblast cell migration and reduced oxidative stress presence. This article describes how a phytoestrogen found in spinach and soy causes impaired placental development during pregnancy.

Introduction

Maternal diet has often been characterized as an important factor in determining the successful outcome of a pregnancy and health of a fetus (Lowensohn RI, Stadler DD et al., 2016). The consumption of vegetarian or vegan diets is especially prevalent with individuals exhibiting a marked interest in meat replacements, such as tofu, nuts, and legumes. Tofu, derived from soybeans, contains multiple compounds including the estrogen-like or phytoestrogen isoflavone called coumestrol (Stopper H, Schmitt E et al., 2005). Found in clover, alfalfa, soy, and spinach, coumestrol uniquely signals directly through the estrogen receptor alpha (Zierau O, Kolba S et al., 2006). Furthermore, recorded intake of isoflavones in the United States and Europe is from ~3 mg/day up to 18 mg/day in health conscious individuals with Asian populations reaching up to 40 mg/day (Messina M, 2010, Zamora-Ros R, Knaze V et al., 2012). Increased total isoflavone intake was also associated with lower metabolic syndrome risk in a Chinese population study (Liu J, Mi S et al., 2018). In pregnant women, consumption of coumestrol was identified in the bloodstream in rare instances (0.5ng/mL) (Todaka E, Sakurai K et al., 2005), but it was not detected in the amniotic fluid or cord blood (Foster WG, Chan S et al., 2002). Interestingly, vegetarian individuals exhibit a decreased risk for preeclampsia (PE), a hypertensive pregnancy disorder characterized by symptoms of hypertension, pre-term birth, and intrauterine growth restriction (Schoenaker DA, Soedamah-Muthu SS et al., 2014, Chaiworapongsa T, Chaemsaithong P et al., 2014). Since consumption of vegetables is often favorable to the development of the placenta and small circulating molecules, especially hormones, can often impact placental growth (Gingrich J, Ticiani E et al., 2020), we hypothesized coumestrol may contribute to trophoblast invasion and placental development but would likely not perturb the fetal environment. Thus, dietary coumestrol can accumulate to quantifiable levels in the blood during pregnancy, but little is known about whether or how coumestrol might affect the developing placenta and fetus. Therefore, we proposed to further investigate the implications of coumestrol exposure during pregnancy.

The placenta develops by the invasion and migration of the blastocyst-derived trophoblast cells through the maternal uterine tissue and the activation of spiral artery remodeling via critical signaling molecules including matrix metalloproteases, cytokines, angiogenic factors, and regulators of proliferation (Chaiworapongsa T, Chaemsaithong P et al., 2014). Reduced gene expression or signaling activation can result in impaired invasion of the trophoblast cells, constricted vessels, and abnormal development of the placenta, leading to fetal growth restriction (Chaiworapongsa T, Chaemsaithong P et al., 2014). In addition, the presence of oxidative stress and hypoxia is especially important in the placenta because they facilitate development by promoting angiogenesis, proliferation, and differentiation (Chiarello DI, Abad C et al., 2020, Pringle KG, Kind KL et al., 2010). In contrast, excessive oxidative stress can lead to reduced trophoblast invasion and PE incidence (Chiarello DI, Abad C et al., 2020). Furthermore, repeated instances of hypoxia early in placental development can also lead to increased reactive oxygen species (ROS) and PE pathologies (Chiarello DI, Abad C et al., 2020). Therefore, placentation is a complex process in which multiple signaling pathways and stressors work in concert to establish the placental network, critical for providing essential nutrients and gas exchange at the fetal-maternal interface.

Upregulated estrogen signaling through the estrogen receptor alpha (ESR1) is important for the progression of pregnancy and growth of the placenta (reviewed in (Berkane N, Liere P et al., 2017)). Placental ESR1 binds to estrogen ligand synthesized from precursors secreted by the maternal and fetal adrenal glands during pregnancy and contributes to trophoblast migration, differentiation, and angiogenic pathways critical for placental development (Berkane N, Liere P et al., 2017). Within placental tissues obtained from women (Fujimoto J, Nakagawa Y et al., 2005, Bukovsky A, Caudle MR et al., 2003), ESR1 levels were found to be increased in the first and second trimesters of pregnancy and decreased in the third trimester (Fujimoto J, Nakagawa Y et al., 2005, Bukovsky A, Caudle MR et al., 2003), while estrogen related receptors (ERRs) which bind estrogenic compounds and promoters of estrogen target genes (Giguere V, 2002), were highly expressed in the placenta mid-gestation until delivery (Fujimoto J, Nakagawa Y et al., 2005). Decreased estrogen ligand and signaling through ESR1 and ERRs is associated with abnormal placental development and hypertensive pregnancy disorders such as PE (Luo J, Sladek R et al., 1997, Hertig A, Liere P et al., 2010). Therefore, naturally occurring estrogenic compounds such as coumestrol may be a valuable treatment option for the prevention of hypertensive pregnancy disorders.

Although other flavonoid compounds, such as quercetin and hesperidin, have been shown to promote trophoblast cell invasion and protect against oxidative stress formation (Ebegboni VJ, Balahmar RM et al., 2019), studies exploring the effects of coumestrol have provided varying, cell-context specific results. Recently, in porcine trophectoderm cells, coumestrol (0-100μM) was shown to promote cellular migration in a dose-dependent manner by activating mitogen-activated protein kinase (MAPK) signaling (Lim W and Song G, 2016). However, reported by the same group in human placental choriocarcinoma epithelial cells, coumestrol (0-100μM) impaired cell invasion and proliferation, induced apoptosis and MAPK signaling, and caused accumulation of peroxides (Lim W, Yang C et al., 2017). These differences may be explained by the type of placentation, as pigs undergo epitheliochorial placentation resulting in surface adhesion between the luminal epithelium and chorion, while women exhibit hemochorial placentation in which the embryonic-derived trophoblast cells robustly invade into the maternal tissue (Geisert RD and Spencer TE, 2021). Despite these differences, coumestrol treatment during pregnancy in rodents was determined to be safe for fetal growth and development (Kramer F, Jensen PS et al., 2003, Elias EA and Kincaid RL, 1984). Contrastingly, in certain livestock species, such as ewes and heifers, long-term grazing on plants such as clover or lucern (alfalfa), which contain high levels of phystoestrogens including coumestrol, is associated with poor fertility and dysregulation of reproductive hormone cycles (Adams NR, 1990). Thus, our current understanding of coumestrol in the context of placental development and offspring health warrants further investigation.

Due to its potential as a possible therapeutic or dietary supplement and its widespread consumption, we chose to characterize the specific effects of coumestrol in trophoblast cell function and placentation. We exposed transformed human trophoblast cells (HTR8/SVneo) and pregnant wildtype mice to a high concentration of coumestrol and examined the effects on placental function, morphology, and fetal health. As described below, we conclude that this dosage of coumestrol impairs trophoblast cell migration, proliferation and apoptosis, promotes accumulation of ROS, and causes significant decreases in fetal and placental weights.

Methods

Cell Culture and Growth Conditions

HTR8/SVneo cells (ATCC #CRL-3271, RRID:CVCL_7162) were obtained from ATCC with a passage of 80 and for these experiments, were passaged between 82-92 and cold stored according to ATCC instructions. Experiments were performed under normal cell culture conditions at 37°C with 5% CO₂. For all cell studies, HTR8/SVneo cells were grown in RPMI-1640 (Gibco #11875093) with 5% fetal bovine serum and 1% penicillin/streptomycin and treated in RPMI-1640 media without phenol red (Gibco #11835030) with 5% fetal bovine serum and 1% penicillin/streptomycin with vehicle or coumestrol (Sigma-Aldrich #27885-50MG) at 65μM. Coumestrol dose was selected based on the induction of estrogen-responsive genes in this cell line (data not shown).

RNA Isolation

For the microarray analysis, RNA was isolated from HTR8/SVneo cells at passage 83 using the Ambion mirVana miRNA Isolation Kit with phenol (Thermo Fisher Scientific #AM1560) for total RNA isolation according to manufacturer’s instructions. For RNA isolation from placental tissue, a placental half was homogenized with TRIzol Reagent (Thermo Fisher Scientific #15596026) in the Precellys 24 (Bertin Technologies) using Lysing Matrix D tubes (MP Biomedicals #6913-100) at 6,000 rpm for two 25 second intervals with a 15 second rest at room temperature, per the manufacturer’s suggestion, in between. The supernatant was removed and chloroform was used to separate the aqueous phase. Isopropanol was used to precipitate the RNA which was pelleted, washed with ethanol (75%), and dried before dissolving in RNase-free water.

Microarray Profiling and Gene Pathway Analysis

For the microarray experiment, confluent cells were exposed for 24 hours in a 6-well plate to vehicle or coumestrol: coumestrol was diluted in DMSO, filtered, and added to the media to reach a concentration of 65μM, while vehicle was filtered DMSO added in a similar amount to the media comparable to the coumestrol treatment. RNA was isolated and the quality was assessed using an Agilent Tapestation System before hybridizing to the human Clariom S Assay (Thermo Fisher Scientific #902927) at the Functional Genomics Core at UNC Chapel Hill. Three arrays or technical replicates were performed per experimental group. RNA samples for the arrays were isolated from one biological replicate experiment consisting of 3 technical replicates. The array results were validated by qRT-PCR of 3 biological replicate experiments each consisting of 3 technical replicates. Raw data was analyzed using Partek Genomics Suite 7.0 using the Gene Expression function. Default RMA Import settings were utilized which included RMA background correction and quantile normalization. Differentially regulated genes were identified with a fold change of 1.4 and a p-value<0.05. Top changed pathways were determined utilizing Ingenuity Pathway Analysis software (Qiagen, Denmark). The microarray data is publicly available on NCBI GEO at accession number GSE183072.

Quantitative Real Time PCR

Total RNA was reverse transcribed using Moloney Murine Leukemia Virus Reverse Transcriptase (M-MLV RT Thermo Fisher Scientific #28025013) according to manufacturer’s instructions with M-MLV added directly to the reverse transcriptase master mix. TaqMan Gene Expression Master Mix (Thermo Fisher Scientific #4369016) or PowerUp SYBR Green Master Mix (Thermo Fisher Scientific #A25741) was used according to manufacturer’s instructions. SYBR probes used included: (human) placental growth factor (PGF) (forward: GAACGGCTCGTCAGAGGTG and reverse: ACAGTGCAGATTCTCATCGCC), and the normalizing control, 18S (forward: ATGCTCTTAGCTGAGTGTCCCG and reverse: ATTCCTAGCTGCGGTATCCAGG). TaqMan probes used included: vascular endothelial growth factor A (human) VEGFA (Thermo Fisher Scientific #Hs00900055_m1), the normalizing control actin beta (human) ACTB (Thermo Fisher Scientific #Hs01060665_g1), vascular endothelial growth factor A (mouse) (Vegfa) (Thermo Fisher Scientific #Mm01281449_m1), platelet and endothelial cell adhesion molecule 1 (mouse) (Pecam1) (Thermo Fisher Scientific #Mm01242584_m1), and (mouse) actin beta (ActB) was used as a normalizing control (Thermo Fisher Scientific #Mm02619580_g1). QuantStudio 7 Flex Real-Time PCR System (Thermo Fisher Scientific #4485690) or the StepOnePlus Real-Time PCR System (Thermo Fisher Scientific #4376600) was used to measure Ct values for all samples. The ddCt method was used to measure relative mRNA levels by first calculating the difference between the housekeeping gene Ct and gene of interest Ct, then normalizing all the values within a gene to the median of the vehicle treated group. The 2^(-ddCt) is then calculated for each value, averaged for each treatment, and plotted accordingly. GraphPad Prism 5 software was utilized for statistics analysis and the unpaired t-test with Welch’s correction was employed to measure significant changes between vehicle and coumestrol relative message levels. Three biological replicate experiments were performed for in vitro qRT-PCR analysis each consisting of three technical replicates per group. For placental qRT-PCR analysis, 5 vehicle and 6 coumestrol treated placentas or biological replicates were examined with two technical replicates for each sample.

Cell Proliferation and TUNEL Assays

HTR8/SVneo cells were plated identically and observed until reaching 60-70% confluency. Cell proliferation was measured in HTR8/SVneo cells by treating with vehicle or coumestrol (65μM) for 24 hours. Edu was incubated for 4 hours (at 20:00 coumestrol treatment) before imaging cells at the end of the 24 hour treatment. Cells were fixed and stained using the Click-It Edu Imaging Kit (Thermo Fisher Scientific #C10339) according to manufacturer’s instructions. In tandem, cell death was measured in the same cells using the In Situ Cell Death Detection Kit, Fluorescein (Roche #11684795910) according to manufacturer’s instructions. At 24 hours of coumestrol treatment, cells were fixed and permeabilized, then stained for Edu incorporation, fragmented DNA puncta via TUNEL reaction, and DAPI nuclear staining. Four images or technical replicates were acquired per treatment well and the number of proliferating, Edu-positive cells and total nuclei were measured using BlobFinder software (Centre for Image Analysis, Swedish University of Agricultural Sciences, Uppsala University, https://www.cb.uu.se/~amin/BlobFinder/index.htm) (Allalou A and Wahlby C, 2009). Proliferating cell number was divided by total cell number to result in the percent proliferating cells. Data were normalized to the mean of the vehicle treated samples within an individual experiment. TUNEL positive puncta, (stained dots representing fragmented DNA), intensity was measured using Fiji. TUNEL intensity was divided by total cell number to represent a normalized TUNEL intensity. Values were normalized to the mean of the control treated samples within an individual experiment and the means from each well were individually plotted. Three individual biological replicate experiments were performed for each assay (n=3) each consisting of at least four averaged technical replicates per treatment group. Unpaired t-test was utilized to compare differences between the vehicle and coumestrol treated groups. Grubb’s Outlier test was employed for analysis of outlier variables.

Cell Migration Assay

HTR8/SVneo Cells were identically seeded onto 6-well plates until 90% confluence was reached. The cells were treated with vehicle or coumestrol (65μM) and two scratches, crossing at the center of the well, were made with a pipette tip. Images of the 4 scratched areas from 4 different regions of the well were captured as technical replicates on the Olympus IX83 inverted microscope with 5% CO2 and 95% humidity in an enclosed chamber to represent time zero. After 6 hours, images were re-taken in the same manner, at the same magnification and in the same location with reference points to the central crossed section at the center of the plate and the cellular landscape. Fiji was used to quantify the distance of the scratched regions at both time points. Relative migration distance was calculated by subtracting the 6 hour measurement from the 0 hour measurement. The experiment was performed with 8-12 technical replicates in three independent biological replicate experiments (n=3). Unpaired t-test with Welch’s correction was utilized to compare differences between the vehicle and coumestrol treated groups. Averages from the wells from each experiment were graphed accordingly.

Measuring Reactive Oxygen Species

HTR8/SVneo cells were seeded on a 24 well plate and incubated until reaching ~60-70% confluence before treatment for 24 hours with 1 mL volume of media containing vehicle or coumestrol (65μM). CellROX Deep Red reagent (Thermo Fisher Scientific #C10422) was added to the cells at 5μM concentration. To stain the nucleus, two drops of Molecular Probes NucBlue Live ReadyProbes Reagent (Fisher Scientific #R37605) was also added to each well. After 30 minutes of 37°C incubation at 5% CO₂, cells were washed with PBS and DAPI and Cy5 channels were measured and acquired on the Olympus IX83 inverted microscope at 20x with 5% CO₂ and 95% humidity. Four different regions were imaged per well as technical replicates. Cy5 intensity was measured for all the images using Fiji. Cell nuclei were counted using BlobFinder software (Centre for Image Analysis, Swedish University of Agricultural Sciences, Uppsala University, https://www.cb.uu.se/~amin/BlobFinder/index.htm) (Allalou A and Wahlby C, 2009). Intensity to nuclei ratio was calculated for each image and the four measurements representing one well were averaged together. Well averages represented a total of 11 averaged technical replicates per group and were plotted from each experiment. The experiment was repeated independently for a total of four biological replicates (n=4). Unpaired t-test with Welch’s correction was used to measure the difference in ratio between vehicle and coumestrol treated cells. Grubb’s Outlier test was employed for analysis of outlier variables.

In vivo Coumestrol Administration

All protocols involving animals were approved by the UNC-CH Institution of Animal Care and Use Committee. Wildtype female mice from the 129/SvEv strain were mated to 129/SvEv males and upon the presence of the vaginal plug (day 0.5 of pregnancy) were administered vehicle or coumestrol (200 μg/kg) in corn oil daily by oral gavage from day 0.5 of pregnancy until day 12.5 of pregnancy. This dosage was selected based on a previously published paper describing a safe dose utilized for treatment of mice during pregnancy with evidence that it represents less than half the recommended soybean isoflavone intake determined for adults by the Japanese Ministry of Labor, Welfare, and Health (30 mg per day or 458-570 μg/kg body weight/day) (Kirihata Y, Kawarabayashi T et al., 2008). For the oral gavage treatments, the coumestrol compound was diluted in warm 100% ethanol and added to corn oil to reach a working solution of 0.04 μg/μL. Vehicle treatment was represented by adding the same amount of 100% ethanol to the same amount of corn oil as the coumestrol treatment. Mice were weighed daily to determine the appropriate oral gavage treatment for a 200 μg/kg dose. Upon euthanasia at day 12.5 in the morning, uteri were surveyed for number of implant sites and resorptions. At necropsy, all embryo and placental pairs were separated, scored for heartbeats/life signs, weighed, and saved for RNA isolation or tissue fixation. Tissue was fixed in 4% paraformaldehyde overnight and then phosphate buffered saline (PBS) for long-term storage. Half of the placental tissue was flash frozen in liquid nitrogen for RNA isolation and half was fixed for embedding in paraffin wax. Embryonic weights and placental weights were measured (n=57 biological replicates for vehicle, 51 for coumestrol). Embryo to placental weight ratios were also tabulated (n=56 biological replicates for vehicle, 50 for coumestrol). Grubb’s Outlier test was employed for analysis of outlier variables in the determination of embryo to placental weight ratios.

Placental Morphological Analysis

Placentas were fixed and embedded in paraffin wax and cut into 5 μm sections on Superfrost Plus microscope slides (Fisher Scientific #12-550-15). Slides were dewaxed in CitriSolv Hybrid Solvent and Clearing Agent (Fisher Scientific #04-355-121) and a decreasing gradient of ethanol before undergoing antigen unmasking using sodium citrate buffer (10mM Sodium citrate tribasic dihydrate (Sigma #S4641), 0.05% Tween 20 (Fisher Scientific #BP337-500), pH=6.0). Placental tissue was permeabilized with PBS containing 0.1% Tween 20 (PBS/T) and blocked for one hour with 5% normal goat serum (NGS) in PBS/T. Primary antibody for PECAM-1 at 1:150 concentration (Millipore #MAB1398Z, RRID:AB_94207) was hybridized overnight in PBS/T in 5% NGS. Next, slides were washed in PBS/T and treated for one hour at room temperature with secondary antibody, Cy3 AffiniPure goat Anti-Armenian Hamster IgG at 1:200 concentration (Jackson ImmunoResearch #127-165-160, RRID:AB_2338989), Hoechst DNA stain at 1:250 concentration (Sigma #B1155), and Dolichos Biflorus Agglutinin (DBA), Fluorescein (Vector Laboratories #FL-1031, RRID:AB_2336394) at 1:100 concentration. Slides were then mounted with Molecular Probes ProLong Gold Antifade Mountant (Thermo Fisher Scientific #P36934) and coverslipped. Fiji was utilized to measure the area and lengths of the maternal decidua, junctional zone, and labyrinth of fixed placentas (n=12 biological replicates). Unpaired t-test with Welch’s correction was utilized to compare differences between vehicle and coumestrol treated placentas.

Determination of Embryonic Sex

A small portion of tissue was obtained from vehicle and coumestrol exposed embryos. Tissue was digested with 0.167 mg/mL proteinase K in an aqueous solution of 100mM Tris (pH 8.0), 5 mM EDTA, 0.2% SDS, and 200mM NaCl in a heat block set at 58°C overnight. DNA was isolated by precipitating with 100% ethanol and eluted in TE buffer (10mM Tris-Cl (pH 8.0), 1mM EDTA). Genotyping for sex was determined using the previously published SX primers (McFarlane L, Truong V et al., 2013): SX_F, 5’-GATGATTTGAGTGGAAATGTGAGGTA-3’, SX-R, 5’-CTTATGTTTATAGGCATGCACCATGTA-3’ with EconoTaq DNA Polymerase according to manufacturer’s instructions (Lucigen #30031-1). PCR parameters included initial denaturation at 94°C for 2 minutes, then a cycling denaturation, annealing, and extension for 35 times with 94°C for 30 seconds, 57°C for 30 seconds, and 72°C for 30 seconds, with a final elongation at 72°C for 5 minutes (McFarlane L, Truong V et al., 2013). Embryo weight was plotted based on sex from the genotyping result (n=25 biological replicates for vehicle treated males, 14 for vehicle treated females, 15 for coumestrol treated males, and 18 for coumestrol treated females). ANOVA with the Unpaired t-test with Welch’s correction was utilized to compare differences between vehicle and coumestrol treated males and females.

Results

Coumestrol Exposure Regulates Pathways Important for Trophoblast Cell Function

To examine the effect of coumestrol exposure on the human trophoblast cell transcriptome, immortalized human trophoblast cells (HTR8/SVneo) were treated for 24 hours with coumestrol or vehicle at a dose of 65μM. The dosage of coumestrol was selected based on the induction of estrogen-responsive genes in this cell line (data not shown). RNA was isolated and hybridized to the Affymetrix Clariom S Human array. Utilizing a fold change of 1.4 with a p-value≤0.05, 1,888 genes were upregulated while 1,191 were downregulated (Figure 1A) (n=3 technical replicates). (Data is publicly accessible at NCBI GEO #GSE183072.) After performing Ingenuity Pathway Analysis, the top changed signaling pathways were ranked by −log(p-value) and plotted by z-score which indicates the fold and directionality of change (Figure 1B). Of the top 20 changed pathways determined using Ingenuity Pathway Analysis, 10 pathways were identified with common themes including oxidative stress related, cell cycle, cell migration, and placental/angiogenic signaling (Figure 1B). The significantly changed gene lists representing these signaling pathways are categorically reported (Figure 1C). Significant increases in placental growth factor (PGF) and vascular endothelial growth factor A (VEGFA), involved in the cell migration pathway, were validated using quantitative real-time PCR (qRT-PCR) (n=3 biological replicates) (Figure 1D-E). Thus, coumestrol administration causes changes in oxidative stress response, cell cycle and migration, and placental signaling pathways in human trophoblast cells.

Fig. 1.

Human trophoblast HTR8/SVneo cells were treated with coumestrol (65μM) for 24 hours and RNA was isolated and hybridized to an RNA microarray (n=3 technical replicates). Differentially regulated genes were determined and graphically depicted on a heatmap (Figure 1A) displaying 1,888 genes upregulated and 1,191 downregulated using a 1.4 fold change with a pvalue≤0.05. Of the top 20 changed pathways, 10 pathways were identified that exhibited similar cellular functions (Figure 1B). The pathways were ordered by –log(p-value) or rank of significance. Z-Score describes the fold and directionality change of each pathway. The significant list of genes representing these changed pathways is depicted categorically in Figure 1C. qRT-PCR relative message levels for placental growth factor (PGF) and vascular endothelial growth factor A (VEGFA) are depicted in Figure1D-E (n=3 biological replicates). Red lines represent mean with SEM. *=p-value≤0.05, ***=p-value≤0.001

Coumestrol Impairs Cell Proliferation and Promotes Cell Death

To further understand how coumestrol impacts cell function and survival, Edu was used to measure proliferation in vehicle and coumestrol treated HTR8/SVneo cells. After a 24 hour treatment of vehicle or coumestrol with an included 4 hour Edu incubation, cells were fixed and permeabilized. Cells were then stained for Edu incorporation, fragmented DNA puncta intensity via TUNEL reaction, and DAPI nuclear staining. Proliferating cell number and TUNEL positive puncta were normalized to the total cell number in the field. Upon coumestrol treatment, the percent of proliferating cells was decreased (Figure 2A-C) and the TUNEL positive puncta percentage was increased (Figure 2D-F) (n=3 biological replicates). Therefore, coumestrol causes a reduction in cell proliferation and increased cell death in trophoblast cells.

Fig.2.

HTR8/SVneo cells were treated with vehicle or coumestrol for 24 hours. Proliferating cells were stained for Edu incorporation, fragmented DNA puncta via TUNEL reaction, and Hoechst for nuclear staining. Proliferating cells divided by total nuclei values were normalized to the mean of the vehicle treated samples within an individual experiment, averaged per well, and plotted (Figure 2A). Inverted images display vehicle and coumestrol treated Edu positive cells (Figure 2B-C). TUNEL intensity divided by total cells per field were also normalized to the mean of the vehicle treated samples within an experiment, averaged per well, and plotted (Figure 2D). Inverted images display vehicle and coumestrol treated TUNEL positive regions (Figure 2E-F). (n=3 biological replicates). Scale Bar=78μm. Red lines represent mean with SEM. ***=p-value≤0.001

Trophoblast Cell Migration is Reduced Upon Coumestrol Treatment

Since coumestrol treatment altered pathways important for cell migration, we directly evaluated the effects of coumestrol on trophoblast cell movement. A cell migration wound healing assay was performed on HTR8/SVneo cells treated with vehicle or coumestrol for 6 hours (Figure 3A-B). After 6 hours, the coumestrol treated cells migrated a significantly shorter distance into the scratched area compared to vehicle treated control cells (Figure 3C-E) (n=3 biological replicates). Thus, coumestrol treatment caused impaired trophoblast cell migration.

Fig. 3.

Cells were treated with vehicle or coumestrol and a scratch was made on the cell plate to represent a wound (Figure 3A-B). Cells were measured at 0 and 6 hours (Figure 3A-D). Migration distance was calculated for vehicle and coumestrol treated cells, averaged per well, and graphed on Figure 3E (n=3 biological replicates). Scale Bar=156μm. Red lines represent mean with SEM. ***=p-value≤0.001

Coumestrol Exposure Promotes Free Radical Accumulation

Since many of the top differentially regulated pathways from the microarray analysis were related to the oxidative stress response, we sought to examine the level of free radicals present in coumestrol treated trophoblast cells. To measure the levels of superoxide anion and hydroxyl radicals, the CellROX Deep Red Reagent was used in live HTR8/SVneo cells treated with coumestrol. After 24 hours, coumestrol treated cells exhibited a slightly larger cellular morphology compared to vehicle (Figure 4A-B with zoomed insets in Figure 4C-D). Moreover, CellROX Deep Red Reagent staining was significantly increased in coumestrol treated cells compared to vehicle, demonstrating that coumestrol induces the accumulation of free radical species within trophoblast cells (Figure 4E) (n=4 biological replicates).

Fig.4.

Upon 24 hours of coumestrol administration, CellROX Deep Red Reagent (depicted as “ROS stain”) was used to stain superoxide anions and hydroxyl radicals in cultured HTR8/SVneo cells. Inverted cell images with positive ROS stain for vehicle (Figure 4A) and coumestrol (Figure 4B) are depicted with magnified insets (Figures 4C-D). The ratio of ROS stain intensity to number of nuclei was calculated, averaged per well, and graphed for each treatment well (Figure 4E), (n=4 biological replicates). Scale Bar=50μm. Red lines represent mean with SEM. ***=p-value≤0.001

Coumestrol Causes Reduced Fetal and Placental Weights at Mid-Gestation

To examine effects in vivo, coumestrol (200 μg/kg) was administered daily to wildtype female mice from day 0.5 to day 12.5 of pregnancy. No difference was observed in the number of resorbed/dead embryos at day 12.5 in coumestrol animals compared to control. However, embryos and placentas were smaller upon treatment with coumestrol (Figure 5A-D) (n=57 biological replicates for vehicle and 51 for coumestrol) with a reduction in embryo to placental weight ratio (Figure 5E) (n=56 biological replicates for vehicle treated and 50 for coumestrol treated). Genotyping for embryonic sex was performed to determine the effect of coumestrol treatment on sex-specific embryonic and placental weight. Coumestrol treatment caused a reduction in fetal weight in both male and female offspring that was not statistically significant (Figure 5F); yet, female weights exhibited pronounced variability and dramatic decreases. Interestingly, coumestrol specifically impaired placental weight in male fetal-placental pairs compared to vehicle (Figure 5G) (n=25 biological replicates for vehicle treated males, 14 for vehicle treated females, 15 for coumestrol treated males, and 18 for coumestrol treated females). Placental tissues were sectioned and stained with platelet and endothelial cell adhesion molecule 1 (PECAM1) to label the labyrinth (lab) and with Dolichos Biflorus Agglutinin (DBA)-lectin to distinguish the uterine natural killer cells present in the maternal decidua (dec) (Figure 5B). The junctional zone (jz) was demarcated by the presence of trophoblast giant cells stained with DAPI and by the absence of PECAM1 and DBA-lectin staining. The ratios and sizes of the placental compartments were measured using Fiji and no difference was identified between the vehicle or coumestrol groups (Figure 5H-M) (n=12 biological replicates). The relative gene expression of Vegfa and Pecam1 from total placenta isolates were measured and determined to be unchanged (Figure 5O-P) (n=5 biological replicates for vehicle, 6 for coumestrol). Thus, coumestrol caused a reduction in fetal and placental weights with placentas from males exhibiting a specific and significant decrease. However, the reduced placenta size represents a proportional decrease across all placental compartments with no changes in relative gene expression of Vegfa or Pecam1.

Fig. 5.

Wildtype dams were treated with vehicle or coumestrol (200 μg/kg) by oral gavage from day 0.5 to day 12.5 of pregnancy. Upon euthanasia at day 12.5, fetal size and morphology were recorded (Figure 5A,C). Placental weights were measured and tissue was stained for morphology with PECAM1 to stain the fetal labyrinth (lab), DAPI to stain trophoblast giant cells in the junctional zone (jz), and DBA-LECTIN to stain uterine natural killer cells found in the maternal decidua (dec) (Figure 5B). Embryo and placental weight differences were tabulated between vehicle and coumestrol treated animals (Figure 5C-D) (n=57 biological replicates for vehicle treated, n=51 for coumestrol treated). Embryo to placental weight ratios were calculated accordingly (Figure 5E) (n=56 biological replicates for vehicle treated, n=50 for coumestrol treated). Embryonic and placental weight was plotted based on sex and treatment (Figure 5F-G) (n=25 biological replicates for vehicle treated males, 14 for vehicle treated females, 15 for coumestrol treated males, and 18 for coumestrol treated females). Placental compartment percent area and length were measured using Fiji software and graphed (Figure 5H-M) (n=12 biological replicates). Relative gene expression was measured for Vegfa and Pecam1 in total placenta tissue (n=5 biological replicates for vehicle, 6 coumestrol). Scale Bar for Figure 5A=1mm, Figure 5B=200μm. Red lines represent mean with SEM.*=p-value≤0.05, **=p-value≤0.01

Discussion

The popularity of vegetarian and vegan diets has risen in recent years, especially the consumption of soy products. Pregnant women are encouraged to consume soy (Miyake Y, Tanaka K et al., 2021), yet our understanding of the positive and negative effects of these soy-based compounds are still in its infancy. Due to this limited understanding, we sought to examine whether coumestrol affects placental cell function and fetal health. Within human trophoblast cells, we observed coumestrol significantly regulates over 3,000 genes including signaling pathways involved in oxidative stress response, cell cycle regulation, cell migration, and angiogenesis. This work demonstrated that coumestrol treatment caused trophoblast cells to migrate and proliferate less, while eliciting an increased presence of fragmented DNA, indicative to cellular death. Furthermore, after coumestrol administration, trophoblast cells accumulated a higher amount of ROS. Upon in vivo administration of coumestrol to wildtype mice during pregnancy, fetal and placental weights were reduced with placental compartments decreased proportionally. Thus, we conclude that the phytoestrogen coumestrol impairs trophoblast cell function and causes increased ROS in vitro; while in vivo, coumestrol reduces the size of the fetus and placenta.

From the microarray data results, it was expected that estrogenic, cell cycle regulation, and placental derived growth factor (PDGF) signaling pathways were upregulated. Since coumestrol is a phytoestrogen and exhibits estrogenic properties, activation of these pathways was of no surprise. In addition, coumestrol administration caused upregulation of the “NRF2-Mediated Oxidative Stress Response” with concurrent increases in “LPS-stimulated MAPK Signaling” and the “Senescence Pathway”, both of which can be activated as a consequence of increased ROS (Chiarello DI, Abad C et al., 2020, Xiao K, Liu C et al., 2020, Cindrova-Davies T, Fogarty NME et al., 2018). Although the ATM signaling pathway is often a sensor of oxidative stress, in this microarray it is decreased significantly. Thus, coumestrol may concomitantly impair the DNA damage response while inducing the oxidative stress response, observed in certain cellular contexts (Barzilai A, Rotman G et al., 2002). Finally, induction of the “Epithelial Adherens Junction” and the “Integrin Linked Kinase (ILK) Signaling” pathways are strongly associated with cell migration and movement, critically important functions of invading trophoblast cells which govern the success of placentation (Pinheiro D and Bellaiotache Y, 2018, Elustondo PA, Hannigan GE et al., 2006). Thus, coumestrol regulates signaling pathways involved in cell cycle regulation, migration, oxidative stress response, and angiogenesis, processes essential for placental development.

Although others have suggested coumestrol exhibits a protective effect in endothelial cells and cardiovascular disease by eliciting activation of antioxidant pathways (Mann GE, Bonacasa B et al., 2009, Kramer F, Jensen PS et al., 2003, Bonacasa B, Siow RC et al., 2011), our data suggest coumestrol elicits trophoblast cell dysfunction by impairing cell migration and proliferation and permitting the accumulation of ROS. Since many of the reported microarray signaling pathways changed were involved in the oxidative stress response, we examined the level of ROS in coumestrol treated cells. We concluded coumestrol significantly increased the amount of hydroxyl radicals and superoxide anions, commonly found oxidative species in the placenta (Wu F, Tian FJ et al., 2015). In accordance with our data, elevated lipid peroxidation and hydrogen peroxides were identified in placental choriocarcinoma cells treated with coumestrol (Lim W, Yang C et al., 2017). Although the presence of ROS is important for healthy placental development, excessive accumulation of ROS is associated with PE and fetal growth restriction (Chiarello DI, Abad C et al., 2020). Thus, in our studies, coumestrol has a negative effect on overall placental growth and development, as it perturbs trophoblast cell migration and causes accumulation of ROS in vitro, two factors tightly associated with the etiology of PE incidence (Wu F, Tian FJ et al., 2015). In concurrence with the in vitro data, administration of coumestrol during pregnancy caused deleterious effects in vivo. Coumestrol treatment resulted in decreased placental and fetal weights, yet the placentas exhibited normal morphology with proportional levels of the placental compartments. Interestingly, in utero coumestrol administration caused a significant and specific reduction in placental weight in male offspring. The presence of coumestrol in the male fetal-placental unit may elicit prioritization of an estrogenic transcriptome profile over an androgenic profile resulting in modulation of placental weight (Meakin AS, Cuffe JSM et al., 2021). No changes were observed in relative gene expression for both angiogenic factors, Vegfa and Pecam1. Therefore, further studies are needed to examine how coumestrol regulates specific cell types in the placenta and to examine the mechanistic pathways resulting in the decreased fetal and placental weights observed in these studies.

As an endocrine disruptor, variable concentrations of coumestrol can result in differing effects depending on the tissue/organ environment. This study is limited to understanding the impact of coumestrol regulation during pregnancy at a high 65 μM concentration, though the precise levels of coumestrol in the serum were not measured. Although deemed safe for humans (Kirihata Y, Kawarabayashi T et al., 2008), this concentration of coumestrol may impart different species effects, including toxicity in mice. As an example of species differences, Lim et al. observed increased proliferation and migration in cultured porcine trophectoderm cells (Lim W and Song G, 2016) but decreased proliferation in cultured human immortalized choriocarcinoma cells (Lim W, Yang C et al., 2017). Others have examined in human populations that increased isoflavone levels correlate with decreased metabolic disease (Liu J, Mi S et al., 2018). Thus, coumestrol in lower doses may exhibit variable effects compared to the findings described in this manuscript. Further work is needed to understand the implications of multiple coumestrol concentrations in an in vivo environment during pregnancy.

Given the popularity of vegetarian diets especially in reproductive aged and pregnant women, it is important to understand the function of environmental endocrine disruptors including soy-based phytoestrogens, such as coumestrol. These studies uniquely provide critical evidence describing the effect of coumestrol on placental function and fetal health. Coumestrol administration not only prevented trophoblast cell migration but also caused the accumulation of dangerous ROS molecules, factors known to contribute to PE pathology. Additionally, in vivo treated dams experienced smaller fetal and placental weights. These results collectively urge a careful evaluation of the effects of consuming high levels of soy-based products during pregnancy.