lable at ScienceDirect

Food and Chemical Toxicology 98 (2016) 50e57

Contents lists avai

Food and Chemical Toxicology

journal homepage: www.elsevier .com/locate/ foodchemtox

miRNAs as common regulators of the transforming growth factor(TGF)-b pathway in the preeclamptic placenta and cadmium-treatedtrophoblasts: Links between the environment, the epigenome andpreeclampsia

Samira A. Brooks a, Elizabeth Martin a, Lisa Smeester a, Matthew R. Grace b, Kim Boggess b,Rebecca C. Fry a, c, *

a Department of Environmental Sciences and Engineering, Gillings School of Global Public Health, 135 Dauer Drive, CB 7431, University of North Carolina,Chapel Hill, NC, USAb Department of Obstetrics & Gynecology, University of North Carolina School of Medicine, University of North Carolina, Chapel Hill, NC, USAc Curriculum in Toxicology, School of Medicine, University of North Carolina, Chapel Hill, NC, USA

a r t i c l e i n f o

Article history:Received 15 June 2016Accepted 21 June 2016Available online 29 June 2016

Keywords:PreeclampsiaTGF-bJEG-3 cellsGene expressionEpigeneticsmiRNA

* Corresponding author. Department of EnvironmenGillings School of Global Public Health, 135 Dauer DNorth Carolina, Chapel Hill, NC, USA.

E-mail address: rfry@unc.edu (R.C. Fry).

http://dx.doi.org/10.1016/j.fct.2016.06.0230278-6915/© 2016 The Author(s). Published by Elsevie).

a b s t r a c t

Preeclampsia (PE) is a pregnancy disorder characterized by high blood pressure and proteinuria that cancause adverse health effects in both mother and fetus. There is no current cure for PE other than deliveryof the fetus/placenta. While the etiology is unknown, poor placentation due to aberrant signaling ofgrowth and angiogenic factors has been postulated as a causal factor of PE. In addition, environmentalcontaminants, such as the metal cadmium (Cd), have been linked to placental toxicity and increased riskof developing PE. Here, we use a translational study design to investigate genomic and epigenomic al-terations in both placentas and placental trophoblasts, focused on the angiogenesis-associated trans-forming growth factor-beta (TGF-b) pathway. Genes within the TGF-b pathway displayed increasedexpression in both the preeclamptic placenta and Cd-treated trophoblasts. In addition, miRNAs thattarget the TGF-b pathway were also significantly altered within the preeclamptic placenta and Cd-treatedtrophoblasts. Integrative analysis resulted in the identification of a subset of Cd-responsive miRNAs,including miR-26a and miR-155, common to preeclamptic placentas and Cd-treated trophoblasts. ThesemiRNAs have previously been linked to PE and are predicted to regulate members of the TGF-b pathway.Results from this study provide future targets for PE treatment.© 2016 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND

license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

1. Introduction

Preeclampsia (PE) is a pregnancy-related condition character-ized by hypertension and proteinuria during gestation, usuallyaccompanied by damage to other organ systems, such as the kid-neys (Chaiworapongsa et al., 2014). PE affects between 5 and 8% ofpregnancies and accounts for approximately 76,000 and 500,000deaths of women and fetuses worldwide, respectively (Berg et al.,2010). While a precise etiology of PE is unknown, it is postulatedthat an underlying factor may be poor placentation due to

tal Sciences and Engineering,rive, CB 7431, University of

r Ltd. This is an open access article

inadequate angiogenesis (Chaiworapongsa et al., 2014; Saito andNakashima, 2014).

An array of ligands and receptors are known to tightly controlangiogenesis. For example, growth factors, chemokines, cytokines,and endogenous angiogenesis inhibitors has been observed in theextracellular matrix (ECM) during vascularization (Brooks andRathmell, 2014). Reduced placental blood flow upregulates theexpression of hypoxia inducible factor 1 alpha (HIF1A), vascularendothelial growth factor (VEGF), VEGF receptor (VEGFR),angiopoeitin-2 (ANGPT2), fibroblast growth factor-3 (FGF-3), nitricoxide synthase (NOS), and transforming growth factors (TGF-a, TGF-b1, TGF-b3) (Harris, 2002). However, in PE, the expression of anti-angiogenic factors such as fms-like tyrosin kinase-1 (FLT1) and itssoluble form (sFLT1) are increased allowing them to bind to pro-angiogenic factors and inhibit angiogenic signaling (Nikuei et al.,

under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/

S.A. Brooks et al. / Food and Chemical Toxicology 98 (2016) 50e57 51

2015). Proper placentation and vascular bed expansion has beenfound to be dependent on the expression of VEGF (Chaiworapongsaet al., 2014). On the other hand, the TGF-beta superfamily has beenobserved to oppose trophoblastic migration (Jones et al., 2006),suggesting an intricate balance of growth factor expression that isfundamental for placental health and embryo development (Nikueiet al., 2015; Conti et al., 2013).

Environmental exposure to toxic metals has been linked toplacental toxicity/dysregulation causing improper vascularizationin the placenta (Esteban-Vasallo et al., 2012; Lazebnik et al., 1989;Pollack et al., 2014). We have recently demonstrated a significantassociation between elevated levels of placental cadmium (Cd) andincreased odds of developing PE (Laine et al., 2015), supporting anexisting literature of this relationship (Pollack et al., 2014). Expo-sure to Cd can occur through multiple sources, including contam-inated food and water, as well as cigarette smoke (Satarug et al.,2010). Alterations to the epigenetic landscape and genomic insta-bility are potential detrimental human health effects of Cd expo-sure. Disruption of genomic stability has been highlighted as afeature of PE, which includes aberrant miRNA expression (Dai et al.,2012; Gao et al., 2012; Gunel et al., 2011; Ishibashi et al., 2012; Lazaret al., 2012; Li et al., 2013; Noack et al., 2011; Wang et al., 2012; Wuet al., 2012; Yang et al., 2011; Zhu et al., 2009) and DNAmethylationalterations of the TGF-b pathway (Martin et al., 2015). Additionalstudies are needed to understand the underlying biology of Cd-associated risk of PE.

In the present study, we used an integrated genomic analysis toexamine Cd-driven regulation of the TGF-b pathway in PE, utilizingboth clinical samples as well as in vitro experimentation inplacental trophoblast cells. Our analysis incorporated placentalexpression of TGF-b pathway-associated genes and miRNA withplacental Cd levels, and Cd-responsive TGF-b pathway-associatedgenes and miRNA in trophoblasts. This study aimed to examinethe direct effects that Cd has on the genome and epigenome ofplacental cells in order to further elucidate the underlying biologythat supports the pathology of PE and to provide future therapeutictargets for treatment.

2. Methods

2.1. Placental sample collection

Placentas from a total of 32 women (16 normotensive controlsand 16 preeclamptic cases) were included in this study. The womenreceived obstetric care at UNC hospitals and consented to collectionof samples at the time of birth as detailed in our previous study(Martin et al., 2015). The American Congress of Obstetricians andGynecologists classified preeclampsia as sustained de novo hyper-tension (>140/90 mmHg) and proteinuria (>300 mg of protein in a24 h urine collection or urine protein/creatinine ratio of 0.3 mg/dL)after 20 weeks of pregnancy. All preeclamptic women displayedsevere features of BP’ > 160/100 with neurologic dysfunction, renaldysfunction, and/or evidence of HELLP syndrome. Women withconfounding conditions such as pre-diabetes, diabetes, and gesta-tional diabetes were excluded from the study. A full-thicknessplacental biopsy was obtained after delivery, avoiding the periph-ery and areas of obvious infarction, flash frozen in liquid nitrogen,and subsequently stored at �80 �C until analysis. Placental Cdconcentrations were measured using ICP-MS as detailed (Laineet al., 2015). This research was approved by the Institutional Re-view Board at the University of North Carolina (#11e2054).

2.2. Cell culture methodology

JEG-3 cells, a human placental trophoblast cell line, were used

for all cell culture analyses. This cell line maintains characteristicsof first trimester trophoblast cells (Matsuo and Strauss, 1994) andhas been used to assess biological effects of exposures to environ-mental toxicants (Bergemann et al., 2003; Canettieri et al., 2008;Ding et al., 2011; Koc et al., 2003; Matscheski et al., 2006;Adebambo et al., 2015; Guan et al., 2013; Kummu et al., 2012;Alvarez and Chakraborty, 2011). Cells were grown in Dulbecco’sEagle Minimum Essential Media with 10% FBS, sodium pyruvate,and penicillin/streptomycin at 37 �C with 5% CO2. Cd chloride(CdCl2) was dissolved into the media by vortex for final concen-trations of 1, 10, and 25 mM.

2.3. Cell viability assay

Cells were seeded in 96-well black clear bottom plates at10,000/mL with a total of 200 mL of media. Dilutions of Cd chloridewere carried out starting at 2000 mM until a final concentration of0.001 mM was reached. Experiments were carried out in biologicaltriplicate. Following a 24 or 48 h incubation, media was removedand cell viability measured by the CellTiter-Glo Luminescent CellViability Assay (Promega, Madison, WI). Luminescence wasmeasured using a Promega GloMax Microplate Luminometer. IC10and IC50 concentrations were established and used for subsequentexperimentation.

2.4. Gene and miRNA expression

For clinical samples, a 0.2 g subsection of placental tissue wascut from each frozen biopsy on dry ice, washed briefly in sterile 1XPBS to remove any residual blood, and homogenized in Buffer RLTwith B-mercaptoethanol (Qiagen, Valencia CA). DNA and RNA se-quences greater than 18 nucleotides in length were collected usingthe AllPrep DNA/RNA/miRNA Universal Kit (Qiagen, Valencia CA)according to manufacturer’s instructions. RNA abundance wasanalyzed using the Affymetrix GeneChip® Human Gene 2.0 ST arrayas described previously (Rager et al., 2014). Data were RMA pro-cessed using Partek Genomics Suite 6.4 (St Louis, Missouri). Thesedata are available at the Gene Expression Omnibus (GEO)(GSE73377). For miRNAs, RNA was amplified, labeled, and hybrid-ized on Agilent Human miRNA (8 � 60 k) Oligo Microarrays. Thegreen median signal was extracted and quantile normalized usingPartek Genomics Suite 6.4 (St Louis, Missouri). These data areavailable at the Gene Expression Omnibus (GEO).

For the in vitro assays, both large and small RNAs were extractedsimultaneously from JEG-3 cells using Qiagen’s Allprep kit allowingthe purification of a single fraction containing all RNA �18 nucle-otides. The quantity of isolated RNA was measured with theNanodrop 1000 spectrophotometer, and its integrity verified by theAgilent 2100 Bioanalyzer. A total of 200 ng of RNA was amplified,labeled, and hybridized on Agilent Human miRNA v.16 (8 � 60 k)Oligo Microarrays. The green median signal was extracted andprobes that expressed values less than 1 standard deviation fromthe mean of the negative controls in more than 50% of the sampleswere removed. Resulting data were quantile normalized usingPartek Genomics Suite 6.4 (St Louis, Missouri) and used for furtheranalyses.

2.5. Real-time PCR analysis

cDNAwas generated using either Qiagen’s RT2 First Strand kit ormiScript II RT Kit for mRNA and miRNA analysis, respectively.Quantitative real-time reverse transcriptase polymerase chain re-action (qRT-PCR) was performed and assessed using the TGF-b/BMP Signaling Pathway RT2 Profiler PCR Array (Valencia, CA), whichtests 84 genes related to TGF-b signal transduction. Raw CT values

S.A. Brooks et al. / Food and Chemical Toxicology 98 (2016) 50e5752

from the RT2 Profiler PCR Arrays were analyzed for subsequentstatistical testing. Validation of miRNAs was completed using Qia-gen’s miScript Primer Assays and SYBRGreen PCR kits. Experimentswere carried out in biological triplicate, and any outliers removedprior to statistical analyses. The miRNA U6 was used as a house-keeping gene.

2.6. Statistical analysis

To assess gene expression and miRNA differences betweenpreeclamptic versus non-preeclamptic placentas, multivariablemodels were applied to the 53,637 probes on the Affymetrix Gen-eChip® Human Gene 2.0 ST array and the 56,044 probes on theAgilent Human miRNA v.16 (8 � 60 k) Oligo Microarrays. Thesemodels controlled for gestational age, maternal age and race. Dif-ferential RNA expression was statistically defined using a falsediscovery rate (FDR)-corrected q� 0.15. Cd-associatedmRNAswereanalyzed separately among cases and controls using multivariableregression models. The models controlled for maternal age, gesta-tional age, race, and smoking status. Statistical significance wasdefined as q � 0.15. A database of miRNAs (n ¼ 651) was identifiedin association with canonical members of the TGF-b pathwayinvolved in signal transduction using TargetScan7.0 (Lewis et al.,2003) and miRBase (Landgraf et al., 2007). Pearson correlationanalysis was conducted to determine associations betweenmiRNAsand their target mRNAs. Statistical significance was set at p < 0.05.

T-tests were performed to assess statistical significance of theTGF-b Signaling Targets RT2 Profiler PCR Array data using Qiagen’sData Analysis Portal. CT values of the miRNA validation were usedto calculate 2̂ (-delta CT) values for cells exposed to 1 mM or 10 mMCdCl2 for 48 h and controls and used for analysis of covariance(ANCOVA) to assess for differential miRNA expression. Differen-tially expressed miRNA transcripts were considered statisticallysignificant if q � 0.15.

3. Results

3.1. Dysregulation of TGF-b pathway-targeting miRNAs in thepreeclamptic placenta

Gene expression was analyzed by genome-wide microarraywith a subsequent focused analysis on 84-targeted TGF-b pathway-associated genes for 16 normotensive (control) and 16 preeclamptic(cases) placentas, which resulted in the identification of a total of 31TGF-b pathway genes that were significantly increased in the pre-eclamptic placenta (Supplemental Table 1). We set out to assesswhether miRNAs may regulate the expression of these TGF-b-associated genes. Of 651 compiled miRNAs that target canonicalmembers of the TGF-b pathway, 109 were identified as differen-tially expressed in the preeclamptic placenta. ThesemiRNAs tendedto be upregulated in the preeclamptic placenta, where 63 miRNAsdisplayed increased expression and 46 miRNAs displayeddecreased expression (Fig. 1A, Supplemental Table 2). These iden-tified TGF-b pathway-associatedmiRNAs were a subset of a broadergenome-wide miRNA analysis where a total of 631 out of 861 (73%)tested human miRNAs were differentially expressed in the pre-eclamptic placenta, 17% (109 out of 631) of which target the TGF-bpathway. Among these miRNAs are 52 that have previously beenshown to be altered in preeclampsia (Harapan and Yeni, 2015; Chenand Wang, 2013) (Supplemental Table 2).

We next set out to examine the relationship between the TGF-bpathway-associated miRNAs and their target genes of interest.Among the 31 target genes were transforming growth factor beta-1(TGFB1), transforming growth factor beta receptor 1 and 2 (TGFBR1/2), SMAD family member 2 (SMAD2), SMAD family member 3

(SMAD3), SMAD family member 4 (SMAD4), SMAD specific E3ubiquitin protein ligase 1 (SMURF1), FBJ murine osteosarcoma viraloncogene homolog (FOS), serpin peptidase inhibitor, clade E,member 1 (SERPINE1), and growth arrest and DNA damage induc-ible beta (GADD45B). When examining the relationship of TGF-btargeting miRNAs to their target genes of interest, Pearson corre-lation analysis identified a total of 128 significant miRNA-mRNArelationships; the identified associations were both negative(n ¼ 50) and positive (n ¼ 78) (Supplemental Table 3). While therelationship between the expression of miRNAs and their targetmRNAs is generally expected to have an inverse association, thesedata highlight the complex mechanisms of miRNA regulation(Supplemental Table 3). Of the 31 TGF-b pathway-associated genes,most (n ¼ 23, 74%) displayed a significant association with a TGFb-associated miRNA suggesting a role for miRNAs as mediators of thispathway.

3.2. Cd-associated dysregulation of TGF-b pathway-targetingmiRNAs in the preeclamptic placenta

Next, we performed multivariable regression analysis to testwhether miRNA expression in the placenta was associated withplacental Cd concentrations. Cd levels were assessed in bothnormotensive subjects (controls; range: 1.01e10.42 ng/g) andpreeclamptics (cases; 0.35e8.01 ng/g) and examined for their re-lationships to altered miRNA expression. As a result, 441 and 344miRNAs were identified in controls and cases, respectively (Fig. 1B).miRNAs that target the TGF-b pathway had robust expressionchanges in relation to Cd levels, with a subset of miRNAs (n ¼ 29)common to both cases and controls (Fig. 1C). Cd levels were asso-ciated with substantial changes in the expression of miRNAs tar-geting the TGF-b pathway in cases and controls, potentiallyhighlighting distinct Cd-induced miRNA regulation that drive thedevelopment of PE.

3.3. Cd induces gene expression of the TGF-b pathway members introphoblast cells

To better understand the relationship between Cd exposure andthe regulation of the TGF-b pathway in the placenta, we performedin vitro experiments utilizing JEG-3 cells, a human placentaltrophoblast cell line. JEG-3 cells were exposed in culture to mini-mally cytotoxic (<5% cell death) concentrations of 1 and 10 mMCdCl2 for 48 h. These concentrations are similar to previous studiesthat have used low Cd concentrations (�1 mM) (Adebambo et al.,2015; Guan et al., 2013; Kummu et al., 2012) to evaluate biolog-ical disruptions in placental cells as a result of treatment, particu-larly anti-migratory effects that can influence risk of PE (Alvarezand Chakraborty, 2011). Gene expression was analyzed by RT-PCRfor 84-targeted TGF-b pathway-associated genes using RNA iso-lated from untreated JEG-3 cells or JEG-3 cells treated with Cd(Supplemental Table 4). Following a 48 h incubation, five uniquegenes were significantly expressed in JEG-3 cells treatedwith either1 mM or 10 mM Cd concentrations compared to untreated cells(Fig. 2A), with most (n ¼ 3) genes displaying increased expressionin response to Cd.

We next analyzed the effects of Cd treatment on the TGF-bpathway-associated genes using the moderately cytotoxic (<50%cell death) concentration of 25 mM CdCl2 for 24 and 48 h (Fig. 2B).This concentration was selected as we aimed to contrast thesecellular responses in vitro to PE, a disease attributed to placentaltoxicity. A total of 52 out of 84 genes tested (62%) displayedsignificantly altered expression after the 24 or 48 h Cd treatments(Fig. 2B). Most of the significantly changed genes displayedincreased expression in response to a 48 h treatment of Cd.

A.

Cases

Expr

essi

on o

f TG

Fβ p

athw

ay

targ

etin

g m

iRN

As (n

=109

)

miRNA z-score

Controls

-0.5 0.00 0.5

Cd-associated miRNAs in cases (n=344) ( )

Cd-associated miRNAs in controls (n=441)

139 miRs

193 miRs

78 miRs

73 miRs

All significant miRNAs

109 miRs

20 miRs 18 miRs

16 miRs

C. Significant TGFβ-targeting miRNAs

36 miRs 29 miRs

B. Cd-associated miRNAs

in cases (n=63) Cd-associated miRNAs

in controls (n=85)

Unique miRs n=56 Unique miRs n=34 Unique miRs (n=248) Unique miRs (n=151)

Fig. 1. miRNA regulation of the TGF-b pathway regulation in the preeclamptic placenta. A. Differentially expressed TGF-b targeting miRNAs (n ¼ 109) from whole-genomemiRNA microarray analysis between normotensive (Controls) and preeclamptic (Cases) placentas. B. All significantly expressed miRNAs associated with Cd exposure in Controls(n ¼ 441) and Cases (n ¼ 344). C. Differentially expressed TGF-b targeting miRNAs in Controls (n ¼ 85) and Cases (n ¼ 63) associated with Cd exposure.

B.

48-hour 25uM

Fold-Change

24-hour 25uM

-4.0 0.00 4.0

Gen

e ex

pres

sion

of

TGFβ

pat

hway

(n=5

2)

A.

10uM

Fold-Change

1uM

-8.0 0.00 8.0

Gen

e ex

pres

sion

of

TGFβ

pat

hway

(n=5

)

AMH

BMP5

TGFBI

TSC22D1

GADD45B

Fig. 2. Cd alters the expression of the TGF-b pathway in placental trophoblast cells. Quantitative PCR results of differentially expressed genes within the TGF-b pathway in JEG-3placental cells following treatment with 1 mM or 10 mM CdCl2 for 48 h (A) and 25 mM CdCl2 for 24 and 48 h (B).

S.A. Brooks et al. / Food and Chemical Toxicology 98 (2016) 50e57 53

Specifically, at the 48 h time point 46 out of 49 genes showed anincrease in expression, while 6 out of 10 genes were increased afterthe 24 h treatment.

3.4. Cd activates the TGF-b pathway targeting miRNAs introphoblast cells

Next, to investigate the regulation of the TGF-b pathway bymiRNAs, a genome-wide miRNA microarray analysis wascompleted using cells treated with 48 h 25 mM CdCl2 since it dis-played the most robust genomic response to Cd. The expressionlevels of 99 out of 651 TGF-b pathway-targeting miRNAs displayedsignificantly differential expression. The majority of these miRNAsdisplaying decreased expression (n ¼ 64) in relation to Cd (Fig. 3A,

Supplemental Table 1). These data demonstrate a general inverserelationship between Cd-induced gene expression of the TGF-bpathway and the Cd-induced repression of miRNAs that target thepathway. However, there were both positive (n ¼ 36) and negative(n ¼ 65) correlations between the expression of TGFB-associatedmiRNAs and their mRNA targets (Supplemental Table 5). Thesechanges were a subset of a broader genome-wide Cd-inducedmiRNA response where 466 miRNAs were significantly altered,with 181 upregulated and 285 downregulated (SupplementalTable 1). Similar to the preeclamptic placenta, approximately 20%(99 out of 466) of the altered miRNAs in Cd-treated placentaltrophoblast cells target the TGF-b pathway.

To further examine Cd-responsive miRNAs in trophoblaststreated with the minimally cytotoxic concentrations, a total of six

Fig. 3. Expression of miRNAs that regulate the TGF-b pathway in Cd-treated trophoblast cells. A. Differential expression of TGF-b targeting miRNAs (n ¼ 99) from whole-genome miRNA microarray analysis in JEG-3 placental trophoblast cells treated with 25 mM CdCl2 for 48 h. B. Quantitative PCR results of selected miRNAs in JEG-3 cells treatedwith 1 mM or 10 mM CdCl2 for 48 h miR-155, miR-187*, and miR-3677 were significantly decreased in the 1 mM group. The 10 mM group displayed significant downregulation of miR-187*. Cells treated with 1 mM or 10 mM CdCl2 were compared to untreated cells to determine fold-change. ANOVA analysis was used to determine significance (*) that was defined asa p-value less than 0.05 and a q-value less than 0.15.

S.A. Brooks et al. / Food and Chemical Toxicology 98 (2016) 50e5754

miRNAs were analyzed by quantitative PCR for their expression inJEG-3 cells treated with 1 mM or 10 mM concentrations of CdCl2 for48 h (Fig. 3B). These miRNAs were selected because they were TGF-b-associated miRNAs significantly expressed in response to Cd inthe preeclamptic placenta (miR-31* and miR-3677-5p), TGF-b-associated miRNAs previously established to be expressed in pre-eclamptic placenta (miR-26a and miR-155), or miRNAs that hadsignificant expression in JEG-3 cells exposed to 25 mM Cd for 48 h(miR-3663-3p and miR-187*). Following ANOVA analysis, miR-155,miR-187*, and miR-3677-5p were significantly downregulatedcompared to control in the 1 mM group, while miR-187* hadsignificantly decreased expression in the 10 mM group compared tocontrol (Fig. 3B). Similar to the gene expression data, the minimallycytotoxic Cd concentrations yielded fewer significantly alteredmiRNAs in Cd-treated trophoblasts compared to the moderatelycytotoxic Cd concentration. Interestingly, the TGFB-associated miR-3677-5p was significantly decreased in the 1 mM group while itspredicted gene target bone morphogenetic proteins 5 (BMP5)(Garcia et al., 2011) was substantially increased (fold-change ¼ 3.8)in response to Cd (Fig. 2A), suggesting a Cd-driven miRNA mech-anism for the regulation of the TGF-b pathway.

3.5. A comparison of Cd-associated dysregulation of the TGF-bpathway genes and targeting miRNAs in the preeclamptic placentaand placental trophoblast cells

After completing miRNA analysis in both placentas andplacental cells, the 109 differentially expressed TGF-b pathway-targeting miRNAs in the preeclamptic placenta and the 99 TGF-bpathway-targeting miRNAs expressed in the 48 h 25 mM Cd-treatedtrophoblasts were compared. Many (n ¼ 58) were common to bothgroups (Fig. 4A). Furthermore, 15 TGF-b pathway-targeting miRNAswere associated with Cd in placental trophoblasts cells, thenormotensive placenta, and the preeclamptic placenta (Fig. 4B).miRNAs expressed in all three Cd-associated groups generally tar-geted ligands that activate the TGF-b pathway such as BMP2 andBMP5, as well as TGF-b pathway members responsible for signaltransduction such as TGFBR1, SMAD1, SMAD2, and SMAD5. Specif-ically, miR-26a and miR-155, which target SMAD1 and SMAD2respectively, were significantly differentially expressed in all threegroups, with miR-155 significantly decreased in 48 h 1 mM Cd-treated trophoblast as well. Furthermore, these miRNAs wereobserved to be associated with PE (Harapan and Yeni, 2015),

providing further support for a possible mechanism underlying Cd-associated PE.

Strong similarities of gene expression involving the TGF-bpathway were observed between placentas and placental cells.Notably, 26 significantly expressed TGF-b pathway genes in Cd-treated placental trophoblast cells were present within the 31genes altered in the preeclamptic placenta. Among these were TGF-b pathway members TGFBR1, TGFBR2, BMP2, and BMP5, which playkey roles in the signal transduction of the TGF-b and BMP signalingpathways, and thus SMAD activation. Moreover, the downstreamtargets of these pathways, SMURF1, FOS, SERPINE1, and GADD45B,displayed increased expression in the placenta and Cd-treatedtrophoblasts (Fig. 5).

4. Discussion

In the present study, we set out to elucidate a potential molec-ular mechanism by which Cd exposure increases the risk for PE.Using a translational approach, we integrated cell culture experi-mentation with clinical sample analysis to establish a novel linkbetween the induction of genes within the TGF-b pathway both inthe preeclamptic placenta and in Cd-exposed placental cells.Furthermore, we highlight miRNA suppression that may underliethe induction of the TGF-b pathway in response to Cd as anepigenetic mechanism that mediates expression. Investigating theunderlying biology of PE and potential environmental influencescan provide insight into the specific effects of toxicants on placentalmolecular phenotypes that increase risk of PE, as well as bettertreatment options for patients.

One of the main hypotheses for the development of PE is poorangiogenesis in the placenta leading to a hypoxic environment.Numerous growth factors exhibit aberrant maternal and placentalexpression in preeclamptic patients as compared to women whoare normotensive during pregnancy (Nikuei et al., 2015). One familyof growth factors is TGF-b, which promotes an anti-migratorysignal in trophoblast cells. Here, we analyzed placentas from pre-eclamptic and normotensive women, as well as JEG-3 placentaltrophoblast cells treated with Cd, in order to investigate the in-duction of the TGF-b pathway. TGF-b receptors I and II were amongthe TGF-b pathway-associated genes that displayed increasedexpression in relation to PE as well as placental cells exposed to Cd.TGF-b receptors I and II activate the signal transduction of the TGF-b pathway upon the binding of the TGF-b superfamily of ligands.

Fig. 4. Venn diagram of TGF-b pathway-targeting miRNAs in control or case placentas and trophoblast cells. TGF-b-targeting miRNAs uniquely expressed or overlapping among25 mM Cd-treated placental trophoblast cells and the preeclamptic placenta (A). TGF-b-targeting miRNAs uniquely expressed or overlapping among Cd-treated (25 mM) placentaltrophoblast cells, and placentas (cases and controls) in association with Cd.

Fig. 5. Dysregulation of the TGF-b pathway in the preeclamptic placenta and Cd-treated placental trophoblast cells. Increased (þ) or decreased (�) gene expression of ca-nonical members of the TGF-b pathway in the preeclamptic placenta and JEG-3 placental trophoblast cells treated with either minimally or moderately cytotoxic CdCl2concentrations.

S.A. Brooks et al. / Food and Chemical Toxicology 98 (2016) 50e57 55

Further studies are needed to demonstrate the specific mechanismby which Cd influences the TGF-b pathway in placentas andplacental cells.

Currently, there are no treatment options for women with PEexcept for premature delivery of the placenta/fetus(Chaiworapongsa et al., 2014). Elucidating the molecular eventsthat lead to PE can facilitate the development of safer treatmentoptions. We hypothesized that not only could Cd alter expression ofthe TGF-b pathway in placentas and in placental trophoblasts, butalso that the induction of the TGF-b pathwaywas regulated throughmiRNA-mediated mechanisms. We observed significant alterationsof miRNAs that target genes associated with TGF-b in both thepreeclamptic placenta and JEG-3 cells following treatment of Cd. Insupport of our findings, related to miRNA dysregulation of thepreeclamptic placenta, we identified a total of 52 miRNAs that aredyregulated in the PE placenta and overlap with other studies(Chen and Wang, 2013). Among these are miRNAs belonging to theprimate-specific miRNA gene cluster (C19MC) imprinted in the

placenta including miR-517c, miR-518c, miR-519d, and miR-520h(Noguer-Dance et al., 2010). Moreover, previous studies haveelucidated associations of miR-26a and miR-155 with PE, twomiRNAs in our study that had significant Cd-altered expression inpreeclamptic placenta, normotensive placenta, and JEG-3 cellstreated with a moderately cytotoxic Cd concentration. SMAD1 andSMAD2, transcription factors that induce downstream TGF-bpathway genes, are predicted targets of miR-26a and miR-155respectively and displayed altered gene expression in the pre-eclamptic placenta and in Cd-treated trophoblasts in this study;highlighting potential focuses for future studies to better under-stand Cd-mediated regulation of the TGF-b pathway.

While this study provides novel information on Cd-inducedexpression and epigenetic reprogramming that can furtherenhance our understanding of the genomic-epigenomic in-teractions that can drive preeclampsia, it is not without limitations.Further studies utilizing a larger clinical cohort of normotensiveand preeclamptic women will be needed to validate these results

S.A. Brooks et al. / Food and Chemical Toxicology 98 (2016) 50e5756

and identify Cd-specific mechanism for miRNA-mediated regula-tion of angiogenic pathways, such as TGF-b. In addition, primaryplacental trophoblast cells should be used in future studies toidentify Cd-specific phenotypes of genetic and epigenetic disrup-tion. Furthermore, future studies should also include a range ofconcentrations of Cd along with other metals to examine the mo-lecular regulation of the TGF-b pathway. Our data suggest higher Cdconcentrations may be needed in vitro to observe comparablebiological features of the preeclamptic placenta. It is also possiblethat a miRNA feedback mechanism may exist to mediate the Cd-induced expression of the TGF-b pathway, highlighting thecomplexity of metal-induced toxicity in the placenta that can in-crease risk of PE.

Taken together our results demonstrate that miRNA dysregula-tion is a feature of PE that is impacted by Cd exposure. In both thepreeclamptic placenta and placental trophoblasts, dysregulatedmiRNA signaling was associated with altered expression signalingof the TGF-b pathway, which may ultimately diminish trophoblastmigratory capabilities. This study provides novel evidence of arelationship between Cd-induced expression and epigeneticreprogramming in the placenta that can further enhance our un-derstanding of the interactions between environmental exposuresand genetics that may drive preeclampsia.

Conflict of interests

The authors claim no competing financial interests.

Acknowledgements

This research was supported by grants from the National Insti-tute of Environmental Health Sciences including R01-ES019315,T32-ES007018, P42-ES005948.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.fct.2016.06.023.

Transparency document

Transparency document related to this article can be foundonline at http://dx.doi.org/10.1016/j.fct.2016.06.023.

References

Adebambo, O.A., Ray, P.D., Shea, D., Fry, R.C., 2015. Toxicological responses ofenvironmental mixtures: environmental metal mixtures display synergisticinduction of metal-responsive and oxidative stress genes in placental cells.Toxicol. Appl. Pharmacol. 289 (3), 534e541.

Alvarez, M.M., Chakraborty, C., 2011. Cadmium inhibits motility factor-dependentmigration of human trophoblast cells. Toxicol. Vitro 25 (8), 1926e1933.

Berg, C.J., Callaghan, W.M., Syverson, C., Henderson, Z., 2010. Pregnancy-relatedmortality in the United States, 1998 to 2005. Obstet. Gynecol. 116 (6),1302e1309.

Bergemann, C., Reimer, T., Muller, H., Hosel, A., Briese, V., Friese, K., et al., 2003.Stimulation of hCG protein and mRNA levels in trophoblast tumour cells Jeg3and BeWo by glycodelin A. Anticancer Res. 23 (2A), 1107e1113.

Brooks, S.A., Rathmell, W.K., 2014. Uniting molecular biomarkers to advance thescience and care of clear cell renal cell carcinoma. J. OncoPathology 1 (4),45e54.

Canettieri, G., Franchi, A., Guardia, M.D., Morantte, I., Santaguida, M.G., Harney, J.W.,et al., 2008. Activation of thyroid hormone is transcriptionally regulated byepidermal growth factor in human placenta-derived JEG3 cells. Endocrinology149 (2), 695e702.

Chaiworapongsa, T., Chaemsaithong, P., Yeo, L., Romero, R., 2014. Pre-eclampsia part1: current understanding of its pathophysiology. Nat. Rev. Nephrol. 10 (8),466e480.

Chen, D.B., Wang, W., 2013. Human placental MicroRNAs and preeclampsia. Biol.Reprod. 88 (5), 130e130.

Conti, E., Zezza, L., Ralli, E., Caserta, D., Musumeci, M.B., Moscarini, M., et al., 2013.Growth factors in preeclampsia: a vascular disease model. A failed vasodilationand angiogenic challenge from pregnancy onwards? Cytokine Growth FactorRev. 24 (5), 411e425.

Dai, Y., Qiu, Z., Diao, Z., Shen, L., Xue, P., Sun, H., et al., 2012. MicroRNA-155 inhibitsproliferation and migration of human extravillous trophoblast derived HTR-8/SVneo cells via down-regulating cyclin D1. Placenta 33 (10), 824e829.

Ding, Y., Ma, L., Wang, X.Z., Zhang, J., Zhao, G.Z., Wang, Z.Q., et al., 2011. In vitro studyon hepatitis B virus infecting human choriocarcinoma JEG3 cells and itsmechanism. Intervirology 54 (5), 276e281.

Esteban-Vasallo, M.D., Aragones, N., Pollan, M., Lopez-Abente, G., Perez-Gomez, B.,2012. Mercury, cadmium, and lead levels in human placenta: a systematic re-view. Environ. Health Perspect. 120 (10), 1369e1377.

Gao, W.L., Liu, M., Yang, Y., Yang, H., Liao, Q., Bai, Y., et al., 2012. The imprinted H19gene regulates human placental trophoblast cell proliferation via encoding miR-675 that targets Nodal Modulator 1 (NOMO1). RNA Biol. 9 (7), 1002e1010.

Garcia, D.M., Baek, D., Shin, C., Bell, G.W., Grimson, A., Bartel, D.P., 2011. Weak seed-pairing stability and high target-site abundance decrease the proficiency of lsy-6 and other microRNAs. Nat. Struct. Mol. Biol. 18 (10), 1139e1146.

Guan, H., Sun, K., Yang, K., 2013. The ERK1/2 signaling pathway regulates 11beta-hydroxysteroid dehydrogenase type 2 expression in human trophoblast cellsthrough a transcriptional mechanism. Biol. Reprod. 89 (4), 92.

Gunel, T., Zeybek, Y.G., Akcakaya, P., Kalelioglu, I., Benian, A., Ermis, H., et al., 2011.SerummicroRNA expression in pregnancies with preeclampsia. Genet. Mol. Res.10 (4), 4034e4040.

Harapan, H., Yeni, C.M., 2015. The role of microRNAs on angiogenesis and vascularpressure in preeclampsia: the evidence from systematic review. Egypt. J. Med.Hum. Genet. 16 (4), 313e325.

Harris, A.L., 2002. Hypoxiaea key regulatory factor in tumour growth. Nat. Rev.Cancer 2 (1), 38e47.

Ishibashi, O., Ohkuchi, A., Ali, M.M., Kurashina, R., Luo, S.S., Ishikawa, T., et al., 2012.Hydroxysteroid (17-beta) dehydrogenase 1 is dysregulated by miR-210 andmiR-518c that are aberrantly expressed in preeclamptic placentas: a novelmarker for predicting preeclampsia. Hypertension 59 (2), 265e273.

Jones, R.L., Stoikos, C., Findlay, J.K., Salamonsen, L.A., 2006. TGF-beta superfamilyexpression and actions in the endometrium and placenta. Reproduction 132 (2),217e232.

Koc, S., Kather, A., Markert, U.R., Durst, M., Schneider, A., Kaufmann, A.M., 2003.Enhancement of immunogenicity of Jeg3 cells by ectopic expression of HLA-A*0201 and CD80. Am. J. Reprod. Immunol. 50 (3), 243e253.

Kummu, M., Sieppi, E., Wallin, K., Rautio, A., Vahakangas, K., Myllynen, P., 2012.Cadmium inhibits ABCG2 transporter function in BeWo choriocarcinoma cellsand MCF-7 cells overexpressing ABCG2. Placenta 33 (10), 859e865.

Laine, J.E., Ray, P., Bodnar, W., Cable, P.H., Boggess, K., Offenbacher, S., et al., 2015.Placental cadmium levels are associated with increased preeclampsia risk. PLoSOne 10 (9), e0139341.

Landgraf, P., Rusu, M., Sheridan, R., Sewer, A., Iovino, N., Aravin, A., et al., 2007.A mammalian microRNA expression atlas based on small RNA librarysequencing. Cell 129 (7), 1401e1414.

Lazar, L., Nagy, B., Molvarec, A., Szarka, A., Rigo Jr., J., 2012. Role of hsa-miR-325 inthe etiopathology of preeclampsia. Mol. Med. Rep. 6 (3), 597e600.

Lazebnik, N., Kuhnert, B.R., Kuhnert, P.M., 1989. Zinc, cadmium, and hypertension inparturient women. Am. J. Obstet. Gynecol. 161 (2), 437e440.

Lewis, B.P., Shih, I.H., Jones-Rhoades, M.W., Bartel, D.P., Burge, C.B., 2003. Predictionof mammalian microRNA targets. Cell 115 (7), 787e798.

Li, P., Guo, W., Du, L., Zhao, J., Wang, Y., Liu, L., et al., 2013. microRNA-29b contributesto pre-eclampsia through its effects on apoptosis, invasion and angiogenesis oftrophoblast cells. Clin. Sci. (Lond) 124 (1), 27e40.

Martin, E., Ray, P.D., Smeester, L., Grace, M.R., Boggess, K., Fry, R.C., 2015. Epigeneticsand preeclampsia: defining functional epimutations in the preeclampticplacenta related to the TGF-beta pathway. PLoS One 10 (10), e0141294.

Matscheski, A., Richter, D.U., Hartmann, A.M., Effmert, U., Jeschke, U., Kupka, M.S.,et al., 2006. Effects of phytoestrogen extracts isolated from rye, green andyellow pea seeds on hormone production and proliferation of trophoblast tu-mor cells Jeg3. Horm. Res. 65 (6), 276e288.

Matsuo, H., Strauss 3rd, J.F., 1994. Peroxisome proliferators and retinoids affect JEG-3 choriocarcinoma cell function. Endocrinology 135 (3), 1135e1145.

Nikuei, P., Malekzadeh, K., Rajaei, M., Nejatizadeh, A., Ghasemi, N., 2015. Theimbalance in expression of angiogenic and anti-angiogenic factors as candidatepredictive biomarker in preeclampsia. Iran. J. Reprod. Med. 13 (5), 251e262.

Noack, F., Ribbat-Idel, J., Thorns, C., Chiriac, A., Axt-Fliedner, R., Diedrich, K., et al.,2011. miRNA expression profiling in formalin-fixed and paraffin-embeddedplacental tissue samples from pregnancies with severe preeclampsia.J. Perinat. Med. 39 (3), 267e271.

Noguer-Dance, M., Abu-Amero, S., Al-Khtib, M., Lefevre, A., Coullin, P., Moore, G.E.,et al., 2010. The primate-specific microRNA gene cluster (C19MC) is imprintedin the placenta. Hum. Mol. Genet. 19 (18), 3566e3582.

Pollack, A.Z., Ranasinghe, S., Sjaarda, L.A., Mumford, S.L., 2014. Cadmium andreproductive health in women: a systematic review of the epidemiologic evi-dence. Curr. Environ. Health Rep. 1 (2), 172e184.

Rager, J.E., Yosim, A., Fry, R.C., 2014. Prenatal exposure to arsenic and cadmiumimpacts infectious disease-related genes within the glucocorticoid receptorsignal transduction pathway. Int. J. Mol. Sci. 15 (12), 22374e22391.

Saito, S., Nakashima, A., 2014. A review of the mechanism for poor placentation inearly-onset preeclampsia: the role of autophagy in trophoblast invasion and

S.A. Brooks et al. / Food and Chemical Toxicology 98 (2016) 50e57 57

vascular remodeling. J. Reprod. Immunol. 101e102, 80e88.Satarug, S., Garrett, S.H., Sens, M.A., Sens, D.A., 2010. Cadmium, environmental

exposure, and health outcomes. Environ. Health Perspect. 118 (2), 182e190.Wang, W., Feng, L., Zhang, H., Hachy, S., Satohisa, S., Laurent, L.C., et al., 2012. Pre-

eclampsia up-regulates angiogenesis-associated microRNA (i.e., miR-17, -20a,and -20b) that target ephrin-B2 and EPHB4 in human placenta. J. Clin. Endo-crinol. Metab. 97 (6), E1051eE1059.

Wu, L., Zhou, H., Lin, H., Qi, J., Zhu, C., Gao, Z., et al., 2012. Circulating microRNAs areelevated in plasma from severe preeclamptic pregnancies. Reproduction 143

(3), 389e397.Yang, Q., Lu, J., Wang, S., Li, H., Ge, Q., Lu, Z., 2011. Application of next-generation

sequencing technology to profile the circulating microRNAs in the serum ofpreeclampsia versus normal pregnant women. Clin. Chim. Acta 412 (23e24),2167e2173.

Zhu, X.M., Han, T., Sargent, I.L., Yin, G.W., Yao, Y.Q., 2009. Differential expressionprofile of microRNAs in human placentas from preeclamptic pregnancies vsnormal pregnancies. Am. J. Obstet. Gynecol. 200 (6), 661 e661e667.