Toxicology and Applied Pharmacology 291 (2016) 84–96

Contents lists available at ScienceDirect

Toxicology and Applied Pharmacology

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

Zebrafish embryos as a screen for DNA methylation modifications after

compound exposure
Manon C. Bouwmeester a, Sander Ruiter a, Tobias Lommelaars a, Josefine Sippel a, Hennie M. Hodemaekers a,
Evert-Jan van den Brandhof b, Jeroen L.A. Pennings a, Jorke H. Kamstra c, Jaroslav Jelinek d, Jean-Pierre J. Issa d,e,
Juliette Legler c, Leo T.M. van der Ven a,⁎
a
Center for Health Protection, National Institute for Public Health and the Environment (RIVM), PO Box 1, 3720 BA Bilthoven, The Netherlands
b
Center for Environmental Quality, National Institute for Public Health and the Environment (RIVM), PO Box 1, 3720 BA Bilthoven, The Netherlands
c
Institute for Environmental Studies (IVM), VU University, De Boelelaan 1085, 1081 HV Amsterdam, The Netherlands
d
Fels Institute for Cancer Research and Molecular Biology, Temple University School of Medicine, Philadelphia, PA, USA
e
Department of Leukemia, The University of Texas MD Anderson Cancer Center, Houston, TX, USA

a r t i c l e i n f o a b s t r a c t

Article history: Modified epigenetic programming early in life is proposed to underlie the development of an adverse adult phe-
Received 27 September 2015 notype, known as the Developmental Origins of Health and Disease (DOHaD) concept. Several environmental
Revised 17 December 2015 contaminants have been implicated as modifying factors of the developing epigenome. This underlines the
Accepted 17 December 2015
need to investigate this newly recognized toxicological risk and systematically screen for the epigenome modify-
Available online 19 December 2015
ing potential of compounds. In this study, we examined the applicability of the zebrafish embryo as a screening
Keywords:
model for DNA methylation modifications. Embryos were exposed from 0 to 72 h post fertilization (hpf) to
Epigenetics bisphenol-A (BPA), diethylstilbestrol, 17α-ethynylestradiol, nickel, cadmium, tributyltin, arsenite,
DNA methylation perfluoroctanoic acid, valproic acid, flusilazole, 5-azacytidine (5AC) in subtoxic concentrations. Both global and
Environmental contaminant exposure site-specific methylation was examined. Global methylation was only affected by 5AC. Genome wide locus-
Screening specific analysis was performed for BPA exposed embryos using Digital Restriction Enzyme Analysis of Methyla-
Zebrafish embryo tion (DREAM), which showed minimal wide scale effects on the genome, whereas potential informative markers
were not confirmed by pyrosequencing. Site-specific methylation was examined in the promoter regions of three
selected genes vasa, vtg1 and cyp19a2, of which vasa (ddx4) was the most responsive. This analysis distinguished
estrogenic compounds from metals by direction and sensitivity of the effect compared to embryotoxicity. In
conclusion, the zebrafish embryo is a potential screening tool to examine DNA methylation modifications after
xenobiotic exposure. The next step is to examine the adult phenotype of exposed embryos and to analyze
molecular mechanisms that potentially link epigenetic effects and altered phenotypes, to support the DOHaD
hypothesis.
© 2015 Elsevier Inc. All rights reserved.

1. Introduction individuals who experienced this famine during their prenatal fetal
life developed an increased incidence of mainly metabolic diseases
There is a worldwide increase in non-communicable diseases later in life (Roseboom et al., 2006). Another example of intrauterine ex-
(NCDs), such as diabetes, obesity, cardiovascular diseases and cancer posure that led to a so-called programmed effect is the pharmaceutical
(World Health Organization, 2014). Exposure to exogenous factors, diethylbestrol (DES), which was used by pregnant women until 1970 to
such as environmental contaminants, diet and stress, during early-life reduce miscarriages. Later in life, their exposed daughters developed
development has been implicated to increase the susceptibility to de- adverse effects, particularly sub-/infertility and cancer in the reproduc-
velop NCDs during adult life. This concept is known as Developmental tive tissues (Reed and Fenton, 2013). The DOHaD phenomenon may be
Origins of Health and Disease (DOHaD) (Gluckman and Hanson, explained by disruption of the sensitive epigenetic tuning of cell func-
2004). One of the first examples of early-life environmental exposure tions, which is active during embryogenesis (Smith and Meissner,
and development of disease later in life is the Dutch Famine cohort; 2013; Tang and Ho, 2007; Wang et al., 2013). Epigenetics is a regulating
mechanism of gene expression, which allows different phenotypes of
⁎ Corresponding author. cells carrying the same genotype. Epigenetic regulation occurs at differ-
E-mail address: [email protected] (L.T.M. van der Ven). ent levels, including DNA methylation, histone modification and non-

http://dx.doi.org/10.1016/j.taap.2015.12.012
0041-008X/© 2015 Elsevier Inc. All rights reserved.
 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96 85

coding RNAs, of which the first is most thoroughly studied (Vaiserman, exposure with BPA was used to identify targets using genome wide
2014). DNA methylation is catalyzed by a family of DNA methyltransfer- locus-specific methylation analysis with the Digital Restriction Enzyme
ases (DNMTs) and involves transfer of a methyl group from the cofactor Analysis of Methylation (DREAM) method, which is based on methyla-
S-adenosyl-methionine (SAM) to a cytosine in a cytosine-guanine pair tion sensitive restriction digestion (Jelinek et al., 2012). Potential
(CpG) at the 5′-ring carbon position. Dense regions of CpG sites, so- DREAM derived targets were verified by pyrosequencing.
called CpG islands, are linked to binding of transcription factors and Three additional targets for locus-specific analysis by pyrosequenc-
gene expression, and methylation of these regions leads to differential ing were derived from literature and included promoter regions of
gene expression (Jones, 2012). The relevance of epigenetic modifica- vasa (new name: ddx4), vtg1 and cyp19a2. Methylation responsiveness
tions as a causative factor for disease has already been implicated in after chemical exposure has been shown for vasa and vtg1 in zebrafish
the onset and progression of cancer (Esteller et al., 2001; Huang and (Fang et al., 2013; Strömqvist et al., 2010) and for cyp19a2 in medaka
Rao, 2014), resulting in identification of diagnostic epigenetic markers (Contractor et al., 2004). Methylation sensitive high resolution melting
and, more recently, in new therapeutic modalities (Katz et al., 2014). Be- (MS-HRM) was included as an additional method to analyze a region
sides epigenetic changes in cancer, other diseases are also associated with several CpG simultaneously.
with epigenetic modifications. For instance, epigenetic marks are asso-
ciated with brain functions and methylation patterns differ in patients 2. Methods
with Alzheimer (M. Bakulski et al., 2012; Woldemichael et al., 2014).
The epigenome can be affected by environmental compounds. Phar- 2.1. Zebrafish maintenance and egg spawning
maceuticals, such as diethylstilbestrol (DES), endocrine active com-
pounds, such as bisphenol-A (BPA), but also metals, such as arsenic Danio rerio adults were commercially obtained wild-type import
(As), have been proven to induce epigenetic alterations (Hou et al., (Ruinemans Aquarium BV, Montfoort, The Netherlands) and main-
2012). However, epigenetic effects induced by compound exposure ob- tained and bred for more than 8 years/generations in our lab. Adult
served in some studies, are not always reproduced in other studies. As zebrafish were kept as groups of 20–30 in 7.5 L tanks in a ZebTec system
an example, Van Esterik et al. (2014a) reviewed programmed effects in- with automatically controlled water parameters at 26.5 ± 1 °C with a
duced by BPA in experimental models and concluded that observed ef- day–night cycle of 14–10 h. Three days before spawning, females were
fects vary with applied analytical tools and study models. To bring separated from males. Small spawning groups of 2 m/2f were composed
insight into the epigenetic programming potential of chemical sub- on the day before spawning, which was triggered by morning light on
stances further, several position papers, supported by international the next day. Eggs were collected within the next 2 h and assessed for
bodies such as the Organization for Economic Co-operation and quality. Spawns with N10% coagulations or deformities, i.e. damaged
Development (OECD) and the European Food Safety Authority (EFSA), membranes or embryo asymmetries, were discarded.
as well as the pharmaceutical industry, emphasized the need to system-
atically characterize the epigenetic modifying effects of substances in a 2.2. Compounds
screening model (Greally and Jacobs, 2013; Rasoulpour et al., 2011).
These authors list several potentially suitable models, ranging from a ro- All compounds were obtained from Sigma-Aldrich (St. Louis, MO,
dent reporter model to a variety of in vitro models, which all have their USA): bisphenol-A (BPA; CAS 80-05-7), 17α-ethynylestradiol (EE2;
specific advantages and disadvantages. We here investigated the appli- CAS 56-53-1), diethylstilbestrol (DES; CAS 50-27-1), nickel chloride
cability of the zebrafish embryo to measure DNA methylation effects (NiCl2; CAS 7718-54-9; further abbreviated as Ni), cadmium chloride
after compound exposure. We selected this model because analysis in (CdCl2; CAS 10108-64-2; further abbreviated as Cd), tributyltin chloride
a whole organism allows the integral measurement of effects in all tis- (TBTC; 1461–22-9; further abbreviated as TBT), sodium arsenite
sues and of processes depending on interaction between tissues which (NaAsO2; CAS 7784-46-5; further abbreviated as As), perfluorooctanoic
only occur in a whole organism. Early developmental stages in the acid (PFOA; CAS 335-67-1), valproic acid (VPA; CAS 1069-66-5),
zebrafish embryos already show metabolic activity and epigenetic flusilazole (FLU; CAS 85509-19-9), and 5-azacytidine (5AC; CAS 320-
changes in zebrafish embryos can be linked to either morphological ef- 67-2). Stock solutions of DES, BPA and FLU were prepared in dimethyl
fects (teratology and delayed development), or, preferably, to affected sulfoxide (DMSO; CAS 67-68-5), of EE2 in ethanol (EtOH) and of Ni,
functions later in life. In our study, we specifically set out to identify pat- Cd, TBT, As, PFOA, VPA, and 5AC in OECD medium (see below); all stocks
terns in epigenetic changes and possible potency differences induced by were diluted with OECD medium (294.0 mg/L CaCl2 ∙2H2O, 123.3 mg/L
compounds. MgSO4 ∙ 7H2O, 64.7 mg/L NaHCO3, 5.7 mg/L KCl (OECD, 2002)) with
Zebrafish embryos were exposed to a range of environmentally rel- 0.2% DMSO, 0.1% EtOH or no solvent in the final concentrations, as
evant reference compounds at subtoxic concentrations following an ex- appropriate.
posure protocol adapted from the OECD Test Guideline 236 (Hermsen
et al., 2011). Test compounds were based on known epigenetic effects. 2.3. Exposure
The estrogenic compounds DES and BPA are associated with program-
ming of disease (Reed and Fenton, 2013; Van Esterik et al., 2014a), as Concentration ranges of the compounds in the epigenetic analysis
well as with epigenetic modifying effects in experimental models experiments were based on range finding experiments including four
(Fernandez et al., 2012; Li et al., 2014), and this group was supplement- concentrations in 10-fold dilutions. The lowest concentration that in-
ed with the synthetic estrogen 17α-ethynylestradiol (EE2). The metals duced morphological effects in the range finding experiments was
nickel (Ni) (Sun et al., 2013), cadmium (Cd) (Wang et al., 2012) and ar- used as highest test concentration in the final exposure experiment in
senite (As) (Bailey and Fry, 2014) are known to induce epigenetic ef- which a narrower spacing of 5-fold dilutions was applied. These final
fects in experimental models and are associated with programming of ranges are shown in the overview Table of the Results (Table 2).
a variety diseases. Tributyltin (TBT) (Li et al., 2011) and The collected fertilized eggs were rinsed with OECD medium. After
perfluorooctanoic acid (PFOA) (Ho et al., 2012) were included as non- rinsing, the eggs were distributed among four test concentrations of
estrogenic PPARγ/RXR agonists. Valproic acid (VPA) and flusilazole compounds plus solvent control and embryos of matched developmen-
(FLU) were included as a control developmental toxicant with and tal stage were distributed over a 24-well plate containing 2 mL test me-
without known epigenetic effects, respectively (Hermsen et al., 2011; dium per well. Each well contained one embryo and in total 20 embryos
Tung and Winn, 2010), and the Dnmt inhibitor 5-azacytidine (5AC) as per test concentration were used. Four internal controls were included
control for induction of hypomethylation. After embryo exposure, in each plate, which consisted of solvent only medium. Each experiment
both global and site-specific methylation changes were assessed. The was performed in triplicate and medium was not renewed during the
86 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96

exposure period. In repeat experiments, experimental units were insensitive. Distinct signatures, 5′-GGG at unmethylated sites or 5′-
established as pools of embryos (n = 10) in petri dish at 2 mL per em- CCGGG at methylated sites were created and ultimately next generation
bryo. Embryos were kept in an incubator at 26.5 ± 1 °C with a day–night sequencing was used to map these sites to the genome. Methylation ra-
cycle of 14–10 h. At 24, 48 and 72 h post fertilization (hpf), the embryos tios for each individual CCCGGG site were calculated as a proportion of
were scored for morphological development (such as development of methylated counts to the sum of unmethylated and methylated counts,
eyes, pectoral fin, somite formation, pigmentation, movement, heart- and subsequently adjusted for differences in restriction enzyme effi-
beat and blood circulation) and teratogenicity (such as malformation ciency using values obtained from spiked in standards.
of the head, tail, heart, sacculi/otoliths and chorda structure, edema in
the heart, yolk sac and eye, scoliosis, rachischisis and yolk deformation) 2.7. Locus-specific methylation
using a standardized scoring (Hermsen et al., 2011). In an experiment
with BPA, exposure was extended to 96 hpf with and without continued Locus-specific methylation analysis was performed through pyrose-
exposure or 120 hpf without continued exposure, to study recovery in quencing of a PCR product of bisulfite converted DNA. During the bisul-
DNA methylation effects. fite treatment only unmethylated cytosines are converted into uracil.
The DNA Methylation-Gold Kit (Zymo Research, Irvine, CA, USA) was
2.4. DNA extraction used according to the manufacturer's instructions (conversion efficien-
cy N 99%), using 500 ng DNA as input. Converted DNA was eluted with
DNA was extracted from the exposed embryos using the DNeasy 14 μL M-Elution Buffer. The concentration of bisulfite converted DNA
Blood & Tissue kit (Qiagen, Hilden, Germany) together with RNase was determined using Nanodrop set to RNA-40.
treatment, according to the manufacturer's instructions, except for elu- Once converted, the target sequences in the promoter region of the
tion of DNA, which was performed by incubating 60 μL H2O on the targets of interest were amplified by PCR using Pyromark PCR Kit
DNeasy membrane for 5 min. The DNA concentration was determined (Qiagen). Position and amplicon sequences of vasa, cyp19a2, and vtg1
at 260/280 nm (DNA-50) using a spectrometer (Nanodrop, Thermo are shown in Supplemental Fig. 1 and Table 1. PCR primers and sequenc-
Fisher Scientific, Waltham, MA, USA). ing primers (Table 1) were designed using PyroMark Assay Design Soft-
ware (Qiagen, version 2.0). Designed primers were selected based on
2.5. Global methylation the highest score, defined by features such as amplicon length, melting
temperature and mispriming sites. For each target one primer set was
Global methylation levels were analyzed using High Pressure Liquid ordered at Invitrogen (Thermo Fisher Scientific, Waltham, MA, USA) of
Chromatography (HPLC), detecting the ratio of unmethylated cytosines which one of the primers was labeled with biotin at the 5′-end. The
(dC) to methylated cytosines (5mdC). Herefore, the extracted DNA was PCR reaction was started by an initial melting step of 15 min at 95 °C,
digested to single nucleotides to allow isocratic separation. The DNA which was followed by 45 (vasa 50) cycles of 94 °C for 30 s, 58 °C for
was treated according to Quinlivan and Gregory (2008) using a single- 30 s and 72 °C for 40 s and a final extension at 72 °C for 10 min. Samples
step digestion protocol. The digestion mix consisted of 300 mU phos- were kept at − 20 °C. Methylation controls for calibration were pre-
phodiesterase I, 250 U Benzonase nuclease and 200 U alkaline phospha- pared by mixing 0 and 100% methylated zebrafish embryo DNA to 25,
tase from bovine intestinal mucosa (all three from Sigma Aldrich) in 50 and 75% methylation before bisulfite conversion. 0% Methylated
5 mL TRIS buffer (20 mM TRIS–HCl, 100 mM NaCl and 20 mM MgCl2, DNA was obtained using REPLI-g Mini Kit (Qiagen), and 100% methylat-
pH 7.9). DNA (1–2 μg in 25 μL nuclease-free water) was incubated over- ed DNA was generated by an incubation step of a reaction mix contain-
night at 37 °C with 50 μL of the digestion mix. After incubation, the sam- ing 5 μL 10× NEBuffer (Cat no. B7002S, New England Biolabs, Ipswich,
ples were diluted with 25 μL mobile phase (25 mM ammonium acetate/ MA, USA), 1 μL 2× diluted SAM 32 mM (Cat no. B9003S, New England
acetic acid in 15% acetonitrile, all from Merck KGaA, Darmstadt, Biolabs), 3 μL M.SssI (Cat no. M0226S, New England Biolabs), 1 μg
Germany, with a pH of 4.8) and transferred to a HPLC vial with insert. gDNA and nuclease free water up to 50 μL reaction mix. After 2 h incu-
Other HPLC parameters were: flow 1.2 mL/min, injection volume bation at 37 °C, 1 μL 2× diluted SAM and 2 μL M.SssI were added and in-
50 μL, run time 15 min, equilibration time 0.2 min, flush volume 30 μL, cubated for 16 h at 37 °C, 20 min at 65 °C and were kept at 4 °C. Both
wash volume 750 μL. After the isocratic separation of the DNA by the reactions were purified using QIAamp DNA Mini Kit (Qiagen) according
HPLC column (Vertex Plus Nucleosil SA, 250 × 4.6 mm × 10 μm, pore to the manufacturer's instructions. Elution was performed using
size 100 μm; Knauer, Berlin, Germany), measurement of the nucleotide 20 μL M-Elution Buffer.
peaks was carried out at 277 nm with the Knauer Smartline 2600 LWL Single-stranded template was prepared for pyrosequencing using
UV detector. Single type nucleotide standards (Deoxynucleotide Tri- Streptavidin-coated Sepharose beads (Streptavidin Sepharose High
phosphate Set (dATP, dGTP, dTTP); PCR Grade, Roche, Germany) were Performance, GE Healthcare, Waukesha, WI, USA). Based on signal
used to identify peaks, and these were digested in the same way as strength in a pilot run, up to 20 μL of the biotinylated PCR product
the samples. Quantification of the dC and 5mdC peak areas was derived was used. Cleaned biotinylated template was annealed to the
from a standard calibration curve, which was produced using a concen- sequencing primer (0.375 μM) at standard conditions. The assay
tration range of 0.0625–4 μg/mL for both 2′-deoxycytidine (Sigma Al- setup and dispensation order of the nucleotides were generated
drich) and 5-methyl-2′-deoxycytidine (MP Biomedicals, Santa Ana, using PyroMark Q24 Software (Qiagen, version 2.0). The dispensa-
USA). The results of the global methylation are given as the percentage tion order was based on the expected sequence in the amplicon
of methylation (%5mdC) calculated from the linear regression of the cal- and in the bisulfite controls, to evaluate the efficiency of the bisulfite
ibration curve. treatment. Pyrosequence was conducted with a PyroMark Q24
System (Qiagen), and the methylation level of each CpG site was
2.6. Genome wide locus-specific analysis calculated as the ratio of peak heights in the pyrogram.

Two concentrations (3 and 12 μM) BPA exposed embryos and non- 2.8. Methylation sensitive high resolution melting (MS-HRM)
exposed controls were analyzed in triplicate for genome wide locus-
specific methylation using DREAM (Digital Restriction Enzyme Analysis In BPA exposed embryos, vasa was also examined using MS-HRM,
of Methylation (Jelinek et al., 2012)) at the Temple University School of which is a potentially equally sensitive method compared to pyrose-
Medicine (Philadelphia, PA, USA). In short, genomic DNA was sequen- quencing, although with less precision because it is an integral mea-
tially digested by SmaI and XmaI, which both recognize the sequence surement of all CpGs contained in a defined amplicon. On the other
CCCGGG. SmaI is methylation sensitive, where XmaI is methylation hand, MS-HRM allows a higher throughput than pyrosequencing,
 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96 87

Table 1
PCR- and sequencing primer sequences.

Target Primer Sequence (5′–3′) Amplicon size (bp) CpG count Source

a2ml Forward GGGAGTTTTTTAGAATATATGGTTGTTGT 263 1 DREAM

Reverse biotin-CCAACTACAATCAATTATATCACATACTC
Sequencing AATTATTTTAGAGTTTAGGGTTTTT
dip2ba Forward TTTGTTTTAGTATGAGGGTTGTAAT 98 1 DREAM
Reverse biotin-ATAATACCCAACATTCCAAACATC
Sequencing TTGTTGTTTTTTAGTGGAG
ehhadh Forward TAGTTGATGATTGGTAGTAAGTTAGT 199 2 DREAM
Reverse biotin-ATTCTATCAACAACCCATCACCTTAACAAC
Sequencing GGTAGTAAGTTAGTTTGGTAT
ghra Forward biotin-AGGGTTTGGGAGAGAGTT 67 1 DREAM
Reverse CTTCCAATAACTCAATATTTCATCACTCA
Sequencing ACTCACCTTTCACAC
LOC100535586 Forward GGTGTTGGTTGTTATGGAAATTATG 107 4 DREAM
Reverse biotin-AACAAAACCATCCACTCAACACTTCTCTA
Sequencing TGATTGGTGTTGAGTT
mhc1zba Forward GGTTTTGTGTGTTAGAAATTGTAATGTTAG 162 2 DREAM
Reverse biotin-ACCCCAAATAACACACTCTC
Sequencing GTTTTATTTAAATATAAATATTTTT
zgc Forward AGGAGTGGGTTTTATTAGGAGTA 259 3 DREAM
Reverse biotin-TCTTCCCCCCTATCTATTCTTTCCTTACAC
Sequencing GAGTAAGAGAGGTTTT
pfkmb Forward TTTGTTTTGGGTTAGAGGATTTTATGAG 130 1 DREAM
Reverse biotin-CTAATTCAAAAAATATTTCATCCCATTTCA
Sequencing GATTTTATGAGTTTAGTGTTTTT
nxnl2 Forward GGGTTTGTGTTTTTGGATGGA 161 3 DREAM
Reverse biotin-ACAAACAATAATTACAATTCCCACATTTC
Sequencing TTGTGTTTTTGGATGGAA
pcdh17 Forward TGTTTTTTGGATTGTTATTTTAGAAGAAA 63 1 DREAM
Reverse biotin-TTCCTCTCCCCAACACCTAT
Sequencing GGATTGTTATTTTAGAAGAAATT
nsfb Forward GTGTGAGATGATTTTTTGTGGTGTGAA 123 3 DREAM
Reverse biotin-AACCAAACCCAATTAAATACTCACC
Sequencing GAAGGTTGTAGGAGT
epha4b Forward GGTGGTTGAGTAAATGGTTAGAAAAGTA 324 2 DREAM
Reverse biotin-AATATCATCCTCCAAAATATAATTCACAA
Sequencing ATATTTTGAGAGAGTAATAT
senp8 Forward AGGGTAGTAGATAGTGTTTTTTGTAATGA 206 6 DREAM
Reverse biotin-ATCTCCTTTTCTCTATCTTCCTCTAAC
Sequencing AGATAGTGTTTTTTGTAATGATT
vasa Forward TGGGAATTTTAGTAGTATATTGATAGT 198 5 Lindeman et al. (2010)
Reverse biotin-ACCTCATTAATCTAAATCAATTATACC
Sequencing I GTTTGAAGAATATTTAAATTAATAA
Sequencing II TTTTTGAGATAGGTTTATTAGT
vtg1 Forward GTTTTGGTTTAATTGAGAGGTTTTAG 174 3 Strömqvist et al. (2010)
Reverse biotin-CTCAACACCATAAAACTCCTCCTTATATCC
Sequencing TTAGTTTTAATTGTGTATAGAATTG
cyp19a2 Forward AGGGTGAGAGAAATGAAGGTTAGT 247 7 Contractor et al. (2004)
Reverse biotin-AACTCATCCCCCTAAACTTC
Sequencing I ATTTTTTTAATTTTTAAAGGTGTTT
Sequencing II AGAGTAATTAATTTTTATGTATATA
Sequencing III GGTGTAGTAGGATTTTTGG

Primers were complementary with bisulfite treated DNA. Sequencing primers of vasa were designed to sequence CpG 1, 2 and 3 (I) and CpG 3, 4 and 5 (II); however, sequencing primer I
induced unexpected peaks, possibly due to its low melting temperature (only As and Ts at its 3′ end) and its distance from the target CpG sites; this primer was not further used. Sequenc-
ing primer II of vasa contains homopolymorphic stretches of T which could limit the accuracy of the % methylation estimation, but was not considered to affect the toxicological conclusion.
Sequencing primer of vtg1 was designed to sequence CpG 1, 2 and 3. Sequencing primers of cyp19a2 were designed to sequence CpG 1 and 2 (I), CpG 3, 4 and 5 (II) and CpG 6 and 7 (III).

while it is based on net temperature shift (NTS) calculations 2.9. Statistical analysis
(Newman et al., 2012). In short, DNA was bisulfite converted and
amplified as described above. Unmethylated control DNA was Concentration response curves were calculated based on the mor-
generated with the repli-G whole genome amplification kit (Qiagen) phology score and methylation ratios at specific sites using the
and fully methylated DNA was generated by methylating zebrafish PROAST software developed by RIVM (Slob, 2002). PROAST applies
DNA with M.SssI CpG methyltransferase (New England Biolabs), fitting an increasingly complex mathematical model to concentration
after which both were bisulfite converted. NTS were determined response data and calculating a specific concentration (the Critical Effect
and percentage methylation was calculated by interpolating Dose; CED, or Benchmark Dose; BMD) to a pre-defined response (the
unknowns from the calibration curve, as described previously Critical Effect Size; CES, or Benchmark Response; BMR). The
(Kamstra et al., 2014). Analyses were performed in triplicate. predetermined response was a 5% change in each endpoint, morpholo-
MS-HRM was performed for vasa and LOC. For the LOC target no gy, teratogenicity or methylation (CES = 0.05). The ratio of upper and
suitable primers could be developed due to severe bias towards lower bound of the confidence interval at CED (CEDU/CEDL) was used
methylated template. as a measure of precision of the result, and considered acceptable at
88 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96

approx. 10 or less. All calculated exponential models were confirmed by Measurement of global methylation did not produce an effect in the
a Hill model analysis with the same parameters. exposed embryos with any of the test compounds (illustrated for BPA in
DREAM data were analyzed for quality, clustering (supervised and Fig. 1B), except for the DNA methylation inhibitor 5AC (not shown).
unsupervised), and concentration differences using bootstrapping, vol- Exposure with BPA was subsequently used to identify informative
cano plot and distribution analysis in R, on subsets of sites with a mini- targets, using the genome wide locus-specific DREAM assay, assessing
mum coverage of 10 or 100 reads in all samples. Finally, selection of differential methylation in 3 and 12 μM BPA exposed embryos com-
potential markers of effect was done on the basis of effect size, signifi- pared to non-exposed controls. After DREAM analysis, corrected meth-
cance, and concentration-effect correlations (see Results for criteria). ylation levels were available for 64,466 sites mapped to the zebrafish
Functional annotation and functional overrepresentation analysis genome. Of these sites, 47,156 had at least 1 read (occurrence in the se-
were performed using DAVID (http://david.abcc.ncifcrf.gov/) (Huang quencing output), 27,251 at least 10 reads and 4,953 of these sites had
et al., 2009; Huang et al., 2008). 100+ reads in all 9 samples. Generally, the data quality was good, as
was illustrated by the uniform methylation levels at the 4,953 sites cov-
ered with 100+ sequencing reads, where only 77 sites had a standard
3. Results deviation of N5%. Unsupervised data analysis showed no systematic
methylation changes between non-exposed and exposed embryos. Su-
As a first step, embryotoxicity of a test compound was determined pervised hierarchical clustering of 1041 CpG sites that showed signifi-
per exposure using a range aiming at morphological effects in the cant differences between BPA 12 μM and controls (P b 0.05 and N 2%
highest concentration, as derived from a range finding experiment. A difference in methylation) assigned the intermediate BPA concentration
concentration response of the morphology score is illustrated for BPA- of 3 μM into a separate cluster (Fig. 2).
exposed embryos in Fig. 1A, where the CED of 20 μM is calculated just Next, the 10 + and 100 + reads sets were separately analyzed for
below the highest concentration in the range of 4.9–27.7 μM, and this potential informative markers of effect, using combined selection
CED is actually determined by effects in the highest concentration. Com- criteria Δ N 5%, P b 0.05, correlation R N 0.7/0.5, respectively. This pro-
parable calculations produced CEDs for embryotoxicity of DES, Ni, Cd, duced a final set of 13 loci, which together with technical criteria related
TBT, As, VPA, FLU and 5AC of 2.4, 183, 67, 0.34, 2900, 6.1 (1.8 for 96 h to primer design were suitable for further pyrosequencing verification,
of exposure) and 307 μM respectively (Table 2). CEDs of embryotoxicity and in the following genes: a2ml, dip2ba, ehhadh, ghra, loc100535586,
were not always exactly reproduced between range-finding and final mhc1zba, zgc:198371, pfkmb, nxnl2, pcdh17, nsfb, epha4b, and senp8 (all
experiments, as was the case with EE2 and PFOA, which did not show in promoter, except zgc:198371, nsfb and epha4b, which were in in-
embryotoxic effects in the final experiment. Frequently observed mor- trons). The pyrosequencing analysis in the same samples with the
phological effects were no hatching (BPA, DES, Ni, Cd, As, FLU) and in three same concentrations as used for DREAM (0, 3, 12 μM) reproduced
some cases no blood circulation (5AC, FLU) or no movement (DES, DREAM methylation percentages in the controls of a2ml, ghra, pcdh17,
FLU). Most severe effects (relative to CED of embryotoxicity) were in- and senp8, but differences up to 20–30 absolute percentage points also
duced by DES, As, and 5AC, which induced coagulation in 35–70% of occurred (dip2ba, mhc1zba, and nxnl2; Fig. 3). Concentration response
the embryos. Observed teratogenic effects were pericardial edema, effects as detected with DREAM were reproduced by pyrosequencing
yolk sac edema (DES, 5AC), malformation of the heart (BPA, DES, VPA, in ghra (1 CpG), loc100535586 (4 CpGs), and nxl2 (2 CpGs), albeit not
5AC) and malformation of the pectoral fin (FLU) or tail (BPA, DES). in the DREAM SmaI CpG in the latter (Fig. 3, Table 3). Further analysis

Fig. 1. Illustration of concentration responses of morphology (A) and global methylation (B) in BPA exposed zebrafish embryos. Small circles are individual data points which overlap in A;
large circles are the mean per concentration. The leftmost data set in each graph represents the blank control, given at a virtual low value and with a dotted connection to graph bridging
the virtual distance. CED in A, Critical effect dose (at 20 μM BPA), with CEDL and CEDU the lower and upper bound of its confidence interval, respectively. Function of the graph is shown on
top of the graph; all exponential functions that were significantly better than y = a (E2–E5) were verified with an analysis using the Hill family of functions. Other parameters of the
PROAST analysis are explained in Slob (2002).
 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96 89

effects. NCEDM/ET values represent calculations with extrapolated CEDM. Multiple values under one target (separated with/) represent replicate experiments, with 96h as additional analysis at 96 hpf for comparison with standard 72 hpf. Concentration

↑, hypermethylation; ↓, hypomethylation. aObserved in a preceding range-finding; beffect was b5%, therefore a CED is not available at CES = 0.05; cadditional analysis with HRM; dCED vasa CpG3 of DES is average of values of 2 primers. –, no observed
Values in the upper part of the Table are the critical effect concentrations (CEDs) calculated at the critical effect size CES = 0.05 (explained under statistical analysis, Methods). CEDs for methylation effects are presented relative to the CED of
embryotoxicity (CEDM/ET). In this way, the CEDs of methylation can be valued as the relative sensitivity of the methylation effect, with values N1, b1 representing methylation effects at higher, respectively lower concentrations than embryotoxic
in the full concentration range only produced a statistically significant

cyp19a2

↓ N1.0
CpG7
concentration-response in the ghra CpG, which was, however, fully de-

↑ 3.8

↑3.2
↓2.5
–
–
–

–
–

–
–
–
termined by a deviating control group, and did not produce a CED at the

The lower two rows show the maximum effect sizes observed in the concentration response curves, with Δ% test giving the range observed over all test compounds, and Δ% AC, values with the control DNA methylation inhibitor 5AC.
5% effect level (Fig. 4A); this target was therefore not considered infor-

–/↓1.1/N1096h

↓3.0–3.6 ↑3.5
mative, and not further explored.

↑ 23.796h
↓ N1.4/–
cyp19a2

Pyrosequencing analysis of methylation in the selected upstream

↓ N1.0
↓ 77.1
CpG6

↓4.0
CpGs of the additional three literature derived targets vasa, vtg1 and
–
–

–
–
–

–
cyp19a2 was also first tested in the BPA exposure experiment. This pro-
duced significant concentration-responses for vasa CpG3 (Fig. 4B,
cyp19a2
CpG3

↑ ()b

Table 2) and cyp19a2 CpG3 (Table 2). In Table 2, the obtained CEDs
–
–

–
–
–
–

–
–
–

–
are presented relative to the CED of embryotoxicity (CEDMethylation/
CEDEmbryotoxicity, CEDM/ET), to enable direct reading of the relative sensi-
cyp19a2

tivity of the methylation effect; values at N1 and b1 represent methyla-

CpG2

↑ 7.9

↑ 1.6
tion effects at higher and, respectively, lower concentrations than at
–
–
–

–
–

–
–
–

–
which embryotoxic effects occurred. Because of the responsiveness of
↓ –/2.1/14.196h
↓ 1.6/–/4.796h

vasa and the suggested relevance of the cyp19a2 and vtg1, further anal-
ysis with these targets was continued with the full compound set, and
↓ 1.0–1.1
cyp19a2

↓ 289.3
CpG1

revealed effects in at least one of these three targets after exposure to

each compound, with the exception of As (Table 2). With BPA, the estro-
–

–
–
–
–

genic compounds EE2 and DES induced reproducible hypomethylation

↑ 20.5/40.996h

in the fragment in CpG3 of vasa, with a CEDM/ET generally showing

↓3.0 ↑2.0

that effects in vasa CpG3 were more sensitive than morphology, at the
↓ 275.0

↓ N1.0
CpG3

↑ 2.4

standard evaluation time point of 72 hpf. The absolute CEDs of these es-
↓2.5
vtg1

–
–
–

–
–

trogen agonists for vasa CpG3 methylation in a single experiment sug-

gested a potency order of DES = EE2 (CEDs 2.4 and 2.8 μM,
↓ 88.3
CpG2

↓8.8
vtg1

respectively) N BPA (CED 3.6 μM). Methylation over the entire fragment
–
–
–

–
–
–
–

in the region of vasa in BPA exposed embryos as examined with MS-

↓ 5.9 ↑2.8–12.0

HRM also revealed hypomethylation (CEDM/ET = 0.6, Fig. 4C), but the
effects in the individual vasa CpG3-5 with EE2, DES and Cd were not
reproduced with MS-HRM. Exposure to the three estrogenic com-
↑ 0.02
CpG1

↑ 1.6

↓ 0.9
vtg1

pounds also induced effects in CpG5 of vasa, however with more varia-
–
–
–

–
–

tion between replicate experiments and between compounds. Variable

results were also observed in cyp19a2, although hypomethylation was
CpG1-5

↓ 0.6c
vasa

consistent with EE2 and DES at CpG1 and CpG6; CEDs for these effects
Overview of concentration-response effects and maximum effect sizes of embryotoxicity and locus-specific methylation.

–
–

were also consistently higher than for embryotoxicity (CEDM/ET N 1;

CEDs outside the exposure range were extrapolated). Unlike the estro-
↓ ()b/1.2/1.296h

↓2.5 ↑2.6–6.7
–/–/↑0.0596h

genic compounds, the metals Ni and Cd both induced hypermethylation

↑ N1.4/–

in CpGs in the region of vasa at CEDs near or higher than for

concentration response; empty cells, not tested; eembryotoxicity with VPA showed as teratogenic effects.
N1.5/–
–/↓2.3

↓12.5
CpG5

↑ 3.3
Locus-specific methylation (CEDM/ET)

embryotoxicity (CEDM/ET after Ni in vasa CpG5 of N 1.4 and after Cd in

vasa

↓0.1
–
–

vasa CpG3 of 0.9, CpG5 of N1.5). Ni also induced hypomethylation in

cyp19a2 (CEDM/ET N 1 in CpG6). TBT induced hypermethylation in vtg1
↓14.7
CpG4
vasa

↓0.2

and cyp19a2. PFOA induced, like the estrogen agonists, hypomethyla-

–
–
–

–
–
–
–

–
–
–

tion at CpG3 in the region of vasa (CEDM/ET 0.7). PFOA also induced
↓6.0–13.6 ↑8.9–33

hypermethylation at CpG1 in vtg1, with high sensitivity (CEDM/ET

responses are visually explained in concentration response graphs for BPA in Fig. 1.
↓1.0d/0.3/N1096h
↓ 0.9/0.5/2.196h

0.02). Both embryotoxic controls VPA and flusilazole produced effects

in all three targets, but in most or all cases with a lower sensitivity com-
↓ 0.2/0.5

↑ 0.9/–

pared with morphological effects — the embryotoxic effect here appar-

↑ 14.3

↓0.05

↓31.3
CpG3

↓ 0.7

↓ 3.2
vasa

ently did not depend on effects on DNA methylation. Embryos exposed

–

–
–

to the control methylation inhibitor 5AC revealed hypomethylation in at

Embryotoxicity (ET) (μM)

least one CpG site in each target, where particularly methylation levels
in the region of vasa appeared more sensitive than embryotoxicity,
with relative CEDM/ET of 0.05 in vasa CpG3, 0.2 in CpG4, 0.1 in CpG5,
compared to N1 in vtg1 CpG3 and cyp19a2 CpG6-7. Table 2 also shows
6.1/1.896h

that observations at 96 hpf with EE2 and DES (compared to the standard
N320a

of 72 hpf) generally reproduced the DNA methylation effects in vasa

N3.2a

2900
0.34

5.6e
183

307
2.4
20

67

CpG3 and cyp19a2, although with lower sensitivity compared to mor-

phology. Effects in vasa CpG5 were more variable.
Test range (μM)

Methylation kinetics was further explored in the vasa CpG3 locus at

a single concentration of BPA (10 μM; Fig. 5), which from earlier exper-
100–3200
0.032–10
4.9–27.7

7.7–240
3.2–100

0.32–32
10–320
10–320

10–320
0.1–3.2
0.1–3.2

iments (Table 2) was known to produce hypomethylation at 72 hpf. In

this particular exposure, the effect at 72 hpf was statistically nearly sig-
nificant, and statistical significance only reached at 96 hpf. This delay of
Compounds

the effect could be explained by a slightly delayed development of the

Cadmium

Arsenite

embryos in this particular experiment, as read from delayed hatching

Δ% 5AC
Δ% test
Nickel
Table 2

PFOA
VPA

5AC
BPA

DES

at 72 hpf. A subgroup, which had BPA removed from the medium at

EE2

TBT

FLU

72 hpf still produced a statistically significant hypomethylation at

90 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96

Fig. 2. Heatmap of DREAM analysis of embryos exposed to BPA (3 and 12 μM). Legend of colors in upper left corner. X-axis: different BPA concentrations; 0, 3, 12 μM; all in triplicate. Y-axis:
Supervised clustering of 1041 sites with 10+ reads, N2% difference in methylation and P b 0.05. Clusters of hypo- and hypermethylation can be observed between non-exposed and 12 μM
BPA exposed embryos. Methylation levels of 3 μM BPA exposed embryos intermediate.

96 hpf, whereas differences were no longer observed at 120 hpf, after overrepresented. Most of these (115) were GO-terms, of which the ma-
48 h of recovery. The induced hypomethylation after 96 h of exposure jority (83) comprised terms related to development/morphogenesis. Of
was also no longer observed at 120 hpf, after 24 h of recovery. these 83 processes, a substantial subgroup consisted of 19
Overall, the methylation levels in the DREAM derived targets in non- neurodevelopmental processes, which also gave the highest overrepre-
exposed embryos was N80% (shown for 1041 sites with differential sentation scores. In the remaining non-developmental 32 GO-terms,
methylation in 12 μM BPA compared to unexposed control in Fig. 2, in- transcription factors dominated (9). With 45 processes, Interpro was
sert), and varied between 6.3% and 98.1% in the DREAM derived targets the second largest database represented in the DAVID analysis, with
selected for pyrosequence verification (Fig. 3). Methylation levels were an overrepresentation of transcription factors/zinc fingers/DNA binding
generally high (above 84%) in all measurable CpG sites of the literature (7 terms). The latter were also highly represented in processes that had
derived targets, except for CpG3 in vasa, which was at 49.8% (Table 4). the highest contribution of the genes in the dataset (15 out of 23 with
Maximum induced effect sizes with the test compounds over all targets N5% of the genes present), and neurodevelopmental and gene expres-
ranged between 1.0 and 13.6% for hypomethylation and between 2.6 sion regulatory processes also made up the majority of the 25 most sig-
and 33.0 for hypermethylation, with vasa CpG3 showing the highest nificantly regulated processes, with a count of respectively 10 and 7.
values, and BPA, PFOA and VPA as compounds inducing effect sizes of
N10% (Table 2). Hypomethylation effects with the control DNA methyl- 4. Discussion
ation inhibitor 5AC were comparable to or stronger than with test
compounds. In this study, we explored the zebrafish embryo as a screening
Functional annotation and overrepresentation analysis was per- model for compound-induced DNA methylation modifications, which
formed using DAVID with 454 DREAM identified loci that showed sig- potentially play a role in developmental origins of adult disease (Tang
nificant differences (p b 0.05) between BPA 12 μM and no BPA and Ho, 2007; Wang et al., 2013). Specifically, we set out to identify sim-
controls of minimal 5%, and a concentration response (R N 0.7) over ilarities in epigenetic changes induced by classes of compounds and de-
the three concentrations. Of these loci 79.7% was in gene promoters, termine possible sensitivity differences compared to direct toxic effects.
16.7% in gene introns, and 3.5% in gene exons. The DAVID analysis To address this, freshly fertilized eggs were exposed to xenobiotic com-
showed that in total 196 functional annotation terms were pounds at subtoxic concentrations, including the estrogenic compounds
 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96 91

Fig. 3. Comparing CpG methylation in DREAM and pyrosequence in 13 targets. In each gene, the first triplet of 0–3.1–12 μM BPA represents the DREAM result, the further triplets are
pyrosequence (PSQ) results, representing all CpGs present in the fragment designed around the DREAM CpG (*), see Table 1; n = 3 per concentration.

EE2, DES and BPA, the metals Ni, Cd, TBT, As, the non-estrogenic PPARγ compound. The predetermined response of 5% for DNA methylation ef-
agonist PFOA, the neuropharmaceutical embryotoxicant VPA, FLU and fects was considered as informative, because the zebrafish embryo is a
the control methylation inhibitor 5AC. To examine compound-induced whole organism and thus includes multiple tissues, which can respond
methylation modifications, global methylation and site-specific methyl- differently to xenobiotic exposure (Gorelick et al., 2014; Hao et al.,
ation levels were analyzed. Targets for site-specific methylation were 2013). In this way, a strong response in a confined tissue or limited
explored using a genome wide site-specific method, the DREAM assay. number of cells can lead to an overall small effect in the whole body.
Three alternative targets, in the promoter region of vasa, vtg1 and This also justifies the low effect size criterium in the selection of
cyp19a2, were assessed additionally. DREAM targets. At the same time, a drawback of this whole organism
Both morphological development (/teratogenic) effects and methyl- model is that important effects may be missed because a significant in-
ation levels of the zebrafish embryos were analyzed in a concentration crease in one region may be compensated by a significant decrease of
response design, including a control and four concentrations of each the same size in another.
All compounds showed embryotoxic potential, because they all in-
duced morphological and/or teratogenic effects in exposed zebrafish
Table 3 embryos, albeit with somewhat varying sensitivity between experi-
Correlation of methylation in 13 selected DREAM and pyrosequencing loci with BPA expo- ments. This is most likely explained by variation in the initiation of the
sure concentration.
exposure or biological differences between batches of embryos (Irmler
Pyrosequenced CpGs et al., 2004). Despite this variability of sensitivity, the morphology of ef-
Locus DREAM
CpG1 CpG2 CpG3 CpG4 CpG5 CpG6 fects per compound was similar between experiments. Together, the
observed lethal effects probably represent non-specific pathology, be-
a2ml −0.82 −0.21*
dip2ba 0.89 0.06*
cause of similar observations in non-exposed embryos (particularly co-
ehhadh −0.89 −0.58* −0.35 agulation), albeit with a lower incidence. Teratogenic effects may
ghra 0.77 0.71* represent more specific pathology, and included pericardial edema,
loc100535586 0.86 0.79* 0.67 0.62 0.76 yolk sac edema (DES, 5AC), malformation of the heart (DES, BPA,
mhc1zba 0.90 0.59 0.55*
5AC), malformation of the pectoral fin (FLU) and tail (BPA, DES). The ob-
zgc:198371 0.86 0.06*
pfkmb 0.92 0.24 0.05* 0.05 served embryotoxicity supported that the compounds were absorbed
nxnl2 0.80 0.60 0.70 0.33* by the embryos.
pcdh17 −0.88 −0.32* As a first screen of DNA methylation effects following exposures,
nsfb −0.90 −0.18 −0.20* global methylation of the exposed embryos was examined. Global
epha4b 0.75 0.30 0.55*
senp8 −0.75 0.35* −0.08 0.04 0.15 −0.06 0.39
methylation changes were revealed only after exposure to 5AC, which
is a methyltransferase inhibitor (Friedman, 1979) and was included as
DREAM loci were selected on the basis of correlation with BPA concentration, significance,
control compound. The absence of effects on the global scale with the
and size of the effect. Pyrosequence regions were designed to include at least the DREAM
SmaI locus (marked with *). Included BPA concentrations were 0, 3 and 12 μM, n = 3 per other tested compounds may be due to induced effects that were too
concentration. Correlations N0.6 are highlighted. small to detect in whole embryos, since we observed site-specific
92 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96

Fig. 4. Concentration responses of locus-specific methylation in BPA. A, pyrosequencing of ghra; B, pyrosequencing of vasa CpG3; and C, HRM of vasa CpG1-5. For explanation of symbols
and concentration-response parameters see Fig.1. Small circles, individual data points (which overlap in A); large circles, the mean per concentration. CED at CES = 5% was not achieved
with ghra, and were 3.6 μM and 11.2 μM for vasa CpG3 (pyrosequence) and in the integral vasa fragment (HRM).

methylation modifications. This is in line with Olsvik et al. (2014), who studies after compound exposure (Fang et al., 2013; Xu et al., 2014),
also showed that small site-specific DNA methylation modifications in but our data do not support this as a suitable general biomarker for epi-
zebrafish were not reproduced on the level of global methylation. Fur- genetic changes.
thermore, the DREAM analysis showed differential hypo- and Alternatively, locus-specific methylation effects could be more infor-
hypermethylated regions, which suggests that opposed methylation mative as biomarker for epigenetic changes, and we therefore used a ge-
changes could balance each other in a full DNA digest single nucleotide nome wide locus-specific assay (DREAM) to identify site-specific
analysis, where site specificity is lost. Still, global methylation is a fre- methylation modifications in BPA-exposed embryos. Even though the
quently used biomarker in epidemiology, and multiple diseases appear DREAM assay has no full CpG coverage, it efficiently detects differential-
to be associated with global methylation changes (Zelic et al., 2015). ly methylated CpGs, as compared to an alternative genome wide de-
Global methylation modifications are also observed in experimental tection method like reduced representation bisulfite sequencing

Fig. 5. Methylation kinetics in vasa CpG3 after BPA exposure (10 μM). Exposure was initiated for 72 or 96 hpf, exposure was discontinued in subgroups, which then received control
medium (recovery), after 72 and 96 hpf.
 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96 93

Table 4 observed varying responsiveness between the examined targets. This

Site-specific methylation levels (%). can be understood because the action of DNMTs is regulated through in-
vasa vtg1 cyp19a2 teraction with methyl-CpG-binding domain proteins (MBDs), which
CpG1 84.5 ± 5.8 94.5 ± 0.5
bind locus-specific (Kimura and Shiota, 2003). MBD expression in turn
CpG2 84.6 ± 2.6 97.8 ± 0.7 can be regulated by compound exposure, as for instance reported for es-
CpG3 49.8 ± 2.2 87.4 ± 1.5 93.8 ± 2.2 trogen induced overexpression of demethylating MBD2 (Tang et al.,
CpG4 88.9 ± 2.1 96.7 ± 1.0 2012).
CpG5 94.1 ± 2.2 99.2 ± 1.2
Unlike vasa, DNA methylation in vtg1 was only affected by TBT,
CpG6 91.3 ± 0.8
CpG7 90.7 ± 0.9 PFOA, VPA, FLU and 5AC exposure. We mostly observed hypermethyla-
tion, which is in line with induced DNMT3 regulation observed with
Values are the mean ± sd in percentage of methylated sites in non-exposed embryos of
several experiments (n = 10). CpG1-2 in vasa not analyzed, see Table 1.
PFOA in human liver cells (Tian et al., 2012). Vtg1 is expressed in the
liver of non-mammalian vertebrates and encodes for the egg yolk pro-
tein vitellogenin. Even though vtg1 is known as an estrogen-
(Jelinek et al., 2012). On the other hand, DREAM only covers 64.5 k of responsive gene which expression can be altered by estrogenic expo-
total 24.2 M CpGs present in the zebrafish genome (Jiang et al., 2013) sure in both zebrafish embryos and adults (Chow et al., 2013; Wang
(0.27%), even with limited reading depth in the majority of the 64.5 k et al., 2005; Zhong et al., 2014), and Strömqvist et al. (2010) showed
CpGs, and relevant targeted CpG modifications may thus be missed. that exposure to EE2 induced methylation modifications in the promot-
Functional annotation analysis (DAVID) of the DREAM results indi- er region of vtg1 in liver of adult zebrafish. Our observations did not con-
cated predominant involvement of processes related to firm such estrogen-induced methylation modifications in zebrafish
(neuro)development and transcriptional/DNA binding mediated gene embryos. The interrogated promoter domain of vtg1 is thus possibly
expression. Although this analysis may provide leads for explorative di- not involved with regulation of gene expression by DNA methylation,
rections of further functional effects, it should be considered indicative, or at least not at this stage of development, or the effect in the embryo
because it is uncertain to what extent the detected DNA methylation liver disappeared in the noise of the whole body.
modifications translate into dysregulation of gene expression. All compounds that induced an effect in cyp19a2 (BPA, DES, Ni, TBT,
Further criteria for confidence and significance and technical feasi- VPA, FLU and 5AC) did so with relative low sensitivity (CEDM/ET N 1).
bility were used to narrow the selection to 13 potential new targets Cyp19a2 encodes brain aromatase, which is involved in E2 synthesis in
for further analyzing by pyrosequencing. Although often discordant re- the brain and known to be estrogen sensitive (Diotel et al., 2010; Le
sults from one CpG to the next within each promoter were observed, the Page et al., 2008; Menuet et al., 2005). DNA methylation of cyp19a2
DREAM results were replicated by pyrosequencing in a limited set of has not been studied in zebrafish before, but sex-dependent methyla-
markers, including ghra, loc100535586, and nxnl2. A subsequent analysis tion modification (hypermethylation in males) was reported in adult
in a full concentration-response, however, only maintained a statistical- medaka (Oryzias latipes) after EE2 exposure (Contractor et al., 2004),
ly significant effect for ghra, which was even due to a deviating control supporting a role of estrogen regulated DNA methylation of expression
group, and with a limited size of the effect. Thus, both DREAM and PSQ of this gene. The observed effects of DES and EE2 support estrogenic reg-
methylation differences after BPA exposures are minimal and not con- ulation of methylation in our embryos, but the low sensitivity and ab-
vincing. This could be interpreted as a minimal effect of BPA on DNA sence of effect of BPA may suggest that BPA is not active in the brain
methylation, but could also be a result of differential effects on the var- of the embryo, or that the locus selection could be optimized.
ious tissues in the mixture of the zebrafish embryo whole body extract. Altogether, we observed locus-specific methylation modifications
In addition, we retrieved targets from literature (see below), located even though no global methylation modifications were detected. Con-
in the promoter region of the vasa, vtg1 and cyp19a2 genes. None of the sidering effects on DNA methylation in the three targets, cyp19a2 was
included CpGs in these targets were SmaI/XmaI sites and these could not a particularly informative marker, because effects on methylation
therefore not be included in the DREAM results. The supported signifi- did not occur at lower concentrations than morphological effects. The
cance by literature of these sites as targets for epigenetic modifications, CpG3 locus in vasa turned out to be the most responsive, and effects
and a clear concentration-response for decreased methylation after ex- in vasa even separated the estrogenic compounds from the metals
posure with BPA in the vasa promotor, encouraged us to test this set through relatively sensitive hypomethylation in the first group. In the
with other xenobiotics. Thus, methylation modifications in the region overall assay set, this locus stands out by its low background methyla-
of vasa were induced by all tested compounds, except As. Targets tion level compared to the other loci, which may contribute to its re-
proved to respond with varying sensitivity of methylation effects per sponsiveness. This is in line with the observation in a genome wide
compound, relative to embryotoxicity (Table 2). However, the effects analysis in liver of mice, showing that variation in methylation induced
in vasa appeared to be linked to individual CpGs instead of overall meth- by BPA departed most from the control group in the range of intermedi-
ylation in the amplicon, because HRM, which analyzed the integrated ately methylated loci (Van Esterik et al., 2014b). On the other hand,
effect in the entire vasa amplicon encompassing 5 CpGs, only none of the 5 (out of total 29 CpGs) in the DREAM associated targets
reproduced the effect in one out of 4 assayed cases (BPA). in the final selection with the intermediate background methylation
Vasa is specifically expressed in the germline and active during de- levels near that of vasa CpG3 (Fig. 3) produced significant
velopment (Nagasawa et al., 2013). Previous studies reported that concentration-responses with BPA.
methylation in the promoter region of vasa in zebrafish embryos Overall, the estrogenic compounds together with PFOA stand out by
could be altered by exposure to benzo[a]pyrene (Fang et al., 2013). high relative sensitivity of DNA methylation effects compared to
However, exact regulation of vasa is unknown (Li et al., 2013). ER-α reg- embryotoxic effects. Although the metals (except As), VPA and FLU pro-
ulates the expression of DNMTs (Shi et al., 2012), which may represent a duced methylation effects, these did so with relative low sensitivity
mode of action of estrogens in regulation of DNA methylation. The compared to embryotoxic effects, suggesting that the epigenetic chang-
metals As and Cd have been reported to interact with DNMT function es were linked to the direct toxic effects, or a bystander effect. The
as well, although with variable effects which may depend on experi- embryotoxic VPA is primarily known to inhibit histone deacetylase, al-
mental model (Arita and Costa, 2009) or on exposure duration though it has also been shown to induce DNA demethylation (Tung
(Takiguchi et al., 2003), while Ni exposure lead to upregulation of the and Winn, 2010). As is known as a modifier of DNA methylation
expression of DNMT1 (Ji et al., 2008). If interaction with DNMTs should (Marsit, 2015), but the absence of effects with As is supported by the
be the major affected pathway of endocrine- and metal-induced meth- minimal As induced tissue-specific DNA methylation alterations ob-
ylation, this regulation is suggested to be target-specific in view of the served in zebrafish embryos by Li et al. (2009), and As may be more
94 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96

active in other, unexplored targets. FLU (and other triazoles) have no biomarker. Effects of BPA as analyzed with DREAM were minimal, and
known epigenetic effects, but these could well be related to the endo- did not produce convincing markers. However, there were informative
crine actions of FLU via cyp26a (retinoid metabolism) and cyp51 (ste- responses in additional specific targets, particularly vasa, and vtg1 and
roid synthesis) (e.g. Urvalek and Gudas, 2014). The estrogenic cyp19a2 to a lesser extent. Some of the observed DNA methylation ef-
compounds included in this study are known to have different E2 po- fects were even more sensitive than morphological effects, and com-
tency (DES N EE2 ⋙ BPA (Mitsui et al., 2007)). Our data confirmed pound or class specific effects were suggested. Taken together, this
DES as the most potent of the three estrogenic agonists, closely followed study showed that the zebrafish embryo may be informative to screen
by EE2, then BPA, by CEDs for embryotoxicity (respectively 2.4, N 3.2, for epigenetic effects after xenobiotic exposure, but that efficacy as a
and 20 μM) and by CEDs for methylation effects in vasa CpG3 (respec- screen very much depends on selection of targets and analytical tools.
tively 2.4, 2.8, and 3.6 μM). As a next step, its applicability as a model needs to be confirmed
In the replicate experiments, the methylation effects showed some through expansion of the compound and target database. Also, further
quantitative variability, which can be explained by the critical timing exploration is needed to elucidate the potential of epigenetic effects in
of both the initiation of the exposure and of the final analysis and the as- the early embryo to predict the adverse phenotype.
sociated high dynamics of DNA methylation during the early stage of
development (Kamstra et al., 2015; MacKay et al., 2007; Mhanni and Transparency document
McGowan, 2004). Exposure was initiated as early as possible, generally
at the 8–16 cell stage (1.2–1.5 hpf), but some variation is unavoidable. The Transparency document associated with this article can be
The effect of time of analysis was indicated by the decreased sensitivity found, in the online version.
of DNA methylation effects relative to embryotoxic effects at 96 hpf
compared to 72 hpf, as tested with EE2, DES and FLU. The early time Acknowledgment
point of 72 hpf was initially selected as informative in view of interest
in early organogenesis. After organogenesis most of the de- and This study has been performed by order and for the account of The
remethylation processes are finalized and the DNA methylome is Netherlands Food and Consumer Product Safety Authority (NVWA),
established for the most part (Jiang et al., 2013; Kamstra et al., 2015; and we thank Dr. Jacqueline JM Castenmiller (NVWA) for her active
Potok et al., 2013). However, the OECD TG236 evaluates until 96 hpf, support. The study was further supported by the RIVM strategic re-
which thus appears to be less optimal for evaluation of DNA methyla- search program, project S/340007. JK and JL were supported by NWO
tion effects, although other targets than the present set may show ASPASIA/864.09.005 and VIDI/864.09.005. We wish to acknowledge
other kinetics. Hans Cremers, Wilmer Wamsteeker and Kevin van Dal (RIVM) for the
The recovery experiment with a single concentration of BPA (Fig. 5), technical support. We thank Wilma Steegenga and Carolien Lute of
where methylation effects in vasa CpG3 were only observed at the the Department Agrotechnology and Food Sciences of Wageningen Uni-
96 hpf time point instead of at 72 hpf, could be explained by an overall versity for using the pyrosequencing facilities.
delayed development. Interestingly, the effect at 96 hpf was not depen-
dent of continued exposure until that time point, indicating that effects
Appendix A. Supplementary data
on DNA methylation induced at 72 hpf were persistent. The comparable
methylation levels at 120 hpf between control and groups recovered
Supplementary data to this article can be found online at http://dx.
after BPA exposure for either 48 h or 24 h can be speculated to be
doi.org/10.1016/j.taap.2015.12.012.
more attributable to masking of effects due to changing relative contri-
butions of various tissues over time, than to actual recovery of original
References
methylation, in view of the changing trend of controls over time.
This study was a first attempt to explore the potential of the Arita, A., Costa, M., 2009. Epigenetics in metal carcinogenesis: nickel, arsenic, chromium
zebrafish embryo to screen for xenobiotic-induced effects on DNA and cadmium. Metallomics 1, 222–228. http://dx.doi.org/10.1039/b903049b.
Bailey, K.A., Fry, R.C., 2014. Arsenic-associated changes to the epigenome: what are the
methylation using dedicated markers. DNA methylation effects were
functional consequences? Curr. Environ. Health Rep. 1, 22–34. http://dx.doi.org/10.
suggested to underlie epigenetic programming of disease later in life. 1007/s40572-013-0002-8.
This is a first step to show that locus specific DNA methylation effects in- Bakulski, K.M., Rozek, L.S., Dolinoy, D.C., Paulson, H.L., Hu, H., 2012. Alzheimer's Disease
and environmental exposure to lead: the epidemiologic evidence and potential role
deed can be detected and that these are not a mere secondary or
of epigenetics. Curr. Alzheimer Res. 9, 563–573. http://dx.doi.org/10.2174/
aspecific effect of embryotoxicity, because they were not observed in 156720512800617991.
all interrogated loci with each xenobiotic at embryotoxic concentra- Chow, W.S., Chan, W.K.-L., Chan, K.M., 2013. Toxicity assessment and vitellogenin expres-
tions. Also the observation that with a subset of compounds, the DNA sion in zebrafish (Danio rerio) embryos and larvae acutely exposed to bisphenol A,
endosulfan, heptachlor, methoxychlor and tetrabromobisphenol A: toxic effects of
methylation effects occurred at lower concentrations than embryotoxic bisphenol A and pesticides on zebrafish embryo and larvae. J. Appl. Toxicol. 33,
effects is of importance, because this supports that potentially persisting 670–678. http://dx.doi.org/10.1002/jat.2723.
epigenetic effects are not necessarily identified by a direct adverse phe- Contractor, R.G., Foran, C.M., Li, S., Willett, K.L., 2004. Evidence of gender-and tissue-
specific promoter methylation and the potential for ethinylestradiol-induced changes
notype, although it remains to be shown whether the observed effects in Japanese medaka (Oryzias latipes) estrogen receptor and aromatase genes. J. Toxic.
can set the conditions for, and thus predict adult disease, and whether Environ. Health A 67, 1–22. http://dx.doi.org/10.1080/15287390490253633.
they are (mechanistically) linked to altered cell and body functions. Diotel, N., Le Page, Y., Mouriec, K., Tong, S.-K., Pellegrini, E., Vaillant, C., Anglade, I., Brion, F.,
Pakdel, F., Chung, B., Kah, O., 2010. Aromatase in the brain of teleost fish: expression,
More comprehensive studies, including more systematic genome- regulation and putative functions. Front. Neuroendocrinol. 31, 172–192. http://dx.
wide DNA methylation explorations such as executed here with the doi.org/10.1016/j.yfrne.2010.01.003.
DREAM method in the case of BPA, and supplemented with molecular Esteller, M., Corn, P.G., Baylin, S.B., Herman, J.G., 2001. A gene hypermethylation profile of
human cancer. Cancer Res. 61, 3225–3229.
studies further down the pathway to altered cell and body functions,
Fang, X., Thornton, C., Scheffler, B.E., Willett, K.L., 2013. Benzo[a]pyrene decreases global
may reveal DNA methylation marker sets that are predictive of adult ad- and gene specific DNA methylation during zebrafish development. Environ. Toxicol.
verse conditions. This study supports that in this context, the zebrafish Pharmacol. 36, 40–50. http://dx.doi.org/10.1016/j.etap.2013.02.014.
Fernandez, S.V., Huang, Y., Snider, K.E., Zhou, Y., Pogash, T.J., Russo, J., 2012. Expression
embryo can be an informative model.
and DNA methylation changes in human breast epithelial cells after bisphenol A ex-
posure. Int. J. Oncol. 41, 369–377. http://dx.doi.org/10.3892/ijo.2012.1444.
5. Conclusion Friedman, S., 1979. The effect of 5-azacytidine on E. coli DNA methylase. Biochem.
Biophys. Res. Commun. 89, 1328–1333.
Gluckman, P.D., Hanson, M.A., 2004. Developmental origins of disease paradigm: a mech-
Analysis of DNA methylation after compound exposure in zebrafish anistic and evolutionary perspective. Pediatr. Res. 56, 311–317. http://dx.doi.org/10.
embryos showed that global DNA methylation is not an informative 1203/01.PDR.0000135998.08025.FB.
 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96 95

Gorelick, D.A., Iwanowicz, L.R., Hung, A.L., Blazer, V.S., Halpern, M.E., 2014. Transgenic Menuet, A., Pellegrini, E., Brion, F., Gueguen, M.-M., Anglade, I., Pakdel, F., Kah, O., 2005.
zebrafish reveal tissue-specific differences in estrogen signaling in response to envi- Expression and estrogen-dependent regulation of the zebrafish brain aromatase
ronmental water samples. Environ. Health Perspect. 122. http://dx.doi.org/10.1289/ gene. J. Comp. Neurol. 485, 304–320. http://dx.doi.org/10.1002/cne.20497.
ehp.1307329. Mhanni, A.A., McGowan, R.A., 2004. Global changes in genomic methylation levels during
Greally, J.M., Jacobs, M.N., 2013. In vitro and in vivo testing methods of epigenomic end- early development of the zebrafish embryo. Dev. Genes Evol. 214, 412–417. http://dx.
points for evaluating endocrine disruptors. ALTEX 30, 445–471. doi.org/10.1007/s00427-004-0418-0.
Hao, R., Bondesson, M., Singh, A.V., Riu, A., McCollum, C.W., Knudsen, T.B., Gorelick, D.A., Mitsui, N., Tooi, O., Kawahara, A., 2007. Vitellogenin-inducing activities of natural, syn-
Gustafsson, J.-Ã., 2013. Identification of estrogen target genes during zebrafish em- thetic, and environmental estrogens in primary cultured Xenopus laevis hepatocytes.
bryonic development through transcriptomic analysis. PLoS One 8, e79020. http:// Comp. Biochem. Physiol. Toxicol. Pharmacol. CBP 146, 581–587. http://dx.doi.org/10.
dx.doi.org/10.1371/journal.pone.0079020. 1016/j.cbpc.2007.07.004.
Hermsen, S.A.B., van den Brandhof, E.-J., van der Ven, L.T.M., Piersma, A.H., 2011. Relative Nagasawa, K., Fernandes, J.M.O., Yoshizaki, G., Miwa, M., Babiak, I., 2013. Identification
embryotoxicity of two classes of chemicals in a modified zebrafish embryotoxicity and migration of primordial germ cells in Atlantic salmon, Salmo salar: characteriza-
test and comparison with their in vivo potencies. Toxicol. In Vitro 25, 745–753. tion of vasa, dead end, and lymphocyte antigen 75 genes. Mol. Reprod. Dev. 80,
http://dx.doi.org/10.1016/j.tiv.2011.01.005. 118–131. http://dx.doi.org/10.1002/mrd.22142.
Ho, S.-M., Johnson, A., Tarapore, P., Janakiram, V., Zhang, X., Leung, Y.-K., 2012. Environ- Newman, M., Blyth, B.J., Hussey, D.J., Jardine, D., Sykes, P.J., Ormsby, R.J., 2012. Sensitive
mental epigenetics and its implication on disease risk and health outcomes. ILAR J. quantitative analysis of murine LINE1 DNA methylation using high resolution melt
53, 289–305. http://dx.doi.org/10.1093/ilar.53.3-4.289. analysis. Epigenetics Off. J. DNA Methylation Soc. 7, 92–105. http://dx.doi.org/10.
Hou, L., Zhang, X., Wang, D., Baccarelli, A., 2012. Environmental chemical exposures and 4161/epi.7.1.18815.
human epigenetics. Int. J. Epidemiol. 41, 79–105. http://dx.doi.org/10.1093/ije/ OECD, 2002. Guidance Document on Transformation/Dissolution of Metals and Metal
dyr154. Compounds in Aqueous Media, OECD Series on Testing and Assessment. OECD
Huang, Y., Rao, A., 2014. Connections between TET proteins and aberrant DNA modifica- Publishing.
tion in cancer. Trends Genet. 30, 464–474. http://dx.doi.org/10.1016/j.tig.2014.07. Olsvik, P.A., Williams, T.D., Tung, H., Mirbahai, L., Sanden, M., Skjaerven, K.H., Ellingsen, S.,
005. 2014. Impacts of TCDD and MeHg on DNA methylation in zebrafish (Danio rerio)
Huang, D.W., Sherman, B.T., Lempicki, R.A., 2008. Systematic and integrative analysis of across two generations. Comp. Biochem. Physiol. Toxicol. Pharmacol. CBP 165,
large gene lists using DAVID bioinformatics resources. Nat. Protoc. 4, 44–57. http:// 17–27. http://dx.doi.org/10.1016/j.cbpc.2014.05.004.
dx.doi.org/10.1038/nprot.2008.211. Potok, M.E., Nix, D.A., Parnell, T.J., Cairns, B.R., 2013. Reprogramming the maternal
Huang, D.W., Sherman, B.T., Lempicki, R.A., 2009. Bioinformatics enrichment tools: paths zebrafish genome after fertilization to match the paternal methylation pattern. Cell
toward the comprehensive functional analysis of large gene lists. Nucleic Acids Res. 153, 759–772. http://dx.doi.org/10.1016/j.cell.2013.04.030.
37, 1–13. http://dx.doi.org/10.1093/nar/gkn923. Quinlivan, E.P., Gregory, J.F., 2008. DNA digestion to deoxyribonucleoside: a simplified
Irmler, I., Schmidt, K., Starck, J.M., 2004. Developmental variability during early embryonic one-step procedure. Anal. Biochem. 373, 383–385. http://dx.doi.org/10.1016/j.ab.
development of zebra fish, Danio rerio. J. Exp. Zoolog. B Mol. Dev. Evol. 302, 446–457. 2007.09.031.
http://dx.doi.org/10.1002/jez.b.21010. Rasoulpour, R.J., LeBaron, M.J., Ellis-Hutchings, R.G., Klapacz, J., Gollapudi, B.B., 2011. Epi-
Jelinek, J., Liang, S., Lu, Y., He, R., Ramagli, L.S., Shpall, E.J., Estecio, M.R.H., Issa, J.-P.J., 2012. genetic screening in product safety assessment: are we there yet? Toxicol. Mech.
Conserved DNA methylation patterns in healthy blood cells and extensive changes in Methods 21, 298–311. http://dx.doi.org/10.3109/15376516.2011.557883.
leukemia measured by a new quantitative technique. Epigenetics 7, 1368–1378. Reed, C.E., Fenton, S.E., 2013. Exposure to diethylstilbestrol during sensitive life stages: a
http://dx.doi.org/10.4161/epi.22552. legacy of heritable health effects. Birth Defects Res. Part C Embryo Today Rev. 99,
Ji, W., Yang, L., Yu, L., Yuan, J., Hu, D., Zhang, W., Yang, J., Pang, Y., Li, W., Lu, J., Fu, J., Chen, J., 134–146. http://dx.doi.org/10.1002/bdrc.21035.
Lin, Z., Chen, W., Zhuang, Z., 2008. Epigenetic silencing of O6-methylguanine DNA Roseboom, T., de Rooij, S., Painter, R., 2006. The Dutch famine and its long-term conse-
methyltransferase gene in NiS-transformed cells. Carcinogenesis 29, 1267–1275. quences for adult health. Early Hum. Dev. 82, 485–491. http://dx.doi.org/10.1016/j.
http://dx.doi.org/10.1093/carcin/bgn012. earlhumdev.2006.07.001.
Jiang, L., Zhang, J., Wang, J.-J., Wang, L., Zhang, L., Li, G., Yang, X., Ma, X., Sun, X., Cai, J., Shi, J.-F., Li, X.-J., Si, X.-X., Li, A.-D., Ding, H.-J., Han, X., Sun, Y.-J., 2012. ERα positively reg-
Zhang, J., Huang, X., Yu, M., Wang, X., Liu, F., Wu, C.-I., He, C., Zhang, B., Ci, W., Liu, ulated DNMT1 expression by binding to the gene promoter region in human breast
J., 2013. Sperm, but not oocyte, DNA methylome is inherited by zebrafish early em- cancer MCF-7 cells. Biochem. Biophys. Res. Commun. 427, 47–53. http://dx.doi.org/
bryos. Cell 153, 773–784. http://dx.doi.org/10.1016/j.cell.2013.04.041. 10.1016/j.bbrc.2012.08.144.
Jones, P.A., 2012. Functions of DNA methylation: islands, start sites, gene bodies and be- Slob, W., 2002. Dose–response modeling of continuous endpoints. Toxicol. Sci. Off. J. Soc.
yond. Nat. Rev. Genet. 13, 484–492. http://dx.doi.org/10.1038/nrg3230. Toxicol. 66, 298–312.
Kamstra, J.H., Aleström, P., Kooter, J.M., Legler, J., 2014. Zebrafish as a model to study the Smith, Z.D., Meissner, A., 2013. DNA methylation: roles in mammalian development. Nat.
role of DNA methylation in environmental toxicology. Environ. Sci. Pollut. Res. Int. Rev. Genet. 14, 204–220. http://dx.doi.org/10.1038/nrg3354.
http://dx.doi.org/10.1007/s11356-014-3466-7. Strömqvist, M., Tooke, N., Brunström, B., 2010. DNA methylation levels in the 5′
Kamstra, J.H., Løken, M., Aleström, P., Legler, J., 2015. Dynamics of DNA flanking region of the vitellogenin I gene in liver and brain of adult zebrafish
hydroxymethylation in zebrafish. Zebrafish 12, 230–237. http://dx.doi.org/10.1089/ (Danio rerio)—sex and tissue differences and effects of 17alpha-ethinylestradiol
zeb.2014.1033. exposure. Aquat. Toxicol. 98, 275–281. http://dx.doi.org/10.1016/j.aquatox.
Katz, T.A., Huang, Y., Davidson, N.E., Jankowitz, R.C., 2014. Epigenetic reprogramming in 2010.02.023.
breast cancer: from new targets to new therapies. Ann. Med. 46, 397–408. http:// Sun, H., Shamy, M., Costa, M., 2013. Nickel and epigenetic gene silencing. Gene 4,
dx.doi.org/10.3109/07853890.2014.923740. 583–595. http://dx.doi.org/10.3390/genes4040583.
Kimura, H., Shiota, K., 2003. Methyl-CpG-binding protein, MeCP2, is a target molecule for Takiguchi, M., Achanzar, W.E., Qu, W., Li, G., Waalkes, M.P., 2003. Effects of cadmium on
maintenance DNA methyltransferase, Dnmt1. J. Biol. Chem. 278, 4806–4812. http:// DNA-(cytosine-5) methyltransferase activity and DNA methylation status during
dx.doi.org/10.1074/jbc.M209923200. cadmium-induced cellular transformation. Exp. Cell Res. 286, 355–365.
Le Page, Y., Menuet, A., Kah, O., Pakdel, F., 2008. Characterization of a cis-acting element Tang, W., Ho, S., 2007. Epigenetic reprogramming and imprinting in origins of disease.
involved in cell-specific expression of the zebrafish brain aromatase gene. Mol. Rev. Endocr. Metab. Disord. 8, 173–182. http://dx.doi.org/10.1007/s11154-007-
Reprod. Dev. 75, 1549–1557. http://dx.doi.org/10.1002/mrd.20892. 9042-4.
Li, D., Lu, C., Wang, J., Hu, W., Cao, Z., Sun, D., Xia, H., Ma, X., 2009. Developmental mech- Tang, W., Morey, L.M., Cheung, Y.Y., Birch, L., Prins, G.S., Ho, S., 2012. Neonatal exposure to
anisms of arsenite toxicity in zebrafish (Danio rerio) embryos. Aquat. Toxicol. 91, estradiol/bisphenol A alters promoter methylation and expression of Nsbp1 and
229–237. http://dx.doi.org/10.1016/j.aquatox.2008.11.007. Hpcal1 genes and transcriptional programs of Dnmt3a/b and Mbd2/4 in the rat pros-
Li, X., Ycaza, J., Blumberg, B., 2011. The environmental obesogen tributyltin chloride acts tate gland throughout life. Endocrinology 153, 42–55. http://dx.doi.org/10.1210/en.
via peroxisome proliferator activated receptor gamma to induce adipogenesis in mu- 2011-1308.
rine 3T3-L1 preadipocytes. J. Steroid Biochem. Mol. Biol. 127, 9–15. http://dx.doi.org/ Tian, M., Peng, S., Martin, F.L., Zhang, J., Liu, L., Wang, Z., Dong, S., Shen, H., 2012.
10.1016/j.jsbmb.2011.03.012. Perfluorooctanoic acid induces gene promoter hypermethylation of glutathione-S-
Li, M., Guan, G., Hong, N., Hong, Y., 2013. Multiple regulatory regions control the tran- transferase Pi in human liver L02 cells. Toxicology 296, 48–55. http://dx.doi.org/10.
scription of medaka germ gene vasa. Biochimie 95, 850–857. http://dx.doi.org/10. 1016/j.tox.2012.03.003.
1016/j.biochi.2012.12.003. Tung, E.W.Y., Winn, L.M., 2010. Epigenetic modifications in valproic acid-induced terato-
Li, Y., Hamilton, K.J., Lai, A.Y., Burns, K.A., Li, L., Wade, P.A., Korach, K.S., 2014. Diethylstil- genesis. Toxicol. Appl. Pharmacol. 248, 201–209. http://dx.doi.org/10.1016/j.taap.
bestrol (DES)-stimulated hormonal toxicity is mediated by ERα alteration of target 2010.08.001.
gene methylation patterns and epigenetic modifiers (DNMT3A, MBD2, and HDAC2) Urvalek, A.M., Gudas, L.J., 2014. Retinoic acid and histone deacetylases regulate epigenetic
in the mouse seminal vesicle. Environ. Health Perspect. 122, 262–268. http://dx.doi. changes in embryonic stem cells. J. Biol. Chem. 289, 19519–19530. http://dx.doi.org/
org/10.1289/ehp.1307351. 10.1074/jbc.M114.556555.
Lindeman, L.C., Winata, C.L., Aanes, H., Mathavan, S., Alestrm, P., Collas, P., 2010. Chroma- Vaiserman, A., 2014. Early-life exposure to endocrine disrupting chemicals and later-life
tin states of developmentally-regulated genes revealed by DNA and histone methyl- health outcomes: an epigenetic bridge? Aging Dis. 5, 419–429. http://dx.doi.org/10.
ation patterns in zebrafish embryos. Int. J. Dev. Biol. 54, 803–813. http://dx.doi.org/ 14336/AD.2014.0500419.
10.1387/ijdb.103081ll. Van Esterik, J.C.J., Dollé, M.E.T., Lamoree, M.H., van Leeuwen, S.P.J., Hamers, T., Legler, J.,
MacKay, A.B., Mhanni, A.A., McGowan, R.A., Krone, P.H., 2007. Immunological detec- van der Ven, L.T.M., 2014a. Programming of metabolic effects in C57BL/6JxFVB mice
tion of changes in genomic DNA methylation during early zebrafish develop- by exposure to bisphenol A during gestation and lactation. Toxicology 321, 40–52.
ment. Genome Natl. Res. Counc. Can. 50, 778–785. http://dx.doi.org/10.1139/ http://dx.doi.org/10.1016/j.tox.2014.04.001.
g07-055. Van Esterik, J.C.J., Vitins, A.P., Hodemaekers, H.M., Kamstra, J.H., Legler, J., Pennings, J.L.A.,
Marsit, C.J., 2015. Influence of environmental exposure on human epigenetic regulation. Steegenga, W.T., Lute, C., Jelinek, J., Issa, J.P.J., Dollé, M.E.T., van der Ven, L.T.M., 2014b.
J. Exp. Biol. 218, 71–79. http://dx.doi.org/10.1242/jeb.106971. Liver DNA methylation analysis in adult female C57BL/6JxFVB mice following
96 M.C. Bouwmeester et al. / Toxicology and Applied Pharmacology 291 (2016) 84–96

perinatal exposure to bisphenol A. Toxicol. Lett. 232, 293–300. http://dx.doi.org/10. World Health Organization, 2014. Noncommunicable Diseases Country Profiles 2014.
1016/j.toxlet.2014.10.021. Xu, N., Chua, A.K., Jiang, H., Liu, N.-A., Goodarzi, M.O., 2014. Early embryonic androgen ex-
Wang, H., Tan, J.T.T., Emelyanov, A., Korzh, V., Gong, Z., 2005. Hepatic and extrahepatic ex- posure induces transgenerational epigenetic and metabolic changes. Mol. Endocrinol.
pression of vitellogenin genes in the zebrafish, Danio rerio. Gene 356, 91–100. http:// Baltim. Md 28, 1329–1336. http://dx.doi.org/10.1210/me.2014-1042.
dx.doi.org/10.1016/j.gene.2005.03.041. Zelic, R., Fiano, V., Grasso, C., Zugna, D., Pettersson, A., Gillio-Tos, A., Merletti, F., Richiardi,
Wang, B., Li, Y., Shao, C., Tan, Y., Cai, L., 2012. Cadmium and its epigenetic effects. Curr. L., 2015. Global DNA hypomethylation in prostate cancer development and progres-
Med. Chem. 19, 2611–2620. sion: a systematic review. Prostate Cancer Prostatic Dis. 18, 1–12. http://dx.doi.org/
Wang, G., Walker, S.O., Hong, X., Bartell, T.R., Wang, X., 2013. Epigenetics and early life or- 10.1038/pcan.2014.45.
igins of chronic noncommunicable diseases. J. Adolesc. Health Off. Publ. Soc. Adolesc. Zhong, L., Yuan, L., Rao, Y., Li, Z., Zhang, X., Liao, T., Xu, Y., Dai, H., 2014. Distribution of vi-
Med. 52, S14–S21. http://dx.doi.org/10.1016/j.jadohealth.2012.04.019. tellogenin in zebrafish (Danio rerio) tissues for biomarker analysis. Aquat. Toxicol.
Woldemichael, B.T., Bohacek, J., Gapp, K., Mansuy, I.M., 2014. Epigenetics of memory and 149, 1–7. http://dx.doi.org/10.1016/j.aquatox.2014.01.022.
plasticity. Prog. Mol. Biol. Transl. Sci. 122, 305–340. http://dx.doi.org/10.1016/B978-
0-12-420170-5.00011-8.