Reproductive Toxicology 49 (2014) 101–116

Contents lists available at ScienceDirect

Reproductive Toxicology
journal homepage: www.elsevier.com/locate/reprotox

Zebrafish embryotoxicity test for developmental (neuro)toxicity:

Demo case of an integrated screening approach system using
anti-epileptic drugs
Anna Beker van Woudenberg a , Cor Snel b , Eke Rijkmans a , Didima de Groot a ,
Marga Bouma a , Sanne Hermsen c , Aldert Piersma c , Aswin Menke b , André Wolterbeek a,∗
a
TNO, Research Group Risk Analysis for Products In Development (RAPID), Zeist, The Netherlands
b
TNO Triskelion B.V., Zeist, The Netherlands
c
Center for Health Protection, National Institute for Public Health and the Environment (RIVM), Bilthoven, The Netherlands

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

Article history: To improve the predictability of the zebrafish embryotoxicity test (ZET) for developmental (neuro)toxicity
Received 16 December 2013 screening, we used a multiple-endpoints strategy, including morphology, motor activity (MA), histopath-
Received in revised form 14 July 2014 ology and kinetics. The model compounds used were antiepileptic drugs (AEDs): valproic acid (VPA),
Accepted 31 July 2014
carbamazepine (CBZ), ethosuximide (ETH) and levetiracetam (LEV). For VPA, histopathology was the
Available online 8 August 2014
most sensitive parameter, showing effects already at 60 ␮M. For CBZ, morphology and MA were the
most sensitive parameters, showing effects at 180 ␮M. For ETH, all endpoints showed similar sensitivity
Keywords:
(6.6 mM), whereas MA was the most sensitive parameter for LEV (40 mM). Inclusion of kinetics did not
Zebrafish embryotoxicity test (ZET)
Developmental (neuro)toxicity
alter the absolute ranking of the compounds, but the relative potency was changed considerably. Taking
Antiepileptic drugs all together, this demo-case study showed that inclusion of multiple-endpoints in ZET may increase the
Integrated test strategy sensitivity of the assay, contribute to the elucidation of the mode of toxic action and to a better definition
Histopathology of the applicability domain of ZET.
Kinetics © 2014 Elsevier Inc. All rights reserved.
Behavior
Gene expression

1. Introduction systems, making it a suitable model organism for medium/high-

Developmental toxicity testing in laboratory animals is required
for the preclinical safety assessment of new pharmaceuticals [1].
Although studies with rodents are still the “gold standard” used to
study developmental toxicity for regulatory purposes, over the last
decades efforts have been made to replace these studies with cell-
based alternatives (e.g. stem cells) [2–5]. Rodent studies are time
consuming, expensive and involve large numbers of laboratory ani-
mals. To date, however, the search for alternatives has not been
very successful, due to the difficulty of investigating the complex-
ity of developmental processes in relatively simple cellular assays
[6]. The zebrafish (Danio rerio) is emerging as a candidate lower
vertebrate animal model capable of filling the gap between high
throughput in vitro cellular assays and conventional preclinical ani-
mal testing [7–10]. The embryonic/larval zebrafish model offers an
intact whole animal model with many of the advantages of in vitro
∗ Corresponding author. Tel.: +31 888665053.
E-mail address: andre.wolterbeek@tno.nl (A. Wolterbeek).
throughput screening [11,12].
Currently, one of the most commonly used endpoints for devel-
opmental toxicity screening in zebrafish is its morphology, which
makes it relatively easy to detect various anomalies [13,14] that
correspond quite well with higher mammalian species [15,16]. In a
previous study [17], we demonstrated that additional histopatho-
logical examination of the larvae contributed to an improved
understanding of the mode of toxicity action, thereby increasing the
predictability of the zebrafish assay for toxicity screening. The use
of zebrafish larvae to test compounds for potential developmental
(neuro)toxicity involves various behavioral tests, such as the motor
activity test (MA). Tests of this kind are based on measuring the
effects on factors such as total distance moved and velocity [18–20].
Recently, an extension of the MA test analysis was proposed, in
which a wider range of endpoints can be evaluated from the same
collected data set [17]. Furthermore, other endpoints than morpho-
logical scoring have been used in the embryonic and larval model to
evaluate the potential toxicity of compounds, including proteomics
[21] and transcriptomics [22]. Although each endpoint evaluation
has its own relevance, in general, they are studied separately,

http://dx.doi.org/10.1016/j.reprotox.2014.07.082
0890-6238/© 2014 Elsevier Inc. All rights reserved.
102 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116
so the results of different studies need to be integrated. In addition,
the measurement of test compound uptake by zebrafish embryos
and larvae has been shown to be a critical factor, one that can influ-
ence the predictability of the assay. Various studies have revealed
an overestimation [23] or an underestimation of exposure, leading
to false positive or false negative assay results.
Here we report on an integrated zebrafish screening approach
based on morphology, histopathology, motor activity and kinetics,
using anti-epileptic drugs (AEDs) as a group of model compounds.
Some AEDs can exert their effects via inhibition of the histone
deacetylases (HDAC) [24,25] by binding to retinoic acid (RA) recep-
tor [26]. Using in situ hybridization (ISH), we also investigated
the gene expression profile of two genes (Aldh1a2 and Cyp26a1)
involved in the retinoic acid (RA) signaling pathway and ways in
which these are affected by AEDs. Aldh1a2 is involved in the syn-
thesis of RA, through the oxidation of retinaldehyde, while Cyp26a1
is involved in the degradation of RA, by the conversion of RA into
more polar metabolites [27].
It is known that, when administered during pregnancy, most
AEDs are capable of inducing developmental (neuro)toxicity
[28,29]. The effects of AEDs on the offspring may range from major
congenital malformations (MCM) to delayed postnatal cognitive
development and growth [30]. The developmental toxic potential
of first generation AEDs, such as valproic acid (VPA) and carba-
mazepine (CBZ), has been clearly demonstrated in humans and in
various animal models [31,32]. Only a limited amount of data has
been published on the developmental effects of other first genera-
tion AEDs such as ethosuximide (ETH) and for newer compounds,
such as levetiracetam (LEV). Despite the limited availability of data,
teratogenicity and cognitive impairment have been described for
ETH in rodents [33,34]. Levetiracetam was found to have develop-
mental effects only at very high dose levels [35]. In addition, it was
recently reported that LEV might be a safer alternative to VPA, as it
involves a lower risk of major congenital malformations in women
with epilepsy who are of childbearing age [36].
The aim of this study was to improve the predictability of
the zebrafish model and to extend its applicability domain by
using an integrated screening strategy, including a wide-range
of endpoints (modules): morphology, behavior (motor activity
parameters), histopathology, kinetics and phenotyping (in situ
hybridization). Four well-known AEDs (valproic acid (VPA), car-
bamazepine (CBZ), ethosuximide (ETH) and levetiracetam (LEV))
with different developmental (neuro)toxic potencies were chosen
as model compounds.
2. Material and methods
2.1. Fish husbandry and egg production
Adult wild type AB line Zebrafish (Zebrafish International
Resource Center, Eugene, OR, USA) were kept in a 12 h dark/12 h
light cycle at 28 ◦ C in self-regulating aquaria (Tecniplast, Tecnilab-
BMI, The Netherlands) at pH 7.5 and a water conductivity of
500 ␮S/cm. The animals were fed twice a day with dry flake food
(SDS) and once a day with live food (Artemia), as recommended
by Westerfield [37] and in accordance with the OECD draft guide-
line on the use of zebrafish for chemical testing [38]. Eggs and
larvae used in this study were derived from a zebrafish colony
held in the animal facility of TNO. Housing and management of the
zebrafish colony fully comply with EU/2010/63 regulations [39].
For egg production, fish of between 7 and 11 months old were
used. On the evening prior to the day of breeding, two male fish
and one female fish were placed in a breeding tank containing
a partition (Techniplast) and egg traps to prevent egg predation.
Next morning, spawning was triggered by turning the light on and
Table 1
Concentration range from the DRF and main experiments.
Compound VPA (␮M) CBZ (␮M) ETH (mM) LEV (mM)
DRF 30–1500 30–720 2–20 0.5–150
Main 30–730 30–480 3–18 20–140
Abbreviations: VPA, valproic acid; CBZ, carbamazepine; ETH, ethosuximide; LEV,
levetiracetam.
removing the partition. Eggs were collected using an 800 ␮m mesh
and transferred to a Petri dish containing aquarium water spiked
with 0.05% methylene blue. Staining indicates that eggs are either
non-fertilized or that they have a damaged membrane. Any such
eggs were discarded. Only batches in which more than 80% of the
eggs were fertilized were used in the experiments.
2.2. Compounds
Valproic acid sodium salt (VPA, CAS No. 1069-66-5), carba-
mazepine (CBZ, CAS No. 298-46-4) and ethosuximide (ETH, CAS No.
77-67-8) were purchased by Sigma-Aldrich (St. Louis, MO, USA).
Levetiracetam (LEV, CAS No. 102767-28-2) was purchased by TCI
Europe N.V. (Zwijndrecht, Belgium). First, a stock solution of each
compound was prepared in dimethyl sulfoxide (DMSO). The stock
solution was then further diluted to the desired concentration, in
aquarium water. The final concentration of DMSO in the water sup-
plemented with the compound was 0.2%. Therefore, 0.2% DMSO in
aquarium water was used as negative control.
2.3. Experimental design
2.3.1. Dose range finding (DRF) experiments
In the DRF experiments, a wide range of concentrations (partly
based on studies from Berghmans et al. [40] and Weigt et al. [41])
was selected for each test compound. Embryos were exposed to the
selected concentrations to narrow down the relevant concentration
range for further testing in main experiments. In the DRF studies,
the embryos were assessed for lethal embryotoxic and/or devel-
opmental toxic effects at 24, 48, 72 and 96 hours post fertilization
(hpf) (results not shown). The concentration ranges of the various
compounds used in the DRF and subsequent main experiments are
shown in Table 1.
2.3.2. Main experiments
The main experiments were performed after the concentration
range of interest, which were the concentrations around the EC50
for morphological endpoints, had been established in the DRF stud-
ies. Selected embryos were at first placed in a Petri dish containing
the test concentrations of interest (see Table 1). Subsequently,
each embryo was transferred to a 12-well plate containing 3 ml of
test medium per well, one embryo per well, unless otherwise indi-
cated. Embryos were kept in an incubator (Binder Germany, Type
BD115) at 26.5 ± 1 ◦ C with a 12 h/12 h dark/light cycle and relative
humidity of 100% (to prevent evaporation). For all experiments,
we used intact embryos, the embryos were not dechorionated
before exposure. Test solutions were not refreshed during the
exposure period. The embryos were subjected to morphological
evaluations at 24 and 48 hpf, primarily for lethal endpoints (data
not shown). Evaluations at 72 and 96 hpf were used for detailed
morphological developmental endpoints (see Section 2.4). Follow-
ing a motor activity assessment (see Section 2.5) the larvae were
euthanized, and processed for histopathological investigations
(see Section 2.6). For morphology and motor activity analysis,
three independent experiments were performed per compound,
unless otherwise indicated. In each independent experiment, 12
embryo/larvae per concentration were used. Histopathology was
 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116
assessed using larvae from one experiment (12 embryo/larvae
per concentration, per compound, unless otherwise indicated).
The analysis of total body burden (kinetics) was performed in
a separate experiment (see Section 2.7), with three samples for
analysis per concentration of each compound. For each sample 18,
25 or 32 larvae were used. The in situ hybridization experiments
(WISH) were performed in two independent experiments for each
transcript (see Section 2.9). In each experiment, embryos and
larvae of different age were exposed to various concentrations
of the test compounds. For each of these experiments, 15–30
embryos/larvae per concentration, per age were used.
2.4. Morphology score
We have investigated the shape and structure (morphology)
of the whole zebrafish embryo. Scoring was performed according
to the method described by Fraysse and coworkers [42], involv-
ing a single observer and a stereomicroscope (Nikon, Japan), at
20–50 times magnification, in non-dechorionated embryos/larvae.
The endpoints were assessed using binary responses (yes/no) as
stipulated in the DIN standard [43]. Any embryos or larvae with
evidence of coagulation and/or no heartbeat were considered to
be dead. Those experiments in which the controls exhibited sur-
vival rates of >90% were considered to be valid. Per compound, in
general, three independent experiments were performed using 12
embryos/larvae per concentration which were evaluated on differ-
ent time points for effects on the various morphological endpoints.
For each endpoint, which was assessed on 72 and 96 hpf, an EC50
was calculated using Graph Pad software (see Section 2.9). The EC50
was defined as the concentration at which there was a 50% increase
(or decrease) of the incidence of a given parameter, in comparison
to the control.
2.5. Motor activity (MA) test
A motor activity (MA) test was performed to assess develop-
mental (neuro)toxicity. Just before the start of the MA test, all larvae
were transferred to 96-well, square well, flat-bottom plates (What-
man, cat no. 7701-1651) in their own test solution (±500 ␮l), one
larvae per well. No selection was made before transfer. However,
larvae scoring positive on any of the developmental toxicity end-
points were measured in the MA test but excluded from further
data analysis in order not to confound MA results. All concentra-
tions from the same compound and the respective control were
tested in the same plate.
The MA test was always performed in the early afternoon, after
the last morphological examination (±100 hpf) as described pre-
viously [17]. Briefly, MA was assessed using a ViewPoint behavior
recording system, together with the associated movement tracking
and analysis software (Zebrabox, ViewPoint, France). The system
itself consisted of a light-box platform and an infrared-sensitive
camera within a closed box. The larvae were maintained at a
constant environmental temperature by passing a thermostati-
cally controlled, continuous water flow (27 ± 1 ◦ C) into the closed
Zebrabox. The larvae were allowed an average of 15 min to accli-
matize in the Zebrabox, under visible light conditions, before MA
recording commenced. The larvae were then exposed to the fol-
lowing lighting conditions: 15 min of darkness (i.e. infrared light),
15 min of light (i.e. visible light plus infrared light) and a final 15 min
period of darkness (i.e. infrared light). These sessions are hereafter
referred to as Dark I, Light and Dark II, respectively. A spherical
sensor was placed in the Zebrabox (with a closed lid), to mea-
sure the visible light intensity in the locomotion apparatus. At the
same light intensity as that used during the experiments, the sen-
sor gave a reading of 700 lx, while the infrared light resulted in a
103
measured intensity of 0 lx. The movements of the zebrafish larvae
were tracked using ViewPoint software.
In each MA test, the tracks of individual larvae were analyzed
using the analysis software (Zebrabox) for Distance Moved (DM)
(mm) as measure of locomotor activity. Tracks were analyzed per
unit time (1 min intervals and 15 min test sessions) as described
previously [17]. The parameters analyzed were: (i) Distance Moved
(DM) per unit time, i.e. DM15min , DM1min ; (ii) Slope of Habituation
(HSL); and (iii) Startle Response (StR). With the exception of the
Startle Response, these parameters were measured during each
of the three 15-min test sessions, i.e.: Dark I (0–15 min); Light
(16–30 min) and Dark II (31–45 min). The Startle Response was
defined as a response by the larvae to a sudden change in the light-
ing conditions. For more details see Beker van Woudenberg et al.
[17].
2.6. Histopathology
After the MA test, all larvae were euthanized using 1 g/L tricaine
solution. They were subsequently fixed overnight in Bouin’s fix-
ative (Klinipath; Duiven, The Netherlands), embedded in paraffin
in the same orientation and sectioned completely into serial 4 ␮m
thick sagittal sections. After staining with hematoxylin and eosin
(H&E), ≥100 sections of each larva were viewed using an Axioskop
2 plus microscope (Zeiss). In general, for each concentration, 12
larvae were used for histopathological analysis. However, since
occasionally, a larva was lost during processing, the exact number
of analyzed larvae is listed in Table 3.
2.7. Kinetics
For the analysis of total body burden (kinetics), a separate exper-
iment was performed. First, embryos were placed in a Petri dish
containing the test compound at the concentration to be inves-
tigated. They were then transferred to 12-well or 48-well plates
containing as average of 3 embryos/ml of the test solution. The
embryos were kept in an incubator for 96 h at 26.5 ± 1 ◦ C with a
12 h/12 h dark/light cycle and relative humidity of 100% (see Sec-
tion 2.3.2). Test solutions were not refreshed during the exposure
period.
The concentrations used in these experiments were: VPA 0, 60,
150, 320 and 730 ␮M; ETH 0, 6, 9 and 12 mM; CBZ 0, 60, 120 and
180 ␮M; LEV 0, 20, 40 and 100 mM which were in the same range
as used for the other endpoints. Three samples, each consisting of
18, 25 or 32 larvae, per concentration of each compound were used.
Before sampling the larvae, they were morphologically evaluated
to confirm that the incidences of effects were comparable to the
main experiments. Then samples were collected in three 1.5 ml safe
lock Eppendorf® tubes, euthanized with tricaine (1 g/L) and washed
three times with aquarium water. After the final wash, the water
was removed and the larvae were frozen (−20 ◦ C) until analysis.
The samples were analyzed using an Acquity ultra-performance
liquid chromatography (UPLC) system (Waters, Belgium) coupled
to a Waters Xevo TQ MS tandem mass spectrometer (Waters-
Micromass, Manchester, UK). Both of these devices were operated
using MassLynx® software.
The UPLC system was equipped with a 2.1 × 100 mm, 1.7 ␮m
ACQUITY BEH C18 column (Waters, Milford, PA, USA) kept at
50 ◦ C. A binary gradient mobile phase was used at a flow rate of
0.4 ml min−1 with solvent A (0.1% FA in MilliQ water) and solvent
B (0.1% FA in MeCN).
The mass spectrometer was operated in electrospray positive
mode, while data acquisition was performed using the multiple
reactions monitoring mode (MRM). The source settings were as
follows: capillary voltage 2.0 kV, source temperature 150 ◦ C, des-
olvation temperature 600 ◦ C, cone voltage 32 V, cone gas flow
104 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116
150 L h−1 , desolvation gas flow 1000 L h−1 , collision gas flow (argon)
0.15 ml min−1 , collision cell pressure 3.4 × 10−3 mbar. The results
were quantified using Target Lynx software. For the four com-
pounds analyzed, the limit of detection was 20 ng/ml and the
recovery was 80–120%. Results of the analysis were expressed as
nanomol test compound per larvae.
2.8. Whole mount in situ hybridization (WISH)
Whole mount in situ hybridization (WISH) was performed to
investigate the expression patterns of two selected genes (Aldh1a2
and Cyp26a1). Being involved in the retinoic acid pathways, these
genes also play a part in mechanisms of early development in
embryos exposed to anti-epileptic drugs (AEDs) [44–46]. Valproic
acid and levetiracetam were selected as model compounds for
these WISH experiments since (based on morphology and motor
activity analyses) these compounds appeared to be the most (VPA)
and least (LEV) toxic compounds tested in the present study.
2.8.1. Synthesis of digoxigenin-labeled probes
Sense and anti-sense digoxigenin (DIG)-labeled probes for
Aldh1a2 (GeneBank: NM 131850.1) and Cyp26a1 (GeneBank:
NM 131146.2) were synthesized following the protocol described
by Thisse and Thisse [47]. Primer sequences for Cyp26a1: sense:
5 -CATTAACCCTC-ACTAAAGGGAAAGCAGGAGGTGAAGAGCGCC-3
and antisense: 5 -TAATACGACTCACTATAGGGAGAG-GTTGCAGTAC
TGGCGGTGGT-3 were used. For Aldh1a2 the following primer
sequences were used; sense: 5 -CATTAACCCTCACTAAAGGGAAGG
CTGCACGTTCGGCCTTCT-3 and antisense: 5 -TAATACGAC-TCACTA
TAGGGAGAGCGGTTGGCCCATAGCCTGG-3 . A digoxigenin(DIG)-
labeled probe for the housekeeping gene, fatty acid binding
protein-10 (Fabp-10), was kindly donated by Dr. C. Esguerra (KU
Leuven, Belgium).
2.8.2. Embryo exposure and WISH protocol
Embryos/larvae were exposed to 0, 150, 320 and 730 ␮M VPA
or to 0, 60, 80, 100 and 140 mM LEV for 24, 48, 72 and 96 hpf. As
a positive control, the expression of Fabp-10 (a gene expressed in
the liver) was tested in larvae at 120 hpf in each experiment. Sense
probes of Cyp26a1 and Aldh1a2 were used as negative controls. A
group without a probe was also used as a negative control. Per tran-
script, two independent experiments were performed (see Section
2.3.2). The WISH protocol was adapted from Thisse and Thisse [47].
Briefly, after treatment, the embryos were mechanically dechorion-
ated (where applicable) and subsequently euthanized at the desired
developmental stages using a 1 g/L tricaine solution. They were
then fixed overnight (O/N) in 4% paraformaldehyde. Embryos at
24 hpf were then immediately dehydrated in 100% methanol and
stored at −20 ◦ C until required. Older embryos and larvae were first
bleached by immerging them in a 3% H2 O2 –5% KOH solution until
all pigmentation disappeared (±1 h). Prior to storage at −20 ◦ C,
the embryos/larvae were gradually dehydrated by being passed
through a series of methanol solutions (25%, 50%, 75% and 100%)
in phosphate buffered saline (PBS).
On the day of the WISH experiments, the embryos/larvae
were rehydrated through successive baths of 75%, 50% and 25%
methanol in PBS. After washing, the embryos/larvae were perme-
abilized using 10 ␮g/ml Proteinase K (Roche Diagnostics, Almere,
The Netherlands). WISH was carried out using DIG-labeled probes
for Aldh1a2 and Cyp26a, which were synthesized as described
above (see Section 2.8.1). Hybridization was performed O/N at
65 ◦ C in a hybridization mix containing 50% formamide (Sigma-
Aldrich), 5× SSC and 40 ␮g/ml Salmon sperm DNA (Invitrogen).
Probes were detected using alkaline phosphatase conjugated anti-
bodies, and colorization was performed O/N with BM purple (Roche
Diagnostics). The embryos/larvae were mounted in 70% glycerol
with PBS, and images were recorded using a Zeiss Stereo Discovery
V8 microscope equipped with a Canon powershot G12 camera. The
embryos/larvae were stored in 4% paraformaldehyde (PFA) at 4 ◦ C.
2.9. Statistical analysis
For morphological examinations, a dose-response curve was
fitted for each relevant endpoint and an EC50 concentration was cal-
culated using Graph Pad (GraphPad Prism 5, version 5.03, December
2009). The following four-parameter logistic equation was used for
curve fitting: y = A + ((B − A)/(1 + ((C/x)D))); where A = minimum y
(constrained to >0), B = maximum y (constrained to <100), C = the
EC50 value, and D = slope factor. All of the derived parameters stud-
ied in the MA test: Distance Moved, Startle Response and Slope of
Habituation were statistically analyzed using a mixed model using
all data of all experiments. Compound/concentration was used as
fixed factor. To correct for plate and row differences and, hence for
instance, to correct for offset differences between controls, loca-
tion (row within a plate) was used as random factor in the mixed
model. All data were log-transformed before analysis. In all of the
tests performed, the hypothesis was rejected at the 5% probability
level (alpha = 0.05). A Dunnett’s post hoc test was performed to cor-
rect for multiple testing. All MA analysis were performed using SAS
9.3 (SAS Institute Inc., Cary, NC, USA). For more details on the MA
analysis, please see Beker van Woudenberg et al. [17]). The statis-
tical significance of the histological findings was evaluated by the
Fisher’s exact probability test.
3. Results
3.1. Morphology
The most sensitive effects (based on EC50 ) on morphology of
zebrafish embryos exposed to antiepileptic drugs (AEDs) at 72 hpf
and 96 hpf are presented in Table 2.
At 72 hpf, in embryos exposed to valproic acid (VPA), a dose-
dependent increase was found in the incidence of non-hatched
embryos. This endpoint appeared to be the most sensitive param-
eter, with an EC50 of 221 ␮M. At this time point, pericardial edema
was also observed in 46% and 50% of the embryos exposed to
VPA at 320 ␮M and 730 ␮M, respectively (data not shown). More-
over, at 96 hpf, the percentage of larvae showing pericardial edema
increased to 92% and 100% at 320 ␮M and 730 ␮M VPA, respec-
tively, and it appeared to be the most sensitive parameter for larvae
of this age, with an EC50 of 165 ␮M. In addition, the group of lar-
vae exposed to 320 ␮M and 730 ␮M VPA had a high incidence of
non-hatching (50% and 67%, respectively).
As with VPA, non-hatching was the most sensitive parameter
in 72 hpf embryos exposed to carbamazepine (CBZ), with an EC50
of 193 ␮M. At 96 hpf, pericardial edema was the most sensitive
parameter, with an EC50 of 220 ␮M. The group of larvae exposed
to 240 ␮M and 480 ␮M CBZ also exhibited a high incidence of
non-hatching (50% and 100%, respectively). At relatively high etho-
suximide (ETH) concentrations, dose-dependent pericardial edema
was observed. This appeared to be the most sensitive parameter at
72 and 96 hpf, with EC50 values of 9.9 mM and 8.9 mM, respectively.
Yolk sac and chorda malformations (defined as larvae showing
a bent spine) were also observed at 96 hpf in larvae exposed to
15 mM ETH (42% and 40%, respectively) and 18 mM ETH (75% and
40%, respectively). Of the 96 hpf larvae exposed to 18 mM ETH, 69%
showed an obviously – i.e. clearly visible – decreased heart rate as
compared to control larvae.
Embryos exposed to levetiracetam (LEV) showed only two
adverse developmental effects, and then only at very high concen-
trations. The first of these was tail malformation at 96 hpf, with
 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116 105

Table 2
Most sensitive effects (based on EC50 ) on morphology of zebrafish embryos/larvae exposed to anti-epileptic drugs (AEDs) at 72 and 96 h post fertilization (hpf).

Abbreviations: VPA, valproic acid; CBZ, carbamazepine; ETH, ethosuximide; LEV, levetiracetam.

an EC50 of 109 mM. This was observed in 58% and 71% of larvae
exposed to 120 mM or 140 mM LEV, respectively. The second was
chorda malformations which had an observed incidence of 12% and
33% (120 mM and 140 mM LEV groups, respectively), not reaching
EC50 levels.
3.2. Motor activity (MA) test
The results of the MA test performed (100 hpf ± 2) on lar-
vae kept in vehicle or exposed to different concentrations of the
AEDs are shown in the left-hand panels of Fig. 1. The right-hand
panels of this figure show the results of the statistical evalua-
tion.
Larvae exposed to valproic acid (VPA) showed a reduction in
total Distance Moved per 15 min session (DM15min ) in Dark I and
Dark II sessions at the highest concentration analyzed (150 ␮M
VPA) compared to the control group. In fact, this 150 ␮M VPA con-
centration group consistently exhibited low levels of activity, and –
consequently – no habituation was observed (Fig. 1A). Moreover, no
Startle Response was observed from the Light to Darkness session
(D-StR) in the 150 ␮M VPA, which was significant lower compared
to the control group. In the other hand, the response observed from
Dark I to Lightness session (L-StR), was significantly higher than
the Startle Response observed in the control group. Also swimming
activity (DM15min ) – although fluctuating – remained increased dur-
ing the Light session, thereby significantly differing from the control
group showing the normal – low – activity in Light. The changes in
the 30 and 60 ␮M VPA groups – reduction of a Startle Response at
the start of the Dark I session and time-shift in the Dark II session
– did not reach the level of statistical significance. Since all larvae
exposed to higher concentrations (320 or 730 ␮M) VPA were either
dead, non-hatched or malformed, they were excluded from the MA
analysis.
Exposure to carbamazepine (CBZ) caused a concentration-
related reduction (significant at concentrations ≥ ␮M CBZ) in total
Distance Moved per 15 min test session (DM15min ) in Dark I and
Dark II, as well as in the Darkness Startle Response (D-StR) (Fig. 1B).
Swimming activity in the 240 ␮M CBZ concentration group was
marginal to even absent and so, the curve slope representing a
measure of habituation was absent as well. For all groups of lar-
vae exposed to <180 ␮M CBZ, the slope of the swimming activity
curve was not affected by CBZ. Since all larvae exposed to 480 ␮M
106 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116

Fig. 1. Effect of anti-epileptic (AEDs) compounds on the swimming activity of zebrafish larvae in the motor activity test. Left panel shows the graphics for the Distance
Moved DM-per minute intervals for larvae exposed to (A) valproic acid (VPA); (B) carbamazepine (CBZ); (C) ethosuximide (ETH) and (D) levetiracetam (LEV). Right panel
presents, respectively, the statistical evaluation of the data for motor activity. Conc = concentration; ns = non-significant; *p < 0.05; **p < 0.01; ***p ≤ 0.001; ****p ≤ 0.0001;
#
total number of independent experiments; ## total number of independent experiments for LEV varied from 2 to 4, depending of the concentration group; ‡ number of larvae
analyzed after exclusion of dead, non-hatched and malformed ones out of the total number of larvae.
 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116
CBZ were either not hatched or dead, this group was excluded from
the MA analysis.
Larvae exposed to ethosuximide (ETH) showed a biphasic
response in total Distance Moved per 15 min test session (DM15min )
(Fig. 1C). Low concentrations (3 mM and 6 mM ETH groups) showed
an increase in (DM15min ) – hyperactivation – compared to the
control group (significant for the 6 mM concentration group in
Dark I). Conversely, in the groups exposed to higher concentra-
tions (≥9 mM), a concentration-related reduction in total Distance
Moved (DM15min ) was observed in both Dark I and Dark II, signif-
icant for the 12 mM ETH group. However, the swimming activity
was slightly increased in the Light session in all concentrations
groups, significant for the 9 mM ETH group. Larvae exposed to
12 mM ETH also startled in response to the sudden Light stimulus
(L-StR), whereas a response to the Darkness stimulus (D-StR) was
marginal or nearly absent, thereby differing from the control group.
Moreover, in groups of larvae exposed to ≥9 mM, we observed a
significant reduction of the Darkness Startle Response (D-StR) and
of the Habituation slope of the Dark II period (DII-HSL). Similarly
to VPA and CBZ, larvae exposed to high concentrations ETH (15
and 18 mM) were all excluded from the MA analysis, due to a high
incidence of malformations.
Although the total Distance Moved (DM15min ) in all groups of lar-
vae exposed to levetiracetam (LEV) (Fig. 1D) was reduced in Dark I
and Dark II sessions when compared to the control group, only in
the 60 mM LEV concentration group this effect reached statistical
significance in the Dark I session. Activity in the ≥40 mM LEV con-
centration groups appeared consistently low throughout the three
sequential 15 min sessions and showed – unlike the control and
20 mM LEV groups – a relative increase in activity during Light.
This increased activity (DM15min ) in Light differed significantly from
the control group in all groups ≥60 mM LEV. The Lightness Startle
Response (L-StR) was significantly increased at the 60 and 100 mM
LEV concentration groups. It was noticed that in all these groups the
activity over time fluctuated more than the activity of the control
and 20 mM LEV concentration groups, and fluctuated more than
usual (compare Fig. 1D with Fig. 1A–C). As a consequence of the
reduced swimming activity, the Habituation Slope in Dark I (DI-HSL)
was significantly reduced – or virtually absent – when larvae were
exposed to ≥40 mM LEV. The Darkness Startle Response (D-StR) was
significantly reduced even at the 40 mM LEV.
3.3. Histopathology
The histopathological findings of larvae exposed to the differ-
ent AEDs used in this study are listed in Table 3 and presented in
Figs. 2–5.
In agreement with the morphological analysis, the histopath-
ology of the control larvae, revealed no abnormalities (Fig. 2A,
Table 3). Larvae exposed to valproic acid (VPA) revealed pericar-
dial edema (Fig. 2B, arrow), non-hatching (Fig. 2C) and coagulation
(death) (Fig. 2D). In addition, a dose-dependent disruption of nor-
mal brain structure was observed (Fig. 2F–H). Already at 60 ␮M
VPA, minimal disruption of normal brain structure was observed,
characterized by small regions of reduced cellularity (Fig. 2F,
arrows). At 150 ␮M VPA, all larvae showed mild disruption of
normal brain structure, characterized by large areas of reduced cel-
lularity, in particular in the developing preoptic region and tectum
opticum (Fig. 2G, arrows). In addition, all larvae showed moderate
disruption of the retinal cell layers with signs of apoptosis (Fig. 2K).
At 320 ␮M VPA, 75% of the larvae did not hatch. All larvae showed
moderate disruption of normal brain structure (Fig. 2H), character-
ized by large areas of reduced cellularity (arrow), disorganization
and signs of apoptosis (Fig. 2H, insert, arrows head). In addition, the
retinal layers were completely disrupted, showing severe apoptosis
and necrosis (Fig. 2L). At 730 ␮M VPA 55% of the larvae were dead
107
and only necrotic tissue was visible (Fig. 2D). The remaining larvae
did not hatch and showed severe disruption of normal brain struc-
ture, characterized by disorganization and apoptosis. Pericardial
edema was occasionally detected in the 150 ␮M VPA and 320 ␮M
VPA group. The heart could not properly be assessed in the 730 ␮M
VPA group.
Histopathological analysis of the larvae exposed to carba-
mazepine (CBZ) showed a dose-dependent increase of a general
developmental arrest and a disrupted brain structure (Fig. 3).
At 240 ␮M, 27% of the larvae failed to hatch. In addition 18%
showed mild disruption of brain structure, characterized by areas
of reduced cellularity (Fig. 3B, arrow). At 480 ␮M, 92% of the lar-
vae failed to hatch (Fig. 3C) and one was necrotic (Fig. 3D). In
addition, 25% showed mild disruption of normal brain structure,
characterized by areas of reduced cellularity. Pericardial edema was
occasionally detected in the 240 ␮M CBZ and 480 ␮M CBZ group.
A dose-dependent increase in the disruption of normal brain
structure was observed in larvae exposed to ethosuximide (ETH)
(Fig. 4A–D). In addition, minimal to mild hepatocellular vacuola-
tion was observed (Fig. 4F–H). At 6 mM ETH (Table 3), one larva
showed hepatocellular vacuolation and mild disruption of brain
structure, characterized by areas of reduced cellularity. At 9 mM
ETH, 50% of the larvae showed mild disruption of brain structure
(Fig. 4B). In addition, 58% of the larvae showed minimal to mild
hepatocellular vacuolation. At 12 mM ETH, all larvae showed mild
disruption of brain structure, characterized by areas of reduced cel-
lularity (Fig. 4C). In addition, 90% of the larvae showed minimal to
mild hepatocellular vacuolation. At 15 mM ETH, 50% of the larvae
showed pericardial edema. In addition, all larvae showed minimal
to mild hepatocellular vacuolation and mild to moderate disrup-
tion of brain structure, characterized by areas of reduced cellularity
and/or apoptosis (Fig. 4D). Similar effects were also observed in lar-
vae exposed to 18 mM ETH (Table 3), however, the percentage of
larvae showing pericardial edema in this group was significantly
higher (92%) compared to the 15 mM group.
Histopathological analysis showed that, of all the compounds
studied, levetiracetam (LEV) had the lowest potency in terms of
embryotoxicity. Only at LEV concentration of ≥ 80 mM (Fig. 5),
minimal to mild hepatocellular vacuolation was observed. At
respectively 80 mM, 100 mM, 120 mM and 140 mM LEV, hepato-
cellular vacuolation was observed in 8%, 36%, 33% and 75% of the
larvae.
3.4. Kinetics
The total body burden of larval zebrafish (96 hpf), as measured
by UPLC, showed a dose-related increase for all tested compounds.
However, uptake efficiency (based on the slope of the curve) was
different for each of the four compounds studied (Fig. 6). The
analysis revealed that carbamazepine (CBZ) had the most efficient
uptake (y = 0.621x) of all the AEDs studied, followed by valproic
acid (VPA) (y = 0.530x), ethosuximide (ETH) (y = 0.194x) and leve-
tiracetam (LEV) (y = 0.0118x). The internal concentrations for the
established EC50 values for developmental toxicity at 96 hpf (see
Section 3.1 and Table 4) were 0.087 (VPA), 0.137 (CBZ), 1.286 (LEV)
and 1.726 (ETH) (nanomoles/larvae), respectively.
3.5. Whole mount in situ hybridization
3.5.1. Aldh1a2 expression
Fig. 7 shows Aldh1a2 expression in the control group (A), the
320 ␮M valproic acid group (VPA) (B) and the 100 mM levetirac-
etam group (LEV) (C) at all of the time points studied (24, 48, 72
and 96 hpf). In the control groups of VPA and LEV, Aldh1a2 was
expressed in the retina, branchial arches and somites at 24 hpf.
At 48 hpf it was expressed in the retina, branchial arches, the
108 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116

Table 3
Histopathological findings of zebrafish larvae exposed to anti-epileptic drugs (AEDs).

Valproic acid (␮M) 0 60 150 320 730

Number of larvae analyzed 12 12 12 12 11

Brain Minimal disruption of brain structure 0 1 0 0 0
Mild disruption of brain structure 0 0 12** 0 0
Moderate disruption of brain structure 0 0 0 12** 0
Severe disruption of brain structure 0 0 0 0 5*

Eye Moderate apotosis and disruption of retinal layers 0 0 12** 0 0

Severe apotosis and disruption of retinal layers 0 0 0 12** 1

Heart Pericardial edema 0 0 3 3 ND

**
General Not hatched 0 0 0 9 5*
Dead 0 0 0 0 6*

Carbamazepine (␮M) 0 30 60 120 180 240 480

Number of larvae analyzed 12 12 12 12 12 11 12

Brain Mild disruption of brain structure 0 0 0 0 0 2 3

Heart Pericardial edema 0 0 0 0 0 3 1

General Not hatched 0 0 0 0 0 3 11**

Dead 0 0 0 0 0 0 1

Ethosuximide (mM) 0 3 6 9 12 15 18

Number of larvae analyzed 12 12 12 12 10 10 12

Liver Minimal to mild hepatocellular vacuolation 0 0 1 7* 9** 10** 9**

Brain Mild disruption of brain structure 0 0 1 6* 10** 2 3

Moderate disruption of brain structure 0 0 0 0 0 8** 8**

Heart Pericardial edema 0 0 0 0 0 5* 11**

General Not hatched 0 0 0 0 0 0 1

Dead 0 0 0 0 0 0 1

Levetiracetam (mM) 0 20 40 60 80 100 120 140

Number of larvae analyzed 12 12 12 12 12 11 12 12

Liver Mild to moderate hepatocellular vacuolation 0 0 0 0 1 4# 4# 9**

General Not hatched 0 0 0 0 0 1 0 0

ND = not determined.
*
p < 0.01.
**
p < 0.0001.
#
p < 0.05.

tectum opticum, spinal cord motor neurons and cerebellum. At
72 hpf Aldh1a2 was weakly expressed in the tectum opticum and
strongly expressed in spinal cord motor neurons and the branchial
arches. At 96 hpf, expression was observed in spinal cord motor
neurons, the tectum opticum and the branchial arches.
Aldh1a2 expression in larvae exposed to 320 ␮M VPA differed
from that seen in control larvae of the same age (Fig. 7B). At
24 hpf, exposure to 320 ␮M VPA induced increased Aldh1a2 expres-
sion in the somites and decreased expression in the branchial
arches. At 48 and 72 hpf, compared to the control, a clear decrease
in Aldh1a2 expression was observed, especially in the tectum
opticum, branchial arches, spinal cord motor neurons and cere-
bellum. At 96 hpf, only weak levels of Aldh1a2 expression were
observed in the spinal cord motor neurons and branchial arches.
In general, it can be stated that down-regulation of Aldh1a2 was
observed in larvae exposed to VPA, especially in those at 48 hpf or
older.
When exposing larvae to 100 mM LEV (Fig. 7C) or to any of
the other concentrations studied, no clear difference in Aldh1a2
expression was observed at any of the time points investigated (not
shown).
3.5.2. Cyp26a1 expression
Fig. 8 shows Cyp26a1 expression in the control group (A), the
320 ␮M valproic acid group (VPA) (B) and the 100 mM levetirac-
etam group (LEV) (C) for each of the time points studied (24, 48,
72 and 96 hpf). In the control groups of VPA and LEV, Cyp26a1 was
generally weakly expressed. The only clear evidence of Cyp26a1
expression was at 24 hpf, in the tail bud and posterior notochord. At
96 hpf, no Cyp26a1 expression was observed in the control groups.
Exposure to 320 ␮M VPA for 24 hpf (Fig. 8B) increased the
expression of Cyp26a1 as compared to control embryos, mainly at
the tail bud and (weakly) at the branchial arches. An up-regulated
expression of Cyp26a1 was also observed in the tail bud of larvae
exposed to 150 ␮M and 730 ␮M VPA (data not shown). At 48 and
72 hpf, as with the control group, the expression of Cyp26a1 in lar-
vae exposed to VPA was less pronounced. At 96 hpf, no expression
of Cyp26a1 was observed in larvae exposed to VPA, as in the control
larvae.
Resembling the expression pattern of Aldh1a2 (Fig. 7C), expo-
sure to 100 mM of LEV (Fig. 8C) and to the other concentrations of
LEV studied (not shown) did not induce any visible alterations in
the expression of Cyp26a1.
4. Discussion
The present study describes an integrated modular approach
to test developmental (neuro)toxicity using the embryonic/larval
zebrafish model. In general, this screening assay is based on mor-
phological screening for developmental toxicity [13,14] and on a
motor activity assay to evaluate potential developmental neurotox-
icity [18–20], in which the different endpoints are usually studied
 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116 109

Fig. 2. Histopathology of larvae exposed to valproic acid (VPA). (A) Control. (B) Histopathological analysis confirmed the morphological observations of pericardial edema
(arrow); (C) non-hatching (chorion; arrow) and (D) death. In addition, disruption of normal brain structure was observed, characterized by areas of reduced cell density (F,
G, H, black arrows), and signs of apoptosis (H-insert; black arrow heads). Histopathological analysis of the eye showed VPA-induced disorganization of the retinal layers and
signs of apoptosis (K, L). (0) Pigment epithelium; (1) photoreceptor cell layer; (2) inner nuclear layer; (3) inner plexiform layer; (4) ganglion cell layer; (5) lens. Magnification
A–D: 100×. Magnification E–H: 200×. Magnification I–L: 400×.

separately. We present an extension to – and a further integra-
tion of – this approach, applying morphological and motor activity
screening in combination with histopathology and kinetics.
The results of this demo-case study with a limited set of
compounds, showed the potency of integration of all measured
endpoints (each with its own distinct specificity and sensitivity)
to enhance the predictability of the zebrafish assay. Further-
more, inclusion of in situ hybridization may well contribute
to further elucidate underlying toxicity pathways by linking
structural/functional endpoints on the one hand and molecular
gene-expression data on the other. Altogether, this may increase
the reliability of the model for potency ranking of developmental
toxicity, while providing a better understanding of the processes
and mode of actions involved in the toxic effects observed. In this
study only a limited number of compounds was tested, therefore, in
order to be able to draw firm conclusions about the specificity and
sensitivity of each of the assays applied in this study, an extended
panel of positive and negative compounds covering different mech-
anisms of action should be studied in future.
By analyzing the larvae treated with valproic acid (VPA)
histopathologically, we observed that 8% of them had reduced brain
cellularity (in the optic regions in particular) already at 60 ␮M
whereas no effects were observed on morphology and motor activ-
ity at this concentration. This indicates that, for VPA, histopathology
appeared to be one of the most sensitive endpoint analyzed. Addi-
tionally, at 150 ␮M VPA, an analysis of the motor activity (MA)

Fig. 3. Histopathology of larvae exposed to carbamazepine (CBZ). (A) Control. (B–D) Examples of CBZ-induced pathology showing disruption of normal brain structure,
characterized by reduced cell density (B; arrow); non-hatching (C) and death (D). Magnification A–D: 100×.
110 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116

Fig. 4. Histopathology of larvae exposed to ethosuximide (ETH). (A) Control. (B and C) Histopathological analysis showed disruption of normal brain structure, characterized
by areas of reduced cell density (arrows), and (D) signs of apoptosis (black arrow heads). (E) Control liver. (F–H) Examples of different degrees of hepatocellular vacuolation
like minimal hepatocellular vacuolation (F), mild hepatocellular vacuolation (G and H). Magnification A1–D1: 100×. Magnification A2–D2: 200×. Magnification E–H: 200×.

test showed a significant reduction in total Distance Moved dur-
ing both of the Darkness sessions and also on the Darkness Startle
Response (Table 4). These effects are indicative of neurodevelop-
mental toxicity [19]. Indeed, various adverse effects of VPA on
neurodevelopment in rodents [48] and in humans (i.e. cognitive,
behavior, verbal IQ) have been extensively described in the lit-
erature [48–51]. By including the histopathological examination,
indeed we were able to demonstrate that 100% of the examined
larvae that had been exposed to ≥150 ␮M VPA exhibited reduced
brain cellularity (especially in the optic regions). However, we
also observed severely damaged retinas. Clearly, in a design that
involves light/darkness sessions (where the integrity of the retina
is clearly of pivotal importance) the observed brain (optic region)
and retinal damage could well account for the failure of the larvae to
react to changes in light intensity, i.e. no effect in Startle Response.
However, since there was a response to light (L-StR) – though not to
darkness (D-StR), the observed changes in brain structure have also
– at least to some extent – contributed to changes in motor activ-
ity behavior, suggesting developmental neurotoxic potency of VPA.
The results of histopathology and motor activity, taken together,
once again, underline the importance of histopathology surveys to
understanding of the mode of toxic action and the importance of
integrating this technique into the ZET test.
In the case of embryos/larvae exposed to carbamazepine
(CBZ), unlike VPA, histopathology does not appear to be the
most sensitive endpoint. This compound only induced observ-
able histopathological effects (reduced cellularity in the brain) at
high concentrations (≥240 ␮M). Instead, embryos/larvae exposed
to 180 ␮M CBZ already showed a delay in the onset of hatching
at 72 hpf, and also a reduced Darkness Startle Response. This is in
agreement with the literature, in which general growth retardation
was observed in rodents embryos exposed to CBZ [52]. Our results

Fig. 5. Histopathology of larvae exposed to levetiracetam (LEV). (A) Control liver. (B and C) Examples of the different degrees of hepatocellular vacuolation found in all larvae
exposed to ≥80 mM LEV, like mild hepatocellular vacuolation (B) and moderate hepatocellular vacuolation (C). Magnification A–C: 200×.
 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116 111

Fig. 6. Total body burden of zebrafish larvae exposed to anti-epileptic (AEDs) measured by ultra-performance liquid chromatography (UPLC). Abbreviations: VPA, valproic
acid; CBZ, carbamazepine; ETH, ethosuximide; LEV, levetiracetam.

are also in agreement with human data, as children exposed to
CBZ in utero presented increased frequency of neurodevelopmental
disorders, compared to non-exposed children [50].
The morphological examination of embryos/larvae exposed to
ethosuximide (ETH) showed developmental toxicity (pericardial
edema) at 72 hpf and 96 hpf, with respective EC50 levels of 9.9 mM
and 8.9 mM. No developmental delay was observed at any of the
concentrations studied (the incidence of larvae that were hatched
at 72 hpf and 96 hpf was not affected). However, histopatho-
logy showed neurodegeneration (reduced brain cellularity) and
hepatocellular vacuolation in 8% of the larvae at a concentration
of just 6 mM ETH. At functional levels (MA test), larvae exposed to
ETH showed a clear biphasic dose-response effect. At low concen-
trations (≤ 6 mM) an increase in the MA parameters (i.e. Distance
Moved, Darkness Startle Response, etc.) was observed, while at
higher concentrations (≥9 mM) the value of these parameters
decreased.
Observations of larvae exposed to levetiracetam (LEV) also
showed that multiple-endpoints are complementary to each
other and that they increase the model’s potential for pinpointing

Table 4
Summarizing results.

Compound Morphologya Motor activity Histopathology Kineticsb (total body burden at EC50 at 96 hpf)

72 hpf 96 hpf

VPA Non-hatching Pericardial edema 150 ␮M 60 ␮M (8% larvae showing reduced (0.530 × 0.165) = 0.087 nmol/larvae
EC10 : 101 ␮M EC10 : 76 ␮M cellularity in brain)
EC20 : 135 ␮M EC20 : 101 ␮M 150 ␮M (100% larvae showing reduced
EC50 : 221 ␮M EC50 : 165 ␮M cellularity in brain and 100% showing
disruption of retinal cell layers)
CBZ Non-hatching Pericardial edema ≥180 ␮M 240 ␮M (18% larvae showing reduced (0.621 × 0.22c ) = 0.137 nmol/larvae
EC10 : 148 ␮M EC10 : 167 ␮M cellularity in brain)
EC20 : 163 ␮M EC20 : 185 ␮M 480 ␮M (25% larvae showing reduced
EC50 : 193 ␮M EC50 : 220 ␮M cellularity in brain)
ETH Pericardial Pericardial ≥6 mM 6 mM (8% larvae showing reduced cellularity (0.194 × 8.9) = 1.726 nmol/larvae
edema edema in brain and hepatocellular vacuolation)
EC10 : 6.6 mM EC10 : 6.6 mM 9 mM (50% larvae showing reduced
EC20 : 7.7 mM EC20 : 7.3 mM cellularity in brain and 58% showing
EC50 : 9.9 mM EC50 : 8.9 mM hepatocellular vacuolation)
12 mM (100% larvae showing reduced
cellularity in brain and 90% showing
hepatocellular vacuolation)
LEV Malformation Malformation ≥40 mM 80 mM (8% larvae showing hepatocellular (0.0118 × 109c ) = 1.286 nmol/larvae
tail tail vacuolation)
EC10 : 69 mM EC10 : 53 mM 120 mM (33% larvae showing hepatocellular
EC20 : 84 mM EC20 : 69 mM vacuolation)
EC50 : 118 mM EC50 : 109 mM 140 mM (75% larvae showing hepatocellular
vacuolation)

Abbreviations: VPA, valproic acid; CBZ, carbamazepine; ETH, ethosuximide; LEV, levetiracetam.
a
Here we also present the morphology EC10 and EC20 for a better comparison with the histopathological findings.
b
The internal concentrations (total body burden) were calculated, for each compound, by multiplying the uptake efficiency (slope of curve, Fig. 6) with the respective
morphology EC50 value at 96 hpf.
c
Observed EC50 concentration for CBZ (220 ␮M) and LEV (109 mM) were outside the range of concentrations used for total body burden measurements as shown in Fig. 6.
Consequently, the values shown in this table were based on extrapolation of the results shown in Fig. 6.
112 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116

Fig. 7. Representative illustrations of Aldh1a2 expression in zebrafish embryos/larvae at 24, 48, 72 and 96 hpf. (A) Control; (B) embryos/larvae exposed to 320 ␮M VPA;
(C) embryos exposed to 100 mM LEV. (1) Retina; (2) branchial arches; (3) somites; (4) tectum opticum; (5) spinal cord motorneurons; (6) cerebellum; *a-specific staining;
Abbreviations: VPA, valproic acid; LEV, levetiracetam; n.d., not done.

the most sensitive parameter for toxicity. The most sensitive
parameter, by far, for this compound was neither morphology nor
histopathology, but motor activity. In embryos/larvae that were
exposed to LEV, effects on morphology (malformation tail) and
histopathology (hepatocellular vacuolation) were only observed
at high concentrations (≥80 mM). Moreover, no structural brain
damage (histopathology) was observed at any of the concentra-
tions studied. However, significant functional effects on motor
activity (i.e. Darkness Startle Response) were observed already
at 40 mM. The remarkable fluctuations observed in swimming
activity at concentrations of ≥60 mM LEV might represent early
alterations in tail morphology that go unrecognized during
 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116 113

Fig. 8. Representative illustrations of Cyp26a1 expression in zebrafish embryos/larvae at 24, 48, 72 and 96 hpf. (A) Control; (B) embryos/larvae exposed to 320 ␮M VPA; (C)
embryos exposed to 100 mM LEV. (1) Notochord; (2) tail bud; (3) branchial arches. Abbreviations: VPA, valproic acid; LEV, levetiracetam; n.d., not done.
114 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116
morphology testing. It should be borne in mind that larvae with
recognized tail aberrations were excluded from motor activity
analysis. This is also an example how integration of different
parameters may help to elucidate the mode of actions underlying
functional toxic effects. In this case, most probably, the effects on
motor activity were related to effects on tail morphology whereas,
for example, in case of VPA the effects on motor activity were
related to histopathological effects on brain and eye.
Taking the results of all endpoints together (Table 4), we note
that valproic acid (VPA) showed significant effects at concentra-
tions ≥150 ␮M (yet histopathology reveals that 8% of the larvae
studied showed a reduction in brain cellularity even at concentra-
tions as low as 60 ␮M VPA). With regard to carbamazepine (CBZ),
clear effects were observed at concentrations ≥180 ␮M, while
zebrafish embryos/larvae exposed to ethosuximide (ETH) showed
significant effects only at millimolar concentrations (≥6 mM). Lev-
etiracetam (LEV) showed the lowest developmental toxic potency
of the AEDs studied. In the MA test, significant effects were detected
at concentrations ≥40 mM LEV, whereas significant effects on
morphology and histopathology were only observed at higher con-
centrations (≥80 mM). No lethal effects whatsoever were observed
at any of the concentrations studied (up to 140 mM).
Therefore, based on the external dose exposure, our results
show that the AEDs can be ranked in terms of their developmental
toxicity/neurotoxicity potency as follows. The most potent com-
pound is valproic acid (VPA), followed by carbamazepine (CBZ),
then ethosuximide (ETH). Finally, levetiracetam (LEV) is the least
potent compound. The above AED developmental (neuro)toxicity
ranking shows good agreement with results from in vivo studies
in animals (see review by Finnel and Dansky [31]) and humans. In
a recent human review study, Tomson and Battino [28] described
that exposure to monotherapy with valproate produced a higher
rate of major congenital malformations (MCM) than was the case
with carbamazepine. This is also in agreement with the results
of a prospective study by Morrow et al. [53], who found that
women receiving VPA monotherapy during pregnancy were at
greater risk of MCM than those given CBZ. Moreover, fetal expo-
sure to antiepileptic drugs at doses lower than those that result
in structural malformations can produce behavioral and cogni-
tive deficits. Indeed, Meador and collaborators [54] reported that
children exposed (in utero) to valproate had a reduced cognitive
outcome (low IQ score) compared those exposed to carbamazepine.
Regarding ETH, our results are in agreement with Sullivan and col-
laborators [55], who performed a study on the ranking potency of
AEDs in mice. These workers found that ETH was less potent than
CBZ in inducing developmental effects. In the present study, LEV
was the least potent compound for developmental (neuro)toxicity.
This is in agreement with a recent study showing that LEV had lower
MCM incidence than valproate when used as monotherapy dur-
ing pregnancy [36]. Moreover, children exposed (in utero) to LEV
achieved significantly higher scores on a mental development scale
than those exposed to valproate [56].
Although uptake is generally dictated by a compound’s phys-
icochemical properties [57], the potency of a compound in the
zebrafish screening assay cannot be predicted from its aqueous con-
centration or log P value alone. For this reason, additional kinetic
analysis is required. In this study, the total body burden (TBB) of
larvae was analyzed (Fig. 6). It was shown that whereas supplemen-
tation of the present approach with a TBB analysis (kinetics) left
the absolute ranking (based on EC50 of morphological endpoints)
unchanged, it altered the relative potency significantly (Table 4).
Accordingly, the addition of kinetics analysis provides the most
relevant and important information on compound uptake, and on
the impact of this uptake in terms of the actual potency of the
compound. For instance, we showed that, based on EC50 of mor-
phological endpoints (at 96 hpf), VPA was found to be 660 times
more toxic than LEV (EC50 VPA: 165 ␮M; LEV: 109 mM). However,
when taking the kinetics (TBB at EC50 at 96 hpf) into consideration,
the difference in potency falls from a factor of 660 to a factor of
just 15 (TBB VPA: 0.087 nM/larvae; LEV: 1.286 nM/larvae) (Table 4
and Fig. 6). Most likely this is due to the fact that the uptake of
LEV (y = 0.0118x) is 45 times lower than that of VPA (y = 0.530x).
The observation that, in the present study, the results of kinetic
analysis did not lead to a change in the absolute ranking of AEDs,
was probably related to the wide range of EC50 values involved (see
Table 4).
In situ hybridization (ISH) was included in the present study in
an attempt to phenotypically relate effects on gene expression pro-
files to histopathological findings. For this purpose, we first studied
the spatial–temporal expression of Aldh1a2 and Cyp26a1, two
important genes involved in the retinoic acid signaling pathway
and therefore involved in the embryo/fetal development. Further
we studied the possible involvement of those genes in the develop-
mental and neurodevelopmental effects of the most potent (VPA)
and least potent (LEV) AEDs (as ranked in this study). Therefore,
we initially performed ISH on Formalin-Fixed Paraffin-Embedded
(FFPE) serial sections. Unfortunately, we did not obtain consistent
results, as the gene expression patterns detected by ISH in these
sections appeared to be very localized. For instance, the expression
of Aldh1a2 was detected in only one or two slides out of an entire
series. The adjacent sections showed only mild expression, or none
at all. Accordingly, far too many sections would have to be ana-
lyzed to obtain absolute certainty about the expression of genes in
histopathological sections and about how this is affected by test
compounds. The follow-up experiments were, therefore, carried
out using whole mount in situ hybridization (WISH) (Fig. 7).
One way in which VPA exerts its teratogenic effect is by inhib-
iting histone deacetylases, by binding to the RA receptor [26]. We
therefore investigated the effects of VPA and LEV (respectively, the
most and least potent AEDs studied) on the expression of the two
major genes involved in RA signaling. In control larvae, at 24 hpf, a
strong expression of Aldh1a2 was observed in the retina, branchial
arches and somites. Concurrently with embryo development (in
time), no further Aldh1a2 expression in the somites was observed
from 48 hpf onwards, while it was still occurring in neuronal tis-
sue up to 96 hpf (Fig. 7). In contrast, control larvae only exhibited a
weak expression of Cyp26a1 at 24 hpf, and in the tail bud and pos-
terior notochord. In addition, unlike the pattern seen in Aldh1a,
no expression of Cyp26a1 was observed in any neuronal tissue
from 48 hpf onwards. This is in agreement with Dobbs-McAuliffe
et al. [58], who reported that, in early development, Cyp26a1
expression occurred either adjacent to, or opposite, tissues express-
ing Aldh1a2. Interestingly, when zebrafish embryos/larvae were
exposed to VPA, they showed an up-regulation of Aldh1a2 during
the first 24 hpf, especially in the somites. However, a general down-
regulation was observed after this period (from 48 hpf onwards).
Similarly to the Aldh1a2 pattern, the expression pattern of Cyp26a1
in embryos exposed to VPA at 24 hpf also showed evidence of
up-regulation. In older embryos/larvae (≥48 hpf), the expression
pattern of Cyp 26a1 in the group exposed to VPA was similar to
that in the control group (i.e. down-regulated). Regarding the expo-
sure of zebrafish embryos/larvae to LEV, we did not observe any
significant effects on the expression of Aldh1a2 and Cyp26a1 at
any of the time-points studied. These results showed a relation
between effects on the expression of RA-related enzymes and apical
endpoints (morphology, motor activity and histopathology). This
indicates that these molecular markers are very useful for pheno-
type prediction.
In conclusion, in this demo-case study the developmental toxic
potential of a series of AEDs was studied at different levels,
i.e. the structural level (whole animal level by morphology and
organ/tissue/cellular level by histopathology), the functional level
 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116
(motor activity test) and molecular level (ISH). The results of the
study showed the potential of adding multiple endpoints in the
zebrafish embryotoxicity test (ZET) that may increase the sensi-
tivity and predictability of the assay and may contribute to the
elucidation of the mode of toxic action. Most probably, the inclu-
sion of – omics – technology will further elucidate the mechanisms
underlying developmental toxicity and deliver relevant biomark-
ers for developmental toxicity. In potency, this will further define
the applicability domain and increase the reliability of effects in
the ZET. The inclusion of analysis of the uptake of test compounds
to measure total body burden will refine the potency ranking of
test compounds. In addition, especially where compounds induce
no effects, it will further increase the predictability of the assay by
ruling out false negative results (and also false positive results, in
cases of overexposure) in the zebrafish embryotoxicity test.
Conflict of interest
None declared.
Transparency document
The Transparency document associated with this article can be
found in the online version.
Acknowledgements
Authors would like to thank Mieke Havekes and Hakim Rahaoui
(TNO, Zeist, The Netherlands) for assembling the RNA probes
for ISH, Carina Rubingh and Ruud Boessen (TNO, Zeist, The
Netherlands) for the statistics support and graphic layout. We
would also like to thank Gerard van Beek and Dick Veldhuysen
(Animal Facilities TNO-Triskelion, Zeist, The Netherlands) for tak-
ing care of the zebrafish. This work was partially financed by the
Netherlands Toxicogenomics Center (NTC).
References
[1] Kultima K, Jergil M, Salter H, Gustafson A-L, Dencker L, Stigson M. Early
transcriptional responses in mouse embryos as a basis for selection of
molecular markers predictive of valproic acid teratogenicity. Reprod Toxicol
2010;30:457–68.
[2] Marx-Stoelting P, Adriaens E, Ahr H-J, Bremer S, Garthoff B, Gelbke H-P, et al. A
review of the implementation of the embryonic stem cell test (EST). Altern Lab
Anim 2009;37:313–28.
[3] West PR, Weir AM, Smith AM, Donley ELR, Cezar GG. Predicting human devel-
opmental toxicity of pharmaceuticals using human embryonic stem cells and
metabolomics. Toxicol Appl Pharmacol 2010;247:18–27.
[4] Piersma AH, Bosgra S, van Duursen MBM, Hermsen SAB, Jonker LRA, Kroese ED,
et al. Evaluation of an alternative in vitro test battery for detecting reproductive
toxicants. Reprod Toxicol 2013;38:53–64.
[5] Breier JM, Gassmann K, Kayser R, Stegeman H, De Groot D, Fritsche E, et al. Neu-
ral progenitor cells as models for high-throughput screens of developmental
neurotoxicity: state of the science. Neurotoxicol Teratol 2010;32:4–15.
[6] Schenk B, Weimer M, Bremer S, van der Burg B, Cortvrindt R, Freyberger A,
et al. The ReProTect Feasibility Study, a novel comprehensive in vitro approach
to detect reproductive toxicants. Reprod Toxicol 2010;30:200–18.
[7] Lieschke GJ, Currie PD. Animal models of human disease: zebrafish swim into
view. Nat Rev Genet 2007;8:353–67.
[8] Barros TP, Alderton WK, Reynolds HM, Roach AG, Berghmans S. Zebrafish: an
emerging technology for in vivo pharmacological assessment to identify poten-
tial safety liabilities in early drug discovery. Br J Pharmacol 2008;154:1400–13.
[9] Sukardi H, Chng HT, Chan ECY, Gong Z, Lam SH. Zebrafish for drug toxicity
screening: bridging the in vitro cell-based models and in vivo mammalian
models. Expert Opin Drug Metabol Toxicol 2011;7:579–89.
[10] Hill A. Zebrafish in drug discovery: safety assessment. In: Vogel HG,
Maas J, Hock FJ, Mayer D, editors. Drug discovery and evaluation: safety
and pharmacokinetic assays. Heidelberg: Springer Berlin; 2013. p. 605–29,
http://dx.doi.org/10.1007/978-3-642-25240-2 22.
[11] Hill AJ, Teraoka H, Heideman W, Peterson RE. Zebrafish as a model vertebrate
for investigating chemical toxicity. Toxicol Sci 2005;86:6–19.
[12] Padilla S. Zebrafish development: high-throughput test systems to assess
developmental toxicity. In: Steinberg P, editor. High-throughput screening
methods in toxicity testing. USA: John Wiley & Sons, Inc.; 2013. p. 371–83.
115
[13] Hermsen SAB, van den Brandhof E, van der Ven LTM, Piersma AH. Relative
embryotoxicity of two classes of chemicals in a modified zebrafish embryo-
toxicity test and comparison with their in vivo potencies. Toxicol In Vitro
2011;25:745–53.
[14] Padilla S, Corum D, Padnos B, Hunter DL, Beam A, Houck KA, et al. Zebrafish
developmental screening of the ToxCastTM Phase I chemical library. Reprod
Toxicol 2012;33:174–87.
[15] Brannen KC, Panzica-Kelly JM, Danberry TL, Augustine-Rauch KA. Development
of a zebrafish embryo teratogenicity assay and quantitative prediction model.
Birth Defects Res B: Dev Reprod 2010;89:66–77.
[16] Selderslaghs IWT, Blust R, Witters HE. Feasibility study of the zebrafish assay
as an alternative method to screen for developmental toxicity and embryo-
toxicity using a training set of 27 compounds. Reprod Toxicol 2012;33:
142–54.
[17] Beker van Woudenberg A, Wolterbeek A, te Brake L, Snel C, Menke A, Rubingh
C, et al. A category approach to predicting the developmental (neuro)toxicity of
organotin compounds: the value of the zebrafish (Danio rerio) embryotoxicity
test (ZET). Reprod Toxicol 2013;41:35–44.
[18] Irons TD, MacPhail RC, Hunter DL, Padilla S. Acute neuroactive drug expo-
sures alter locomotor activity in larval zebrafish. Neurotoxicol Teratol
2010;32:84–90.
[19] de Esch C, van der Linde H, Slieker R, Willemsen R, Wolterbeek A, Woutersen R,
et al. Locomotor activity assay in zebrafish larvae: influence of age, strain and
ethanol. Neurotoxicol Teratol 2012;34:425–33.
[20] Selderslaghs IWT, Hooyberghs J, Blust R, Witters HE. Assessment of the devel-
opmental neurotoxicity of compounds by measuring locomotor activity in
zebrafish embryos and larvae. Neurotoxicol Teratol 2013;37:44–56.
[21] Shrader EA, Henry TR, Greeley Jr MS, Bradley BP. Proteomics in zebrafish
exposed to endocrine disrupting chemicals. Ecotoxicology 2003;12:
485–8.
[22] Hermsen SAB, Pronk TE, van den Brandhof E-, van der Ven LTM, Piersma AH.
Transcriptomic analysis in the developing zebrafish embryo after compound
exposure: individual gene expression and pathway regulation. Toxicol Appl
Pharmacol 2013;272:161–71.
[23] Van den Bulck K, Hill A, Mesens N, Diekman H, De Schaepdrijver L, Lammens L.
Zebrafish developmental toxicity assay: a fishy solution to reproductive toxic-
ity screening, or just a red herring. Reprod Toxicol 2011;32:213–9.
[24] Phiel CJ, Zhang F, Huang EY, Guenther MG, Lazar MA, Klein PS. Histone deacety-
lase is a direct target of valproic acid, a potent anticonvulsant, mood stabilizer,
and teratogen. J Biol Chem 2001;276:36734–41.
[25] Lloyd KA. A scientific review: mechanisms of valproate-mediated teratogene-
sis. Biosci Horizons 2013;6.
[26] Menegola E, Di Renzo F, Broccia ML, Giavini E. Inhibition of histone deacetylase
as a new mechanism of teratogenesis. Birth Defects Res C: Embryo Today Rev
2006;78:345–53.
[27] Rhinn M, Dollé P. Retinoic acid signalling during development. Development
2012;139:843–58.
[28] Tomson T, Battino D. Teratogenic effects of antiepileptic drugs. Lancet Neurol
2012;11:803–13.
[29] Ornoy A. Neuroteratogens in man: an overview with special emphasis
on the teratogenicity of antiepileptic drugs in pregnancy. Reprod Toxicol
2006;22:214–26.
[30] Samrén EB, Van Duijn CM, Koch S, Hiilesmaa VK, Klepel H, Bardy AH, et al.
Maternal use of antiepileptic drugs and the risk of major congenital malforma-
tions: a joint European prospective study of human teratogenesis associated
with maternal epilepsy. Epilepsia 1997;38:981–90.
[31] Finnell RH, Dansky LV. Parental epilepsy, anticonvulsant drugs, and repro-
ductive outcome: epidemiologic and experimental findings spanning three
decades; 1: Animal studies. Reprod Toxicol 1991;5:281–99.
[32] Pennell PB, Klein AM, Browning N, Baker GA, Clayton-Smith J, Kalayjian LA,
et al. Differential effects of antiepileptic drugs on neonatal outcomes. Epilepsy
Behav 2012;24:449–56.
[33] Fabro S, Shull G, Brown NA. The relative teratogenic index and teratogenic
potency: proposed components of developmental toxicity risk assessment.
Teratog Carcinog Mutagen 1982;2:61–76.
[34] Ponnusamy R, Pradhan N. The effects of chronic administration of ethosuximide
on learning and memory: a behavioral and biochemical study on nonepileptic
rats. Behav Pharmacol 2006;17:573–9.
[35] Long L. Levetiracetam monotherapy during pregnancy: a case series. Epilepsy
Behav 2003;4:447–8.
[36] Mawhinney E, Craig J, Morrow J, Russell A, Smithson WH, Parsons L, et al.
Levetiracetam in pregnancy: results from the UK and Ireland epilepsy and
pregnancy registers. Neurology 2013;80:400–5.
[37] Westerfield M. The zebrafish book: a guide for the laboratory use of
zebrafish (Danio rerio). 5th ed. Eugene, OR, USA: University of Oregon Press;
2007.
[38] OECD. Guideline for the testing of chemicals. Draft proposal for a new guideline
Fish Embryo Toxicity (FET) test; 2006.
[39] European Commission (EC). EU. Directive 2010/63/EU of the European parlia-
ment and of the council of 22 September 2010 on the protection of animals
used for scientific purposes.; 2010. p. 33–79.
[40] Berghmans S, Hunt J, Roach A, Goldsmith P. Zebrafish offer the potential for a
primary screen to identify a wide variety of potential anticonvulsants. Epilepsy
Res 2007;75:18–28.
[41] Weigt S, Huebler N, Strecker R, Braunbeck T, Broschard T. Zebrafish (Danio rerio)
embryos as a model for testing proteratogens. Toxicology 2011;281:25–36.
116 A. Beker van Woudenberg et al. / Reproductive Toxicology 49 (2014) 101–116
[42] Fraysse B, Mons R, Garric J. Development of a zebrafish 4-day embryo-larval
bioassay to assess toxicity of chemicals. Ecotoxicol Environ Saf
2006;63:253–67.
[43] DIN. Basivalidierung genormter verfahre zur wasser-,abwasser-,und schlam-
muntersuchung validierungsdokument (38415-T6); 2002. p. 27.
[44] Nau H, Tzimas G, Mondry M, Plum C, Spohr H-L. Antiepileptic drugs alter
endogenous retinoid concentrations: a possible mechanism of teratogenesis
of anticonvulsant therapy. Life Sci 1995;57:53–60.
[45] Duester G. Retinoic acid synthesis and signaling during early organogenesis.
Cell 2008;134:921–31.
[46] Chuang C-M, Chang C-H, Wang H-E, Chen K-C, Peng C-C, Hsieh C-L, et al. Val-
proic acid downregulates RBP4 and elicits hypervitaminosis a-teratogenesis—a
kinetic analysis on retinol/retinoic acid homeostatic system. PLoS ONE 2012:
7.
[47] Thisse C, Thisse B. High-resolution in situ hybridization to whole-mount
zebrafish embryos. Nat Protoc 2008;3:59–69.
[48] Kaindl AM, Asimiadou S, Manthey D, Hagen MVD, Turski L, Ikonomi-
dou C. Antiepileptic drugs and the developing brain. Cell Mol Life Sci
2006;63:399–413.
[49] Roullet FI, Lai JKY, Foster JA. In utero exposure to valproic acid and
autism—a current review of clinical and animal studies. Neurotoxicol Teratol
2013;36:47–56.
[50] Dean JCS, Hailey H, Moore SJ, Lloyd DJ, Turnpenny PD, Little J. Long term health
and neurodevelopment in children exposed to antiepileptic drugs before birth.
J Med Genet 2002;39:251–9.
[51] Rihtman T, Parush S, Ornoy A. Developmental outcomes at preschool age after
fetal exposure to valproic acid and lamotrigine: cognitive, motor, sensory and
behavioral function. Reprod Toxicol 2013;41:115–25.
[52] Piersma AH, Verhoef A, Opperhuizen A, Klaassen R, Van Eijkeren J, Olling
M. Embryotoxicity of carbamazepine in rat postimplantation embryo culture
after in vitro exposure via three different routes. Reprod Toxicol 1998;12:
161–8.
[53] Morrow J, Russell A, Guthrie E, Parsons L, Robertson I, Waddell R, et al.
Malformation risks of antiepileptic drugs in pregnancy: a prospective study
from the UK Epilepsy and Pregnancy Register. J Neurol Neurosurg Psychiatry
2006;77:193–8.
[54] Meador KJ, Baker GA, Browning N, Clayton-Smith J, Combs-Cantrell DT, Cohen
M, et al. Cognitive function at 3 years of age after fetal exposure to antiepileptic
drugs. N Engl J Med 2009;360:1597–605.
[55] Sullivan FM, McElhatton PR. A comparison of the teratogenic activity of the
antiepileptic drugs carbamazepine, clonazepam, ethosuximide, phenobarbital,
phenytoin, and primidone in mice. Toxicol Appl Pharmacol 1977;40:365–78.
[56] Shallcross R, Bromley RL, Irwin B, Bonnett LJ, Morrow J, Baker GA. Child
development following in utero exposure: levetiracetam vs sodium valproate.
Neurology 2011;76:383–9.
[57] Diekmann H, Hill A. ADMETox in zebrafish. Drug Discov Today: Dis Models
2013;10:e31–5.
[58] Dobbs-McAuliffe B, Zhao Q, Linney E. Feedback mechanisms regulate retinoic
acid production and degradation in the zebrafish embryo. Mech Dev
2004;121:339–50.