The toxic effect of Amiodarone on valve formation in the developing heart of zebrafish embryos

Contents lists available at SciVerse ScienceDirect

Reproductive Toxicology

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / r e p r o t o x

The toxic effect of Amiodarone on valve formation in the developing heart of zebrafish embryos

Ying-Hsin Chen ^a,b,1 , Hung-Chieh Lee ^c,1 , Ren-Jun Hsu ^c , Ta-Yuan Chen ^c , Yu-Kai Huang ^c , Hao-Chan Lo ^c , Sheng-Chuan Hu ^a,d , Horng-Jyh Harn ^e , Jing-Ren Jeng ^f , Chi-Kuang Sun ^g , Shinn-Zong Lin ^h , Huai-Jen Tsai ^c,∗

a

Institute of Medical Sciences, Buddhist Tzu Chi University, Taiwan

b

Department of Emergency Medicine, Tri-Service General Hospital, National Defense Medical Center, Taiwan

c

Institute of Molecular and Cellular Biology, National Taiwan University, Taiwan

d

Department of Emergency Medicine, Buddhist Tzu Chi General Hospital, Taiwan

e

Departments of Pathology, China Medical University and Hospital, Taiwan

f

Division of Cardiology, Department of Internal Medicine, Buddhist Tzu Chi General Hospital, Tzu Chi University, Taiwan

g

Department of Electrical Engineering and Graduate Institute of Photonics and Optoelectronics, National Taiwan University, Taiwan

h

Center for Neuropsychiatry, China Medical University and Hospital and Beigang Hospital, Taiwan

a r t i c l e i n f o

Article history:

Received 12 August 2011

Received in revised form 8 December 2011 Accepted 12 December 2011

Available online 29 December 2011

Keywords:

Zebrafish Amiodarone Valve development Versican

cdh5

Epithelial-mesenchymal transition

a b s t r a c t

Background: Amiodarone is a class D drug given to treat arrhythmia, including pregnant women, but its effects on the developing heart have not been studied. Although some studies have suggested that this drug is safe for fetuses, they have been conducted on mothers with fetuses at or beyond six months of gestational age.

Results: The occurrence of valve defect was positively proportional to Amiodarone concentrations over 9 ␮M, but not lower than 6 ␮M. Ectopic overexpression of versican was observed at the atrioventricular canal of the Amiodarone-treated embryos at 15 ␮M (EC

). VE-cadherin (cdh5), normally downregulated at the endocardial cushion, was also ectopically overexpressed in the Amiodarone-treated embryos. Knock- down of either versican or cdh5 in the Amiodarone-treated embryos could rescue the valve defect caused by Amiodarone.

Conclusions: By inducing versican ectopical overexpression, leading, in turn, to cdh5 ectopical overexpres- sion, Amiodarone treatment causes failure of cardiac valve formation in zebrafish embryos.

© 2011 Elsevier Inc. All rights reserved.

1. Introduction

Amiodarone is categorized as a type III antiarrhythmic drug [1,2]. It is considered a broad-spectrum antiarrhythmic agent [3] because it has multiple and complex effects on the electrical activity of the heart. As such, Amiodarone is effective in treating tachyarrhymias, including re-entry supraventricular tachycardias, ventricular tachycardia, atrial arrhythmias and ventricular fibril- lation [4]. While Amiodarone is considered the antiarrhythmic treatment of choice, it is classified as a category D drug. Conse- quently, caution should be exercised before using Amiodarone for pregnant women, as it causes embryonic hypothyroidism and hyperthyroidism [5,6]. In contrast, Valensise et al. [7] showed that long-term use of Amiodarone has no effect on embryogenesis.

∗ Corresponding author. Tel.: +886 2 3366 2487, fax: +886 2 2363 8483.

E-mail address:

(H.-J. Tsai).

1

These two authors contributed equally.

In a clinical study [8], no developmental effects, except one case of hypothyroidism, was demonstrated. However, these studies focused on embryos older than 6 months when most organs, including the heart, have been completely formed. There is no epidemiological data for women who were taking Amiodarone when they inadvertently became pregnant. Still, based on the evidence at hand, it is important to know whether Amiodarone could be toxic for embryos at an early developmental stage because the half-life of Amiodarone is reported to be 26–107 days [9], and Amiodarone could be prescribed in the case of undetected pregnancy or pregnancy within the gestational period.

To study the toxicity of Amiodarone on heart development, we used zebrafish as a system model since the transparency of embryos allows us to directly observe cardiac development without invasive procedures. As such, zebrafish is an excellent organism to study car- diovascular genetics and defects [10]. Two zebrafish heart-specific fluorescence transgenic lines are available for the in vivo study of cardiac development: Tg(cmlc2:EGFP) with heart-specific green flu- orescence [11] and Tg(cmlc2:HcRFP) with heart-specific infrared 0890-6238/$ – see front matter © 2011 Elsevier Inc. All rights reserved.

doi:10.1016/j.reprotox.2011.12.008

emission [12,13]. In addition, early-stage cardiac development of zebrafish is similar to that of human in many respects, such as migration of cardiac precursor cells towards the central line, heart tube formation, early chamber formation and the looping process.

In our studies, we found that Amiodarone caused defects in cardiac valve formation of zebrafish embryos. We also revealed that the molecular mechanism underlying such defects involves overexpression of both versican and VE-cadherin5 (cdh5) at the atri- oventricular canal (AVC). Therefore, we concluded, that ectopic expression of versican and cdh5 induced by Amiodarone caused defects impeding normal heart valve development in zebrafish.

2. Materialsandmethods

2.1. Observation of zebrafish transgenic lines and heart development

The zebrafish AB strain, as well as transgenic lines Tg(cmlc2:HcRFP) and Tg(cmlc2:EGFP), were cultured as previously described. Heart formation was observed under fluorescent stereomicroscopy (MZ FLIII, Leica). Valve formation was observed in vivo by using Harmonic Generation Microscopy assisted by Two-Photon Fluorescence Microscopy (HGM/2PF)

The excitation source was a femtosecond Cr:forsterite laser with an output wavelength of 1230 nm. Second harmonic gener- ation (SHG) and third harmonic generation (THG)

were applied to observe valve development in transgenic zebrafish Tg(cmlc2:HcRFP) in vivo. RFP marked the myocardial cells, while THG (410 nm) marked all cells in yellow, and SHG (615 nm) marked the skeletal and cardiac muscles in green. Paraffin sectioning with Hema- toxylin and Eosin (H&E) staining was also used to perform histochemical analysis of the heart.

2.2. Drug treatment with zebrafish embryos

Amiodarone (Sigma) was dissolved in water at 65

C for 2 h and stocked as 900 ␮M at 4

C. Before use, the solution was re-dissolved at 65

C for 1 h. In the control group, 100 embryos were placed in a 9 cm dish filled with a volume of 30 ml embryo medium containing 0.2 mM 1-phenyl-2-thio-urea (Sigma). In the experi- mental group, the protocol was identical to the control group, except that embryos at different stages were treated with concentrations of Amiodarone that ranged from 3 to 30 ␮M, and embryos were exposed to treatment from 12 to 84 h (Supplementary

Long-term treatment during 12–72 hpf included the specification stage of valve formation at 36–55 hpf and the invagination stage of valve formation at 55 hpf.

Treatment during 12–48 hpf was used to examine the gene markers versican and cdh5 which expressed at the AVC. During treatment, Amiodarone was refreshed every 24 h, and after treatment, embryos were washed twice with embryo medium, collected into a new 9 cm dish, and then incubated at 28

C.

Ion channel inhibition was achieved by treating embryos with 3.5 mM 4- Amiopyridine (potassium channel blocker; Sigma), 200 mg/L Nifedipine (calcium channel blocker; Sigma), 15 mg/L Lidocaine (sodium channel blocker; Sigma) or 20 mg/L Propranolol (beta-adrenergic receptor blocker; Sigma), and then fixing the embryos by 4% paraformaldehyde. 100 embryos were placed in a 9 cm dish filled with a volume of 30 ml embryo medium containing these inhibitors from 12 to 48 hpf.

2.3. Knockdown experiments

The following morpholino nucleic acid oligomers (MOs) were purchased from GeneTools (USA): versican-MO (CTGAAACACCCATGGGAGTGGACAT); cdh5-MO (TTTACAAGACCGTCTCCTTTCCAA)

and standard control-MO (CCTCTTAC- CTCAGTTACAATTTATA). All MOs were prepared at a stock concentration of 1 mM and diluted to the desired concentration, specifically, 8, 12 and 16 ng for versican- MO; 4, 8, 12 and 16 ng for control-MO; and 0.8, 1.2, 1.6 and 2 ng for cdh5-MO. The standard control-MO served as negative control (Supplementary

2.4. Whole-mount in situ hybridization (WISH)

WISH was performed as previously described

Riboprobe of cdh5 was pre- pared by cloning its partial DNA fragment, while riboprobe of versican was provided by Haramis

2.5. Western blot analysis

The embryos were dechorionated and deyolked with two extra washing steps as described in Link et al.

Deyolked samples were dissolved in 2 ␮l of 2× SDS sample buffer for each embryo and incubated for 5 min at 95

C. After full-speed centrifugation for 1 min in a microcentrifuge to remove insoluble particles, total proteins extracted from embryos were analyzed on a 12% SDS-PAGE gel, and West- ern blot analysis was performed

using antiserum against mouse Cdh5 (15;

1:10,000). Anti-␣-tubulin and anti-␤-actin served as a protein loading control.

Table1

The percentages of defective phenotypes of zebrafish embryos treated with different concentrations of Amiodarone for various exposure times.

Concentration Treatment duration The percentages of phenotypes Blood

regurgitation

Pericardiac edema

0 ␮M – 2/200 (1%) 7/200 (3.5%)

3 ␮M (12 hpf–48 hpf) 4/320 (1.3%) 5/313 (1.6%) 6 ␮M (12 hpf–48 hpf) 20/416 (4.6%) 15/384 (3.9%) 9 ␮M (12 hpf–48 hpf) 168/398 (42.2%) 72/398 (18.1%) 12 ␮M (12 hpf–48 hpf) 177/331 (53.5%) 53/304 (17.4%) 15 ␮M (12 hpf–48 hpf) 351/638 (55.0%) 166/638 (26.0%) 30 ␮M (12 hpf–48 hpf) 188/295 (63.7%) 78/237 (32.9%) 3 ␮M (12 hpf–72 hpf) 3/317 (0.9%) 11/317 (3.5%) 6 ␮M (12 hpf–72 hpf) 27/403 (6.7%) 43/372 (11.6%) 9 ␮M (12 hpf–72 hpf) 189/394 (47.9%) 174/368 (47.3%) 12 ␮M (12 hpf–72 hpf) 103/216 (47.7%) 116/237 (48.9%) 15 ␮M (12 hpf–72 hpf) 78/156 (50%) 79/156 (50.6%)

30 ␮M (12 hpf–72 hpf) –

–

The medium and drugs were renewal every 12 h.

a

All embryos were lethal.

2.6. Statistics

All values for statistical significance represent the mean ± standard deviation (S.D.). We arrived at means by t-test with significant difference of P < 0.05. For dose response curve, the formula we used is Y = (Top − Bottom)/(1+10log EC

− X)Hill slope. The variable Hill slope describes the steepness of the curve. A stan- dard dose response curve has a hill slope of ±1. Means and standard errors were determined according to at least three independent experimental replicates. Regres- sion curves are generated using the MasterPlex non-linear regression analysis for 4 parameter logistic models software.

3. Results

3.1. The toxicity and lethal dosage of zebrafish embryos treated with Amiodarone

First, we studied the half-lethal concentration (LC

) of Amio- darone treatment at different concentrations and under different exposure times. The percentage of embryos suffering acute toxicity from lethal dosage at each treatment concentration was calculated.

As shown in Supplementary Fig. 2, the rate of lethality resulting

from Amiodarone dosages from 9 ␮M to 18 ␮M for an exposure of

36 h was no more than 10%. In contrast, the rates of lethality result-

ing from Amiodarone dosages at 15 ␮M and 18 ␮M for an exposure

of 60 h were 36.5 and 38%, respectively. When embryos were

treated with 30 ␮M Amiodarone, the rate of lethality exceeded

80%, making it impossible to observe heart development during

later stages. Thus, when higher concentrations of Amiodarone are

coupled with longer exposure times, the rates of lethality among

treated embryos increased substantially. Additionally, we also cal-

culated the occurrence percentages of such heart defects as slow

heartbeat, blood regurgitation and pericardiac edema in zebrafish

embryos treated with different concentrations of Amiodarone at

different exposure times. As shown in Table 1, compared to con-

trol group, no increase in heart defects was observed in embryos

treated with lower concentrations of Amiodarone, such as 3 ␮M

and 6 ␮M for 36 and 60 h. However, the occurrence percentages

of pericardiac edema and blood regurgitation were dramatically

increased if embryos were treated with higher concentrations of

Amiodarone, such as 9 ␮M, 12 ␮M and 15 ␮M for 36 and 60 h. We

determined that the half maximal effective concentration (EC

)

of Amiodarone is 15 ␮M and that the exposure time is either 36 h

(from 12–48 hpf) or 60 h (from 12–60 hpf). We chose a 15 ␮M con-

centration of Amiodarone for further experiments in this study

for the following reasons: (I) Zebrafish heart development is com-

pleted at 72 hpf. (II) The LC

of Amiodarone for an incubation of

60 h ranged from 18 to 24 ␮M. (III) The EC

of Amiodarone is

15 ␮M, which is below the LC

(Supplementary Fig. 3). (IV) This concentration is close to the EC

which avoids defects that are caused by nonspecific toxicity, and it is also easy for us to quantify.

3.2. Amiodarone caused abnormal heartbeat and defective valve formation in zebrafish embryos

Embryos were treated with 15 ␮M Amiodarone to observe its effect on zebrafish development. In contrast with control embryos, embryos treated with 15 ␮M Amiodarone from 12 to 48 hpf had

shorter axes, smaller heads, and more delayed development (Fig. 1A vs. D). No obvious defects in the pericardiac cavity (Fig. 1B vs. E) were observed, but 82.3% of the embryos showed tail shrinkage (Fig. 1C vs. F). If embryos were continuously treated with 15 ␮M Amiodarone from 12 to 72 hpf, the embryos showed phenotypes similar to those treated at 48 hpf. However, these embryos were smaller (Fig. 1G vs. J) and showed pericardiac edema (Fig. 1H vs. K), some tail shrinkage (Fig. 1I vs. L) and some ulceration to the venous plexus (Fig. 1L). We noted that pericardiac edema and tail shrink- age were both dependent on the concentration of Amiodarone

Fig.1.

Defective phenotypes of embryos caused by Amiodarone treatment. Zebrafish embryos were treated with 15 ␮M Amiodarone starting at 12 hpf, and the phenotypes of the control (A–C and G–I) and treated (D–F and J–L) groups were compared. Compared to the control group, embryos treated with Amiodarone during 12–48 hpf displayed defective phenotypes, such as shorter axes, smaller heads, developmental delay (A vs. D), and cell necrosis at the pericardial membrane (B vs. E, triangle) and tail-fin (C vs.

F, arrow). Embryos treated with Amiodarone from 12–72 hpf displayed phenotypes (G vs. J; H vs. K; and I vs. L) similar to the embryos treated at 12–48 hpf, except with

higher ratio of phenotype occurrence and a swollen pericardial cavity (K). The occurrence of defective phenotypes of embryos treated with Amiodarone from 12–72 hpf was

calculated (L). The rate of heartbeat (in bpm) was counted in the embryos treated with 15 ␮M Amiodarone for 36 and 60 h (M). Data are presented as mean ± S.D. Embryos

derived from transgenic line Tg(cmlc2:EGFP) were observed under fluorescent microscopy, and the incomplete looping of developing heart could be observed in embryos

treated with 15 ␮M Amiodarone during 48 hpf (O vs. P).** indicates the significant difference at the level of P < 0.05. ve, ventricle; at, atrium.

(Fig. 1M). We found that the heart rate was also greatly reduced in the embryos treated continuously with 15 ␮M Amiodarone for 36 and 60 h (Fig. 1N). In embryos treated with Amiodarone for 36 h, there was evidence of arterial/venous occlusion and ventric- ular arrhythmia. In a morphological examination of the heart, we also found that looping of the developing heart was incomplete in embryos treated with 15 ␮M Amiodarone during 48 hpf (Fig. 1O vs.

P). Specifically, the endocardial cushion, a subset of cells found in the developing heart tube that gives rise to the heart’s valves and septa, begins to function at 72 hpf, which is when we noticed blood regurgitation between atrium and ventricle in the Amiodarone- treated embryos (see attached movie 1 vs. movie 2).

Endocardial cushion cells proliferate at the AV canal, and the process of specification begins during 36 to 55 hpf. Therefore, we treated embryos with 15 ␮M Amiodarone during 22–34, 34–46 and 46–58 hpf and then examined for blood regurgitation both at 48 and 72 hpf. When we examined embryos at 48 hpf, we found that the occurrence of blood regurgitation was dramatically increased in the embryos treated with Amiodarone during 34–46 hpf and dur- ing 46–58 hpf, compared to control embryos (Supplementary Table 1). Such results suggested that Amiodarone causes blood regurgita- tion of zebrafish embryos during, but not prior to, valve formation.

Importantly, when Amiodarone treatment was stopped either at 36 or 58 hpf, most defective embryos displaying the blood regurgi- tation phenotype could be rescued by 72 hpf (Supplementary Table 1), indicating that heart defect induced by Amiodarone can also be abolished in the absence of Amiodarone.

Since the most important function of valves is the blockage of blood regurgitation from ventricle to atrium, we further examined whether a defect of valve development occurred in the process of heart development. Histochemical staining on the paraffin section- ing of 72-hpf zebrafish hearts revealed that the AVC endocardial cushion-forming region (ECFR) showed a bulged structure in the untreated embryos (Fig. 2B). These endocardial cushion cells orig- inated from endocardial cells migrating from the AVC towards the extracellular matrix (ECM), where these cells started to aggregate and proliferate. However, in the Amiodarone-treated embryos, the ECRF failed to form at the corresponding position of AVC, and the endocardial cells remained a structure formed by a single layer of cells (Fig. 2D). These results indicated that Amiodarone treatment caused abnormal valve development in zebrafish embryo hearts.

3.3. Cardiac valves of zebrafish were directly observed in vivo

By using HGH/2PF to examine the zebrafish embryos derived from transgenic line Tg(cmlc2:HcRFP), we could easily observe the dynamics of valve development in vivo. For example, at 48 hpf, there was a single layer of cells in the endocardium at the AVC (Fig. 3A).

At 72 hpf, a bulged structure was observed at the AVC (Fig. 3B), and the endocardial cells continued to move towards cardiac jelly and gradually elongated to form the structure of valves at 96 hpf.

Finally, at 87 hpf, the elongated structure was protruding towards the ventricle (Fig. 3C). However, if we treated these embryos during 12–87 hpf with 300 ␮M CsA, a drug known to repress epithelial- mesenchymal transition (EMT) and thus cause valve defect [21], we found that valves were not formed, which, again, could be clearly observed at 87 hpf under HGH/2PF (Fig. 3H). Interestingly, when we treated zebrafish embryos with 15 ␮M Amiodarone during 12–48 hpf, no difference between treated and untreated embryos was noted in endocardial cells at the AVC where only a single layer of cells was observed (Fig. 3A vs. E). However, when embryos were treated with the same dosage of Amiodarone during 12–72 hpf, only a small aggregation of cells could be seen at the upper valve site and no valve structure at the lower site (Fig. 3B). Furthermore, compared to untreated embryos (Fig. 3C), embryos treated with

Amiodarone during 12–87 hpf displayed almost no cellular aggre- gation at the valve region (Fig. 3G).

3.4. Loss of endocardial cushion cells resulted from Amiodarone treatment, not apoptosis

Long-term treatment of embryos with Amiodarone resulted in defective valves (Fig. 2D). This defect might have been caused by Amiodarone-induced apoptosis, which, in turn, would reduce the number of caudal fin and endocardial cushion cells. Therefore, to confirm Amiodarone-induced apoptosis in the endocardial cushion areas, the TUNEL assay was applied. In the 72-hpf control group without Amiodarone treatment, the cardiac and marginal regions were transparent without any apoptotic signals (Supplementary Figs. 4A and B). Treatment with 15 ␮M Amiodarone from 12 hpf showed some signals of apoptosis in the marginal region of the trunk (Supplementary Fig. 4D), but stronger signals in the tail. At the same time, however, no apoptotic signals were observed in the cardiac area (Supplementary Fig. 4C). Similarly, there were no apoptotic signals in the 87-hpf control group (Supplementary Figs.

4E and F). At 87 hpf, the embryos that had been treated with 15 ␮M Amiodarone from 12 hpf showed severe apoptosis in the tail region, whereas apoptotic signals were still undetectable in the cardiac region (Supplementary Fig. 4G). Thus, the valve defect observed in embryos undergoing long-term Amiodarone treatment did not result from Amiodarone-induced apoptosis, which, in turn, would have reduced the number of caudal fin and endocardial cushion cells. Instead, it appears that Amiodarone acts directly to cause the loss of endocardial cushion cells.

3.5. Amiodarone treatment caused ectopic overexpression of versican

Versican, a gene involved in cell migration, is expressed in the myocardium [22,23] and endocardium [18], respectively, at AVC in zebrafish at 48 hpf. Therefore, we performed WISH to observe the expression of versican, which is indicative of valve forma- tion, in embryos treated with Amiodarone. In wild-type embryos, versican was normally expressed at AVC at 48 hpf (Fig. 4A). How- ever, embryos treated with 15 ␮M Amiodarone during 12–48 hpf showed massive ectopic overexpression of versican in ventricle, atrium, and at AVC (Fig. 4B). Similarly, versican was expressed restrictedly at AVC at 72 hpf in the untreated embryos (Fig. 4C).

However, in the embryos treated with 15 ␮M Amiodarone during 12–72 hpf, versican was again found to be expressed ectopically in ventricle, atrium, and at AVC (Fig. 4D). We note that the ectopic overexpression of versican at AVC was observed irrespective of the embryonic stage during which Amiodarone treatment was begun (12–7 2hpf, 36–72 hpf, 36–48 hpf, 48–60 hpf, or 60–72 hpf) (Fig. 4F–J). In fact, versican was overexpressed ectopically at AVC, even in embryos treated with Amiodarone at 84–96 hpf, after valves had been formed (Fig. 4K vs. L). This line of evidence indicated that Amiodarone treatment enables embryonic overexpression of the genes involved in endocardial cell migration in zebrafish heart, which, in turn, suggests that Amiodarone may impede the normal development of heart valve formation.

3.6. Amiodarone treatment at specification and invagination stages caused versican to overexpress at AVC and repressed valve development

Guided by the hypothesis that Amiodarone impedes proper

heart valve formation by causing the overexpression of versican,

we further investigated the critical stage(s) at which Amiodarone

affects heart valve development. To accomplish this, we treated

embryos with Amiodarone at different developmental stages and

examined the embryos for versican expression at AVC with WISH

Fig.2.

Amiodarone caused developmental abnormality of cardiac valves in zebrafish embryos. To observe cardiac valve development, zebrafish embryos at 72 hpf were subjected to paraffin sectioning followed by H&E staining. (A and B), the heart of wild-type zebrafish embryos underwent invagination, and the superior endocardial cushion- forming region (ECFR) started to form endocardial cushion (arrow). (C and D) Embryos treated with 15 ␮M Amiodarone from 12–72 hpf displayed neither invagination progression nor endocardial cushion formation (arrow). ve, ventricle; at, atrium.

aided by paraffin sectioning and H&E staining. We observed that valve development was repressed in 72-hpf embryos treated with 15 ␮M Amiodarone during 12–72 hpf. Next, we focused our atten- tion on the specification (36–55 hpf) and invagination (after 55 hpf) stages (Supplementary Fig. 1). At 72 hpf, the wild-type embryos showed the expression of versican at the cardiac AVC (Fig. 5A), and sectioning results revealed the existence of an endocardial cushion, the precursor structure of valve formation at the inter- section of ventricle and atrium (Fig. 5E). However, 72-hpf embryos treated with 15 ␮M Amiodarone during 12–72 hpf showed that versican was ectopically overexpressed at the AVC (Fig. 5B), and no endocardial cushion at AVC was observed (Fig. 5F). In addi- tion, 72-hpf embryos treated with Amiodarone at the specification stage showed that versican was ectopically expressed at the car- diac region (Fig. 5C), and no endocardial cushion structure was observed (Fig. 5G). Similarly, at the invagination stage, versi- can ectopic overexpression was observed in the 72-hpf embryos treated with Amiodarone (Fig. 5D), and no endocardial cushion structure was found (Fig. 5H). In summary, our results demon- strated that embryos treated with 15 ␮M Amiodarone during 12–72 hpf, including both specification and invagination stages, ectopically overexpressed versican in the cardiac region with no formation of the endocardial cushion. These lines of evidence sug- gest that Amiodarone treatment causes heart valve defect at both specification and invagination stages.

3.7. Amiodarone treatment increases cdh5 expression at zebrafish cardiac AVC

Cdh5 is an adhesion molecule expressed in endocardial cells [24], and decreased expression of cdh5 promotes EMT [25].

To clarify whether Amiodarone affects endocardial cells dur- ing invagination, we applied WISH to detect the expression of

cdh5, an EMT-related gene, at AVC in embryos treated with 15 ␮M Amiodarone during 12–72 hpf, including both the spec- ification and invagination stages. The results showed that cdh5 expressed restrictedly at AVC in the untreated embryos at 72 hpf (Fig. 6A). However, in the embryos treated with Amiodarone dur- ing 12–72 hpf, cdh5 was overexpressed at AVC when observed at 72 hpf (Fig. 6B). When treated at both specification (Fig. 6C) and invagination (Fig. 6D) stages, cdh5 was also overexpressed in the 72-hpf embryos. Thus, it is plausible that Amiodarone could induce the overexpression of cdh5, which would then inhibit endocardial cells from undergoing invagination and suppress the develop- ment of endocardial cushion as well. As further confirmation of this hypothesis, we performed Western blot analysis using Cdh5 antibody to determine whether Cdh5 protein is also increased in the Amiodarone-treated embryos. The results showed that Cdh5 protein was greatly increased in zebrafish embryos after Amio- darone treatment (Fig. 6E), suggesting that Amiodarone enhances the expression of Cdh5 protein. These lines of evidence demon- strated that Amiodarone not only induces the overexpression of versican but also induces the overexpression of cdh5, a downstream regulator of versican involved in EMT in zebrafish embryos.

3.8. Reduction of cdh5 translation by cdh5-MO injection rescues the valve defect caused by Amiodarone treatment

To understand the relationship between the valve defect caused by Amiodarone treatment and cdh5 expression, cdh5-MO was injected into embryos to specifically inhibit the translation of cdh5 transcripts [15,16]; (Supplementary Fig. 5C). The endocardial cush- ion was then observed with paraffin sectioning and H&E staining.

In the untreated and uninjected embryos, the endocardial cush-

ion was observed at AVC (Fig. 7A). In embryos injected with 1.6 ng

cdh5-MO, the endocardial cushion was bulged at 72 hpf, indicating

Fig.3.

In vivo observation of the cardiac valve defects in the Amiodarone-treated zebrafish embryos. HGM/2PF was applied to observe the valve development of transgenic zebrafish Tg(cmlc2:HcRFP) in vivo. RFP marked the myocardial cells, while third harmonic generation (THG, shown in yellow) marked all cells. Second harmonic generation (SHG, shown in green) marked the cardiac muscles. Valve development in zebrafish started around 36–40 hpf. Cardiac epithelial cells transformed from squamous to columnar cells (A, arrow) at 48 hpf. Columnar cells concentrated towards ECM to form the endocardial cushion through the process of invagination at 72 hpf (B, arrow). The endocardial cushion started to protrude towards the atrium and form the valve structure (C, arrow) at 87 hpf. Interestingly, the cardiac epithelial cells in embryos treated with 15 ␮M Amiodarone from 12–48 hpf were squamous cells (E, arrow), similar to those in non-treated embryos. However, when embryos were treated with Amiodarone from 12–72 hpf, the superior endocardial cushion-forming region (ECFR) formed a smaller and incomplete endocardial cushion (E, blue arrow), and inferior ECFR had no endocardial cushion formation (F, white arrow). The endocardial cushion was not observed to form (G, arrow), even when embryos had developed at 87 hpf, the maximal time we could observe living Amiodarone-treated embryos. Two control groups were designed: embryos treated with propranolol, an antiarrhythmic drug, from 12–87 hpf, which was followed by normal endocardial cushion development (D), and embryos treated with CsA, a known valve development inhibitor, which was not followed by endocardial cushion formation (H). ve, ventricle; at, atrium.

Table2

The rescue experiments of Amiodarone-induced defects by knockdown of either versican (versican-MO) or cdh5 (cdh5-MO).

Materials Amount

(ng)

Amiodarone treatment (␮M)

No. of survival embryos among injected eggs

No. of wild-type like phenotype

No. of reduced cdh5 expression

No. of ectopic cdh5 expression

No. of blood regurgitation

Non-injection

0 0 38/40 94.7% 0 0% N/A

Control-MO

4 0 97/112 99% 1% 0% N/A

Control-MO

8 0 122/142 99.2% 0.8% 0% N/A

Control-MO

12 0 83/93 100% 0% 0% N/A

Control-MO

16 0 106/114 98.1% 1.9% 0% N/A

Versican-MO

8 0 102/113 53.9% 46.1%

0% N/A

Versican-MO

12 0 110/123 41.8% 58.2%

0% N/A

Versican-MO

16 0 117/128 29.9% 70.1%

0% N/A

Non-injection

0 15 112/175 2.7% 0% 97.3% N/A

Versican-MO

12 15 114/164 20.2% 0% 79.8% N/A

Versican-MO

16 15 113/182 41.6% 4.4% 54%

N/A

Control-MO

4 0 26/30 100% N/A N/A 0%

non-injection

0 15 114/192 43% N/A N/A 57%

Versican-MO

12 15 104/173 55.8% N/A N/A 44.2%

Versican-MO

16 15 157/214 59.9% N/A N/A 40.1%

cdh5-MO

0.8 15 106/173 45.3% N/A N/A 54.7%

cdh5-MO

1.2 15 118/181 52.5% N/A N/A 47.5%

cdh5-MO

1.6 15 101/184 66.8% N/A N/A 33.2%

cdh5-MO

2.0 15 56/136 62.5% N/A N/A 37.5%

N/A, not available.

*

Indicates P < 0.05.

**

Indicates P < 0.01.

a

Embryos were treated with Amiodarone starting at 12 hpf for 60 h, and then fixed by paraformaldehyde for WISH to detect cdh5 expression.

b

Embryos were treated with Amiodarone starting at 12 hpf for 60 h, and then observed the direction of blood flow in the heart of zebrafish embryos.

Fig.4.

Zebrafish embryos treated with Amiodarone manifested ectopic overexpression of versican in the cardiac region. Whole mount in situ hybridization was applied to indicate the expression of versican (A)–(D), which expressed at the intersection of ventricle and atrium, i.e. the AVC where the valves normally form. At 48 hpf, versican was ectopically overexpressed in the cardiac region by Amiodarone treatment, as indicated by comparing embryos without (A and C) and with (B and D) 15 ␮M Amiodarone treatment during 12–48 hpf. At 72 hpf, the ectopic expression of versican was also observed in the cardiac region, as shown by comparing embryos without (C) and with (D) 15 ␮M Amiodarone treatment during 12–72 hpf. The ectopic overexpression of versican at AVC was observed irrespective of the embryonic stage at which Amiodarone treatment was begun (12–72 hpf (F), 36–72 hpf (G), 36–48 hpf (H), 48–60 hpf (I), or 60–72 hpf (J)). Versican was overexpressed ectopically at AVC, even in embryos treated with Amiodarone at 84–96 hpf, after valves had already been formed (L) (A: atrium; V: ventricle).

incomplete formation of valves (Fig. 7B). Embryos treated with 15 ␮M Amiodarone during 12–72 hpf without cdh5-MO injec- tion showed no formation of an endocardial cushion at AVC (Fig. 7C). However, embryos treated with 15 ␮M Amiodarone dur- ing 12–72 hpf with 1.6 ng cdh5-MO injection presented a normally formed endocardial cushion at AVC (Fig. 7D). Indeed, when we injected cdh5-MO to reduce the overexpression of cdh5 induced by Amiodarone treatment in embryos, we found that the valve defect caused by Amiodarone was rescued (Table 2). Thus, Amio- darone causes cdh5 to overexpress, resulting in the failure of valve formation by inhibition of the invagination process at AVC.

3.9. Amiodarone-induced valve defect can be rescued by knockdown of either versican or cdh5 in embryos

Since versican and cdh5 are both genes involved in valve devel- opment, we performed experiments to understand how versican

and cdh5 act together to affect valve formation in Amiodarone-

treated embryos. First, we microinjected versican-MO to reduce

the expression of versican, and then we examined cdh5 expression

using WISH (Table 2). Paraffin sectioning and H&E staining were

used to observe the development of cardiac valves. In untreated

and uninjected 72-hpf embryos, cdh5 expressed at AVC (Fig. 8A),

and an endocardial cushion was observed by sectioning and H&E

staining (Fig. 8E). Embryos microinjected with 16 ng versican-MO

showed cdh5 down-regulation at AVC at 72 hpf (Fig. 8B), result-

ing in an abnormally bulged endocardial cushion (Fig. 8F). In the

embryos treated with 15 ␮M Amiodarone during 12–72 hpf, cdh5

was overexpressed at AVC at 72 hpf (Fig. 8C), resulting in the com-

plete absence of endocardial cushion formation (Fig. 8G). In the

embryos injected with 16 ng versican-MO and also treated with

15 ␮M Amiodarone during 12–72 hpf, the expression of cdh5 was

reduced, compared to embryos treated with Amiodarone alone

(Fig. 8C vs. D; Table 2). In addition, cdh5 expression in these embryos

Fig.5.

Versican expression and endocardial cushion development in zebrafish embryos. Amiodarone-treated zebrafish embryos during the developmental stages of spec- ification and invagination demonstrated ectopic overexpression of versican in the cardiac region and hypoplasia of endocardial cushion. After treating the embryos with 15 ␮M Amiodarone at these two stages, the expression of versican at 72 hpf (A)–(D) was observed with whole mount in situ hybridization, and the developmental status of endocardial cushion (E)–(F) was monitored with paraffin sectioning and H&E staining. In embryos without treatment, versican was expressed precisely at the AVC (A), and the endocardial cushion was developed normally (E, arrow). Long-term treatment of Amiodarone caused ectopic overexpression of versican in the cardiac region (B), and the development of endocardial cushion was suppressed (F, arrow). Amiodarone treatment during 36–55 hpf, which is the stage when endocardial cells start the specification process, also resulted in the ectopic overexpression of versican (C) and the repression of endocardial cushion development (G, arrow). Amiodarone treatment after 55 hpf, which starts the invagination process upon migration of endocardial cells, also resulted in the ectopic overexpression of versican (D) and the repression of endocardial cushion development (H, arrow) (A: atrium; V: ventricle).

was similar to wild-type embryos (Fig. 8A vs. D), and the endocar- dial cushion was also observed (Fig. 8H). These results demonstrate that the down-regulation of versican could rescue the valve defect caused by overexpression of cdh5 at AVC, which was initially caused by Amiodarone treatment. Based on this evidence, we can postu- late that Amiodarone-induced valve defect results from the ectopic overexpression of not only versican, but also cdh5, which is the downstream effector of versican. This finding suggests, in turn, that versican might positively regulate the expression of cdh5, thus influ- encing the development of endocardial cushion.

During the knockdown experiments of injecting either versican- MO or cdh5-MO, we also injected in parallel with a standard control-MO to serve as a negative control. Similar to wild-type group, no defective phenotypes were found in embryos injected with control-MO (Supplementary Fig.5A vs. B). The defective phe- notypes induced by either versican-MO or cdh5-MO were dosage dependent, and injection of either versican-MO or cdh5-MO could effectively rescue the valve defects induced by Amiodarone treat- ment, proving that versican-MO and cdh5-MO we used are specific.

3.10. Embryos treated with a potassium channel blocker that mimics an ectopic overexpression of versican

The pharmacological mechanisms of Amiodarone are complex.

Specifically, Amiodarone blocks the potassium, sodium and cal- cium channels and a beta-adrenergic receptor [24]. To determine which pharmacological mechanism is involved in Amiodarone- induced cardiac valve defect in zebrafish embryos, we treated

zebrafish embryos with the following four drugs during 12–48 hpf and observed the expression of versican at 48 hpf: 4-Amiopyridine as a potassium channel blocker, Lidocaine as a sodium channel blocker, Nifedipine as a calcium channel blocker, and Propra- nolol as a beta-adrenergic receptor blocker. Similar to untreated embryos, embryos treated with Lidocaine, Nifedipine, and Pro- pranolol (Supplementary Figs. 6C–E) showed a focally expressed pattern of versican at the AVC (Supplementary Fig. 6A). However, the 4-Amiopyridine-treated embryos displayed ectopic overex- pression of versican at the AVC (Supplementary Fig. 6B), which was similar to the embryos treated with Amiodarone (Fig. 4B vs. Supplementary Fig. 6B). Since the potassium channel blocker caused an ectopic overexpression of versican, it is suggested that Amiodarone could perturb valve development-related signal trans- duction by blocking the flow of potassium, which, in turn, would suppress cardiac valve development in zebrafish embryos.

4. Discussion

4.1. Valve defect caused by Amiodarone is specific and unrelated to toxicity

We analyzed heart defects in embryos treated with various

concentrations of Amiodarone. As shown in Table 1, compared to

control group, no increase in heart defects was observed in embryos

treated with lower concentrations of Amiodarone, such as 3 ␮M and

6 ␮M for 36 and 60 h. However, the occurrence percentages of peri-

cardiac edema and blood regurgitation were dramatically increased

Fig.6.

Amiodarone treatment at the specification and invagination stages, respectively, resulted in cdh5 overexpression at cardiac AVC. The expression of cdh5, an EMT inhibition gene, in embryos treated with 15 ␮M Amiodarone was observed at 72 hpf. The embryos without treatment showed normal cdh5 expression at AVC (A), whereas those with long-term Amiodarone treatment showed overexpression of cdh5 at AVC (B). In addition, embryos treated with Amiodarone at the specification stage only showed overexpression of cdh5 (C), which was the same as the embryos treated only at the invagination stage (D). Western blot analysis of Cdh5 protein extracted from the zebrafish embryos with or without Amiodarone treatment (E). A: atrium; V: ventricle.

if embryos were treated with higher concentrations of Amiodarone, such as 9 ␮M, 12 ␮M and 15 ␮M for 36 and 60 h. The occurrence percentages of pericardiac edema and tail shrinkage were posi- tively proportional with the treated concentration of Amiodarone.

The rate of heartbeat was also greatly reduced and looping of the

developing heart was incomplete in embryos treated with 15 ␮M Amiodarone.

Moreover, as shown in Supplementary Table 1, embryos were treated with 15 ␮M Amiodarone during 22–34, 34–46 and 46–48 hpf and then examined for blood regurgitation both at 48 hpf

Fig.7.

Reduction of Cdh5 expression by cdh5-MO microinjection promoted valve development of zebrafish embryos and was able to rescue the hypoplasia of valve caused by

Amiodarone treatment. Paraffin sectioning and H&E staining were applied to observe the development of endocardial cushion. In embryos absent treatment with Amiodarone

and 1.6 ng cdh5-MO microinjection, endocardial cushion formation was observed at AVC, the intersection of ventricle and atrium (A, arrow). However, after microinjection

with 1.6 ng cdh5-MO, the endocardial cushion bulged abnormally (B, arrow). Amiodarone treatment at 12–72 hpf resulted in inhibition of endocardial cushion development

(C, arrow). Embryos with 1.6 ng cdh5-MO microinjection and 12–72 hpf Amiodarone treatment showed restoration of endocardial cushion development to a normal state

(D, arrow). A: atrium; V: ventricle.

Fig.8.

Microinjection of versican-MO to repress versican expression resulted in down-regulation of cdh5 and was able to rescue cdh5 overexpression and malformation of values caused by Amiodarone treatment. After microinjecting 16 ng versican-MO at the one-cell stage, the expression of cdh5 was observed with whole mount in situ hybridization (A)–(D) at 72 hpf, and the developmental status of endocardial cushion (E)–(F) was also observed with paraffin sectioning and H&E staining. In embryos without Amiodarone treatment and 16 ng versican-MO, versican was expressed at AVC precisely (A), and the structure of endocardial cushion was observed (E, arrow).

Embryos microinjected with 16 ng versican-MO reduced the expression of cdh5 at AVC (B), and endocardial cushion bulged abnormally (F, arrow). Embryos treated with Amiodarone at 12–72 hpf increased cdh5 expression at AVC (C), and the development of endocardial cushion was suppressed (G, arrow). However, embryos with both 16 ng versican-MO microinjection and 12–72 hpf Amiodarone treatment showed levels of cdh5 expression (D) similar to untreated embryos, and the endocardial cushion developed normally (H, arrow). A: atrium; V: ventricle.

and at 72 hpf. When we examined embryos at 48 hpf, the occur- rence of blood regurgitation was dramatically increased in the embryos treated with Amiodarone either between 34 and 46 hpf or between 46 and 58 hpf, compared to control group. Such results suggested that Amiodarone causes blood regurgitation of zebrafish embryos during, but not prior to, valve formation. Moreover, we noticed that blood regurgitation was observed at 48 hpf in the one-third of embryos treated from 46 to 58 hpf, indicating that Amiodarone can rapidly act on embryos within 2 h treatment.

Importantly, when Amiodarone treatment was stopped either at 46 or 58 hpf, most defective embryos displaying the blood regurgi- tation phenotype could be rescued by 72 hpf, indicating that heart defects induced by Amiodarone can also be abolished in the absence of Amiodarone. This line of evidence demonstrated that the heart defects caused by Amiodarone are specific and unrelated to any cytotoxicity against endocardial cells.

4.2. Valve defect caused by Amiodarone does not result from lower velocity of blood cells

Hemodynamics is a necessary component of valve develop- ment. Endocardium at the AV canal responds to blood flow-induced forces and stimulates endocardial cells to rearrange and change the genetic expression profiles [26]. After treatment with Amiodarone, embryonic heart rate was reduced, which reduces the velocity of blood flow. Therefore, to clarify whether Amiodarone-induced valve defects result from the reduction of blood flow and whether such slower blood flow is common to other antiarrhythmic drugs, we observed and compared the effect of Amiodarone and Propranalol on valve development. Propranalol is known to reduce blood flow speed. Embryos treated with Propranalol

during 12–87 hpf were observed at 87 hpf with HGM/2PF. Results showed that these embryos had normal cardiac valve development (Fig. 3D). Furthermore, while hemodynamics may theoretically affect heart development, heart muscle contractions may influence valve development. To rule out this possibility, we treated embryos with either trican or knockdown of cardiac troponin T2a (tnnt2a) by injection of tnnt2a-MO into one-celled embryo to silence heartbeat. Notwithstanding these treatments, valve development proceeded normally (data not shown). Using WISH, we also found that the expression of versican in the AVC of tnnt2a-MO-injected embryos was similar to that in the AVC of untreated wild-type embryos (data not shown), suggesting that the expression of versican in AVC is unaffected by either the reduction in blood flow or silencing of heartbeat. Thus, we believe that valve defects do not arise from side effects of Amiodarone. Rather, the valve defects are specific to ectopic versican expression in the heart induced by Amiodarone. In addition, since this phenomenon specifically occurred with Amiodarone treatment, it is not an effect that can be associated with other antiarrhythmic drugs.

4.3. Amiodarone begins to influence zebrafish cardiac development during the invagination process and Amiodarone-induced valve defect results from abnormal invagination process

It was shown in this study that the heart valve defects caused by Amiodarone are not manifestations of developmental delay.

Instead, by the evidence of blood regurgitation and absence of

valvular structure, Amiodarone acts directly to repress zebrafish

embryonic cardiac valve development. Furthermore, the fact that

embryos only display defects in endocardial cushion at 72 hpf,

when invagination starts [16], suggested that Amiodarone treat- ment begins to affect zebrafish heart valve development during invagination.

When we treated embryos with 15 ␮M Amiodarone, ver- sican was ectopically overexpressed at AVC during 12–72 hpf, i.e. the period covering stages of specification and invagination (Fig. 5A–D). Concurrently, the results of paraffin sectioning/H&E staining showed that zebrafish embryonic cardiac valve develop- ment was suppressed by the treatment of Amiodarone (Fig. 5E–H).

Furthermore, there was no cell aggregation in the endocardial cush- ion region of the Amiodarone-treated embryos at 72 hpf (Fig. 3B vs.

F), which, in turn, prevented endocardial cells from transforming into mobile cells to form endocardial cushion. In addition, com- pared to the valve defect at 72 hpf caused by Amiodarone treatment during the specification stage (Fig. 5G), treatment during the invagi- nation stage repressed endocardial cushion development at 72 hpf (Fig. 5H). These results suggested that Amiodarone directly affects invagination to cause the defect. In further support of this inference, we found that cdh5, an EMT determinant gene, was overexpressed at AVC in the embryos treated with Amiodarone. Since cdh5 is nor- mally a repressive gene, its overexpression results in the inhibition of EMT, leading to the failure to form normal valve structures.

These lines of evidence indicated that invagination is affected by Amiodarone, and thus, the endocardial cushion fails to develop properly.

4.4. Amiodarone induces ectopic overexpressions of versican and cdh5 in the AVC to inhibit valve formation

During embryogenesis, cells switch back and forth between different cellular phenotypes via mesenchymal-epithelial transi- tion (MET) and its reverse process, EMT [27]. More specifically, in mice and chicks, the atrioventricular endocardial cells undergo EMT, as they migrate into the cardiac jelly and proliferate to form endocardial cushions. These cushions form the AV valvu- loseptal complex that divides the ventricular inflow into a right and a left AV valve [28]. Formation of the atrioventricular valve of zebrafish is similar to that of higher vertebrates [22,29]. In our studies, we demonstrated that embryos treated with Amio- darone (I) fail to form cardiac valves and (II) present ectopic overexpression of versican and cdh5 at AVC. We showed that knockdown of either versican or cdh5 can rescue the valve defects caused by Amiodarone (Figs. 7 and 8 and Table 2). These find- ings suggest that the ectopic overexpression of versican and cdh5 induced by Amiodarone significantly represses EMT in valves, resulting in defective phenotypes. Sheng et al. [30] reported that expression of versican can induce MET in NIH3T3 fibrob- lasts. This in vitro study may suggest that versican may repress EMT through inducing MET in the heart. Thus, we speculate that Amiodarone-induced overexpression of versican might seriously repress EMT at AVC. Meanwhile, Timmerman et al. [31] reported that the constitutive activation of Notch not only induces exces- sive endocardial EMT, but it is also required to reduce cdh5 expression in zebrafish, indicating that cdh5 must be reduced when EMT is processed. Thus, we speculate that Amiodarone- induced overexpression of cdh5 might also seriously inhibit EMT at AVC.

Additionally, we knocked down cdh5 in the Amiodarone-treated embryos and found that the valve defects caused by Amiodarone were rescued (Fig. 7). Moreover, knockdown of versican by a spe- cific MO could reduce cdh5 expression and rescue the valve defect caused by Amiodarone (Fig. 8 and Table 2). Therefore, we concluded that Amiodarone-induced ectopic expressions of versican and cdh5 at AVC of embryos are specific and that Amiodarone, which induces overexpression of both versican and cdh5 in the heart, leads to the

repression of EMT resulting the loss of endocardial cell motility, and finally disrupts cardiac valve development.

4.5. Amiodarone treatment begun at specification still causes cardiac valve defect during the key developmental stage of invagination

In this study, treatment of 15 ␮M Amiodarone only at the speci- fication stage during 36–55 hpf showed that both versican and cdh5 ectopically overexpressed at zebrafish embryonic cardiac AVC at 72 hpf. The results of paraffin sectioning/H&E staining also indi- cated cardiac valve defect. Yet, these results were similar to those of embryos treated with Amiodarone at the invagination stage after 55 hpf. Therefore, while Amiodarone treatment affects embryos as early as the specification stage, the invagination process is still affected by the long half-life of Amiodarone. Although the pro- cess of specification occurs during 36–55 hpf, this study based its conclusions on observations recorded at 72 hpf, which, in this case, allows only a 17 h window between treatment and obser- vation. However, the half-life of Amiodarone is 26–107 days [9].

In addition, one of the metabolites after Amiodarone is degraded, such as N-monodesethylamiodarone, still has bio-activity similar to Amiodarone [32]. Therefore, even though the drug is washed out at 55 hpf, there might still be some residual amount left in the treated embryos. Thus, the residual dosage of Amiodarone might be sufficient to continuously affect the invagination process in the embryos, even though treatment was begun at the specification stage.

4.6. Amiodarone-induced valve defect results from the overexpression of versican and cdh5 to suppress the process of invagination

In this study, zebrafish embryos treated with Amiodarone showed overexpression of versican at AVC. Furthermore, cdh5, an EMT-related gene, is also overexpressed at AVC (Fig. 6). Indeed, knockdown of cdh5 caused the endocardial cushion to grow larger than normal, suggesting that the reduced expression of cdh5 might result in valve enlargement (Fig. 7). On the other hand, Amiodarone treatment caused cdh5 to overexpress at AVC (Fig. 6B), resulting in valve defect in zebrafish embryos. However, this valve defect induced by Amiodarone treatment could be rescued by injection of cdh5-MO to knock down cdh5 expression. These results suggested that Amiodarone might cause zebrafish embryonic valve defect by greatly enhancing the expression of the otherwise repressive cdh5 gene. Furthermore, Mjaatvedt et al. [33] reported that versican is involved in the development of cardiac chambers and valves in a mouse model system, which strengthens our hypothesis that the abnormal expressions of cdh5 and versican cause valve defect in the zebrafish system by Amiodarone treatment, particularly during the critical invagination stage. To clarify the relationship between Ver- sican and Cdh5, versican-MO was injected to zebrafish embryos, and the results showed that both cdh5 and versican expressed in the same manner. Moreover, Versican knockdown could rescue the valve defect caused by Amiodarone (Fig. 8).

5. Conclusions

In summary, using zebrafish embryos as an experimental animal

model in this study, we determined that the EC

of Amiodarone is

15 ␮M and that the exposure time is between 34 and 58 h. We also

demonstrated that the heart defects are not observed in embryos

treated with lower concentrations of Amiodarone, such as 3 ␮M

and 6 ␮M. However, defective hearts characterized by failure of

valve formation, incomplete looping, blood regurgitation, pericar-

diac edema, and ventricular arrhythmia are dramatically increased

if embryos are treated with higher concentrations of Amiodarone, such as 9 ␮M, 12 ␮M and 15 ␮M. The rate of lethality exceeds 80% if embryos are treated with 30 ␮M Amiodarone. Under these condi- tions, women exposed to Amiodarone in early pregnancy might be at high risk for human embryonic cardiac valve defect. To decrease the rate of newborn congenital heart disease, this study might pro- vide cardiologists and obstetricians with a reference for monitoring cardiac septum and valve development of the fetus in instances where the mother has been treated with Amiodarone 2–3 months before or during pregnancy.

Ethical statement

This study was conducted according to the guidelines recom- mended by the National Laboratory Animal Center. The animal use protocol was approved by the Institutional Animal Care and Use Committee of National Taiwan University (Permit Number: No.

(96)-42, January-23-2008).

Acknowledgments

This work was supported by the National Science Council, Republic of China, under grant number NSC 100-2321-B-002-036, and partially supported by the NHRI funding with a number NHRI- EX101-9936EI.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.reprotox.2011.12.008.

References

[1] Heger JJ, Solow EB, Prystowsky EN, Zipes DP. Plasma and red blood cell concen- trations of amiodarone during chronic therapy. Am J Cardiol 1984;53:912–7.

[2] Singh BN. Amiodarone: historical development and pharmacologic profile. Am Heart J 1983;106:788–97.

[3] Mason JW. Amiodarone. N Engl J Med 1987;316:455–66.

[4] Helmy I, Herre JM, Gee G, Sharkey H, Malone P, Sauve MJ, et al. Use of intra- venous amiodarone for emergency treatment of life-threatening ventricular arrhythmias. J Am Coll Cardiol 1988;12:1015–22.

[5] Matsumura LK, Born D, Kunii IS, Franco DB, Maciel RM. Outcome of thy- roid function in newborns from mothers treated with amiodarone. Thyroid 1992;2:279–81.

[6] Basaria S, Cooper D. Amiodarone and the thyroid. Am J Med 2005;118:706–14.

[7] Valensise H, Civitella C, Garzetti GG, Romanini C. Amiodarone treatment in pregnancy for dilatative cardiomyopathy with ventricular malignant extra systole and normal maternal and neonatal outcome. Prenat Diagn 1992;12:

705–8.

[8] Plomp TA, Vulsma T, de Vijlder JJ. Use of amiodarone during pregnancy. Eur J Obstet Gynecol Reprod Biol 1992;43:201–7.

[9] Kaski JC, Girotti LA, Messuti H, Rutitzky B, Rosenbaum MB. Long-term man- agement of sustained, recurrent, symptomatic ventricular tachycardia with amiodarone. Circulation 1981;64:273–9.

[10] Stainier DY. Zebrafish genetics and vertebrate heart formation. Nat Rev Genet 2001;2:39–48.

[11] Huang J, Tu CT, Hsiao CD, Hsieh FJ, Tsai HJ. Germ-line transmission of a myocardium-specific GFP transgene reveals critical regulatory elements in the cardiac myosin light chain 2 promoter of zebrafish. Dev Dyn 2003;228:

30–40.

[12] Sun CK, Chu SW, Chen SY, Tsai TH, Liu TM, Lin CY, et al. Higher harmonic generation microscopy for developmental biology. J Struct Biol 2004;147:

19–30.

[13] Tsai TH, Lin CY, Tsai HJ, Chen SY, Tai SP, Lin KH, et al. Biomolecular imaging based on far-red fluorescent protein with a high two-photon excitation action cross section. Opt Lett 2006;7:930–2.

[14] Hsieh CS, Ko CY, Chen SY, Liu TM, Wu JS, Hu CH, et al. In vivo long term con- tinuous observation of gene expression in zebrafish embryos nerve systems by using harmonic generation microscopy and morphant technology. J Biomed Opt 2008;13:064041.

[15] Larson JD, Wadman SA, Chen E, Kerley L, Clark KJ, Eide M, et al. Expression of VE- cadherin in zebrafish embryos: a new tool to evaluate vascular development.

Dev Dyn 2004;231:204–13.

[16] Mitchell IC, Brown TS, Terada LS, Amatruda JF, Nwariaku FE. Effect of vascular cadherin knockdown on zebrafish vasculature during development. PLoS One 2010;5:e8807.

[17] Hsu RJ, Lin CY, Hoi HS, Zheng SK, Lin CC, Tsai HJ. Novel intronic microRNA represses zebrafish myf5 promoter activity through silencing dickkopf-3 gene.

Nucleic Acids Res 2010;38:4384–93.

[18] Hurlstone AF, Haramis AP, Wienholds E, Begthel H, Korving J, Van Eeden F, et al. The Wnt/beta-catenin pathway regulates cardiac valve formation. Nature 2003;425:633–7.

[19] Link V, Shevchenko A, Heisenberg CP. Proteomics of early zebrafish embryos.

BMC Dev Biol 2006;6:1.

[20] Hsu RJ, Lin CC, Su YF, Tsai HJ. dickkopf-3-related gene regulates the expres- sion of zebrafish myf5 gene through phosphorylated p38a-dependent Smad4 activity. J Biol Chem 2011;286:6855–64.

[21] Ranger AM, Grusby MJ, Hodge MR, Gravallese EM, de la Brousse FC, Hoey T, et al. The transcription factor NF-ATc is essential for cardiac valve formation.

Nature 1998;392:186–90.

[22] Walsh EC, Stainier DY. UDP-glucose dehydrogenase required for cardiac valve formation in zebrafish. Science 2001;293:1670–3.

[23] Camenisch TD, Schroeder JA, Bradley J, Klewer SE, McDonald JA. Heart-valve mesenchyme formation is dependent on hyaluronan-augmented activation of ErbB2–ErbB3 receptors. Nat Med 2002;8:850–5.

[24] Kodama I, Kamiya K, Toyama J. Cellular electropharmacology of amiodarone.

Cardiovasc Res 1997;35:13–29.

[25] Lampugnani MG, Resnati M, Raiteri M, Pigott R, Pisacane A, Houen G, et al. A novel endothelial-specific membrane protein is a marker of cell–cell contacts.

J Cell Biol 1992;118:1511–22.

[26] Hove JR, Köster RW, Forouhar AS, Acevedo-Bolton G, Fraser SE, Gharib M.

Intracardiac fluid forces are an essential epigenetic factor for embryonic car- diogenesis. Nature 2003;421:172–7.

[27] Thiery JP, Sleeman JP. Complex networks orchestrate epithelial-mesenchymal transitions. Nat Rev Mol Cell Biol 2006;7:131–42.

[28] Armstrong EJ, Bischoff J. Heart valve development: endothelial cell signaling and differentiation. Circ Res 2004;95:459–70.

[29] Beis D, Bartman T, Jin SW, Scott IC, D’Amico LA, Ober EA, et al. Genetic and cellular analyses of zebrafish atrioventricular cushion and valve development.

Development 2005;132:4193–204.

[30] Sheng W, Wang G, La Pierre DP, Wen J, Deng Z, Wong CK, et al. Versican mediates mesenchymal-epithelial transition. Mol Biol Cell 2006;17:2009–20.

[31] Timmerman LA, Grego-Bessa J, Raya A, Bertrán E, Pérez-Pomares JM, Díez J, et al. Notch promotes epithelial-mesenchymal transition during car- diac development and oncogenic transformation. Genes Dev 2004;18:99–

115.

[32] Kaski JC, Girotti LA, Messuti H, Rutitzky B, Rosenbaum MB. Long-term man- agement of sustained, recurrent, symptomatic ventricular tachycardia with amiodarone. Circulation 1987;64:273–9.

[33] Mjaatvedt CH, Yamamura H, Capehart AA, Turner D, Markwald RR. The Cspg2

gene, disrupted in the hdf mutant, is required for right cardiac chamber and

endocardial cushion formation. Dev Biol 1998;202:56–66.