Amiodarone Induces Overexpression of Similar to Versican b to Repress the EGFR/Gsk3b/Snail Signaling Axis during Cardiac Valve Formation of Zebrafish Embryos

Although Amiodarone, a class III antiarrhythmic drug, inhibits zebrafish cardiac valve formation, the detailed molecular pathway is still unclear. Here, we proved that Amiodarone acts as an upstream regulator, stimulating similar to versican b (s-vcanb) overexpression at zebrafish embryonic heart and promoting cdh-5 overexpression by inhibiting snail1b at atrioventricular canal (AVC), thus blocking invagination of endocardial cells and, as a result, preventing the formation of cardiac valves. A closer investigation showed that an intricate set of signaling events ultimately caused the up-regulation of cdh5. In particular, we investigated the role of EGFR signaling and the activity of Gsk3b. It was found that knockdown of EGFR signaling resulted in phenotypes similar to those of Amiodarone-treated embryos. Since the reduced phosphorylation of EGFR was rescued by knockdown of s-vcanb, it was concluded that the inhibition of EGFR activity by Amiodarone is s-vcanb-dependent. Moreover, the activity of Gsk3b, a downstream effector of EGFR, was greatly increased in both Amiodarone-treated embryos and EGFR-inhibited embryos. Therefore, it was concluded that reduced EGFR signaling induced by Amiodarone treatment results in the inhibition of Snail functions through increased Gsk3b activity, which, in turn, reduces snail1b expression, leading to the up-regulation the cdh5 at the AVC, finally resulting in defective formation of valves. This signaling cascade implicates the EGFR/Gsk3b/Snail axis as the molecular basis for the inhibition of cardiac valve formation by Amiodarone.


Introduction
, NFAT [23], Wnt [24] and ErbB [25] signaling, as reported in mouse, suggesting that the major cellular and molecular events of cardiac valve development in zebrafish are largely conserved to amniotes.
Amiodarone is an effective and widely used anti-arrhythmic drug. Cardiac arrhythmia is also common in pregnant women, and the toxic effect of Amiodarone on embryogenesis has been controversial and is an important issue. We previously reported that Amiodarone inhibits zebrafish cardiac valve formation by inducing the expression of versican and chromodomainhelicase-DNA-binding protein 5 tumor suppressor gene (cdh5; VE-cadherin) at the AVC of zebrafish embryos [26]. Since cdh5 can be overexpressed by Amiodarone, this finding suggested that Amiodarone may interfere with EMT processes during cardiogenesis. However, compared to mammalian Versican, zebrafish Versican is largely unknown, except for its restriction to the AVC endocardium by 48 hpf [24]. Furthermore, we do not know the molecular mechanism underlying Amiodarone's inhibition of cardiac valve formation. Here, our results implicated the EGFR/Gsk3b/Snail signaling axis as the molecular basis for Amiodarone's inhibition of cardiac valve formation. In brief, Amiodarone induces ectopic expression of only one Versican isoform, similar to versican b (s-vcanb), which represses EGFR activity, thereby increasing Gsk3b activity, which, in turn, down-regulates the expression of snail1b, leading, finally, to the up-regulation of cdh5 at the AVC and improper development of cardiac valves of zebrafish heart.

Material and Methods
Zebrafish husbandry, animal studies, and microscopy observation Zebrafish AB strain was maintained in aquaria according to standard procedures described by Westerfield et al. [27]. The experiments and treatments of this animal have been reviewed and approved by the National Taiwan University Institutional Animal Care and Use Committee with ethics approval number NTU-101-EL-115. Fluorescence was visualized with a fluorescent stereomicroscope (MZ FLIII, Leica) and a confocal spectral microscope (TCS SP5, Leica).

Drug treatment with zebrafish embryos
EGFR inhibitor AG 1478 (CalBiochem) was dissolved in DMSO and stocked as 1 mM at -20°C. The working concentration used in this study was 7 μM. Amiodarone (Sigma) was dissolved in water at 65°C for 2 h and stocked as 900 μM at 4°C. Before use, the solution was dissolved again 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 experimental 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 μM, and embryos were exposed to treatment from 12 to 84 h. Long-term treatment was performed during 12-72 hpf, in which the specification stage of valve formation at 36-55 hpf and the invagination stage of valve formation at 55 hpf were included. Treatment during 12-48 hpf was used to examine the expressions of versican and cdh5 genes. 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.

Plasmid construction
The zebrafish s-vcanb contains a 5-kb coding region. Therefore, we used eight primers to amplify four fragments containing s-vcanb. The first 1.5 kb s-vcanb fragment was amplified by Z-s-vcanb ClaI F1 and Z-s-vcanb MfeI R1; the 1.5 to 2 kb fragment was amplified by Z-s-vcanb MfeI F2 and Z-s-vcanb HpaI R2; the 2-3.5 kb fragment was amplified by Z-s-vcanb HpaI F3 and Z-s-vcanb XbaI R3; and the 3.5-5 kb fragment was amplified by Z-s-vcanb XbaI F4 and Zs-vcanb AfeI R4. These four DNA fragments were ligated using their specific restriction enzymes' cutting sites. Finally, the full length zebrafish s-vcanb was inserted into pCS2+ vector. The zS-vcanb-mE was previously described in Lee et al. [31]. To prove the specific effectiveness of vcana-MO, we designed a synthetic aMO-target-eGFP mRNA, in which the vcana cDNA, including vcana-MO target sequence, is fused in frame with egfp cDNA.

Whole-mount in situ hybridization (WISH)
To examine the expression patterns of these three versican family genes, we designed specific probes to perform WISH. For example, the coding region 1.0-2.0 kb of vcana cDNA served as a vcana probe, and the coding region 3.3-3.9 kb of vcanb cDNA served as a vcanb probe, while the coding region 0.01-1.5 kb of s-vcanb served as s-vcanb probe. The full length coding sequence of snail1a, snail1b, snail2, and snail3 were isolated by RT-PCR, inserted into plasmid pGEMTeasy (Promega) and confirmed by sequencing. After we cloned the partial DNA fragments of the desired genes, the probes were labeled by Digoxigenin (DIG). After permeabilization, embryos were hybridized overnight. Then, embryos were incubated with anti-DIG antibody (Roche; 1:8,000), stained and observed under a fluorescent stereomicroscope (MZ FLIII, Leica). Numbers shown in the bottom corner indicated number of embryos with similar staining patterns out of total number of embryos examined. All the primers used in this study were listed in S1 Table.

Statistical analysis
Statistical analysis of the intensity of band shown on Western blot was averaged from three independent experiments, and presented as mean ± S.D., and difference levels were analyzed using Student's t-test. ÃÃ P<0.01 and ÃÃÃ P<0.005 indicated the levels of significant difference.

Results
Amiodarone causes overexpression of s-vcanb resulting in decreased snail1b and, consequently, increased cdh5 in zebrafish embryos Since Amiodarone can induce cdh5 transcripts at the AVC [26], the Snail family of proteins, which functions as transcriptional repressors of the cdh5 gene [32], may also be influenced by Amiodarone. To confirm this, four members of the Snail family of repressors were analyzed in zebrafish for their expression pattern using WISH. Only snail1b was expressed in the AVC ( Fig  1C'). When embryos were treated with Amiodarone, snail1b transcripts were lost at the AVC (Fig 1D'). Meanwhile, knockdown of snail1b by specific MO increased cdh5 expression at the AVC ( Fig 1I). Western blot analysis also demonstrated that the protein level of Cdh5 was increased in the Amiodarone-treated embryos and the snail1b-MO-injected embryos ( Fig 1J and  S2A Fig), but reduced in snail1b-mRNA-injected embryos. Importantly, it was found that overexpression of snail1b in Amiodarone-treated embryos reduced the Cdh5 protein level ( and S2A Fig), which was corresponding with that the reduced Cdh5 protein level was observed in the overexpression of snail1b alone in embryos. Thus, we suggested that Amiodarone causes the ectopic expression of cdh5 through its repression of snail1b expression.
It has been reported that the ectopic expression of cdh5 caused by Amiodarone treatment is, in fact, dependent on versican overexpression [26]. Moreover, as determined by the knockdown experiments above, the repressor snail1b is lost at the AVC, resulting in increasing cdh5 expression. Therefore, it can be concluded that the effects of Amiodarone on snail1b/cdh5 signaling are dependent on versican overexpression [26]. Based on the NCBI database, the Versican family genes in zebrafish include three isoforms: vcana, vcanb and s-vcanb. The vcana gene is located at chromosome 5 and usually serves as a paraxial mesoderm marker during early gastrulation [33] and an AVC marker during heart development [20,24]. The vcanb gene, located at chromosome 10, plays roles on dermal bone development [34]. Cheng et al. [35] reported that HNF factor could induce vcanb to express in liver tissue. Many previous studies have focused on vcana and vcanb genes. However, the s-vcanb gene is understudied in the literature.
To examine the expression patterns of these three versican family genes and to further determine which type of Versican is affected by Amiodarone, we performed WISH and found that only vcana (Fig 2A) and s-vcanb ( Fig 2E) were expressed at the AVC during zebrafish cardiac development. Furthermore, Amiodarone-treated embryos displayed ectopic expression of vcana ( Fig 2B) and s-vcanb (Fig 2F) at the AVC. The vcanb was not detected at the AVC during 72 hpf (Fig 2C), and its expression pattern remained unchanged after Amiodarone treatment ( Fig 2D).
Interestingly, however, knockdown of vcana by injection of vcana-specific MO (aMO) did not alter the expression of snail1b (Fig 2J vs. 2H) or cdh5 (Fig 2N vs. 2L), whereas s-vcanb-MO (vMO)-injected embryos exhibited increased expression of snail1b (Fig 2R vs. 2P), but greatly reduced cdh5 expression (Fig 2V vs. 2T). Additionally, knockdown of vcana in Amiodaronetreated embryos did not rescue the expression patterns of snail1b (Fig 2K vs. 2H) or cdh5 ( Fig  2O vs. 2L) at the AVC, whereas knockdown of s-vcanb blocked the effects of Amiodarone because snail1b was still present (Fig 2S vs. 2P), and cdh5 was not ectopically expressed (Fig 2w  vs. 2T). Western blot analysis also showed that the expression of S-Vcanb protein was increased in the Amiodarone-treated embryos, but it was reduced when s-vcanb was knocked down ( Fig  2G and S2B Fig), indicating that Amiodarone is able to induce s-vcanb overexpression, and Amiodarone effects on cdh5 overexpression is dependent on overexpressive s-vcanb, as stated above.

The EGF motif of S-vcanb is involved in regulating snail1b and cdh5 expression
We employed PCR-based in vitro mutagenesis and transgenic assays to test whether the EGF motif of S-Vcanb is involved in Amiodarone-mediated inhibition of zebrafish cardiac valve development. The zebrafish S-Vcanb EGF motif contains eight defined cysteine residues to form specific disulfide bridges responsible for the secondary structure. Following the report of Schrijver et al. [36], we mutated eight cysteine residues into arginine to disrupt the specific disulfide bridges (Fig 3A) and termed the resulting mutated protein as S-Vcanb-mE. Plasmid containing wild-type (WT) S-Vcanb or S-Vcanb-mE, which is driven by CMV promoter, was injected into one-celled embryos and analyzed at 72 hpf. The s-vcanb signal was ectopically expressed at the AVC of embryos injected with WT S-Vcanb (Fig 3E vs. 3B), indicating that the injected plasmid DNA was transcribed in the cardiac cells. Similar to the pattern shown in the Amiodarone-treated embryos (Fig 3C, 3G and 3K), snail1b expression was absent (Fig 3I), while, at the same time, ectopic expression of cdh5 (Fig 3M) was observed in the s-vcanb-overexpressed embryos, suggesting that the effects of Amiodarone on cardiac valve development are S-Vcanb-dependent. At the AVC of embryos injected with mutated S-Vcanb-mE, s-vcanb signaling was still ectopically expressed (Fig 3D), and snail1b was also expressed (Fig 3H), while cdh5 expression was limited to the AVC with no ectopic distribution (Fig 3L). Western blot analyses confirmed our WISH data in which the level of Snail1b protein was reduced both in the Amiodarone-treated embryos and s-vcanb-overexpressed embryos, while it remained unchanged in the s-vcanb-mE-injected embryos. Additionally, the level of Cdh5 protein was increased in both the Amiodarone-treated embryos and s-vcanb-overexpressed embryos, while it remained unchanged in the s-vcanb-mE-injected embryos (Fig 3N and S2C Fig). Since neither snail1b expression nor cdh5 expression was affected in S-Vcanb-mE-injected embryos, the EGF motif of S-Vcanb is involved in regulating the expressions of Snail1b and Cdh5.

Role of EGFR signaling
To prove that EGFR signaling regulates Snail1b and Cdh5 during zebrafish embryogenesis, we treated zebrafish embryos at 55 to 72 hpf with an EGFR inhibitor, AG1478, at a concentration of 7 μM. Results showed that the expression of snail1b was absent (Fig 4B vs. 4A) and that the expression of cdh5 was ectopically expressed (Fig 4D vs. 4C) at the AVC. These phenotypes were similar to those of Amiodarone-treated embryos and s-vcanb-overexpressed embryos. While Amiodarone was able to increase EGFR dimerization (Fig 4E), we noted that the phosphorylation of EGFR was reduced in Amiodarone-treated embryos in a dose-dependent manner (Fig 4F). The pEGFR(Y845) was reduced 79% in embryos treated with 30 uM Amiodarone, while it was reduced 52% in embryos treated with 15 uM Amiodarone. Importantly, knockdown of s-vcanb could rescue the reduced EGFR phosphorylation caused by In Amiodarone-treated embryos, s-vcanb (C) and cdh5 (K) were ectopically expressed, but snail1b (G) was lost. In embryos overexpressed in the mutated form of s-vcanb-mE, s-vcanb (D) was increased, but snail1b (H) and cdh5 (L) levels were similar to those of control group (F and J). In embryos overexpressed with the wild-type form of s-vcanb, s-vcanb was increased, but snail1b was lost, while cdh5 was greatly ectopically expressed. (N) Western blot analysis of Snail1b and Cdh5 in wild-type embryos, 72 hpf zebrafish embryos treated with 15 μM Amiodarone from 55 to 72 hpf (15 A), overexpressed mutation form of s-vcanb-mE, and wild-type s-vcanb embryos. Overexpressive wild-type s-vcanb embryos displayed reduced Snail1b and increased Cdh5 patterns. The number of embryos displaying a similar pattern was indicated in each figure. The relative intensities of each protein were as indicated. The α-tubulin was used as an internal control. Amiodarone, whereas knockdown of vcana could not (Fig 4G and S2D Fig), suggesting that inhibition of EGFR activity by Amiodarone is dependent on s-vcanb expression.
To further investigate the molecular mechanism(s) underlying the effects of Amiodarone on EGFR activity, we carried out a detailed analysis of the downstream effectors of EGFR signaling, in particular Gsk3b which can phosphorylate Snail1b, leading to its degradation. While the total protein level of Gsk3b remained unchanged in WT, Amiodarone-treated and EGFRinhibited embryos, its activity was greatly increased in both Amiodarone-treated and EGFRinhibited embryos (Fig 4H and S2E Fig). However, knockdown of s-vcanb could block the effects of Amiodarone such that Gsk3b activity remained unchanged. In contrast, knockdown of vcana caused an increase in Gsk3b activity (Fig 4H and S2E Fig). Reduced EGFR signaling then inhibits Snail functions through increased Gsk3b activity, thereby up-regulating Cdh5 at the AVC. Knockdown of vcana in Amiodarone-treated embryos did not rescue the level of Snail1b or Cdh5, whereas knockdown of s-vcanb blocked the effects of Amiodarone because the protein levels of Snail1b and Cdh5 were not changed compared to Wt embryos (Fig 4H  and S2E Fig). This line of evidence suggests that Amiodarone induces ectopic expression of S-Vcanb, which, in turn, interacts with EGFR by its EGF motif, resulting in reducing EGFR signaling, then inhibits Snail functions through increased Gsk3b activity, finally up-regulating Cdh5 at the AVC (Fig 5).

Discussion
In this study, we originally found that Amiodarone treatment causes the inhibition of zebrafish cardiac valve formation, hypothesizing that the EMT processes in cardiac endocardium might be repressed. Upon further investigation, we demonstrated that Amiodarone induces ectopic expression of s-vcanb at the AVC, resulting in the inhibition of EGFR/Gsk3b/Snail signaling axis and, finally, the increase of Cdh5. We also demonstrated that that s-vcanb, but not vcana, affects the expression of downstream genes in endocardium. The expression patterns of snail1b and cdh5 remained unchanged in the heart of embryos injected with dominant negative svcanb, which is absent of EGF-like domain's function, indicating that s-vcanb in myocardial cell affected the expression of snail1b and cdh5 in endocardial cells through EGF-like domain. Furthermore, we found that Amiodarone increases EGFR dimers formation, but decreases EGFR phosphorylation. The phosphorylation of Gsk3b decreases in the Amiodarone-treated embryos, but it did not change in the Amiodarone-treated embryos injected with s-vcanb-MO, indicating Amiodarone-mediated Gsk3b activity is s-vcanb-dependent. Taken together, our results suggested that Amiodarone represses EGFR/Gsk3b/Snail signaling, which, in turn, upregulates cdh5 at the AVC, and causes the defective formation of cardiac valves. Each Versican isoform plays a separate and distinct role during cardiac valve development Versican belongs to a family of hyaluronan aggregating chondroitin sulfate proteoglycans [37,38], which are highly expressed during endocardial cushion formation [39,40]. In transgenic mice carrying a homozygous insertional mutation to disrupt the versican gene, the endocardial cushion fails to become populated with mesenchymal cells [41]. Since the ability of Versican to influence cellular proliferation has been reported [15,16,38,42,43], it was thought that Versican acts as a positive effector for the growth of cushion mesenchymal cells derived from EMT. Many studies also demonstrated that different isoforms of Versican could have opposing functions. For example, V1 isoform promotes neurite outgrowth [15], whereas V2 isoform inhibits axonal and neurite growth [14]. This functional "balance" among isoforms was also supported by Sheng et al. [16], who demonstrated that the V1 isoform is able to enhance cell proliferation, while the V2 isoform inhibits cell proliferation. In the present study, we showed, for the first time, that the inhibition of EMT processes of zebrafish cardiac valve formation caused by Amiodarone results from the ectopic expression of s-vcanb, a negative VCAN isoform effector which plays an important role during the process where endocardial cushion becomes populated with mesenchymal cells. Collectively, we proposed that the dynamically balanced expressions of two Versican isoforms may provide a suitable extracellular environment for normal proliferation and differentiation of valve primordial cells.

Zebrafish S-vcanb inhibit EGFR signaling through G3 and CSα domains
It is well known that Versican possesses one chondroitin sulfate (CS)-attachment domain and two globular domains, which are the N-terminal G1 domain and the C-terminal G3 domain. Both V1 and V2 isoforms contain G1 and G3 domains, whereas V1 isoform possesses CSβ and V2 isoform possesses CSα. The G1 domain of Versican can interact with Hyaluronic acid (HA), allowing the cells to regulate cellular responses through the hyaluronan receptors in coordination with CD44 [44]. Then, CD44 associates with the Erbb family and activates the EGFR signaling [45,46]. The G3 domain contains two EGF motifs which can interact with EGFR [38,47,48]. Sheng et al. [16] proved that CSβ domain is responsible for the activation of EGFR signaling; but the CSα domain is responsible for the suppression of EGFR signaling. Therefore, each domain of Versican may have potential to influence EGFR signaling. In this report, we found that repression of EGFR signaling by S-vcanb is dependent on its EGF motif. The function of S-vcanb on repression of EGFR signaling suggests that zebrafish S-vcanb may be functionally conserved to mammalian Versican V2, which also displays negative effect on EGFR activity [16]. Together, we suggested that the repression of EGFR signaling by S-vcanb is dependent on CSα domain of EGF motif. However, the role of G1 domain on zebrafish Svcanb requires further study.

Clinical implications
The toxic effect of Amiodarone on cardiac valve development has been well proven in the present study, with the knowledge of its detailed molecular signaling mechanism. Therefore, it can be reasonably concluded that Amiodarone should be avoided in the pregnant women, especially those in the first trimester.  Figs 1J, 2G, 3N, 4G and 4H were densitometrically quantified and performed statistical analysis using Student's t-test, which were illustrated in panels A, B, C, D and E, respectively. Data are presented as mean±SD. ÃÃ P<0.01 and ÃÃÃ P <0.005 indicated the levels of significant difference. (DOCX) S1