A Variant of Fibroblast Growth Factor Receptor 2 (Fgfr2) Regulates Left-Right Asymmetry in Zebrafish

Many organs in vertebrates are left-right asymmetrical located. For example, liver is at the right side and stomach is at the left side in human. Fibroblast growth factor (Fgf) signaling is important for left-right asymmetry. To investigate the roles of Fgfr2 signaling in zebrafish left-right asymmetry, we used splicing blocking morpholinos to specifically block the splicing of fgfr2b and fgfr2c variants, respectively. We found that the relative position of the liver and the pancreas were disrupted in fgfr2c morphants. Furthermore, the left-right asymmetry of the heart became random. Expression pattern of the laterality controlling genes, spaw and pitx2c, also became random in the morphants. Furthermore, lefty1 was not expressed in the posterior notochord, indicating that the molecular midline barrier had been disrupted. It was also not expressed in the brain diencephalon. Kupffer's vesicle (KV) size became smaller in fgfr2c morphants. Furthermore, KV cilia were shorter in fgfr2c morphants. We conclude that the fgfr2c isoform plays an important role in the left-right asymmetry during zebrafish development.


Introduction
The bodies of most adult animals have left-right symmetry. However, some organs are not symmetrical, including the heart, liver, spleen, stomach, and pancreas [1]. When laterality is disrupted, many defects can result, such as abnormal position of organs, skeletal malformation and failure of neural tube closure [2]. Mechanisms involved in the regulation of laterality for various animal species have been identified. These include motor proteins, ion channel, cytoskeleton, serotonin, cell-cell junction, Ca 2+ , and cilia [3]. For example, in the mouse, the leftward movement of fluid at the ventral node, called nodal flow, is the critical process for left-right asymmetry [4]. The nodal flow is generated by the clockwise rotation of nodal cilia. This directional flow causes some morphogens to concentrate at left side of the node and leads to left-right polarization [5,6].
In zebrafish, Kupffer's vesicle (KV) is equivalent to the mouse node and is important for left-right development [15]. KV is a fluid-filled ciliated organ. Recent results indicate that most cilia are located on the dorsal side and are distributed along the anteriorposterior axis unequally [16]. Interestingly, unlike the leftward fluid flow in mice, fluid flow generated in the KV of zebrafish has a counter-clockwise rotation [17]. Recently, Fgf signaling regulation of laterality has been reported in zebrafish [18,19,20]. In ace/fgf8 mutant, the asymmetric visceral organs and the proper symmetric craniofacial skeleton are disrupted. Furthermore, the KV morphogenesis is defective in ace/fgf8 mutant fish [18]. Moreover, Fgf signaling can regulate the length of cilia through the Fgf8/ Fgf24-Fgfr1 pathway [20]. The downstream effectors of Fgf signaling, Ier2 and Fibp1, are also identified in the process of KV ciliogenesis [19].
We studied the role of Fgfr2 in liver development (manuscript in preparation). Unexpectedly, we detected that left-right asymmetry of visceral organs was randomized in the different fgfr2 morphants, especially for the fgfr2c. Furthermore, normal heart jogging and looping were disrupted in the morphants. The expression of specific left-sided genes, such as spaw, pitx2c and lefty1, was affected in fgfr2c morphants. The expression of spaw and pitx2c was randomized in left lateral plate mesoderm (LPM) of fgfr2c morphants. However, the expression of lefty1 was absent in most fgfr2c morphants. Furthermore, we found that ciliogenesis was defective in fgfr2c morphants: the cilia length was shorter in fgfr2c morphants. This phenomenon was similar in fgfr1, fgf8, and fgf8/ fgf24 morphants. These results suggest that Fgfr2c is important in the regulation of left-right asymmetry.

Results
Visceral Organ Laterality was Affected by fgfr2 Inhibition Fgf signaling pathways have been proposed to regulate liver specification [21]. However, the critical Fgf receptor(s) that participate in this process have not been fully characterized. We found the expression of earliest marker for developing liver, hematopoietically expressed homeobox gene (hhex), was absented in more than half of fgfr2 morphants (data not shown).
Unexpectedly, the disruption of left-right asymmetry in the fgfr2-ATG morphants was noticed from the expression pattern of foxA3 at 48 hours post fertilization (hpf). In 94.1% (n = 101) of the wild type embryos, the liver bud was located on the left side and the pancreatic bud was located on the right side in zebrafish ( Fig. 1A  and 1O). In 5.9% of the wild type embryos, abnormal left-right asymmetry was observed, and the relative locations of liver and pancreas were reversed ( Fig. 1E and 1O). In fgfr2-ATG morphants, the left-right pattern was affected, and the abnormal percentages increased with dosages ( Fig. 1B, C, F, G, and 1O; 8 ng/egg: 14.6% abnormal embryos, n = 41; 16 ng/egg: 30.6% abnormal embryos, n = 49). These results were confirmed by a splice-blocking MO, fgfr2-I4E5 MO (Fig. 1D, 1H, and 1O; 29.6% abnormal embryos, n = 27). In order to verify which fgfr2 variant controlled left-right asymmetry, we designed specific splicing blocking MOs that targeted fgfr2b and fgfr2c, respectively. According the cDNA sequence data (AB094118 and AB083105), isoform-specific exonic regions could be identified. We further confirmed these regions in fgfr2b and fgfr2c were exon8 and exon9, respectively, according to annotated zebrafish Zv9 assembly. In order to specifically inhibit the splicing of fgfr2b and fgfr2c, morpholino target sites were located at E8I8 (fgfr2b) and E9I9 (fgfr2c). The specificity was verified by RT-PCR analysis and sequencing (supplementary Fig. S1). The predicted translated products of fgfr2b and fgfr2c had in frame deletions of 13 and 17 amino acids, respectively. The deleted region of Fgfr2c consists of two critical amino acid residues (I350 and Y352) Figure 1. The left-right asymmetry of visceral organs was randomized in fgfr2 morphants. The expression pattern of foxA3 in liver (l) and pancreas (p) was shown in wild type, fgfr2-ATG morphants, fgfr2-I4E5 morphants, fgfr2c morphants, fgfr2b and fgfr2c-5 mm morphants (A,D, I,K). Abnormal pattern of reverse visceral organs was also observed in these embryos (E,H, L,N). All pictures were dorsal view. The bar charts showed the percentage of left-right asymmetry of visceral organs (O). doi:10.1371/journal.pone.0021793.g001 which form hydrophobic groove to interact with Fgf ligands [22]. We found 29.6% (n = 375) of embryos to be abnormal in fgfr2c morphants (Fig. 1I, 1L, and 1O). However, only 7.4% (n = 27) of embryos were observed to be abnormal in fgfr2b morphants (Fig. 1J, 1M, and 1O). The laterality of visceral organ was not affected in fgfr2c 5-base mismatch (fgfr2c-5 mm) morphants (Fig. 1K, 1N and 1O; 3% abnormal embryos, n = 133). The phenotype in fgfr2c morphants could be rescued with morpholino-resistant fgfr2c mRNA (16.9% abnormal embryos, n = 118). These results indicate that fgfr2c is the major fgfr2 isoform that regulates the left-right pattern of visceral organs.

Heart Laterality was Randomized by fgfr2 Inhibition
Since the laterality of visceral organs was affected in fgfr2c morphants, we wanted to analyze whether the left-right asymmetry of developing heart was also randomized. We examined heart jogging at 30 hpf and looping at 48 hpf using Line 544 (cmlc2:EGFP/b-actin2-mCherry) fish, in which GFP was specifically expressed in heart. Heart jogging occurred after heart-tube formation, and heart looping took place following heart jogging. In 90.2% (n = 41) of the un-injected transgenic line 544 embryos we observed, heart corn migrated toward the left-anterior and formed heart tube (L-jog, Fig. 2A and 2R), and in 92.7% (n = 41) of the uninjected line544 embryos we observed, the atrium was located at the left side of ventricle (D-loop, Fig. 2J and 2R). When laterality was disrupted, the direction of heart jogging became random in 9.8% of the un-injected transgenic embryos (heart corn migrated toward the right-anterior, R-jog; heart corn migrated toward mid line, mid-jog, Fig. 2D, 2G, and 2R), and in 7.3% of the un-injected transgenic embryos, the location of the atrium turned toward the right side of Figure 2. The laterality of heart jogging and looping was randomized in fgfr2c morphants. The development of heart was followed using Line 544 (cmlc2:EGFP/b-actin2-mCherry) transgenic fish. Normal direction of heart jogging was toward left side (A,C, L-jog). Randomization resulted in abnormal patterns of jogging (D,I, R-jog and mid-jog). Normal heart looping (J,L, D-loop) and abnormal heart looping (M,Q, L-loop and noloop) were detected in un-injected transgenic line 544 embryos, fgfr2c and fgfr2c-5 mm morphants. All pictures were ventral-anterior view. The bar charts showed the percentage of different types of heart jogging and looping (R). A: atrium, V: ventricle. Left-right axis was indicated as labeled. doi:10.1371/journal.pone.0021793.g002 ventricle (L-loop, Fig. 2M and 2R). In fgfr2c morphants, the numbers of embryo with R-jog or mid-jog increased to 49% (n = 51) ( Fig. 2E, 2H, and 2R). Furthermore, 60% (n = 45) of fgfr2c morphants exhibited abnormal heart looping (L-loop or no-loop, Fig. 2N, 2P, and 2R). Embryos injected with fgfr2c-5 mm MO were relatively normal in heart jogging ( Fig. 2C and 2R; 88.2% L-jog) and heart looping ( Fig. 2L and 2R; 86.2% D-loop). According to these results, we conclude that Fgfr2c signaling is required for leftright pattern of heart in zebrafish.
It's known that foxj1 is important for the motile ciliogenic program [25]. Because the left-right asymmetry is also randomized in foxj1 morphants [24], we analyzed the expression of foxj1a in 90% epiboly embryos. In wild type embryos, foxj1a was normally expressed in the KV ( Fig. 4I and I9, 73%, n = 30). Some embryos (17%) had a scattered expression pattern of foxj1a (Fig. 4J and J9), and some (10%) had a reduced signal (Fig. 4K and K9). In fgfr2c morphants, the number of embryos with normal expression pattern of foxj1a was slightly reduced (Fig. 4L and L9, 63%, n = 30). The morphants with abnormal pattern, including scattered expression patterns, reduced signal, and no signal, were increased to 37% (Fig. 4M,4O). Consistent to our finding, a recent study also indicated that the expression of foxj1 was downregulated in fgfr1 morphants. Accordingly, cilia length was also reduced in fgfr1 morphants [20]. These results indicate that the reason for the reduced cilia length of fgfr2c morphants was due to affected expression of foxj1a.

Discussion
In this study, we found that the orientation of asymmetric organs was randomized after knocking down Fgfr2. We further identified fgfr2c as the main fgfr2 variant that regulates the leftright asymmetry. Expression patterns of spaw, pitx2c and lefty1 were abnormal in fgfr2c morphants. These results indicate that the molecular midline barrier was disrupted and further affects the asymmetric expression of spaw and pitx2c. Importantly, the cilia length was reduced in the KV of fgfr2c morphants.
In mice, chickens, and rabbits, left-right asymmetry requires FGF8 [9,10,27]. In zebrafish, Fgf8 signaling can regulate morphogenesis of the KV. KV is lost in about 30% of ace mutant embryos, and the laterality of visceral organs, including heart and brain, is also disordered [18]. Recently, some evidence has indicated that Fgf signaling could regulate ciliogenesis in the KV to determine left-right asymmetry [19,20,28]. In the KV of fgfr1 morphants, length of cilia is shorter. This results from the reduction of ciliogenic transcription factors, foxj1 and rfx2, and intraflagellar transport gene ift88 [20]. In ace mutant embryos, the length of cilia is not affected. However, the cilia length is reduced in ace mutant embryos injected with fgf24 MO [20]. These results suggest that fgf8, fgf24, and fgfr1 are important for ciliogenesis. In the other hand, two Fgf8 signaling target genes, ier2 and fibp1, have been identified [19]. The cilia number in KV of ier2 and fibp1 morphants is also reduced. When ier2 and fibp1 mRNA are injected in fgf8 morphants, the cilia number is restored. Therefore, these two genes can mediate Fgf8 signaling in ciliogenesis and are essential for the establishment of laterality. A recent report indicates that Fgf4 signaling is important for left-right asymmetry [28]. The left-right asymmetry of visceral organs and heart are randomized in fgf4 morphants. Furthermore, the expression of lefty1 is absent in the posterior notochord, and the cilia length is reduced despite normal quantities of cilia in fgf4 morphants. In our studies, reducing cilia length rather than cilia number might result in randomizing the left-right asymmetry in fgfr2c morphants. In order to verify which Fgf ligands regulate left-right asymmetry through Fgfr2c, we analyzed the possible synergistic effect of Fgfr2c and the above three mentioned ligands. We examined the heart looping in different low dosage MO combinations, including fgfr2c-fgf4, fgfr2c-fgf8 and fgfr2c-fgf24 ( Supplementary Fig. S4). These preliminary results showed Fgfr2c and Fgf ligands did not have obviously synergistic effect except Fgfr2c-Fgf8. So we suggested that Fgfr2c could functionally interact with Fgf8, whereas Fgf4 and Fgf24 were parallel pathways with Fgfr2c for left-right asymmetry. The disruption of laterality in DFC fgf24 MO morphants is not known, whereas left-right asymmetry of visceral organs and the heart was affected in DFC fgfr2c MO but not in DFC fgf4 MO morphants [28]. This observation also supports the independence of Fgfr2c and Fgf4 signaling. . KV formation is very important for left-right asymmetry. The cellular origin of KV is DFCs which migrate at the leading edge of the blastoderm margin [29]. When DFCs is ablated by laser or the KV morphogenesis is disrupted, the expression pattern of leftright asymmetry genes, including spaw, lefty1 and lefty2, becomes random [15,30]. In addition to shorter cilia length, the morphology of KV is changed and its area is reduced significantly in fgfr2c morphants (Fig .4F and data not shown). This phenomenon has not been reported in fgfr1 and fgf4 morphants [20,28]. To investigate whether the reduced KV area was due to changes of DFC numbers in fgfr2c morphants, we used casanova (cas) probe to highlight DFCs. Preliminarily we found that the number of cas expressed cells was not reduced in the morphants compared to wild type embryos (33.860.8 cells in fgfr2c moprhants, n = 195, and 33.460.9 cells in wild type, n = 140; P = 0.7738). Whether the cell size is affected in the morphants needs to be further examined. Notably, we did find the DFC morphology was obviously different in fgfr2c morphants (Supplementary Fig. S5). So, disorganized DFC pattern may cause defects of KV formation. For Fgf related genes in laterality, the ier2 and fibp1 have been indicated to affect KV formation starting at the time of DFCs formation [19]. Since these two genes mediate Fgf8 signaling in left-right asymmetry patterning and Fgf8 and Fgfr2c signaling could have functional interaction, we suggest that Fgf8, Fgfr2c and Ier2/Fibp1 may be the same pathway to regulate DFC patterning. Further examination of fgfr2c MO specifically targets to DFCs also reveals the disruption of laterality of visceral organs and heart ( Supplementary Fig. S3). Therefore, we suggest that Fgfr2 may function cell-autonomously in KV to regulate the organization of DFCs during the laterality establishment. However, the detailed mechanism remains unclear. In addition to KV morphogenesis, the expression of lefty1 in midline is also important for the left-side expression of spaw in LPM. In this study, we found that the expression of lefty1 in midline was absent in fgfr2c morphants. The abnormal expression of spaw in the LPM of fgfr2c morphants could be due to the loss of lefty1 in midline. Taken together, we conclude that Fgfr2c signaling controls left-right patterning through regulating the cilia length and controlling the expression of lefty1 to set up a molecular midline barrier. These suggest that Fgfs have multiple roles in left-right patterning.

Ethics Statement
All embryos were handled according to protocols approved by the Institutional Animal Care and Use Committee of Tzu Chi University, Hualien, Taiwan (approval ID: 97062).

Zebrafish
The zebrafish (Danio renio) were raised as described in the Zebrafish Book [31]. The AB wild type strain was used for morpholino injection and other experiment. Line 544 (cmlc2:EGFP/ b-actin2-mCherry) was generated by Dr. Chung-Der Hsiao.

Plasmid Construction
Tol2 kit was used to rapidly assemble expression vectors by twofragment gateway recombination cloning. The p5E-b-actin2 59 entry clone contains 5.3 kb upstream regulatory sequences of b-actin2 gene that sufficient to target transgene ubiquitously express. The pME-mCherry middle entry clone contains mCherry fluorescent reporter gene. The p3E-polyA 39 entry clone contains late polyA sequence from SV40 virus. Finally, p5E-b-actin2, pME-mCherry and p3E-polyA were assembled together with pDestTol2CG2 by LR reaction to create expression vectors of pDestTol2CG2bactin2-mCherry-pA.

Creation of cmlc2:EGFP/beta-actin2-mCherry transgenic zebrafish
For generation of transgenic zebrafish, we mixed expression constructs of pDestTol2CG2bactin2-mCherry-pA (50 ng/ml) with in vitro transcribed transposases mRNA (50 ng/ml) and injected about 1-3 nl DNA solution into the animal pole of one-cell stage embryos. The injected embryos were raised to adulthood and the putative founders were screened according to the green fluorescent signals in the heart of their F1 progenies. We totally identified 10 independent lines out of 89 crosses and used the most robust expression line Tg(cmcl2:EGFP; bactin2:mCherry) cy1 for the following experiments.

Whole mount in situ hybridization
The following in situ probes were used: cas [37], cmlc2 [38], fgfr2 [36], foxA3 [39], foxj1a [25], lefty1 and spaw (both were provided by Dr. Karuna Sampath, The National University of Singapore), and pitx2c [40]. The DIG-labeled probes were generated by in vitro transcription using a DIG RNA labeling kit (Roche). For whole mount in situ hybridization, DIG-labeled probes were used to hybridize the embryos overnight at 65 or 70uC and then washed with high stringency condition. The embryos were treated with blocking buffer (Roche) and incubated with AP-counjugated anti-DIG antibody overnight at 4uC (1:8000, Roche). Excess antibody was washed and the embryos were colored with NBT/BCIP.

Immunofluorescence
Embryos were fixed overnight in 4% paraformaldehyde at 4uC. Fixed embryos were washed with PBST (containing 0.3% TritonX-100) and treated with 10 mM Tris, 1 mM EDTA, 0.05% Tween20 for 5 minutes in 95uC. The embryos were subsequently blocked in PBST containing 4% BSA for one hour. Embryos were incubated in mouse anti-acetylated tubulin (Sigma T-6793, 1:200) and rabbit anti-aPKC (Santa Cruz sc-216, 1:100) at 4uC for overnight. After washed with PBST, embryos were incubated in goat anti-mouse Alexa Fluor 488 (Molecular Probes A-11029, 1:200) and goat anti-rabbit Alexa Fluor 647 (Molecular Probes A-21245, 1:200) at 4uC for overnight. After washed with PBST, embryos were mounted in SlowFade Gold antifade reagent with DAPI (Molecular Probes S-36938). Embryos were imaged using a LEICA TCS SP2 AOBS confocal microscope. Ciliary length and number were measured using Leica Confocal Software. KV size was analyzed by ImageJ using arbitrary unit. Two-tailed Student's t-test was used for analyzing on cilia length and number. Figure S1 The effects of fgfr2b and fgfr2c specific MOs. (A and B): Blue arrows were primer sites for RT-PCR to detect the splicing products. Red thick lines were MO target sites. Injection of fgfr2b (4 ng per embryo) or fgfr2c (1, 2 or 4 ng per embryo) MOs caused partial deletion of exon 8 (b) and exon 9 (c), respectively, that had been confirmed by sequencing. The original splice donor sites were blocked and cryptic splice donor sites in exon8 and exon9 were activated (indicated by bottom red lines) by the corresponding MOs. The partial cDNA sequences were shown (exon7, 8, 10 and 11 for fgfr2b and exon 7, 9 10 and 11 for fgfr2c). Underline indicated the primer sequence. The deleted regions were highlighted. (EPS) Figure S2 Expression pattern of fgfr2. The expression of fgfr2 was detected in marginal YSL (A,C, arrow, 95%,100% epiboly) and in the area near KV (arrowhead in D9, 5 somitestage). Boxed area shown in panel D is enlarged in panel D9. (EPS) Figure S3 The effects of fgfr2c MO specific on DFCs. The normal expression pattern of foxA3 in liver (l) and pancreas (p) was shown in DFC fgfr2c MO and DFC fgfr2c-5 mm MO morphants (A and C). Abnormal pattern of visceral organs was also observed in these embryos (B and D). The development of heart was examined using cmlc2 probe (E,H). Normal (E and G) and abnormal heart looping (F and H) can be observed in both morphants. The bar charts showed the percentage of embryos with different expression distribution of foxA3 or cmlc2 in both morphants (I). Panel A to D were dorsal view and panel E to H were ventral-anterior view. A: atrium, V: ventricle. (EPS) Figure S4 The percentage of abnormal heart looping in fgfr2c and fgf ligand morphants. In order to test the synergistic effect of Fgfr2c and Fgf ligands (Fgf4, Fgf8 and Fgf24), different combinations of low dosage fgfr2c MO and fgf MOs were injected into Line 544. Double morphants of fgfr2c and fgf4 (2, 1 or 0.5 ng/embryo for fgfr2c MO; 34, 22.5 or 12 ng/embryo for fgf4 MO) did not have synergistic effect on the abnormal heart looping, including L-loop and no loop pattern (A). Co-injection with fgfr2c MO (0.5 ng/embryo) and fgf24 MO (1.25 ng/embryo) also did not greatly increase the abnormal percentage (B). In contrast to above results, a synergistic effect was detected in fgfr2c-fgf8 double morphants (0.5 ng/embryo for fgfr2c MO and 1 ng/embryo for fgf8 MO; C). (EPS) Figure S5 The cas expression pattern in fgfr2c morphants. Embryos at 90% epiboly were stained with cas probe for labeling DFCs. The morphology of normal DFC cluster in wild type was shown in panel A (79%, n = 140). The mild and severe disorganization of DFC pattern could also be detected. However, the percentages of abnormal pattern were increased in fgfr2c morphants (B and C, 56.9%, n = 195). (EPS)