Protease-activated receptors (PARs) are G protein-coupled receptors (GPCRs) that play critical roles in hemostasis, thrombosis, vascular development, and the inflammatory response (Camerer et al.,2000; Macfarlane et al.,2001; Hollenberg and Compton,2002; Ramachandran and Hollenberg,2008; Peters and Henry,2009). PARs are activated by multiple proteases, through a unique and irreversible proteolytic mechanism that involves receptor binding by a protease and subsequent protease-mediated cleavage of the receptor N-terminus. The new N-terminus generated by this event acts as a tethered ligand that subsequently interacts intramolecularly with the second extracellular loop of the cleaved receptor, leading to its activation (Vu et al.,1991; Coughlin,2000; Macfarlane et al.,2001; Hollenberg and Compton,2002). Notably, synthetic peptides that mimic the tethered ligand sequences can activate PARs in the absence of proteolysis (Macfarlane etal.,2001).
Four PARs (1–4) have been identified in mammals. PARs 1, 3, and 4 can be activated by thrombin and are regarded as thrombin receptors. PAR2 is not activated by thrombin but by multiple trypsin-like serine proteases, including trypsin itself and tryptase (Nystedt et al.,1994,1995; Ishihara et al.,1997; Molino et al.,1997; Xu et al.,1998; Camerer et al.,2000). PAR1 and PAR3 have similar affinity for thrombin; PAR4 has a low affinity for thrombin and is activated only at much higher concentrations of thrombin. It has been proposed that PAR3 serves as a cofactor for PAR4, facilitating PAR4 activation without itself activating a downstream molecule (Nakanishi-Matsui et al.,2000). PAR1 is regarded as the prototype of the thrombin receptor. It mainly triggers Gαq/11 to activate phospholipase C, which leads to Ca2+ mobilization and PKC activation. In addition, PAR1 can activate other G proteins, including Gαi/o, Gα12/13, and Gβγ (Macfarlane et al.,2001; Hollenberg and Compton,2002; Traynelis and Trejo,2007). It has been reported that PAR2 and PAR4 signal through Gαq and Gα12/13 (Traynelis and Trejo,2007).
Thrombin is generated at sites of vascular injury, where it triggers the coagulation cascade and serves as the main regulator of platelet secretion and aggregation during clot formation. PAR1, PAR3, and PAR4 are expressed prominently in cells of the circulatory system, including platelets, endothelial cells, fibroblasts, leukocytes, and smooth muscle cells, and have been shown to play critical roles in blood hemostasis and thrombosis (Dery etal.,1998; Coughlin,2000; Traynelis and Trejo,2007). Notably, whereas human platelets express PAR1 and PAR4 (Kahn etal.,1999; Faruqi etal.,2000), mouse platelets express PAR3 and PAR4 (Kahn et al.,1998; Coughlin,2005; Traynelis and Trejo,2007). It also has been reported that, whereas PAR1 and PAR3, but not PAR4, are highly expressed in human endothelial cells, PAR1 and PAR4 are the predominant receptors in mouse endothelial cells (Macfarlane et al.,2001; Kataoka et al.,2003; Traynelis and Trejo,2007). In the case of PAR3, expression is not limited to the vascular system; the PAR3 mRNA is present in a wide range of human tissues, including stomach, small intestine, trachea, and bone marrow (Ishihara et al.,1997). In mouse, by contrast, PAR3 is restricted to certain tissues such as bone marrow (megakaryocytes) and spleen (Ishihara et al.,1997). Similarly, in humans the PAR4 mRNA is expressed in a variety of tissues; the highest levels outside the circulatory system are in the lung, pancreas, thyroid gland, and intestine (Xu et al.,1998).
In the case of PAR2, expression at both the transcript and protein levels is high in the gastrointestinal tract (including the stomach, intestine, and colon), in digestive-tract organs such as the pancreas and liver (both human and mouse), and in the kidneys (Nystedt et al.,1994,1995; Bohm et al.,1996). Consistent with the fact that trypsin is most abundant in the intestines, in situ hybridization revealed that mouse Par2 is highly expressed on the luminal surface of epithelial cells of the small intestine (Bohm et al.,1996). This expression profile has implicated PAR2 in gut motility, inflammation, and secretory function. In addition, both mouse and human PAR2 are expressed in the smooth muscle, where they contribute to vascular relaxation (Hwa et al.,1996; Molino etal.,1997; D'Andrea etal.,1998; Ramachandran and Hollenberg,2008).
PAR transcripts are also detected in other organ systems, including heart, central nervous system, airway, and epidermis, as well as in many cell types of the immune system (e.g., macrophages, monocytes, neutrophils, and mast cells; Macfarlane et al.,2001; Miotto et al.,2002; Ramachandran and Hollenberg,2008; Rattenholl and Steinhoff,2008; Sokolova and Reiser,2008). The expression of PAR proteins suggests that their functions are diverse. However, except in the context of the circulatory system, the roles of these proteins remain to be characterized in depth.
PAR proteins also appear to play an important role in embryonic development. PAR1 has been implicated in blood vessel formation; Par1-knockout mice display bleeding and disorganized blood vessel formation during embryonic development, with 50% fatality occurring by midgestation (Connolly et al.,1996; Griffin et al.,2001). This phenotype is rescued by expression of a Par1 transgene in endothelial cells, demonstrating that PAR1 is critical for proper endothelial-cell function in developing blood vessels (Connolly et al.,1996; Griffin et al.,2001). Of interest, mice in which the Gα13-encoding gene was disrupted specifically in endothelial cells displayed the same phenotype as Par1-deficient mice, suggesting that Gα13 is the main mediator of PAR1 signaling during vessel development (Ruppel etal.,2005). Mice in which other Par genes are knocked out developed normally to adulthood (Hirano,2007).
Zebrafish has become a useful vertebrate model for studying embryonic developmental processes, due to the transparency and rapid external development of the embryos, and the availability of a broad array of genetic tools (Grunwald and Eisen,2002). Notably, with respect to the circulatory system, it has been shown that trypsin is expressed in zebrafish (mouth, nose, gill, and pancreas), and that active trypsin obtained from zebrafish water can efficiently induce thrombocyte activation in vitro (Kim et al.,2009). Furthermore, the zebrafish genome has been reported to harbor 27 genes encoding Par proteins, with some of these genes expressed in thrombocytes and responding to thrombin and trypsin (Kim etal.,2009). These findings suggest that the role of Par proteins in hemostasis has been conserved over the course of evolution. However, the expression patterns of the par genes have not been reported, and the reported gene classification is inconsistent with the most recent version of the zebrafish genome. Here, we present our identification and cloning of the zebrafish par genes, our characterization of their genomic organization and phylogenetics, and our analyses of the protein primary structures, internalization, and spatiotemporal expression patterns during various stages of development.
RESULTS AND DISCUSSION
Identification and Cloning of the Zebrafish par Genes
In mammals, four PAR isoforms (PARs 1–4) have been identified. With respect to the 27 par-like genes previously reported to exist in the zebrafish genome (Kim etal.,2009), we found that many are in fact identical to one another, and that some of the genomic information provided in the study is not represented in the current version of the database. These discrepancies are probably due to mistakes and redundancies in the early versions of the zebrafish database used in that study (zebrafish genomic assembly version (Zv) 6 and 7). For our analysis, we initially conducted a new search of the expressed sequence tag (EST) and Ensembl Genome (Zv7) databases, using the human PAR gene sequences. We found four par-related genes (pars 1, 2a, 2b, 3)—corresponding to human PAR1, PAR2, and PAR3—in zebrafish, but failed to identify a homolog of the human PAR4 gene (Fig. 2A). We also failed to find PAR4 homologs in the Medaka and Fugu genomes (data not shown). However, in the course of revising this manuscript, we discovered a gene encoding a Par4-like protein (XP_001333055) in the newest version of the zebrafish genome database (Zv8). Notably, despite 65% similarity with mammalian PAR4 at amino acid level, syntenic analysis suggests that the genomic organization of this par4-like gene is not similar to that of mammals. In addition, this gene lacks a similar tethered peptide that is found in human PAR4 (data not shown). Thus, it remains to be determined if this is a true zebrafish counterpart of the mammalian PAR4 gene. The zebrafish par1 gene (GenBank accession no. GQ325252) contains an open reading frame (ORF) encoding a protein of 413 amino acids, which shares 46% identity and 70% similarity with human PAR1 (Fig. 1A). Two closely related zebrafish par2 genes, designated as par2a and par2b (GenBank accession nos. GQ325254 and GQ 325253), were also found. The predicted encoded proteins share 55% identity and 71% similarity with one another, and 50%/48% (respectively) identity and 70%/66% similarity with human PAR2, at the amino acid level (Fig. 1B). Also, using the new version of the Zv8 database, we identified a par3 gene (GenBank accession no. XM_002662087). It encodes a predicted protein that is 367 amino acids long and shares 37% identity and 59% similarity with human PAR3 (Supp. Fig. S1; which is available online). However, we were able to clone only the first 774 base pairs of the cDNA, which encode the first 258 amino acids of the predicted Par3 protein (GenBank accession no. GQ325255).
Given the discrepancies between our results and those reported in the earlier study (Kim et al.,2009), we re-examined the data on which that study was based. First, we found that many of the genes that had been classified into particular groups based on similarity in fact represent the same gene (see Supp. Table S1). Briefly, five groups of Par genes had been identified, with group 1–3 representing par1 homologs and group 4–5 representing par2 homologs (Kim et al.,2009). The group1 genes are identical to the par1 that we describe. The group 2 genes likewise represent a single gene, which is predicted to encode a Par2-like protein (XP_0013 34308, Zv8). Although it was suggested that PAR1-21A (group 2) encodes a thrombin receptor based on the fact that the protein could be activated modestly by human thrombin (Kim et al.,2009), whether it represents a second Par1 isoform remains to be determined since, first, the tethered peptide of this protein (PTQQTL) differs significantly from that found in mammalian Par1 (SFLLRN and SFFLRN in human and mouse, respectively) and, second, there is no direct evidence that thrombin-based activation of this protein involves cleavage of the peptide. Similarly, the group 3 genes represent a single gene encoding a protein that is predicted to be Par2-like (XP_001096094, Zv8). This protein probably cannot be classified as a thrombin receptor because the transcripts are not expressed in thrombocytes, and a synthetic peptide based on its sequence failed to induce thrombocyte activation (Kim et al.,2009). In case of the group 4 and group 5 genes, which were thought to encode Par2 proteins according to the previous analysis, we again found that the members of each group likely represent a single protein. In the case of group 4, they are similar to the par2b described here, and in the case of group 5, they are similar to our par2a. With regard to the remaining two representatives of the Par family reported in the earlier study, we found that PAR2-14A represents the Par4-like protein described above, whereas Par2-19A is now annotated as a rCG59702-like protein (XP_0026655 01, Zv8) and unrelated to Par proteins. Given these considerations, and the fact that the findings presented here are further supported by our analyses of genomic organization and phylogenetics (see below), we believe that our characterization of the Par genes is accurate.
Characterization of the Zebrafish Par Proteins
Analysis of multiple sequence alignments for each identified zebrafish par gene revealed that the predicted Par proteins contain seven putative transmembrane domains (indicated by solid lines above the sequences, Fig. 1, Supp. Fig. S1), a characteristic feature of G protein-coupled receptors. The predicted amino acid sequences of these domains in the zebrafish Pars are highly conserved with those of their human and murine PAR counterparts, supporting the notion that zebrafish par genes are the orthologs of mammalian PAR-encoding genes (Fig. 1, S1). Interestingly, each zebrafish Par protein has the conserved “CHD motif” in its second extracellular loop (indicated by asterisks, IT/VTCHDV in Par1 and Par 2, Fig. 1; and LVTCHDI in Par 3, Supp. Fig. S1); this region has been implicated in the ligand docking that leads to receptor activation (Gerszten et al.,1994; Xu etal.,1998).
Mammalian PARs are known to trigger cellular responses through a unique activation mechanism where by a protease cleaves the N-terminus of the receptor, generating a new N-terminus that acts as a tethered ligand. This newly generated autoligand then binds to and activates the receptor (Camerer et al.,2000). Consistent with the conservation of such a mechanism in zebrafish, Par1 harbors a putative protease cleavage site (Arg-28/Ser-29, indicated by the arrow, Fig. 1A), as well as a putative tethered ligand sequence (SFSGFF, dashed line box, Fig. 1A), within its extracellular N-terminal region. This peptide has been shown to induce thrombocyte activation (Kim et al.,2009). These features are similar to those of human and mouse PAR1 proteins, in which Arg-41/Ser-42 function as the cleavage site, and SFLLRN and SFFLRN function as the tethered ligand sequence in human and mouse, respectively.
These findings support the notion that zebrafish Par1 is a protease-activated receptor. Also, a putative protease cleavage site (Lys-37/Thr-38) and a tethered ligand (TFTGTI) similar to those in human and mouse PAR3 were identified in zebrafish Par3 (Supp. Fig. S1). Although the N-terminus of mammalian PAR2 also contains a protease cleavage site (Arg-36/Ser-37, indicated by the arrowhead, Fig. 1B) and tethered ligand sequences (SLIGKV in human, SLIGRL in mouse, indicated by dashed line boxes, Fig. 1B), we failed to identify precisely the same features within the N-termini of zebrafish Par2a and Par2b. However, in Par2b, we found a similar potential protease cleavage site (Arg-27/Met-28, indicated by the arrow, Fig. 1B) and a tethered ligand sequence (MMIVIQ, dashed line box, Fig. 1B). This tethering peptide is different from that reported previously (SLV VIQ; Kim et al.,2009), probably due to differences in the genetic backgrounds of zebrafish strains. Although a cleavage site (Ser-31/Gly-32, indicated by the arrow, Fig. 1B) and tethered ligand sequence (GFTEET, dashed line box, Fig. 1B) have also been proposed in Par2a, a synthesized version of this tethering peptide failed to induce thrombocyte activation (Kim et al.,2009). Thus, activation of these zebrafish Par2 receptors may involve cleavage sites different from those that have been defined for the mammalian homologs (Fig. 1B).
Phylogenetic and Syntenic Analyses of Zebrafish par Genes
We examined the evolutionary relationships between zebrafish Pars and their mammalian counterparts by building a phylogenetic tree using the Neighbor-Joining (NJ) method (http://www.ebi.ac.uk/Tools/clustalw2). Our results indicate that the zebrafish Par proteins are closely related to their mammalian orthologs (Fig. 2A). Syntenic analyses indicate that the genomic organization of the genes encoding Par proteins is similar in zebrafish and mammals. The human and mouse PAR1–3 genes are clustered in a small region (∼200 kb) of a single chromosome, on 5q13 and 13D1, respectively (Fig. 2B and O'Brien etal.,2001), with PAR2 and PAR1 separated from PAR3 by the IQGAP2 gene (encoding the IQ motif containing GTPase activating protein 2). The PAR3 gene is flanked by the IQGAP2 and SV2C genes (encoding synaptic vesicle glycoprotein 2C), and the PAR2 gene is positioned between the CRHBP (encoding corticotropin-releasing hormone binding protein) and S100Z genes on one side, and the PAR1 gene on the other. In both human and mouse, the PAR4 gene is located on a second chromosome (human, 19p12; mouse, 8B3.3; O'Brien et al.,2001). In a striking synteny, the zebrafish par1 and par3 genes on chromosome 5 are clustered and separated by iqgap2. In addition, par3 is flanked by the iqgap2 and sv2c genes (Fig. 2B), although par1 is not flanked by par2. In fact, both par2a and par2b are located on chromosome 21, next to the crhbp gene (Fig. 2B). These observations suggest that zebrafish par2 underwent a tandem duplication and was translocated, along with crhbp, from par1 and par3 to a different chromosome.
Internalization of Zebrafish Par Proteins
Many GPCRs undergo internalization in response to agonist stimulation, i.e., they are redistributed from the plasma membrane to the cytosol. Such internalization of the receptors plays a critical role in terminating receptor signaling (Hoxie et al.,1993; Trejo,2003). Similarly, internalization of mammalian PAR1 is triggered by thrombin (Trejo,2003), and the previous study of zebrafish Par proteins had shown that human thrombin can stimulate zebrafish Par1 to induce a Ca2+ response in Xenopus oocytes (Kim et al.,2009). To evaluate Par1's response to thrombin, we generated a zebrafish Par1-GFP (green fluorescent protein) fusion protein in which GFP is attached to the C-terminus of the receptor and transiently expressed it in HEK293 cells. As shown in Figure 3A, in cells in the basal state, Par1-GFP was mainly detected on the plasma membrane. However, in cells stimulated with thrombin for 30 min, the majority of the receptor was internalized into intracellular compartments that were visible as discrete punctate accumulations (Fig. 3B), suggesting that Par1 can respond to thrombin to undergo internalization. By contrast, treatment with trypsin did not change the membrane localization of Par1-GFP (Fig. 3C). These results are consistent with the notion that zebrafish Par1 is a thrombin receptor.
Mammalian PAR2 is not activated by thrombin, but rather by trypsin-like serine proteases (Dery et al.,1999). Similarly, zebrafish Par2b showed a significant Ca2+ response upon trypsin activation (Kim et al.,2009). To investigate directly how zebrafish Par2 responds to thrombin and/or trypsin, we constructed a Par2b-GFP fusion protein and expressed it in HEK293 cells. Like Par1-GFP, Par2b-GFP was detected largely on the plasma membrane (Fig. 3D). Not surprisingly, the distribution of Par2b-GFP was not altered upon exposure to thrombin; the fusion protein was still present on the cell membrane 60 min after stimulation (Fig. 3E). In contrast, 10 μM trypsin induced internalization similar to that of mammalian PAR2 (Fig. 3F), whereas 100 nM trypsin failed to promote this internalization (data not shown). These results are consistent with the finding that the concentration of bovine trypsin required to reduce gill bleeding is much higher than that of zebrafish trypsin, because the latter is 10-fold more active than bovine trypsin when applied to zebrafish (Kim et al.,2009). Another possibility is that the protease cleavage site of the zebrafish Par2b receptor is different (see above) from that of mammalian PAR2, and that it does not respond well to thepig trypsin used in our study. Taken together, these results support previous findings that Par1 is a thrombin receptor and Par2b is a trypsin receptor.
Expression of the Zebrafish par1 Gene
We performed whole-mount in situ hybridization to examine the spatiotemporal expression of each of the par genes during zebrafish embryogenesis. The par1 transcript was detected at both the eight-cell and blastula stages (Fig. 4A,B), indicating that it was maternally deposited. During the gastrula stage, par1 expression was detected ubiquitously throughout the embryo (Fig. 4C–E), suggesting that this gene may play a role during early embryogenesis. At the 10-somite stage, par1 expression was limited to the posterior and ventral mesoderm (Fig. 4G,H). By 1 day postfertilization (dpf), its expression was confined to the heart and veins. Specifically, par1 expression was found in the common cardinal vein (CCV), posterior cardinal vein (PCV), caudal vein (CV), and intermediate cell mass of mesoderm (ICM; Fig. 4I–N), suggesting that par1 might play a specific role in vein development and/or function. At 2 dpf, in addition to being expressed in the heart, veins, and caudal hematopoietic tissue (CHT), par1 RNA was present in the branchial arches, notochord, otic vesicle, gut, pancreas, and pronephric duct (Fig. 4O–S). At 3 dpf, par1 expression persisted in the branchial arches, pancreas, heart, and vein (Fig. 4T–U). Overall, these results indicate that par1 expression is ubiquitous during early development, and then becomes confined to particular organs, including the cardiovasculature (heart, vein), digestive tract (gut, pancreas), and developing kidney. This expression pattern resembles that of the mammalian PAR1 gene (Vu et al.,1991; Dery et al.,1998; Coughlin,2000), suggesting that the function of Par1 and the regulation of its expression are conserved among vertebrates.
Expression of the Zebrafish par2a and par2b Genes
As in the case of the par1 gene, par2a and par2b transcripts were detected at as early as the eight-cell stage, suggesting that they are also maternally deposited. Both were expressed ubiquitously during the blastula and gastrula stages (Fig. S2). However, by 1 dpf, par2a expression was dramatically decreased, to a level that could not be detected by in situ hybridization (data not shown), whereas par2b transcript persisted in certain tissues, specifically in the pronephric ducts and otic vesicles (Fig. 5A–E). Transverse and sagittal sections revealed that the par2b transcript was enriched on the apical membrane of pronephric-duct epithelial cells (Fig. 5C,D). At 2 dpf, par2b expression persisted in those locations but was also observed in the gut and notochord (Fig. 5F–J). At 3 dpf, par2b expression in the gut became more prominent, and was strongest on the apical membrane of the gut epithelial cells. In addition, par2b was detected in the pharynx, branchial arches, and liver (Fig. 5K–P). Thus, as in the case of par1, the expression of par2b is similar to that of its mammalian counterpart (Nystedt etal.,1995; Bohm et al.,1996; Grandaliano et al.,2003; MacNaughton,2005). Notably, both human and zebrafish Par2 are highly expressed in epithelial cells of the intestinal surface (Bohm et al.,1996). Collectively, these data suggest that zebrafish Par2b is likely a functional counterpart of human PAR2, and that Par2a may not contribute to organogenesis.
Expression of the Zebrafish par3 Gene
Although a reverse transcriptase-polymerase chain reaction (RT-PCR) approach enabled us to detect par3 transcripts in 3 dpf embryos (Fig. S3), whole-mount in situ hybridization in embryos ranging from blastula stage through 3 dpf failed to detect any par3 expression (data not shown). These findings suggest that the expression levels during this period are below the level of detection for this method. Furthermore, given that PAR3 has been shown to act as a cofactor for PAR4 activation rather than to directly generate a signal (Nakanishi-Matsui etal.,2000), and that we did not identify a convincing Par4 homolog in zebrafish, par3 may not be required during early stages of zebrafish development.
We have identified and cloned four zebrafish par genes. Analyses of the PAR family phylogenetic tree and the genomic organization of the PAR genes indicate that the newly discovered zebrafish genes are homologs of mammalian PARs—PARs1–3. Our characterization of internalization supports the previous findings that zebrafish Par1 is a thrombin receptor and Par2b is a trypsin receptor. In situ hybridization studies demonstrated that the zebrafish par1–2 genes display distinct spatial and temporal expression patterns during development, with par1 being expressed in the vascular system, gut, and developing kidney, much like its mammalian counterpart. Although both par2a and par2b are expressed ubiquitously during the blastula and gastrula stages, par2a expression diminishes after segmentation is complete, whereas par2b is detected in the gut, pronephric duct, and liver in embryos from 1 to 3 dpf; this expression profile is also similar to that of mammalian PAR2. Overall, our analysis of the expression patterns of the zebrafish par genes suggests that different Par proteins play distinct roles during zebrafish development. Our findings lay the foundation for future studies that will elucidate the functions of Par proteins during embryonic development.
Embryo Generation and Maintenance
WT zebrafish of AB*, India, TL, and hybrid backgrounds used in this study were maintained either in the Animal Care Facility at the University of Iowa or the zebrafish facility at Vanderbilt University, as described previously (Lin et al.,2005). Embryos were obtained from natural matings, and staged according to morphology or by days postfertilization (dpf) at 28.5°C, as described (Kimmel et al.,1995). Embryos older than 1 dpf were treated with 0.003% phenylthiourea (PTU) to prevent pigmentation.
BLAST Search, Sequence Alignments, and Phylogenetic Analysis
PCR primers flanking the ATG initiation and stop codons of the zebafish par genes were designed to amplify the full coding sequences. For zebrafish par3, primers flanking the EST sequence (GenBank accession no. EE306722.1) were used to amplify the partial cDNA. RT-PCR (Titanium™ One-Step RT-PCR kit, Clontech) was performed using RNAs prepared from zebrafish embryos. PCR products were cloned into the PCR II-TOPO vector (Invitrogen), and sequences were verified. Subsequently, all par genes were subcloned into the pCS2 expression vector using the Gateway (Invitrogen) strategy. GFP fusion constructs (par1-gfp and par2b-gfp) in which the gfp sequence was inserted at the 3′ end of the par genes were generated using the MultiSite Gateway (Invitrogen) kit.
Whole-Mount In Situ Hybridization
Digoxigenin-UTP–labeled sense and antisense RNA probes for each par gene were generated as described previously. Whole-mount in situ hybridization was performed as described (Lin etal.,2005). Frozen sections were generated from embryos labeled by whole-mount in situ hybridization using a cryostat.
Internalization of Par Receptors in HEK293 Cells
HEK (human embryonic kidney) 293 cells were transiently transfected with plasmid constructs encoding Par1-GFP or Par2b-GFP. At 24 hr posttransfection, cells were starved overnight in serum-free medium, and then stimulated with thrombin (5 nM, Sigma) or trypsin (100 nM and 10 μM, Amresco; or left untreated) for 30–60 min. Cells were then fixed with 4% paraformaldehyde (PFA) and mounted in Vectashield mounting medium (Vector Laboratories). Confocal images were acquired on a Leica Inverted DMI 6000B microscope.
Embryos subjected to whole-mount in situ hybridization were mounted in 75% glycerol/PBT, and photographed using an Axiophot2 microscope (Zeiss) and an Axiocam digital camera (Zeiss), or a Leica M165FC stereofluorescent microscope with a Leica DFC290 color digital camera. For confocal imaging, a Leica Inverted DMI6000B microscope with a BD CARV II Spinning Disk Confocal Imager was used. Images were acquired using a ×63 oil objective and an EMCCD camera equipped with the IPlab software. All images acquired were compiled and edited using Adobe Photoshop and Illustrator software.
This study was initiated in the laboratories of Heidi Hamm and Lila Solnica-Krezel at Vanderbilt University, and completed at the University of Iowa. We acknowledge Joshua Clanton, Heidi Beck, and Amanda Bradshaw of the Vanderbilt fish facility and the zebrafish care team of the Animal Care Facility at the University of Iowa for excellent fish care. We are grateful to Drs. Chi-Bin Chien (University of Utah) and Nathan Lawson (University of Massachusetts Medical School) for sharing the Gateway constructs. F.L. was supported by the NIH/NCRR and L.S.K. and H.H. were funded by NIH/NIGMS.