R-Spondin 3 Regulates Dorsoventral and Anteroposterior Patterning by Antagonizing Wnt/β-Catenin Signaling in Zebrafish Embryos

The Wnt/β-catenin or canonical Wnt signaling pathway plays fundamental roles in early development and in maintaining adult tissue homeostasis. R-spondin 3 (Rspo3) is a secreted protein that has been implicated in activating the Wnt/β-catenin signaling in amphibians and mammals. Here we report that zebrafish Rspo3 plays a negative role in regulating the zygotic Wnt/β-catenin signaling. Zebrafish Rspo3 has a unique domain structure. It contains a third furin-like (FU3) domain. This FU3 is present in other four ray-finned fish species studied but not in elephant shark. In zebrafish, rspo3 mRNA is maternally deposited and has a ubiquitous expression in early embryonic stages. After 12 hpf, its expression becomes tissue-specific. Forced expression of rspo3 promotes dorsoanterior patterning and increases the expression of dorsal and anterior marker genes. Knockdown of rspo3 increases ventral-posterior development and stimulates ventral and posterior marker genes expression. Forced expression of rspo3 abolishes exogenous Wnt3a action and reduces the endogenous Wnt signaling activity. Knockdown of rspo3 results in increased Wnt/β-catenin signaling activity. Further analyses indicate that Rspo3 does not promote maternal Wnt signaling. Human RSPO3 has similar action when tested in zebrafish embryos. These results suggest that Rspo3 regulates dorsoventral and anteroposterior patterning by negatively regulating the zygotic Wnt/β-catenin signaling in zebrafish embryos.

The activity of the canonical Wnt signaling pathway is regulated by a number of secreted proteins, including DKK1 and R-spondin (RSPO) proteins [14,[17][18][19][20][21][22][23][24]. While DKK1 inhibits the canonical Wnt signaling, RSPO3, a member of the RSPO family, has been suggested to activate Wnt signaling activity in mice and Xenopus [22,25]. Mammalian and amphibian RSPO3/Rspo3s contain two furin-like cysteine-rich (FU) domains near the N-terminus, a thrombospondin type I (TSP1) domain in the central region, and a positively charged C-terminal region [26,27]. Knockdown of rspo3 causes ventral edema and vascular defects in Xenopus [28]. Rspo3null mice suffer from severe vascular defects and are embryonic lethal [28,29]. In Xenopus, the effects of Rspo3 gain-and loss-offunction on expression of hematopoietic markers were similar to that of Wnt8 [28]. Loss of Rspo3 in mice resulted in reduced Wnt reporter activity [28]. These results have lead to the notion that Rspo3 promotes Wnt signaling activity in mice and Xenopus. To date, the structure and function of RSPO3/Rspo3 in other vertebrates such as fish have not been reported.
In this study, we have determined the structure of the Rspo3 in zebrafish and several ray-finned fish species and discovered that all ray-finned fish Rspo3 orthologs contain three furin-like (FU) domains. In contrast, a cartilagous fish, elephant shark Rspo3 has two FU domains, like mammalian RSPO3/Rspo3. Using zebrafish as an experimental model, we investigated the developmental role of Rspo3. Unexpectedly, our results suggest that Rspo3 regulates dorsoventral and anteroposterior patterning in zebrafish embryos by negatively regulating the zygotic Wnt/b-catenin signaling.

Chemicals and Reagents
Restriction enzymes were purchased from New England BioLabs (Ipswich, MA, USA). Oligo(dT) 18 was purchased from Sangon Biotech (Shanghai, China). iQ SYBR Green Supermix was purchased from Bio-Rad (Hercules, CA, USA). DIG-UTP and Anti-Digoxigenin-AP were purchased from Roche (Indianapolis, IN, USA). PCR primers were synthesized by Sangon Biotech and their sequences are shown in Table S1.

Experimental Animals
Wild-type zebrafish (Danio rerio, Tübingen and AB strains) were maintained on a 14-h light/10-h dark cycle at 28.5uC and fed twice daily. Embryos obtained by natural cross were kept in embryo rearing solution and staged according to standard methods [30]. In some experiments, 2-phenylthiourea [0.003% (w/v)] was added to prevent embryonic pigment formation. Animal manipulation was performed under tricaine for anesthesia of fish, and all efforts were made to minimize suffering. All experimental protocols were approved by and conducted in accordance with the Ethical Committee of Experimental Animal Care, Ocean University of China (Permit Number: 11001).

Molecular Cloning and Sequence Analysis
The full-length cDNA sequence of zebrafish rspo3 was determined by 59-and 39-rapid amplification of cDNA ends (RACE) using the SMART RACE cDNA Amplification Kit (Clontech Laboratories, Mountain View, CA, USA) following the manufacturer's instructions. The sequence of spotted gar, medaka, fugu, and stickleback Rspo3 were retrieved from Ensembl (www. ensembl.org) and that of elephant shark Rspo3 from http:// esharkgenome.imcb.a-star.edu.sg/. The amino acid sequence alignment was performed using the GeneDoc software (Free Software Foundation). The phylogenetic tree was constructed using the Neighbor-Joining method with MEGA 4 software (The Biodesign Institute, Tempe, AZ, USA). The bootstrap analyses were run on 1,000 replicates with amino acid substitutions of the JTT model. The genomic structure of the elephant shark, spotted gar, zebrafish, medaka, fugu, and stickleback rspo3 gene was obtained using the Blat program (http://genome.ucsc.edu/cgibin/hgBlat) and GENSCAN (http://genes.mit.edu/GENSCAN. html).

Plasmid Construction
For functional analysis, cDNA encoding the zebrafish rspo3 open reading frame (ORF) was amplified by reverse transcriptionpolymerase chain reaction (RT-PCR) using KOD plus DNA polymerase (TOYOBO, Shanghai, China) and cloned into the pCS2+ enhanced green fluorescent protein (EGFP) expression vector.

RT-PCR and Whole Mount in situ Hybridization
Total RNA was isolated from zebrafish embryos using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and then reverse transcribed into first-strand cDNA using M-MLV (Promega,  Madison, WI, USA) with Oligo(dT) 18 as primer. RT-PCR was carried out using premix Taq DNA polymerase (Takara, Dalian, China). Quantitative real-time RT-PCR (RT-qPCR) was performed in an iCycler iQ Multicolor real-time PCR detection system (Bio-Rad Laboratories). Samples from 3 independent experiments were collected and each sample was measured in duplicate. The levels of mRNA of the gene of interest were calculated using the 2 2DDCt method and normalized by b-actin mRNA levels [31].
The plasmid containing the rspo3 partial ORF and 39 untranslated region (UTR) was used to generate sense and antisense riboprobes using DIG RNA labeling mix (Roche, Indianapolis, IN, USA) following standard procedures. The specificity of the riboprobes was verified using dot-blot assay. In situ hybridization was performed as described previously [32].

Luciferase Assays
Luciferase assays were performed as reported previously [33]. Briefly, one-to two-cell stage embryos were injected with morpholinos and/or mRNA plus 100 pg Topflash DNA and 20 pg Renilla plasmid DNA, and raised to the shield stage. Two independent groups of embryos (each with more than 15 embryos) were lysed. The luciferase reporter assay was performed using a Dual-Luciferase Assay Kit (Promega). The Topflash luciferase activity was normalized by the Renilla luciferase activity.

Statistical Analysis
Data are presented as Means+S.E. Differences among groups were analyzed by one-way ANOVA followed by Tukey's Multiple Comparison Test, Chi-Square test or unpaired t-Test using GraphPad Prism version 5.01 (San Diego, CA, USA). Statistical significance was accepted at P,0.05 or smaller p values.

Zebrafish and Other Bony Fish rspo3 Orthologs have Unique Structural Features
By searching public databases and performing 59-and 39-RACE experiments, we cloned the full-length zebrafish rspo3 cDNA and determined its gene structure (zgc: 162040). While human and mouse RSPO3/Rspo3 genes have 5 exons, zebrafish rspo3 contains 6 exons (Fig. 1A). Zebrafish rspo3 encodes a protein that shares high sequence identities to known RSPO3s. The overall sequence identities of zebrafish Rspo3 to its human and Xenopus orthologs are 45% and 48% (Fig. 1B). Phylogenetic analyses suggested that the zebrafish Rspo3 is a bona fide Rspo3 (Fig. 1C). Zebrafish Rspo3 has an overall domain arrangement similar to its mammalian and amphibian orthologs (Fig. 1B). Unlike their mammalian and Xenopus counterparts, however, zebrafish Rspo3 has three FU domains. We also determined the rspo3 gene and protein structure in medaka, fugu, and stickleback. They all likewise have an additional FU3 domain encoded by an extra exon ( Fig. 1A and 1B). Searching these teleost genomes suggested that they contain only one rspo3 gene. To determine when the FU3 domain was evolved, we obtained the rspo3 sequence from the spotted gar (a non-teleost bony fish) and elephant shark (a cartilaginous fish). As shown in Fig. 1B, spotted gar Rspo3 also has three FU domains. In contrast, elephant shark Rspo3 has two FU domains. These findings suggest that the third FU domain was gained in the rayfinned fish lineage.

Developmental Expression Pattern of Zebrafish rspo3
RT-PCR analysis results showed that zebrafish rspo3 mRNA was expressed in all of the examined stages ranging from 1-cell to 144 hours post fertilization (hpf, Fig. 2A). Whole-mount in situ hybridization results indicated that the rspo3 transcript was expressed in a ubiquitous manner from the 1-cell stage to 12 hpf. At 14 hpf and 18 hpf, the rspo3 transcript began to be highly expressed in telencephalon, metencephalon, cephalic floor plate, and otic vesicle. At 24 hpf, strong rspo3 mRNA signal was observed in telencephalon, diencephalon, metencephalon, rhomb-encephalon, cephalic floor plate, lateral line precordium, and hypochord. At 48 hpf, strong signals were detected in telencephalon, diencephalon, metencephalon, rhombencephalon, lateral line primordium, branchial arches, palatoquadrate, and hypochord (Fig. 2B).

Overexpression and Knockdown Studies Reveal that rspo3 Regulates Dorsoventral and Anteroposterior Patterning in Zebrafish
To investigate the developmental role(s) of zebrafish Rspo3, we performed mRNA injection experiments. Injection of rspo3 capped mRNA increased dorsoanterior phenotypes ( Fig. 3A and 3B). These abnormal embryos were classified morphologically into three groups: mild, medium, and severe (Fig. 3A). Embryos in the mild group exhibited shortened body axis (Fig. 3A). Embryos in the medium group exhibited truncated posterior body axis and curved tail (Fig. 3A). Embryos in the severe group lacked the posterior body axis (Fig. 3A). As shown in Fig. 3B, injection of rspo3 mRNA resulted in a 23%, 24%, and 12% increase in the severe, medium, and mild group. We next performed in situ hybridization with emx1 (labeling forebrain) and rx1 (labeling retina). As shown in Fig. 3C and 3D, injection of rspo3 mRNA resulted in 78% and 33% embryos with enlarged brain and eyes, respectively.
Next, knockdown experiments were carried out using two independent translation-blocking antisense MOs. The efficacy of these rspo3 targeting MOs was verified by co-injecting an rspo3 5'-UTR-GFP expression construct. Both MO1 and MO2 blocked the reporter GFP expression (Fig. 3E). Knockdown of rspo3 by either MO1 or MO2 resulted in an increase in the number of embryos displaying enhanced ventral-posterior phenotypes ( Fig. 3F and 3G). In addition, knockdown of rspo3 resulted in lateral expansion of somites as indicated by myoD mRNA expression at 14 hpf (Fig. 3H). The abnormal embryos were classified into medium and severe groups ( Fig. 3F and 3G). Embryos in the medium group exhibited smaller eyes, slightly reduced head, and curved body axis (Fig. 3F). Embryos in the severe group exhibited smaller eyes, reduced brain, and shorter and curved body axis (Fig. 3F). 50% and 42% of the MO1injected embryos were in the severe and medium group (Fig. 3G). Likewise, 40% and 38% of the MO2-injected embryos were in the severe and medium group (Fig. 3G). Importantly, co-injection of rspo3 mRNA with MO1 or MO2 markedly reduced the MOinduced abnormal phenotypes from ,90% to ,40% (MO1) and ,80% to ,25% (MO2), respectively (p,0.0001) (Fig. 3G).
Next, we performed in situ hybridization using a number of dorsoventral marker genes. The dorsal markers chordin (chd) and gooscoid (gsc) were expressed on the dorsal embryonic shield in wildtype and control MO (cMO)-injected embryos (Fig. 4A-C). Their expression domains were reduced in MO1-injected embryos ( Fig. 4A-C). In contrast, the expression domains of even-skipped-1 (eve1) and ventral edema (ved), two ventral mesoderm marker genes, were increased in the morphants (Fig. 4A, 4D, and 4E). Forced expression of rspo3, on the other hand, increased chd and gsc mRNA expression domains and reduced eve1 and ved mRNA 3E). Knockdown of rspo3 by either MO1 or MO2 resulted in an increase in the number of embryos displaying enhanced ventral-posterior phenotypes ( Fig. 3F and 3G). In addition, knockdown of rspo3 resulted in lateral expansion of somites as indicated by myoD mRNA expression at 14 hpf (Fig. 3H). The abnormal embryos were classified into medium and severe groups ( Fig. 3F and 3G). Embryos in the medium group exhibited smaller eyes, slightly reduced head, and curved body axis (Fig. 3F). Embryos in the severe group exhibited smaller eyes, reduced brain, and shorter and curved body axis (Fig. 3F). 50% and 42% of the MO1-injected embryos were in the severe and medium group (Fig. 3G). Likewise, 40% and 38% of the MO2injected embryos were in the severe and medium group (Fig. 3G). Importantly, co-injection of rspo3 mRNA with MO1 or MO2 markedly reduced the MO-induced abnormal phenotypes from ,90% to ,40% (MO1) and ,80% to ,25% (MO2), respectively (p,0.0001) (Fig. 3G)

Rspo3 Does Not Promote Maternal Wnt Signaling in Zebrafish Embryos
In zebrafish embryos, maternal and zygotic Wnt/b-catenin signaling manifests different biological effects [1,3,10,[34][35][36]. While maternal Wnt/b-catenin signaling in dorsal marginal cells establishes dorsal cell fates before gastrulation, zygotic Wnt/bcatenin signaling in ventrolateral regions initiates ventral cell fate after gastrulation [3,4,8,37,38]. Either promotion of the maternal Wnt activity or inhibition of the zygotic Wnt activity could result in the phenotypes observed above [4,6,39]. To distinguish these two possibilities, we first investigated the effects of Rspo3 on bozozok (boz) expression at the sphere stage when zygotic Wnt/bcatenin is not yet functional [5,7]. In zebrafish, boz is a direct maternal Wnt signaling target gene [5,7]. Knockdown of rspo3 by MO1 or MO2 did not decrease boz mRNA levels (Fig. 5A). In fact, injection of MO1 even increased boz mRNA levels (Fig. 5A). Forced expression of rspo3 did not change boz mRNA levels (Fig. 5B). We also examined the chd expression at the sphere stage. Expression of chd at the sphere stage is another read-out of maternal b-catenin activity [40]. Consistent with the boz results, knockdown of rspo3 by MO1 actually increased chd mRNA levels (Fig. 5C). MO2 injection increased chd mRNA levels but the change did not reached statistical significance (Fig. 5C). Forced expression of rspo3 did not change chd mRNA levels (Fig. 5D). Collectively, these results indicate that Rspo3 does not promote maternal Wnt/b-catenin activity.

Rspo3 Inhibits Zygotic Wnt/b-catenin Signaling in Zebrafish Embryo
The phenotypes that resulted from rspo3 overexpression resembled those caused by the loss of Wnt ligands and the overexpression of the Wnt inhibitor dkk1 [36,[41][42][43]. The rspo3 morphants resembled the dkk1 morphant phenotypes [44]. We postulated that Rspo3 may negatively regulate the zygotic Wnt/bcatenin signaling. To test this hypothesis, we injected a Wnt reporter construct (Topflash), in which Wnt-responsive elements drive the expression of the luciferase reporter [45], together with Wnt3a, Dkk1, and Rspo3 mRNA. Forced expression of wnt3a in zebrafish embryos resulted in a robust increase in the Topflash reporter activity (Fig. 6A). Forced expression of rspo3 or dkk1 decreased the basal Topflash reporter activity (Fig. 6A), suggesting that Rspo3 plays an inhibitory role in regulating Wnt/b-catenin signaling. The role of endogenous Rspo3 in regulating the zygotic Wnt/b-catenin signaling was investigated next. Compared with the control group, knockdown of rspo3 by either MO1 or MO2 increased the Wnt reporter activity in a dose-dependent manner at the shield stage (Fig. 6B). More importantly, knockdown of rspo3 resulted in increased expressions of vent and sp5l, two direct Wnt target genes [46,47], at the 90% epiboly stage ( Fig. 6C and 6D).
Next, we performed co-injection experiments. Injection of wnt3a mRNA resulted in dorsalized phenotypes in more than 80% of the injected embryos at the 5-somite stage. Co-injection of rspo3 mRNA with wnt3a mRNA reduced the percentages of dorsalized embryos to 30% (Fig. 6E and 6F). Similarly, co-injection of dkk1 mRNA with wnt3a mRNA reduced the percentages of dorsalized embryos to 16% (Fig. 6F). Like Dkk1, co-expression of Rspo3 with Wnt3a abolished the Wnt3a-induced Topflash reporter activity (Fig. 6G), suggesting that Rspo3 inhibits the action of Wnt3a. Injection of b-cateninDN, which encodes a constitutively active bcatenin lacking the first 45 N-terminal residues, led to dorsalized embryos at the 5-somite stage (Fig. 6H). Co-injection of rspo3 with b-cateninDN mRNA did not reduce the percentages of dorsalized embryos (Fig. 6I). Likewise, co-injection of rspo3 with b-cateninDN mRNA did not block Wnt reporter activity induced by b-cateninDN (Fig. 6J). These results indicated that Rspo3 inhibits the zygotic Wnt/b-catenin signaling, likely acting at a step upstream of bcatenin.
One critical role of the zygotic Wnt/b-catenin signaling is to induce posterior neural fates [9,10,46,48]. If Rspo3 indeed inhibits the zygotic Wnt/b-catenin signaling pathway in zebrafish, then knockdown of rspo3 should increase posteriorization and forced expression of rspo3 should lead to anteriorization. Indeed, knockdown of rspo3 by either MO1 or MO2 resulted in a marked reduction in the expression of the anterior neuroectoderm marker otx2 and a concomitant expansion of the posterior neural marker hoxb1b region (Fig. 6K and 6M). In contrast, forced expression of rspo3 resulted in an expansion in the otx2 mRNA expression and a reduction in the hoxb1b mRNA expression (Fig. 6L and 6M). These results support the notion that Rspo3 promotes dorsoanterior development by negatively regulating the Wnt/b-catenin signaling pathway in zebrafish.
In addition to Wnt/b-catenin signaling, Nodal and Fgf signaling have been shown to regulate dorsoventral patterning in zebrafish [49][50][51]. We examined the possible effects of the rspo3 knockdown and overexpression on the expression of Nodal ligand sqt and Fgf ligands. Neither knockdown nor forced expression of rspo3 changed the levels of sqt mRNA (Fig. 7A and 7A'). Knockdown of rspo3 by both MO1 and MO2 increased the levels of fgf3 mRNA, while both MO1 and MO2 injection decreased the levels of fgf8 mRNA (Fig. 7B). However, forced expression of rspo3 did not result in any significant changes in the levels of fgf3 or fgf8   (Fig. 7B'). Mkp3, which has similar expression domains with the Fgf ligands, also regulates dorsovental patterning in zebrafish embryos [52]. Knockdown of rspo3 by either MO1 or MO2 did not affect expression of mkp3 but forced expression of rspo3 decreased mkp3 mRNA levels ( Fig. 7C and 7C').

Human RSPO3 Has a Similar Inhibitory Effect in Zebrafish Embryos
To examine whether the inhibitory action of Rspo3 is due to the structural difference between teleost Rspo3 and mammalian RSPO3, we forced expression of human RSPO3 in zebrafish embryos. Forced expression of human RSPO3 resulted in similar phenotypes as zebrafish Rspo3 (Fig. 8A). Forced expression of human RSPO3 also resulted in notable increases in chd and gsc expression domains (Fig. 8B a-d, 8C, and 8D) and decreased eve1 and ved expression (Fig. 8B e-h, 8E, and 8F). These data suggest that human RSPO3 has inhibitory activity that is similar to that of zebrafish Rspo3.

Discussion
In this study, we determined the structure of the Rspo3 protein and gene sequences in five ray-finned fish and a cartilagous fish. We mapped the spatial and temporal expression pattern of rspo3 in zebrafish and determined its developmental role through overex-pression and MO-mediated knockdown experiments. Our results suggest that bony fish Rspo3 has unique structural features and plays a previously unrecognized role in regulating dorsoventral and anterior-posterior patterning by antagonizing the zygotic Wnt/b-catenin signaling pathway in zebrafish embryos.
Both gain-and loss-of-function evidence supports our conclusion. Forced expression of rspo3 promoted dorsoanterior patterning in zebrafish embryos. Forced expression of rspo3 increased the expression of chd and gsc mRNA, while it reduced eve1 and ved mRNA expression. Furthermore, forced expression of rspo3 increased the expression of the anterior marker gene otx2 but reduced the expression of posterior neural marker hoxb1b. Knockdown of rspo3 had the opposite effects. These results suggest that Rspo3 promotes dorsoanterior patterning and inhibits ventral-posterior patterning in zebrafish.
A major finding made in this study is that Rspo3 functions as a negative regulator in the zygotic Wnt/b-catenin signaling pathway in zebrafish embryos. This notion is supported by several lines of experimental evidence. First, forced expression of rspo3 resulted in dorsoanteriorized embryos, resembling those affected by the loss of wnt3a and/or wnt8 [10,36] and the overexpression of Wnt inhibitors, e.g., frzb or dkk1 [15,41,42,53]. When co-overexpressed with Wnt3a, Rspo3 abolished the exogenous Wnt3a-induced Topflash reporter expression. Overexpression of rspo3 also inhibited the basal Wnt signaling activity. Second, knockdown of rspo3 resulted in ventral-posterior phenotype, which resembles those seen in the dkk1 knockdown embryos [44]. Importantly, knockdown of rspo3 increased the Wnt reporter activity in a dosedependent manner and increased the expression of vent and sp5l. In zebrafish embryos, the zygotic Wnt/b-catenin signaling plays a prominent role in the anterior-posterior neuroectoderm patterning. Our data showed that knockdown of rspo3 caused neuroectodermal posteriorization, while rspo3 overexpression led to neuroectodermal anteriorization. These findings strongly support the notion that Rspo3 inhibits the zygotic Wnt/b-catenin signaling. It has been well documented that maternal and zygotic Wnt/b-catenin signaling play opposite roles during zebrafish embryogenesis [1,[4][5][6][7]. While the maternal b-catenin regulates the formation of the dorsal organizer before gastrulation, the zygotic Wnt/b-catenin signaling initiates ventral cell fates after gastrulation [8,9]. We tested the possibility that Rspo3 may promote maternal Wnt signaling in zebrafish embryos. This hypothesis was not supported by the results. We found that forced expression of rspo3 did not affect the expression of boz and chd. Knockdown of rspo3 did not decrease boz and chd expression. In fact, MO1 injection actually increased their expression. Instead, our results suggest that Rspo3 does not promote maternal Wnt signaling in zebrafish embryos.
The finding that Rspo3 inhibits Wnt signaling in zebrafish embryos is unexpected. Several in vitro and in vivo studies using other model organisms suggest that RSPOs enhance the canonical Wnt signaling by interacting with Lgr4/5/6 and ZNRF3 [18,19,21,54,55]. The exact reason(s) underlying the different roles of Rspo3 observed in zebrafish compared with mice and Xenopus are not clear at present, but there are several plausible explanations. Unlike its mammalian and amphibian counterparts, zebrafish Rspo3 contains three FU domains. This additional FU3 domain is present in all five ray-finned fish species studied but not in the elephant shark, suggesting the FU3 domain is a structural feature that evolved and was conserved in the ray-finned fish lineage. The FU domains are known to be indispensable for mammalian RSPO3/Rspo3 protein activity [22,25]. A recent report demonstrated that the FU1 and FU2 domains of human RSPO1 are involved in its binding to ZNRF3 and LGR4, respectively [56]. We therefore speculated that its inhibitory role might be attributable to the unique 3 FU structure. This idea, however, was not supported because human RSPO3 had an inhibitory effect similar to that of zebrafish Rspo3 when tested in zebrafish embryos. This finding also indicated that the context of the zebrafish embryo is critical. In zebrafish embryos, rspo3 mRNA is maternally deposited and has a ubiquitous expression from the 1-cell stage to 12 hpf. After that, however, it displays a tissuespecific expression pattern. This expression pattern in zebrafish differs considerably from those reported in Xenopus and mice. In Xenopus and mice, the expression of Rspo3 mRNA is initially detected at the gastrulation stage and in the primitive streak at E7.5, respectively [29,57]. Future studies will be needed to determine whether the different roles of Rspo3/RSPO3 are related to the different spatial and temporal expression patterns among these different model organisms. It should be mentioned that our rspo3 mRNA expression result is different from the expression pattern shown in the zebrafish resource ZFIN, which shows no maternal expression and a restricted expression in the forebrain at the 12 hpf stage. An important difference between these two studies lies in the probes utilized. While a 569 bp partial ORF region (95-663 nt) was used in the study deposited in the ZFIN database, our probe covers a 482 bp 39-UTR sequence plus 322 bp ORF region (633-1436 nt). It is plausible that the longer probe containing a UTR region is more sensitive and has greater specificity. It is noted that Rspo3 morphants had a shorter body axis and a marked lateral expansion of the myoD expression domains, suggesting Wnt/PCP signaling may be altered. Studies in Xenopus suggest that Rspo3 alter both Wnt/b-catenin and Wnt/ PCP signaling [58]. In addition, Wnt/PCP signaling also inhibits the canonical Wnt pathway [59]. It remains to be determined whether Rspo3 indeed alters the Wnt/PCP signaling and whether this possible regulation inhibits the Wnt/b-catenin signaling pathway in zebrafish.
In addition to the Wnt/b-catenin signaling pathway, the Fgf and Nodal signaling pathways are also implicated in the dorsoventral and anterior-posterior patterning in zebrafish embryos [49][50][51]. The Nodal ligand sqt gene is expressed in a dorsal region of the blastula and forced expression of sqt resulted in expanded or ectopic dorsal mesoderm [49]. Overexpression of fgf3 dorsalizes zebrafish embryos [50]. The fgf8 lost-of-function mutant acerebellar displayed mild dorsoventral patterning defects [51]. Rspo2, a member of the Rspo family, has been shown to inhibit Nodal signaling in Xenopus [57]. We found that overexpression or knockdown of rspo3 had no significant effect on the mRNA levels of the Nodal ligand sqt. The effects of rspo3 on the Fgf ligands are more complicated. Knockdown of rspo3 by both MO1 and MO2 increased the expression of fgf3 mRNA. In the case of fgf8, both MO1 and MO2 injected embryos had reduced fgf8 mRNA levels. However, overexpression of rspo3 had little effect on the fgf3 and fgf8 mRNA levels. In zebrafish, overexpression of mkp3 ventralizes while knockdown of mkp3 dorsalizes embryos [52]. Knockdown of rspo3 has little effect on the expression of mkp3 mRNA but overexpression of rspo3 decreases its expression. These changes in Fgf ligands and mpk3 expression cannot explain the phenotypic changes observed in the rspo3 overexpression or knockdown embryos.
In summary, ray-finned fish Rspo3 has a unique structural feature and Rspo3 plays an important role in regulating dorsoventral and anterior-posterior patterning in zebrafish embryos. We have provided evidence suggesting that Rspo3 plays a negative role in regulating Wnt/b-catenin signaling in zebrafish embryos. During the revision of this manuscript, Wu et al. (2014) reported that human RSPO2, another member of the RSPO family, plays an inhibitory effect on Wnt/b-catenin signaling in colorectal cancer cells [60]. These new studies suggest that the roles of Rspo/RSPO proteins in the Wnt/b-catenin signaling pathway may be more complex. Future studies will elucidate the molecular mechanisms underlying the inhibitory role of Rspo3/ RSPO3 in regulating Wnt/b-catenin signaling. These studies will provide novel insights into Wnt/b-catenin signaling in vertebrates.