Wnt signaling pathways in vertebrates use the phosphoprotein Dishevelled(Dvl). The cellular responses to Wnt signaling may in part be modulated by Dvl-associated proteins, including Dapper (Dpr). We have cloned and characterized the zebrafish Dpr paralogs Dpr1 and Dpr2. Loss-of-function studies reveal that endogenous Dpr1 but not Dpr2 is required to enhance Wnt/β-catenin activity in zebrafish embryos that are hypomorphic for Wnt8. Conversely, Dpr2 but not Dpr1 is required for normal convergence extension movements in embryos that are hypomorphic for Stbm or Wnt11,supporting a functional interaction of Dpr2 with Wnt/Ca2+-PCP signaling. In gain-of-function experiments, Dpr1 but not Dpr2 induces Wnt/β-catenin target genes. Dpr1 synergizes with zebrafish Dvl2, and with the Dvl-interacting kinases CK1ϵ, Par1 and CK2, in activating target genes. We conclude that two Dvl-associated paralogs, Dpr1 and Dpr2,participate in distinct Wnt-dependent developmental processes.
Wnt signaling through Frizzled (Fz) serpentine receptors controls a plethora of developmental and adult processes in metazoa(Lustig and Behrens, 2003; Myers et al., 2002; Veeman et al., 2003a; Wodarz and Nusse, 1998). However, the number of distinct Wnt/Fz pathways in vertebrates is unresolved. The best characterized is the Wnt/β-catenin pathway, which regulates cell proliferation, survival and fate in a context-dependent manner. For example,this pathway controls developmental events in frog and fish embryos including early dorsoventral and anteroposterior patterning(Erter et al., 2001; Kelly et al., 2000; Larabell et al., 1997; Lekven et al., 2001). Another vertebrate Wnt pathway is the Wnt/Ca2+pathway, which may functionally and mechanistically overlap with a vertebrate counterpart of the Drosophila planar cell polarity (PCP) pathway. In Drosophilathe PCP pathway controls wing hair, ommatidia and bristle polarity. In Xenopus and zebrafish, the Wnt/Ca2+pathway and orthologs of the Drosophila PCP pathway control proper cell movements during gastrulation that are required for convergence and extension (CE) (reviewed by Veeman et al., 2003a). To reflect shared functions in regulating CE, we refer to these pathways as the Wnt/Ca2+-PCP pathway, leaving open the issue of whether they are mechanistically unique or overlapping pathways.
In addition to a Wnt ligand, Wnt/β-catenin signaling involves a seven transmembrane receptor, Fz, and a transmembrane co-receptor, LRP receptor-related protein-5 or -6 (LRP5/6)(Mao et al., 2001; Tamai et al., 2000). In the absence of Wnt signaling, β-catenin is phosphorylated in a complex of proteins that includes the scaffolding proteins Axin and APC, and the kinases GSK-3β and CK1α. Phosphorylation of β-catenin leads to its ubiquitination and degradation (Liu et al., 1999; Marikawa and Elinson, 1998). In response to a Wnt signal, Dishevelled (Dvl) is activated, promoting the stabilization of β-catenin. As β-catenin accumulates in the cytoplasm and nucleus, it associates with the LEF/TCF transcription factors to activate transcription of targets genes(Huelsken and Behrens, 2002). In vertebrates, the Wnt/Ca2+-PCP pathway also appears to involve a Wnt ligand and Fz, but does not involve LRP-5/6 co-receptors. Instead,stimulation of this pathway leads to the association of Dvl with proteins that control cytoskeletal remodeling, including Rac and the Rho-interacting protein Daam (Habas et al., 2003; Habas et al., 2001). Vertebrate Dvl also interacts with orthologs of Drosophila proteins that regulate PCP signaling, including Strabismus/Trilobite (Stbm/Tri) and Prickle(Pk) (reviewed by Wharton,2003). However, the general mechanisms of Wnt/Ca2+-PCP signaling remain unclear (reviewed by Veeman et al., 2003a).
The phosphoprotein Dvl is therefore involved in the Wnt/β-catenin and Wnt/Ca2+-PCP pathways. Both genetic and biochemical screens have identified a host of Dvl-interacting proteins (reviewed by Wharton, 2003). In addition to those proteins previously mentioned, this list includes Naked cuticle (Nkd)(Rousset et al., 2001), Casein kinase 1ϵ (CK1ϵ) (Peters et al.,1999), Protein kinase CK2 (CK2)(Willert et al., 1997), Par1(Sun et al., 2001), GBP(Li et al., 1999), FRODO(Gloy et al., 2002) and Dapper(Dpr; Dact – Zebrafish Information Network)(Cheyette et al., 2002). CK1ϵ, CK2, Par1, GBP and FRODO have been described as positive regulators of Wnt/β-catenin signaling, while Nkd and Dpr have been characterized as inhibitors of Wnt signaling. A number of these Dvl-interacting proteins have been proposed to act as switches that regulate Dvl functions in different Wnt/Fz signaling pathways. For example, it has been proposed that Par1 promotes Wnt/β-catenin signaling at the expense of Wnt/Ca2+-PCP signaling, while vertebrate Nkd has been proposed to do the opposite (Sun et al.,2001; Yan et al.,2001). Clearly an understanding of Dvl-associated proteins is central to understanding how Wnt pathways are regulated.
We have previously reported the characterization of a Dvl-interacting protein, Xenopus Dpr1 (Cheyette et al., 2002), and presented evidence that it is required for notochord formation. More recently, Hikasa and Sokol(Hikasa and Sokol, 2004) have presented evidence that a Dpr homolog, FRODO, is involved in aspects of organizer formation and neural patterning. To understand better the functions of the Dpr family in vertebrate embryos, in the present study we have characterized two zebrafish Dpr paralogs, Dpr1 and Dpr2, which we analyze through their loss and gain of function. Our data suggest that the Dpr1 and Dpr2 paralogs have different roles in early vertebrate development.
Materials and methods
Cloning, expression analysis and phylogenetic analysis
A fragment of zebrafish Dpr1 was cloned from a degenerate PCR screen of cDNA. cDNAs encoding the complete open reading frames for zebrafish Dpr1, Dpr2 and Dvl2 were cloned using the Clontech SMART RACE Kit (BD Biosciences, Palo Alto, CA). In situ procedures were performed as previously described(Oxtoby and Jowett, 1993). RT-PCR was performed using cDNA made from RNA of 25 embryos of each stage using the Thermoscript kit (Invitrogen, Carlsbad, CA). All sequences of primers are available upon request. Alignment of sequences was performed using ClustalW via MacVector. Neighbor joining phylogenetic analysis was performed using PAUP* 4.0.
All genes used except for β-galactosidase, CK1ϵ, CK1D→N, Drosophila Par1 and Par1-KN were cloned into pCS2P+. The Xenopus CK1ϵ and CK1D→N in pCS2 were gifts from J. Graff(Peters et al., 1999). CK2α and β subunits, gifts from D. Seldin, were subcloned into pCS2P+. Drosophila Par1 and Par1-KN were in pCS2+, as reported previously (Sun et al., 2001). GBP RNA was a gift of D. Kimelman and H. Gist Farr. The zebrafish Dvl2ΔPGB construct deletes amino acids 231-252. Capped mRNAs for injection were prepared using the Message Machine kit (Ambion, Austin, TX) and dissolved in water.
RT-PCR analysis of Wnt/β-catenin target genes
For analysis of the Wnt/β-catenin target genes siamois(sia) and Xenopus nodal-related 3 (Xnr3), Xenopus embryos were injected, animal caps isolated, and RT-PCR conducted essentially as described(Cheyette et al., 2002). Controls omitting RT were conducted for all experiments (not shown).
Morpholino injections of zebrafish
Stock morpholinos (MO; Genetools, Philomath, OR) were dissolved in 1×Danieu's buffer to a concentration of 10 ng/nl(Nasevicius and Ekker, 2000). The diluted morpholino stocks (2 nl) were injected into one-cell zebrafish embryos. strabismus/trilobite MO(Park and Moon, 2002), wnt11/silberblick MO (Lele et al., 2001), wnt8-orf1 and wnt8-orf2 MOs(Lekven et al., 2001) have been reported previously. Morpholino sequences for dpr1 and dpr2 are as follows:
dpr1 MO, 5′-AACAAATACTAACCAGAGCTTGGCC;
dpr1 MO(2), 5′-TATAAATCCAATGAAAGCAATAGAC;
dpr2 MO, 5′-TCAGACGTGCCGTTTAGACATATCC;
dpr2 MO(2), 5′-GGCATGTGTAGTCACCTGAACTGG.
For immunocytochemistry, Xenopus embryos were injected, and animal caps were isolated, fixed in phosphate-buffered saline and processed by standard methods (Cheyette et al.,2002). The anti-Myc antibody was the monoclonal 9E10 and the secondary antibody was the goat-anti-mouse Alexa Fluor 488 (Molecular Probes,Eugene, OR).
Xenopus embryo lysates and immunoprecipitation
Embryos injected with ∼3 ng of each RNA were lysed with 150 mM NaCl, 20 mM Tris-HCl pH 7.5, 0.5% CHAPS, 1 mM EDTA, 1 mM EGTA and Complete protease cocktail (Roche, Indianapolis, IN). For immunoprecipitation (IP), the lysis buffer did not include EDTA and EGTA. Anti-HA antibody was a commercial monoclonal antibody.
Cell culture and reporter assays
Human embryonic kidney 293T cells were transfected using Lipofectamine Plus(Invitrogen) with the reporter construct, Super(8X)TOPFLASH(Veeman et al., 2003b), an internal control construct encoding Renilla luciferase and the indicated plasmids and empty vector to maintain constant amounts of DNA. Luciferase activity was measured 24 hours after transfection, using the Dual-Luciferase Reporter Assay System (Promega).
GenBank Accession numbers
Zebrafish Dpr1, AY545443; zebrafish Dpr2, AY545444; zebrafish Dvl2,AY552332.
Vertebrate genomes have multiple Dpr paralogs
We initially conducted phylogenetic analysis of the two Dpr homologs. Alignment of the sequences using ClustalW, followed by neighbor-joining(shown) or maximum parsimony (not shown), which showed identical relationships, revealed two ortholog groups(Fig. 1A and see Fig. S1 in the supplementary material). The Dpr homologs that formed a monophyletic group and clustered with the previously published Dpr members, XDpr1 and FRODO, were assigned as the Dpr1 orthologs. The other monophyletic group was assigned Dpr2. Based on the available genomic sequences, each vertebrate species,except for Xenopus, has only two full-length Dpr paralogs. The phylogenetic analysis revealed that FRODO and XDpr are a duplication event(Graf and Kobel, 1991) that is specific to Xenopus, hence we refer to XDpr1 as XDpr1a, and FRODO as FRODO/XDpr1b. The Dpr family has seven conserved domains(Fig. 1B and see Fig. S1 in the supplementary material). Only two conserved domains have defined motifs: the leucine zipper and the PDZ binding motif, which is required to interact with the Dvl PDZ domain.
Expression patterns of zebrafish Dpr1 and Dpr2
By RT-PCR, dpr1 is expressed only zygotically, while dpr2is expressed both maternally and zygotically(Fig. 2). With regard to expression patterns, both genes are expressed on the future dorsal side of the embryos at sphere stage (Fig. 2C, part a; Fig. 2D, part a). At shield stage, both dpr1 and dpr2are expressed in the shield as well as the margin(Fig. 2C, part b; Fig. 2D, part b). dpr1is more highly expressed in the involuted hypoblast, but is expressed at low levels in the marginal hypoblast and epiblast (compare Fig. 2C, part b with 2D, part b). dpr2,however, is highly expressed throughout the shield and immediately lateral areas (Fig. 2D, part b). At 80%epiboly, dpr1 and dpr2 are expressed in the axial and lateral mesoderm and the ectoderm (Fig. 2C, part c; 2D, part c).
At approximately tailbud stage, dpr1 is expressed in specific ectodermal and mesodermal domains. dpr1 is expressed in a stripe in the brain anterior to pax2.1 (arrows; Fig. 2C, part d). dpr1staining is also seen in the posterior hindbrain and anterior spinal cord(arrowheads; Fig. 2C, part d). To determine the localization of dpr1 expression in the brain, in situ hybridization was performed with probes for dpr1 (blue), pax6.1 (red), a diencephalic marker, and krox20 (red), a marker for rhombomeres 3 and 5 (arrowheads; Fig. 2C, part e). At tailbud stage, dpr1 expression completely overlaps that of pax 6.1(not shown). By the five-somite stage, dpr1 (black arrow; Fig. 2C, part e) and pax6.1 (red arrow; Fig. 2C, part e) resolve into separate domains, with dpr1anterior and pax6.1 posterior(Fig. 2C, part e). After tailbud stage, dpr1 is no longer expressed in the axial mesoderm, but is expressed in specific areas of the somites during somitogenesis, as is evident at the 12 somite stage. The anterior, older 3 or 4 somites have dpr1 staining throughout the somites (black arrow; Fig. 2C, part f). The more posterior, younger somites have dpr1 staining in the posterior of the somites (black arrowhead; Fig. 2C, part f). In the posterior, dpr1 stains the paraxial mesoderm (asterisk), but is excluded from the tailbud (yellow arrow; Fig. 2C, part f). Later expression patterns for dpr1 are presented in Fig. S2 in the supplementary material.
At tailbud stage, dpr2 is maintained in the axial mesoderm and in the anterior brain. This brain expression is transient and is predominantly lost by three somites (not shown). dpr2 expression is maintained in the lateral mesoderm, and occurs in a banding pattern that is reminiscent of wnt11 (arrowhead; Fig. 2D, part d) (Kilian et al.,2003). At the three somite stage, dpr2 is expressed posterior to a sharp boundary in the posterior hindbrain and anterior spinal cord. Double staining with dpr2 (blue) and krox20 (red)reveals this sharp boundary is adjacent to r5 at three somites (red arrow; Fig. 2D, part e). Similar to dpr1, dpr2 expression is quite specific during somitogenesis(Fig. 2D, part f). By the 12 somite stage, dpr2 is upregulated in the anterior of the older somites (Fig. 2D, part f,arrow), but there is little dpr2 expression in the nascent somites(arrowhead; Fig. 2D, part f). dpr2 is also expressed in the posterior presomitic mesoderm and tailbud. Later expression patterns for dpr2 are presented in Fig. S3 in the supplementary material. The data of Gillhouse et al.(Gillhouse et al., 2004),published during revision of this manuscript, support these patterns of expression of dpr1 and dpr2.
Dpr1 and Dpr2 have unique functions in development
In order to study the requirements for endogenous Dpr1 and Dpr2 in zebrafish development, we used morpholino antisense oligonucleotides (MOs)(Draper et al., 2001; Nasevicius and Ekker, 2000). The dpr1 MO was designed to block the proper splicing of the second intron/exon boundary. The dpr1 MO was able to significantly abrogate proper splicing by the appearance of an ∼1700 bp band as analyzed by RT-PCR (Fig. 3A). The morpholino to dpr2 was designed to the 5′-UTR. To test if this specifically blocked translation, the 5′-UTR was cloned upstream of luciferase (5′Dpr2-luc). The dpr2 MO specifically blocked the in vitro translation of the 5′-UTR Dpr2-luciferase construct, relative to an internal control β-galactosidase(Fig. 3B).
During gastrulation dpr1 morphants (morpholino-injected embryos)have apparently normal domains of expression of expression of bozozok, a dorsal marker and β-catenin target gene (data not shown). At approximately the 10 somite stage, dpr1 morphants are slightly smaller (n=162/178 – 91%), but still do not have significant problems with specification of cell fate, as assayed by a cocktail of probes for in situ hybridization containing catL (hatching gland), opl (telencephalon), en2 (midbrain-hindbrain boundary), krox20 (rhombomeres 3 & 5), and myod (adaxial cells and somites) (compare Fig. 3C, part b with 3C, part a). The most noticeable defect is that the brain anterior to the midbrain-hindbrain boundary is smaller. This can been seen by comparing the distance between the catL and opl, and opl and en2 (compare distances between arrowheads in Fig. 3C, part b and 3D, part a).
dpr2 morphants (n=258/289 – 89%) have a different phenotype that is indicative of aberrant convergent extension movements during gastrulation (Fig. 3C, part c; 3D, parts c,d). The body axes are short and wide, and there are no major changes in cell fates. Consistent with this observation, the anterior of the embryos does not extend as far around the yolk (compare arrowheads in Fig. 3D, part c with part a), the notochord is wider and somites are elongated (compare distance between small, posterior arrowheads in Fig. 3C, part c with those in 3C, parts a,b; note elongated somites in Fig. 3D, part d compared with those in 3D, part b). Similar results were obtained during somitogenesis with a different MO, dpr2 MO(2), that abrogates proper splicing of this gene (data not shown).
To see if Dpr1 and Dpr2 are functionally redundant we injected both dpr morpholinos. The phenotype (n=121/137 – 88%) was strictly additive (Fig. 3C,compare part d with parts a-c). We conclude that dpr1 morphants have a subtle phenotype leading to reduction of the midbrain and diencephalon,while dpr2 appears to regulate convergent extension movements. However, repeating the morpholino injections in sensitized, hypomorphic,backgrounds proved to be informative, as discussed below.
Dpr1 is an enhancer of Wnt/β-catenin signaling
Although dpr1 morphants do not have a severe phenotype, the pattern of dpr1 expression could be consistent with it playing a role in early Wnt-8/β-catenin signaling that is required for ventralization and posteriorization of embryos (Lekven et al., 2001). To determine if dpr1 functionally interacts with the Wnt8 signal during early zebrafish development, we injected the dpr1 MO with a low amount of two wnt8 morpholinos(MO2, ORF1 MO+ORF2 MO), a dose that produces no phenotype or a weak hypomorphic phenotype (Lekven et al.,2001). Embryos were first examined at 24 hours, and scored on a 4 point scale with 0 being wild type, 1 having modest enlargement of the telencephalon, 2 having enlargement of the anterior and reduction of the posterior, and 3 having severe enlargement of the anterior and reduction of the posterior (Fig. 4A). Injection of wnt8 MO2 plus control MO yielded embryos with a slightly larger telencephalon than controls(Fig. 4A, arrow in panel 1 and graph). dpr1 MO plus control morpholino injected embryos are equivalent to dpr1 MO alone (Fig. 4A, graph). However, when the dpr1 MO was injected with the wnt8 MO2, the morphants phenocopied strong loss of Wnt8 function (Fig. 4A, graph). These embryos had enlarged telencephalons (arrowhead in panel 3), eyes, and reduced posterior (trunk and tail). We confirmed these results using a second dpr1 morpholino (dpr1 MO2, not shown).
To ensure that this phenotype was due to problems in specification of ventral and posterior cell fates attributable to wnt8 ORF1+2 function, embryos were examined at the one- to two-somite stage by in situ hybridization using probes for opl, pax2.1 and tbx6(ventrolateral mesoderm). In agreement with the 24 hour phenotype and a role for Dpr1 in early specification events, co-injection of suboptimal amounts of wnt8 MO2 plus dpr1 MO yielded embryos that had large expansion of opl coupled with reduction of tbx6,phenocopying strong Wnt8 loss of function (compare Fig. 4B, parts g-j with 4B, parts o,p). The embryos depicted in Fig. 4B, parts g,h are representative of a score of 2, while 4B, parts i,j are representative of a score of 3 at 24 hours. wnt8 MO2 plus a control morpholino yielded embryos with a moderate expansion of anterior tissues as indicated by the posterior expansion of opl and pax2.1 (compare Fig. 4B, parts c,d with parts a,b). Injection of dpr1 MO plus a control morpholino yielded embryos with no major changes in cell fate(compare Fig. 4B parts e,f with 4B, parts a,b). Although the dpr1 morphant phenotype is subtle, when dpr1 perturbation is coupled with moderate wnt8 attenuation, the resulting dpr1 +wnt8 morphants phenocopy strong loss of wnt8 function. We conclude that Dpr1 is required as an enhancer of Wnt8 signaling.
Given that dpr2 morphants have convergence extension problems, we would not expect to see the same interaction with Wnt8. To test this hypothesis, we injected the hypomorphic dose of wnt8 MO2plus a maximal dose of the dpr2 MO. In dpr2 MO plus control morpholino, the embryos did not extend as far around the yolk and their bodies are wider (Fig. 4B, parts k,l). Injection of dpr2 MO plus wnt8 MO2 is strictly additive (compare Fig. 4B,parts m,n with parts k,l and parts c,d). The embryos are shorter and wider,coupled with modest expansion of opl, like wnt8MO2 alone. We conclude that Dpr2 does not play a role in Wnt8 signaling in early zebrafish development.
Dpr2 functionally interacts with the Wnt/Ca2+-PCP pathway(s)
The dpr2 morphant phenocopies loss of PCP components like strabismus/trilobite (stbm), kynpek (kny)and prickle (pk) (Fig. 5A) (Jessen et al.,2002; Park and Moon,2002; Topczewski et al.,2001; Veeman et al.,2003b). At 24 hours dpr2 morphants are noticeable shorter, with block-like somites (Fig. 5A, arrow) and they have eyes that have not properly separated owing to improper migration of underlying mesodermal tissues(Fig. 5A, arrowhead). To test if dpr2 functionally interacts with stbm, we injected a low hypomorphic dose of the stbm MO with a low hypomorphic dose of the dpr2 MO. Embryos were then scored on a scale of 0 to 2, with 0 being wild-type, 1 having mild convergence extension (CE) problems and 2 having strong CE problems (Fig. 5B). Injection of either morpholino alone (plus control morpholino) yielded 10% and 15% of the embryos with mild CE problems for stbm MO and dpr2 MO, respectively (Fig. 5C). None of the embryos had strong CE problems(Fig. 5C). However,co-injection of the stbm MO and dpr2 MO yielded 30% of the embryos having mild CE problems and 55% of the embryos with strong CE defects. Thus, we conclude there is a functional interaction between Dpr2 and Stbm.
dpr2 morphants have anteriorly improperly placed eyes, but they rarely have cyclopia (Fig. 5D,parts g,h). Although perturbations in genes affecting CE may not individually have a high percentage of cyclopia, in combination they can genetically interact to yield a high percentage of cyclopic embryos(Marlow et al., 1998). To test if Dpr2 functionally interacts with Wnt11 by increasing the percentage of cyclopic embryos, we used a wnt11 MO that phenocopies wnt11/slb mutants (Lele et al.,2001). A suboptimal dose of wnt11 MO was injected with a suboptimal dose of dpr2 MO. Embryos were scored on a cyclopia index of 0 to 2, with 0 being wild type, 1 having forwardly positioned narrow eyes,and 2 being cyclopic and having fusion of the lenses(Fig. 5D, parts a-f). Injection of wnt11 MO plus control morpholino yielded a low percentage of cyclopic embryos (17%; Fig. 5E). However, 65% of embryos co-injected with wnt11 MO and dpr2 MO were cyclopic (Fig. 5E). The percent of cyclopic embryos was increased without affecting the posterior of the embryos (not shown). Therefore, we conclude that dpr2 functionally interacts with wnt11/slb as well as with stbm/tri.
To test if these functional interactions with stbm/tri and wnt11/slb are specific to Dpr2 and not Dpr1, we co-injected embryos with a higher dose of dpr1 MO(2) and either the stbm MO or wnt11 MO (Fig. 5C,E). The scoring of these embryos was almost identical to the stbm MO or wnt11 MO plus control morpholino. Therefore, there is not a functional interaction between stbm or wnt11 and dpr1.
Association with zebrafish Dvl2 and subcellular localization of Dpr orthologs
After injection of dpr1-myc or dpr2-myc RNA with zebrafish dvl-HA RNA into Xenopus embryos, followed by immunoprecipitation with anti-HA antibodies, both Dpr1-myc and Dpr2-myc proteins were detected by western blot(Fig. 6A,B). Thus, both zebrafish Dpr family members can associate either directly or indirectly with Dvl2, as reported for Xenopus Dpr family members(Cheyette et al., 2002; Gloy et al., 2002).
To determine the subcellular localization of ectopic Dpr1 and Dpr2,immunochemistry was performed using Xenopus animal caps injected with the dpr1-myc or dpr2-myc RNA. In this heterologous system,chosen in order to take advantage of the large size of the cells, both ectopic proteins are cytoplasmic (compare Fig. 6C with 6D). However, Dpr2 is also visible in the nuclei of some cells (small arrows) and has more intense staining at the cell membrane (large arrow).
Gain of function of Dpr1 and Dpr2 in Xenopus and HEK293T cell assays
Overexpression of signaling proteins is useful in establishing the activity of a protein in a specific context, and thus learning its potential involvement in a given pathway. We asked whether gain-of-function of Dpr1 and Dpr2 might affect Wnt/β-catenin signaling. Unfortunately the overexpression studies could not be conducted in zebrafish, as injection of Dpr RNAs into embryos leads to a high percentage of early embryonic lethality and non-specific toxicity. Therefore, the gain-of-function assays were conducted in Xenopus embryos and in HEK293T cells, both of which are responsive to Wnt signals.
We first tested whether Dpr1 and Dpr2 induce Wnt/β-catenin target genes in Xenopus animal caps and found that dpr1 but not dpr2 RNA induces sia and Xnr3(Fig. 6E, compare lane 1 with lanes 2 and 3). Consistent with this result, ventral injection of dpr1 RNA induced partial secondary axes in 36% of Xenopusembryos (n=78), while dpr2 RNA induced partial secondary axes in only 5% of Xenopus embryos (n=52) (not shown). Injection of less than 4 ng of dpr1 or dpr2 RNA did not induce Wnt/β-catenin target genes or secondary axes (not shown).
While injecting embryos for these animal cap experiments we observed that injection of dpr2 RNA into the animal pole resulted in aggregation of pigment around the site of injection, whereas injection of dpr1 RNA and β-gal control RNA did not (see Fig. S4 in the supplementary material). This result, which we do not understand in terms of mechanisms,further suggests functional differences in the activities of the two Dpr proteins.
Despite differences in the ability of injected dpr1 and dpr2 RNAs to induce Wnt/β-catenin target genes, we found that both Dpr orthologs synergize with zebrafish Dvl2 to induce Wnt/β-catenin target genes in Xenopus animal caps(Fig. 6F, compare lanes 2-7 with control lane 8; see Fig. S5 in the supplementary material for similar results with Xenopus Dpr orthologs). However, dpr1synergizes more strongly (n=6 experiments). This is probably not due to preferential expression of Dpr1, as Dpr1 and Dpr2 translate at similar levels in vitro (Fig. 6G) and tagged proteins Dpr1-myc and Dpr2-myc, which are functional equivalent to untagged Dpr1 and Dpr2, are expressed similarly in frog embryos (not shown and Fig. 6G). This synergistic activation of Wnt/β-catenin target genes by Dpr1 and Dvl2 was blocked by co-expression with Axin (Fig. 6H), consistent with its being a β-catenin-dependent synergy.
As Dpr1 and Dpr2 synergize with Dvl2, a component of Wnt signaling pathways, we next asked if either would synergize with Wnt itself to activate Wnt/β-catenin target genes. Neither zebrafish ortholog synergized with Wnt8, nor did either inhibit activation of target genes(Fig. 6I). We then turned to an independent assay system to investigate effects of co-expression of Dpr orthologs with Dvl2 or Wnt. HEK293T cells are responsive to Wnt/β-catenin signaling, which can be monitored by the luciferase activity of Super(8X)TOPFlash, a β-catenin-responsive reporter construct(Veeman et al., 2003b). Consistent with what we observed in Xenopus animal cap experiments,both zebrafish Dpr1 and Dpr2 synergize with zebrafish Dvl2 after transient transfection in HEK293T cells (Fig. 6J). However, both Dpr orthologs were inhibitory to the Wnt-mediated activation of the luciferase reporter in this cell line (see Fig. S6 in the supplementary material).
As the Dvl-associated protein Dpr1 synergizes with Dvl2, and does so to a greater extent than Dpr2, we next investigated whether Dpr1 could synergize with other proteins that interact with Dvl. We focused on CK1ϵ, protein kinase CK2 (here referred to as CK2), Par1 and GBP/FRAT-1. We expected that these Dvl-interacting proteins should synergize with Dvl2 in animal cap assays to activate Wnt/β-catenin target genes, which is what we observed(Fig. 7A, compare lanes 2-6 with lane 1).
We next asked whether Dpr1 synergizes with any of three different Dvl-associated kinases beginning with CK1ϵ, which is a potent activator of Wnt/β-catenin signaling (Hino et al., 2003; McKay et al.,2001; Peters et al.,1999). We found that CK1ϵ and Dpr1 synergize in inducing Wnt/β-catenin target genes in Xenopus animal caps(Fig. 7B, compare lanes 2 and 3 with lane 1). To determine whether this Dpr1-CK1ϵ synergy requires the kinase activity of CK1ϵ, we co-injected dpr1 RNA with ck1-D→N (a less active form of CK1ϵ) RNA. The ck1-D→N was substantially less active in synergizing with Dpr1 (Fig. 7B, compare lanes 4 and 5 with lanes 2 and 3). Like its antagonization of the synergy between Dpr1-Dvl2, Axin was also able to inhibit Dpr1-CK1ϵ synergy (not shown).
We then tested whether Dpr1 could synergize with a second Dvl-associated kinase, CK2, to induce Wnt/β-catenin target genes. As CK2 functions as a hetero-tetramer, `CK2' indicates that we injected RNA encoding both αand β subunits (Pinna,2002) unless otherwise noted. Although ck2 alone did not induce the target genes, co-injection of ck2 RNA with dpr1RNA induced β-catenin target genes(Fig. 7C, compare lanes 2 and 3 with lane 1). As CK2 functions as a hetero-tetramer, with the β subunit being required to stabilize the α subunit(Pinna, 2002), we tested the specificity of the Dpr1-CK2 synergy by co-injecting just the ck2α subunit RNA with dpr1 RNA. No synergy was observed (data not shown).
We next tested whether there is a functional interaction between Dpr1 and the Dvl-associated kinase Par1. At any dose tested, par1 alone did not activate Wnt/β-catenin target genes in animal caps. However, when par1 RNA was co-injected with dpr1 RNA, there was a synergistic activation of Wnt/β-catenin target genes(Fig. 7D, compare lanes 2 and 3 with lane 1). To determine whether this functional interaction was dependent upon Par1 kinase activity, we co-injected dpr1 RNA with a higher dose of par1-KN (a kinase dead form of Par1) RNA. Dpr1 did not synergize with Par1-KN (Fig. 7D, compare lanes 4 and 5 with lanes 2 and 3).
Finally, we tested whether GBP/FRAT-1 and Dpr1 synergize to activate Wnt/β-catenin target genes. We only observed a mild enhancement between Dpr1 and GBP/FRAT-1 when co-injected (Fig. 7E). As Par1 synergizes with Dpr1, while the GBP/FRAT-1-mediated gene activation was weakly enhanced by Dpr1, we deleted the Par1-GBP/Frat-1-binding domain (Hino et al., 2003) in zebrafish Dvl2, resulting in the construct Dvl2ΔPGB (ΔPar1 and GBP Binding domain). In principle, this should result in a Dvl2 protein that would be unable to interact with endogenous GBP or Par1. Like Hino et al, (Hino et al.,2003), we found this construct could activate Wnt/β-catenin target genes alone (not shown). However, we found that Dvl2ΔPGB did not synergize with Dpr1 in the Xenopus animal cap assay(Fig. 7F, compare lane 2 with lane 1).
In Xenopus explant assays, the above data show that Dpr1 can functionally interact with Dvl2 and the Dvl-interacting proteins CK1ϵ,Par1 and CK2, but it does not functionally interact as well with GBP. We next turned to prospective loss-of-function studies using the CK1ϵ inhibitor CKI-7 (Peters et al., 1999). CKI-7 had no apparent effect on gene induction by stabilized β-catenin, a positive control for inducing sia and Xnr3(Fig. 8A, compare lane 5 with lane 6). Whether gene activation by Dvl requires CK1ϵ has not been explored. In addressing this, we found that Dvl-mediated induction of target genes was not affected by CKI-7 injection(Fig. 8A, compare lane 3 with lane 4), though it has been reported that Dvl-mediated axis duplication in Xenopus can be inhibited by CKI-7(Peters et al., 1999). We then tested if Dpr1-Dvl2 synergy requires CK1ϵ. Dpr1 and Dvl2 did not synergize in gene induction when CKI-7 was co-injected(Fig. 8B, compare lane 1 with lane 2). Thus, Dpr1 requires endogenous CK1ϵ to enhance Dvl activity. We also found that the synergy of Dpr1-Par1 requires endogenous CK1ϵactivity (Fig. 8B, compare lane 5 with 6). Thus, Dpr1 is dependent on CK1ϵ for synergistic interactions with both Dvl2 and Par1.
Last, we determined if Dpr1-CK2 synergy in inducing Wnt/β-catenin target genes in Xenopus explants requires CK1ϵ, and found that it does not (Fig. 8B, compare lane 3 with lane 4). However, CK2-Dvl2 synergy is dependent upon CK1ϵ(Fig. 8A, compare lane 1 with lane 2). Therefore, both Dpr1 and CK2 synergistically interact with Dvl2 in a CK1ϵ-dependent manner, but excess Dpr1 and CK2 overcomes a requirement for CK1ϵ to enhance activation of Wnt/β-catenin target genes(summarized in Fig. 8C).
Potential Dpr mechanisms
We further attempted to determine any mechanisms by which Dpr1 and Dpr2 might affect Wnt pathways. With respect to the Wnt/Ca2+-PCP pathway, Rho activation has been proposed to be a potential indicator of this poorly understood pathway(s) (Habas et al.,2001). However, neither Dpr1 nor Dpr2 affected Rho activation in HEK 293T cells or in Xenopus embryo lysates (not shown). With respect to Wnt/β-catenin signaling, CK1ϵ has been shown to increase GBP association with Dvl-1 (Hino et al.,2003). However, we did not find that CK1ϵ enhanced Dpr1 or CK2 association with Dvl2, nor did we find that Dpr increased the association of other Dvl-interacting molecules with Dvl2 (not shown). Increased CK2 activity is correlated with increased Dvl-2 stability(Song et al., 2000). However,we did not find that Dpr or any other Dvl-interacting protein (including CK2)could enhance the stability of Dvl (not shown). Finally, Dvl has been proposed to undergo an electrophoretic mobility shift in correlation with Wnt or Fz activation (Lee et al., 1999; Rothbacher et al., 2000; Sun et al., 2001). We also did not find that Par1 or CK2, in the presence or absence of ectopic Dpr,increased the phosphorylation state of Dvl, as monitored by electrophoretic mobility (not shown). With the caveats associated with any negative data, we suspect that these potential mechanisms of action of Dpr orthologs are not the most likely to explain how Dpr family members participate in Wnt signaling pathways.
Zebrafish Dpr paralogs can regulate different Wnt-dependent pathways
Paralogous vertebrate Wnt and Fz genes have been found to function in Wnt/β-catenin and/or in Wnt/Ca2+-PCP pathways. For example,zebrafish Wnt8 is thought to activate β-catenin signaling and to be required for regulation of gene expression and cell fates(Erter et al., 2001; Lekven et al., 2001). Wnt/Ca2+-PCP signals, which are elicited by Wnt-11/Silberblick and Wnt-5/Pipetail, are not involved in specification of cell fates, but control the cell movements during gastrulation that allow the embryo to elongate(Heisenberg et al., 2000; Rauch et al., 1997). In this report, we have provided data consistent with zebrafish Dpr1 functioning as an enhancer of Wnt8/β-catenin signaling that specifies ventral and posterior cell fates. We have also provided evidence that Dpr2 functions as an enhancer of the Wnt/Ca2+-PCP components Stbm/Tri and Wnt-11/Slb, thereby controlling cell movements during gastrulation. The primary conclusion of this study is that the two zebrafish Dpr paralogs appear to be required for processes regulated by β-catenin or Wnt/Ca2+-PCP pathways. This parallels the requirement for Wnt and Fz proteins in each of these pathways.
Gain- and loss-of-function of Dvl, as well as structure-function analysis are consistent with the idea that specific Wnt and Fzs somehow activate Dvl in both vertebrate Wnt pathways. How Dvl regulates multiple Wnt pathways is not known. One hypothesis, based primarily on gain-of-function assays in cell culture, predicts that Dvl-interacting proteins direct Dvl function into a specific Wnt/Fz pathway (Sun et al.,2001; Yan et al.,2001). Although this is an attractive hypothesis, there is yet little evidence to support it. Most loss-of-function perturbations of the Wnt/Ca2+-PCP pathway do not produce gain-of-function phenotypes of the Wnt/β-catenin pathway and vice versa. Examples include dshmutants in Drosophila, wnt11/slb and stbm/tri mutants in zebrafish, and daam1 Xenopus morphants(Axelrod et al., 1998; Habas et al., 2001; Heisenberg et al., 2000; Jessen et al., 2002). In addition, Drosophila nkd is not required for the PCP pathway(Rousset et al., 2001). Nevertheless, a few recent reports provide some evidence that in specific contexts there may be opposition of these pathways, though they do not point to the opposition occurring at Dvl (Topol et al., 2003; Westfall et al.,2003). Thus, in some contexts there might be mechanisms by which Dvl is free to regulate either β-catenin or Wnt/Ca2+-PCP pathway without affecting the other.
In the current study, we provide evidence that the zebrafish Dpr paralogs represent examples of Dvl-interacting paralogs that appear to be functionally associated with separate Wnt pathways. Importantly, they do not appear to have redundant functions, yet they have largely overlapping expression patterns in the early zebrafish development and both can interact with zebrafish Dvl2. Therefore, the Dpr morphant phenotypes are consistent with a hypothesis whereby the Wnt/β-catenin and Wnt/Ca2+-PCP pathways can co-exist independently of each other.
Zebrafish Dprs are enhancers of Wnt signaling
The loss-of-function data in zebrafish support the general conclusion that Dpr orthologs are enhancers of multiple Wnt pathways. The Dpr1 morphants have subtle phenotypes that do not lead to any discernable loss of cell fates, yet they have strong Wnt loss-of-function phenotype in a mildly hypomorphic Wnt8 background. These results are reminiscent of those obtained in CK1 RNAi experiments, which also acts as a signaling enhancer in nematodes(Peters et al., 1999). Alone,RNAi of CK1 gives a low percentage of worms with the more-mesoderm (MOM)phenotype. However, when combined with loss of another component of the Wnt pathway, there is a strong MOM phenotype. Likewise, while loss of Dpr2 has a phenotype similar to Wnt/Ca2+-PCP mutants, the strongest affected dpr2 morphants are not as severe as the strongest affected stbm/tri alleles or morphants(Jessen et al., 2002; Park and Moon, 2002). Nevertheless, there is a strong functional interaction between dpr2and stbm/tri or wnt11/slb, indicating that dpr2 may act to promote Wnt/Ca2+-PCP pathway activity.
How is it possible that the zebrafish dpr2 morphant indicates a loss of Wnt/Ca2+-PCP activity, but when overexpressed it can synergize with Dvl2 to activate Wnt/β-catenin targets? We do not interpret these results as being mutually exclusive. Wnt components have been demonstrated to be context-dependent activators of Wnt/β-catenin or Wnt/Ca2+-PCP pathways. Wnt5a/pipetail is probably the best example. Wnt5a overexpression in Xenopus leads to improper convergent and extension movements (Du et al.,1995). Similarly, a loss-of-function mutation of wnt5/pptin zebrafish leads to improper convergence extension movements in the tail(Rauch et al., 1997; Westfall et al., 2003). Thus,the overt phenotype of loss or gain of Wnt/Ca2+-PCP components is similar. However, overexpression of Wnt5a along with Fz5 or LRP6 activates Wnt/β-catenin target genes in animal caps and duplicates axes in Xenopus embryos (He et al.,1997; Tamai et al.,2000). Furthermore, expression of high levels of Wnt11 and some Fzs can weakly duplicate axes and induce β-catenin target genes,emphasizing a caveat for overexpression of paralogous proteins(Du et al., 1995; Ku and Melton, 1993; Sheldahl et al., 1999).
Comparison with the Xenopus Dpr family
We have previously reported that Xenopus Dpr1a could function as an inhibitor of Wnt/β-catenin signaling(Cheyette et al., 2002). Gloy et al. (Gloy et al., 2002)initially reported that the related FRODO/XDpr1b acted like an activator of Wnt signaling. In an effort to resolve the apparent discrepancies between the first two studies, the Sokol laboratory has compared the signaling activities of the two Dpr family members side by side in gain- and loss-of-function studies. They conclude that both XDpr1a and FRODO/XDpr1b can act as either activators or inhibitors, depending on the level at which the pathway is activated, and the doses of the proteins(Hikasa and Sokol, 2004). To understand better the normal functions of Dpr family members we undertook the present study in zebrafish in part to exploit the ability to conduct loss-of-function of Dpr orthologs in sensitized backgrounds. Although our new experiments support the hypothesis that endogenous Dpr orthologs are positive modulators of Wnt signaling in zebrafish under the conditions assayed, we do not dispute the conclusions of Cheyette et al.(Cheyette et al., 2002) and Hikasa and Sokol (Hikasa and Sokol,2004) that Xenopus Dpr family members can act as inhibitors in some contexts. We propose that further analysis of Dpr family members in other contexts besides frog and fish embryos may help determine the mechanisms of action of these interesting proteins.
In conclusion, we provide loss-of-function evidence that Dpr1 but not Dpr2 is required for proper dorsoventral and anteroposterior patterning in zebrafish that are mildly hypomorphic for Wnt8, indicating that Dpr1 probably acts as an enhancer of Wnt8/β-catenin signaling in early development. Dpr2 but not Dpr1 acts as an enhancer of stbm/tri and wnt11/slb in the Wnt/Ca2+-PCP pathway(s). Thus, in early zebrafish development, loss-of-function evidence indicates Dpr1 and Dpr2 are not redundant, and they function to regulate Wnt/β-catenin and Wnt/Ca2+-PCP pathways, respectively. Regarding mechanisms of action, how Dpr2 functions is unclear, largely because of the lack of understanding of the Wnt/Ca2+-PCP pathway and the lack of robust assays. With regard to Dpr1 orthologs, in the majority of contexts they function as positive regulators of Wnt/β-catenin signaling, though they can inhibit this pathway in various overexpression assays. We extend our analysis of the Dvl-associated Dpr1 orthologs by demonstrating that zebrafish Dpr1 synergizes in gene induction with three Dvl-associated kinases,CK1ϵ, Par1 and CK2. Moreover, we demonstrate that gene regulation by Dpr1 in some contexts is dependent upon CK1ϵ. We look forward to genetic and proteomic approaches being employed to further test these findings on the functions and possible mechanisms of action of Dapper family members.
Note added in proof
During the final preparation of this manuscript, Zhang et al.(Zhang et al., 2004) reported dpr2 morphants have a similar phenotype as reported here.
We thank D. Raible, C. Thorpe and J. Ragland; the reviewers; and the editor who handled the manuscript, Jim Smith, for advice and comments. We thank J. Webster and S. Angers for technicalassistance; T. Tran, M. Moffett and E. Prince for frog and fish care; and H. Roelink for CKI-7. We are indebted to A. Kaykas, J. Miller and S. Sokol for communicating unpublished results. J.S.W. was in part supported in part by Training Grant NIH 2T32 HD07183-21. R.T.M. is an Investigator of the HHMI. This work was supported by NIH RO1 HD27525.