The maternal Toll signaling pathway sets up a nuclear gradient of the transcription factor Dorsal in the early Drosophila embryo. Dorsal activates twist and snail, and the Dorsal/Twist/Snail network activates and represses other zygotic genes to form the correct expression patterns along the dorsoventral axis. An essential function of this patterning is to promote ventral cell invagination during mesoderm formation,but how the downstream genes regulate ventral invagination is not known. We show here that wntD is a novel member of the Wnt family. The expression of wntD is activated by Dorsal and Twist, but the expression is much reduced in the ventral cells through repression by Snail. Overexpression of WntD in the early embryo inhibits ventral invagination,suggesting that the de-repressed WntD in snail mutant embryos may contribute to inhibiting ventral invagination. The overexpressed WntD inhibits invagination by antagonizing Dorsal nuclear localization, as well as twist and snail expression. Consistent with the early expression of WntD at the poles in wild-type embryos, loss of WntD leads to posterior expansion of nuclear Dorsal and snail expression,demonstrating that physiological levels of WntD can also attenuate Dorsal nuclear localization. We also show that the de-repressed WntD in snail mutant embryos contributes to the premature loss of snail expression, probably by inhibiting Dorsal. Thus, these results together demonstrate that WntD is regulated by the Dorsal/Twist/Snail network,and is an inhibitor of Dorsal nuclear localization and function.
Mesoderm is the middle germ layer formed during gastrulation. In Drosophila, the mesoderm arises from the invagination of the ventral cells of the blastoderm. The mesoderm provides the precursor cells for muscles, hemocytes, lymph glands, the somatic gonad and the heart. The maternal Toll signaling pathway has a crucial role in establishing the ventral cell fate and thus mesoderm formation(Anderson, 1998; Roth, 2003; Stathopoulos and Levine, 2002; Wasserman, 2000).
Toll is a single-pass transmembrane receptor and is activated by a series of upstream serine proteases that processes the ligand Spätzle(Hashimoto et al., 1988; Hu et al., 2004; LeMosy et al., 1999; Morisato, 2001; Weber et al., 2003). The activated Toll recruits the cytoplasmic components MyD88, Tube and Pelle to regulate the nuclear transport of the transcription factor Dorsal(Charatsi et al., 2003; Kambris et al., 2003; Sun et al., 2004). Dorsal, a NF-κB homolog, is normally retained in the cytoplasm by Cactus, an IκB homolog. Toll signaling causes the phosphorylation and degradation of Cactus, thereby allowing Dorsal to enter the nucleus and regulate gene expression (Belvin et al.,1995; Bergmann et al.,1996; Fernandez et al.,2001; Reach et al.,1996). These signaling components are ubiquitously distributed,but the pathway is activated only in the ventral side of the embryo(LeMosy et al., 1999; Roth, 2003). Thus, activation of Toll by the diffusible Spätzle leads to the formation of a nuclear gradient of Dorsal, with the highest concentration in ventral nuclei(Anderson, 1998; Roth, 2003; Roth et al., 1989; Rushlow et al., 1989; Stathopoulos and Levine, 2002; Steward, 1989; Wasserman, 2000).
A single gradient of nuclear Dorsal can generate multiple patterns of zygotic gene expression along the dorsoventral axis(Jiang and Levine, 1993; Stathopoulos et al., 2002). Dorsal acts as both a transcriptional repressor and activator. For example, zerknüllt and decapentaplegic are repressed by Dorsal and therefore can be expressed only in the dorsal side of the embryo where the dorsal ectoderm is formed (Huang et al.,1993; Ip et al.,1991; Jiang et al.,1992; Pan and Courey,1992). Meanwhile, Dorsal activates other zygotic genes, such as twist, snail, rhomboid, short gastrulation, lethal of scute and single-minded (sim). Depending on the affinity of the Dorsal-binding sites and on the presence of co-activator sites on their promoters, these target genes are activated by different thresholds of the Dorsal gradient, and thus have ventral expression with variable lateral limits(Stathopoulos and Levine,2002).
High levels of nuclear Dorsal activate the expression of twist and snail, and the Dorsal/Twist/Snail network regulates ventral cell invagination to form the mesoderm (Ip and Gridley, 2002; Leptin,1999; Stathopoulos and Levine,2002). In dorsal, twist or snail mutants, no ventral invagination occurs and no mesodermal tissues are formed. Twist is a basic helix-loop-helix transcription factor and acts as a co-activator for Dorsal to optimally activate other zygotic target genes, including snail. Snail contains five zinc fingers and functions as a transcriptional repressor (Hemavathy et al., 2000; Nieto,2002). A model for this gene regulatory network in promoting mesoderm formation is that Dorsal/Twist activates multiple zygotic genes that are expressed in the ventral region with different lateral limits. These target genes may promote the ventral (mesodermal) cell fate or the lateral(neuroectodermal) cell fate. Snail specifically represses those genes that are not compatible with mesoderm formation. Consistent with this model, many genes, including rhomboid, sim, lethal of scute, short gastrulation,crumbs, Delta and Enhancer of split, are repressed by Snail in the ventral region and their expression is, therefore, restricted to the lateral regions. In snail mutant embryos, these genes are de-repressed into the ventral region. However, it has not been demonstrated that any of these Snail target genes can directly inhibit ventral invagination and mesoderm formation (Hemavathy et al.,1997).
To identify novel components in the dorsoventral pathway, we carried out a microarray assay using embryos derived from gain-of-function and loss-of-function mutants of the Toll pathway. Among the novel genes identified, we analyzed the expression and function of wntD because the Wnt family of secreted proteins regulates patterning, cell polarity and cell movements (Nelson and Nusse,2004; Veeman et al.,2003). Our results show that wntD is activated by Dorsal and Twist but repressed by Snail. Increased expression of WntD in wild-type early embryos inhibits ventral invagination. Thus, wntD is the first Snail target gene shown to have an interfering function in mesoderm invagination. We also demonstrate that the overexpressed WntD blocks invagination by inhibiting Dorsal nuclear localization. Loss-of-function analyses also show that physiological levels of WntD can attenuate Dorsal nuclear localization and function. Therefore, wntD is a novel downstream gene of the Dorsal/Twist/Snail network and can feed back to inhibit Dorsal.
Materials and methods
Drosophila stocks and genetics
Control strains used were OregonR or y w. Transgenic lines were generated by microinjection of the P element plasmid together with the Δ2-3 transposase helper plasmid into y w embryos. The UAS-wntD flies used for most of the experiments were homozygous for the transgenes on both second and third chromosomes. The pCaSpeR-wntDtransgenic flies and the Df(3R)l26c flies were mated with a double balancer chromosome strain, and then mated together to establish a stable line containing the homozygous rescue construct on the second chromosome and Df(3R)l26c on the third chromosome, with a balancer. The snailHG31, twistID96, dorsalH,Df(2L)TW119, Toll10b, twistIIH snailIIGdouble mutant, DeltaB2 and nanos-Gal4 stocks were used.
Plasmids and cloning
OregonR genomic DNA was used as the template for PCR amplification of the wntD ORF. To generate the pUAST-wntD for embryonic expression experiments, the primers GATCGCGGCCGCTCAGTCGATCTAACGACATCGCAG and GATCGGTACCGTTGTGGTAATAAATTAGAGGTGG were used to amplify the wntD ORF together with 58 bp 5′ and 117 bp 3′ of the ORF. This fragment was subcloned into the NotI and Asp718 sites of pBluescript KS(+). This entire fragment was then excised with NotI and Asp718, and subcloned into the NotI and Asp718 sites of pUAST. To generate the pCaSpeR-wntD genomic rescue construct, a 5 kb genomic DNA region was amplified in two fragments using PCR. The 5′ fragment of 3184 bp was amplified using the primers GATCGGTACCGATCTGGTCGGTGGCCTCTTCAAC and GATCGGTACCGTTGTGGTAATAAATTAGAGGTGG, and then digested with Asp718 and NcoI. The 3′ fragment of 2842 bp was amplified using the primers GATCGCGGCCGCTCAGTCGATCTAACGACATCGCAG and GATCGCGGCCGCCAGACATCGACTTGTGCGACTGGC, and then digested with NcoI and NotI. The fragments were then ligated into the Asp718 and NotI sites of pBluescript KS(+). The 5 kb genomic clone was then digested with ApaI and AgeI and blunted with Klenow polymerase. This yielded a 2721 bp fragment that included the wntDORF plus 1558 bp of 5′ flanking sequence and 233 bp 3′ flanking sequence. This region does not contain any other annotated ORF. This 2721 bp fragment was blunt-end ligated into the XbaI site of the pCaSpeR vector.
Embryo in situ and antibody staining
Embryo in situ hybridization using digoxigenin-labeled probes was carried out as previously described (Hemavathy et al., 2004; Hemavathy et al.,1997). Double in situ hybridization for the simultaneous detection of snail and wntD transcripts was performed using digoxigenin-labeled snail (diluted 1:200) and biotin-labeled wntD (diluted 1:100) probes together during hybridization. After washing with buffers, embryos were incubated overnight at 4°C with sheep anti-digoxigenin Fab fragments conjugated to peroxidase (Roche, 1:1200 dilution). The peroxidase stain was developed using 0.5 mg/ml DAB in 1×PBS and 0.006% H2O2. Embryos were then washed five times in 1×PBT to remove the DAB/H2O2. They were incubated overnight with anti-Biotin Fab fragments conjugated to alkaline phosphatase (Roche, 1:2000 dilution) at 4°C. The alkaline phosphatase staining was developed with NBT/BCIP. Embryos were mounted in Permount and visualized under Nomarski optics. The monoclonal antibody 7A4 was used to stain for Dorsal, using the procedure as described(Hemavathy et al., 2004; Hemavathy et al., 1997). Goat anti-mouse IgG (Fab fragments) conjugated with the Alexa 488 fluorochrome(Molecular Probes) was used and the samples were mounted in Vectashield with DAPI (Vector Laboratories).
Drosophila wntD is expressed in a complex pattern during early embryogenesis
We used microarray chips (Affymetrix) to study the gene expression profiles of gastrulating embryos derived from wild-type, dorsal–/– and Toll10bflies. Toll10b codes for a gain-of-function Toll receptor(Schneider et al., 1991). Many known target genes, such as twist, snail, short gastrulation, tinmanand mef2, showed lower expression in the dorsal–/– sample and increased expression in the Toll10b sample, as predicted (data not shown). Among the novel targets, we selected the annotated gene CG8458 for further study because it encodes a member of the Wnt family, and Wnt proteins have been implicated in controlling cell polarity and cell movement in many organisms(Nelson and Nusse, 2004; Veeman et al., 2003).
The predicted amino acid sequence of CG8458 has the closest homology to cephalochordate Wnt8, chicken Wnt8C, and zebrafish Wnt8. Sequence alignment and pair-wise comparison show that CG8458 is a distal member of this subfamily(Fig. 1A,B). The average identity between CG8458 and other Wnt8 molecules is approximately 27%, while the identity among other members is higher than 50%. Nonetheless, 20 out of the 22 characteristic cysteine residues of Wnt proteins are conserved in CG8458 (Fig. 1A, asterisks). FlyBase(http://flybase.bio.indiana.edu/)has named this gene Drosophila Wnt8. However, a recent report suggests that this gene may not be an ortholog of vertebrate Wnt8 but instead an orphan Wnt gene (Kusserow et al., 2005). Based on our functional analysis, we elected to use the name Drosophila wntD for the annotated gene CG8458, and the encoded protein WntD (Wnt inhibitor of Dorsal). A similar microarray analysis was reported, but Drosophila wntD was not included probably because of the different criteria used for selection(Stathopoulos et al.,2002).
In situ analysis reveals that wntD mRNA is expressed in a dynamic pattern in the early embryo. There is no detectable maternally deposited RNA and the earliest zygotic expression is present at the anterior and posterior poles of early stage 4 embryos (Fig. 1C). Soon after, wntD is expressed in a few patches of ventral cells (Fig. 1D). This low level of expression remains in the ventral cells throughout the blastoderm stage. Meanwhile, expression arises in two lines of cells abutting the mesoderm (Fig. 1E,F). These two lines of staining coincide with the mesectoderm, the precursor of ventral midline cells (see Fig. 2). The expression of wntD in the mesectoderm persists during germ band extension and gradually disappears (Fig. 1G,H). De novo expression appears around stage 8 in the ventral neuroectodermal cells adjacent to the midline(Fig. 1G,H). This expression continues in the neuroectoderm through stages 9 and 10(Fig. 1I,J), and is reduced to an undetectable level by stage 11. No expression was detected in other stages of embryonic development. The expression pattern of wntD is largely different from other Drosophila Wnt genes(Eisenberg et al., 1992; Graba et al., 1995; Janson et al., 2001; Russell et al., 1992).
Genetic control of wntD expression by the dorsoventral pathway
We first performed double in situ hybridization to determine the location of the two lateral lines of wntD expression(Fig. 2A-F). Previous reports demonstrate that sim is expressed in the mesectoderm(Fig. 2C,D) and is strongly repressed by Snail in ventral cells (Kasai et al., 1992; Kosman et al.,1991; Leptin,1991). Embryos that contained both wntD and simin situ probes (Fig. 2E,F)showed essentially the same lateral pattern as embryos that contained either probe (Fig. 2A-D),demonstrating that the two patterns overlap. Double in situ hybridization also showed that, similar to sim, the lateral expression of wntDis at the border of the snail pattern (see Fig. 6A). Thus, we conclude that the two lateral lines of wntD expression are in the mesectoderm.
To understand the regulation of wntD, we analyzed its expression in various genetic mutants. No signal was observed in embryos derived from dorsal–/– mothers(Fig. 2G), demonstrating that the expression in both the trunk and the poles is absolutely dependent on Dorsal. In embryos derived from Toll10b mothers, the expression of wntD was expanded into the dorsal side but the overall staining was not stronger than wild type(Fig. 2H), probably as a result of both activation by Dorsal and repression by Snail (see below). In conclusion, the mRNA staining in dorsal–/– and Toll10b embryos corroborates the results of the microarray analysis.
In snail homozygous mutant embryos, a higher level of wntD expression was present throughout the ventral region but mesectodermal expression was not obvious(Fig. 2J). We also observed that in some heterozygous embryos there was normal mesectodermal staining but higher ventral expression of wntD(Fig. 2I). Gene expression in the mesectoderm is regulated by a complex interaction between the Notch pathway and Snail, such that the mesectodermal expression of sim also requires the positive input of Snail(Cowden and Levine, 2002; Morel et al., 2003; Morel and Schweisguth, 2000). The mesectodermal expression of wntD in both wild-type and snail mutant embryos is similar to that of sim, suggesting that wntD and sim are regulated by a similar mechanism. More importantly, the results demonstrate that Snail also represses wntDexpression in the ventral cells.
In twist mutant embryos, wntD showed a narrower version of the wild-type pattern, centered on the ventral midline(Fig. 2K). The Snail pattern is significantly reduced in twist mutant embryos(Kosman et al., 1991; Leptin, 1991; Ray et al., 1991). Therefore,the narrower Wnt8 pattern in twist mutants can be explained by the reduced expression of the repressor Snail. In twist snaildouble-mutant embryos, the expression of wntD was weak and only present in the ventral-most cells (Fig. 2L). We speculate that high levels of Dorsal are sufficient to activate this weak expression of wntD in the ventral nuclei. However,the overall ventral staining of wntD in the double mutant was much weaker than that in snail mutants (compare with Fig. 2J), suggesting that wntD is weakly activated by Dorsal and strongly activated by Dorsal/Twist cooperation, as has been shown for other target genes of the dorsoventral pathway (Ip et al.,1992b; Jiang and Levine,1993; Shirokawa and Courey,1997). A stronger activation by the Dorsal/Twist combination may also explain the detectable expression of wntD in the ventral cells of wild-type embryos despite the repression by Snail. Within 1.6 kb of the 5′ flanking sequence of wntD, there are seven sites that are similar to the Snail-binding consensus and five sites that are similar to Dorsal-binding consensus (data not shown). However, the demonstration of whether wntD is a direct target requires further evidence.
wntD expression in the neuroectoderm depends on Delta. In zygotic Delta mutant embryos, the early wntD pattern was largely unaffected but the late pattern during germ-band extension was reduced and subsequently lost (Fig. 2M,N). Early embryos contain a significant maternal load of Delta gene products. As a result, the expression of target genes such as simremains unaffected until later stages(Martin-Bermudo et al., 1995). The regulation of wntD by Delta in the neuroectoderm may depend on a similar mechanism.
Increased expression of WntD blocks presumptive mesoderm invagination
An essential biological function of the Dorsal/Twist/Snail network is to promote invagination of the ventral cells to form the mesoderm(Ip and Gridley, 2002; Leptin, 1999; Stathopoulos and Levine,2002). Although the repressor function of Snail is required for ventral invagination, none of the known target genes normally repressed by Snail has been directly implicated in disrupting ventral invagination(Hemavathy et al., 2004; Hemavathy et al., 1997). Because wntD is repressed by Snail, we increased wntDexpression in wild-type embryos in an attempt to phenocopy the defects in snail mutant embryos. The maternal nanos-Gal4 line was used to direct the ubiquitous expression of UAS-wntD in early embryos. We found that approximately 50% of these embryos at the gastrulation stage had observable defects in ventral invagination. Approximately one quarter of these defective embryos had completely lost the ventral furrow(Fig. 3B), and the others showed varying degrees of invagination with the anterior regions always being worse than the posterior regions. The ventral invagination defect is not a result of general problems in cell shape changes or cell movements because cephalic furrow formation and germ-band extension occurred normally in these embryos. Tissue sectioning confirmed the phenotype that the mesoderm was largely missing in gastrulating embryos(Fig. 3D).
The nanos-Gal4 female flies deposit maternally the Gal4gene products, which direct the UAS-dependent WntD expression ubiquitously in pre-blastoderm stage embryos. We also tested the rhomboid-Gal4 driver; this rhomboid promoter contains mutations in its Snail-binding sites and directs zygotic Gal4 expression in the ventral half of the blastoderm(Ip et al., 1992a). In these experiments, approximately 5% of embryos at gastrulation stage showed slightly defective invagination (data not shown). The rhomboid promoter, as well as other ventral zygotic promoters, is activated by Dorsal. Thus, the expression of WntD by zygotic promoters may be too late to induce a substantial phenotype. This speculation is consistent with the mechanism of feedback inhibition of Dorsal by WntD as shown below.
WntD blocks invagination by disrupting the expression of mesoderm determinants
The Wnt family of secreted proteins regulates cell fate, cell polarity,cytoskeleton and cell movement (Nelson and Nusse, 2004; Veeman et al.,2003). To elucidate the mechanism that underlies the invagination defect induced by WntD, we stained for various markers of cell fate and cell shape. We were surprised to find that twist and snailexpression became highly abnormal in the nanos-Gal4-driven WntD-expressing embryos. The twist pattern was narrower than 12-cell widths along the circumference, compared with 20-cell widths in wild-type embryos (Fig. 3E,F)(Kosman et al., 1991). The snail expression pattern was even more severely affected. A total of 93% (n=147) of WntD-expressing embryos at the blastoderm and gastrulation stages showed abnormality in the snail expression pattern. The abnormality is variable and ranges from a few cells narrower to an almost complete disappearance of the pattern(Fig. 3G-J). The anterior expression was always more severely affected, and the posterior expression was affected to various extents in different embryos. We quantitated the phenotype by counting the width of the snail expression domains at 50%egg-length. In WntD-overexpression embryos that we assigned to have a phenotype, the snail pattern varied from zero to 11 cells, with an average width of seven cells. For wild-type embryos, the width of the snail domain is 13 to 17, with an average of 15 cells. Thus, all the embryos that we assigned to have a phenotype showed quantitative defects.
sim and rhomboid are normally repressed by Snail in the presumptive mesoderm (Fig. 3K,M) (Kasai et al.,1992; Kosman et al.,1991; Leptin,1991). In WntD-overexpression embryos the sim pattern disappeared in the anterior region and the lateral rows of staining came closer in the posterior region (Fig. 3L). As described above, Snail represses sim expression in the ventral cells but the expression and positioning of sim in the presumptive mesectoderm also requires Snail(Cowden and Levine, 2002; Morel et al., 2003; Morel and Schweisguth, 2000). Therefore, the abnormal sim pattern follows exactly the reduced snail expression. rhomboid showed similar narrowing of the pattern (Fig. 3N), consistent with the model that Snail is a simple repressor of rhomboid. In conclusion, increased WntD expression causes highly reduced twist and snail expression, leading to abnormal expression of other genes in ventral cells. Even though increased WntD expression may also cause other defects, the reduced twist and snail expression is probably sufficient to account for the loss of invagination.
Negative regulation of Dorsal nuclear localization by WntD
The direct activator of twist and snail expression in the blastoderm is Dorsal (Ip et al.,1992b; Jiang et al.,1991). Therefore, we examined the distribution of the Dorsal protein. In wild-type blastoderm and gastrulating embryos, Dorsal shows the characteristic ventral nuclear pattern(Fig. 4A-C). By contrast, WntD overexpression caused low-level staining around the periphery of the whole embryo (Fig. 4G), and high-resolution imaging showed that the ventral cells had Dorsal proteins predominantly in the cytoplasm (Fig. 4H,I). This phenotype was similar to that of embryos derived from a gastrulation-defective mutant(Fig. 4D-F), which causes no activation of the Toll pathway. Embryos derived from the opposite Toll10b gain-of-function mutant showed nuclear staining all around the embryo (Fig. 4J,K). The phenotype induced by WntD overexpression was different from that in the Dorsal protein null mutant, which essentially showed no staining (Fig. 4L). These results together suggest that the overexpressed WntD inhibits Dorsal nuclear localization.
Specific mutants of wntD are not yet available. Therefore, we examined a few deficiency strains by staining for wntD mRNA expression in the embryo and confirmed that Df(3R)l26c, which has the 87E1-87F11 region deleted, has uncovered wntD. We used this deficiency to assess whether endogenous WntD regulates Dorsal. In wild-type blastoderm, the Dorsal nuclear gradient extends into the neuroectoderm and the posterior end (Stathopoulos and Levine,2002). Before the onset of gastrulation, the posterior Dorsal staining is normally retracted (Fig. 4M-O). However, in embryos derived from the Df(3R)l26cstrain, Dorsal nuclear staining expanded in the posterior region(Fig. 4P-R). Because the earliest wntD expression is at the anterior and posterior regions(Fig. 1C), the loss of wntD in the deficiency could be the underlying reason for the posterior expansion of Dorsal in these embryos.
WntD attenuates the function of Dorsal
The posterior expression of snail in wild-type embryos is retracted and shows a sharp pattern before the onset of gastrulation(Fig. 5A,B). snailexpression in Df(3R)l26c mutant embryos, however, expanded into the posterior region (Fig. 5C,D). Double staining shows that, in wild-type embryos, the posterior gene huckebein is complementary to the snail pattern(Fig. 5F). Moreover, we did not detect a change in the huckebein pattern in the Df(3R)l26cmutant embryos (data not shown). Using huckebein expression as a position marker, we found that the snail pattern expanded into the posterior region so that it overlapped with that of huckebein in the deficiency mutant embryo (Fig. 5E). Quantitation by using the snail pattern revealed that 24% (n=55) of all gastrulating embryos from Df(3R)l26cheterozygous parents showed the posterior expansion. Based on Mendelian ratios, this result represents an almost full penetrance. Thus, there is a posterior expansion of snail expression that correlates with the posterior expansion of nuclear Dorsal shown in Fig. 4. Subtle broadening of the snail pattern was also observed in the anterior region(Fig. 5C,D), suggesting that there is an increase of nuclear Dorsal in the anterior region, but the increase was not detectable by immunofluorescence staining.
Because Df(3R)l26c removes a number of genes in addition to wntD, we performed a genetic rescue experiment to confirm the involvement of wntD. We generated a transgenic line that contained a wntD genomic fragment, which showed all the normal expression patterns of wntD in the early embryo (data not shown). When we crossed this genomic construct into the Df(3R)l26c mutant, the posterior and anterior expansion phenotype of snail was completely rescued (Fig. 5G). We also did not observe posterior Dorsal expansion in any of the embryos derived from the wntD-rescued Df(3R)l26c strain (data not shown). The rescue experiment demonstrates that the deletion of wntD in the deficiency strain is responsible for the observed Dorsal and snail expression phenotypes. We also performed RNA interference of wntD by injecting double-stranded RNA into wild-type pre-blastoderm stage embryos. Approximately 10% of these injected embryos at late blastoderm stage had a mild posterior expansion of snail (Fig. 5I,J), and none of the embryos injected with buffer alone showed such a phenotype. This result further supports the idea that loss of WntD causes posterior expansion of snail expression.
We then examined whether the de-repressed WntD expression in the ventral cells of snail mutant embryos can inhibit Dorsal function. A previous report demonstrated that in mutant embryos that produced non-functional Snail proteins, the expression of snail mRNA disappeared prematurely(Ray et al., 1991). This premature loss of snail mRNA expression could be due to the inhibition of Dorsal function by the de-repressed WntD. We surmised that the removal of wntD in a snail mutant should lead to enhanced Dorsal function. Thus, we established a snail;Df(3R)l26cdouble-mutant strain. The double homozygous embryos were identified by the lack of wntD mRNA staining (blue color in Fig. 6) and the lack of a ventral furrow. The snail mRNA pattern (brown color) in snail mutant embryos exhibited a fuzzy border around the onset of gastrulation (Fig. 6E). By contrast, double-mutant embryos showed a snail pattern with sharp borders (Fig. 6F), similar to that observed in wild-type or Df(3R)l26c embryos(Fig. 5 and Fig. 6D). The developmental stages of these embryos were very similar based on the position of the pole cells, the degree of cellularization, and the cephalic furrow formation(Fig. 6G-J), supporting our argument that the genetic defect is the cause for the change of snailpattern. By mid-germ band extension, the snail mRNA staining became very weak in snail mutant embryos(Fig. 6K,M), but in double-mutant embryos the snail mRNA level was better sustained(Fig. 6L,N). The establishment and maintenance of the sharp snail pattern requires Dorsal(Ip et al., 1992b; Kosman et al., 1991; Ray et al., 1991). The Df(3R)l26c deficiency strain has many genes deleted and the effect cannot be attributed directly to the loss of wntD, but the result is consistent with our speculation that deleting wntD in the snail mutant embryo allows Dorsal to function more efficiently in activating target genes.
We have shown that Drosophila wntD is a novel downstream gene, and a negative regulator, of the Dorsal/Twist/Snail network. The dynamic pattern of wntD expression in the early embryo is a combined result of activation by Dorsal/Twist and repression by Snail. Overexpressed WntD negatively regulates Dorsal nuclear localization, leading to an inhibition of ventral cell invagination. Physiological levels of WntD can also negatively regulate Dorsal, as loss of WntD leads to detectable expansion of both Dorsal nuclear localization and snail expression in the posterior regions. Furthermore, de-repressed WntD expression in the ventral region of snail mutant embryos can also attenuate Dorsal function. However, the loss of WntD could not rescue the invagination defect of the snailmutant embryo, suggesting that in the snail mutant embryo there are other de-repressed genes that can interfere with ventral invagination.
The wntD loss-of-function phenotype correlates with the expression of wntD at the poles of pre-cellular blastoderms(Fig. 1C). wntD is also expressed a bit later in the mesectoderm, and weakly in the mesoderm. Because WntD can inhibit Dorsal, one speculation is that WntD in the early mesectoderm may help to establish the sharp snail expression at the mesectoderm-neuroectoderm boundary (Kosman et al., 1991). However, we did not detect any changes in the Dorsal protein gradient or snail pattern in the trunk regions of the Df(3R)l26c embryos. We speculate that the timing of early expression of wntD, which may have additional input from the Torso pathway at the poles, is important for the feedback inhibition of Dorsal. By the time of cellularization, the Dorsal protein gradient is well established. This well-established Dorsal gradient activates the wntD gene in the trunk regions, but the subsequently translated WntD protein may not be capable of exerting a strong negative-feedback effect on the already formed Dorsal gradient. This timing argument is supported by the results of WntD-overexpression experiments. The use of maternal nanos-Gal4 caused a strong inhibition of Dorsal nuclear localization and of ventral invagination, whereas the use of zygotic promoters did not result in a significant phenotype.
Snail acts as a transcriptional repressor for at least 10 genes in the ventral region where mesoderm arises(Hemavathy et al., 2000; Kosman et al., 1991; Leptin, 1991; Ray et al., 1991). In snail mutant embryos, all of these target genes are de-repressed in the ventral cells, concomitant with severe ventral invagination defects. However, no direct evidence has been reported on whether these de-repressed genes interfere with invagination(Hemavathy et al., 2004; Hemavathy et al., 1997). We show here for the first time that a target gene of Snail, namely wntD, can block ventral invagination when overexpressed. If de-repressed WntD is solely responsible for inhibiting ventral invagination,we would expect that, in the snail;Df(3R)l26c double-mutant embryos,ventral invagination will appear again. We did not observe a rescue of ventral invagination in the double-mutant embryos(Fig. 6), suggesting that wntD is not the only de-repressed target gene that inhibits invagination. Nonetheless, the de-repressed WntD can attenuate Dorsal function(Fig. 6), and may contribute to the ventral invagination defect.
Previous reports have shown that overexpression of String/Cdc25 leads to early mitosis in the ventral cells and a block in ventral invagination(Grosshans and Wieschaus, 2000; Mata et al., 2000; Seher and Leptin, 2000). The zygotic transcription of string in the ventral region is activated by the Dorsal/Twist/Snail network. Meanwhile, the String protein is kept at a low level in the ventral cells by Tribbles through protein degradation, and this process requires the positive input of Snail(Grosshans and Wieschaus, 2000; Mata et al., 2000; Seher and Leptin, 2000). Therefore, in the snail;Df(3R)l26c double-mutant embryos, the ventral cells should have increased String protein, as well as many other de-repressed gene products (Fig. 7). Perhaps the cumulative effect contributed by many of these snail target genes causes the severe invagination defect observed in the snail mutant embryo (Hemavathy et al.,2004); the simultaneous deletion of wntD and other interfering genes may be required to suppress the ventral invagination phenotype in snail mutants.
WntD may inhibit a component in the Toll pathway, or a component in the nuclear import/export pathway, leading to the cytoplasmic localization of Dorsal. However, the downstream mediators of Drosophila WntD signaling are not known. Being the closest homologs of DrosophilaWntD, vertebrate Wnt8 proteins regulate mesoderm patterning, neural crest cell induction, neuroectoderm patterning, and axis formation(Hoppler and Moon, 1998; Lekven et al., 2001; Lewis et al., 2004; Popperl et al., 1997). These vertebrate Wnt8 proteins may transmit the signal through the canonical pathway, but the exact mechanism remains unclear(Lekven et al., 2001; Lewis et al., 2004; Momoi et al., 2003). We examined Drosophila embryos that lacked maternal and zygotic functions of both Frizzled 1 and Frizzled 2 but did not observe any obvious defects in Dorsal or snail expression. A similar experiment using a dishevelled null mutant also did not reveal any such defects (data not shown). Furthermore, overexpression of Dishevelled or dominant-negative Gsk3 did not cause a detectable change of dorsoventral patterning (data not shown). These results suggest that Drosophila WntD may use other components for signaling. Wnt molecules employ multiple receptors and pathways to regulate various processes (Nelson and Nusse, 2004; Veeman et al.,2003). For example, Drosophila Wnt5 interacts with the receptor tyrosine kinase Derailed to regulate axon guidance(Yoshikawa et al., 2003). There are seven Wnt proteins and five Frizzled receptors in Drosophila, and WntD showed detectable affinity towards Frizzled 4 in cell culture assays (Wu and Nusse,2002), but the in vivo relevance of this interaction is not clear. It is important to elucidate how Drosophila WntD transmits its signal. Equally important is to find out whether WntD interacts with the Toll pathway, and whether the interaction also occurs in processes such as the immune response and cancer progression in other organisms.
This work was supported by an NIH grant (HD36240). We acknowledge the discussion and the agreement with Dr Roel Nusse for the use of the gene name wntD in place of Drosophila Wnt8. We thank Dr James Ooi for a partial wntD cDNA, and Dr Ruth Steward for making available the Dorsal antibody through the hybridoma bank (The University of Iowa). We thank Dr Gregory Hendricks for technical help in embryo sectioning. Core resources supported by the Diabetes Endocrinology Research Center grant DK32520 were also used.