BIOLOGICAL OVERVIEW

Wnt-4 is 60 kb upstream from wingless and is transcribed from the opposite strand of DNA, away from wingless. Wnt-4 was isolated from a library of genomic fragments that associate with Ultrabithorax proteins, in the hope of discovering genes regulated by homeodomain transcription factors. Expressed early in the head, and in an ectodermal segmental array, Wnt-4 is expressed in the visceral mesoderm later in embryonic development.

In most instances the targets of homeotic proteins are not known. However, Wnt-4 was isolated in a search for genes whose promoters bind homeodomain proteins; happily, investigators isolated a gene regulated by homeodomain proteins. Abdominal-A is directly required for WNT-4 induction, while Ultrabithorax is required indirectly.

Expression of Wnt-4 in the visceral mesoderm coincides with locations where two major morphological events will take place; formation of the gastric caeca (parasegment 4), and formation of the second midgut constriction (between parasegments 7 and 8). This overview will deal with the induction of the second midgut constriction. Ultrabithorax acts indirectly by first stimulating dpp production in parasegment 7. DPP then diffuses and induces Wnt-4 in parasegment 8. abdominal-A is required directly in parasegment 8 for Wnt-4 production.

Wnt-4 is inhibited in the other parasegments of the ventral mesoderm by the actions of abdominal-A, Antennapedia and Ultrabithorax, again demonstrating that Wnt-4 is downstream of homeotic genes. Wnt-4 is not the only target of DPP. DPP from parasegment 7 also induces wingless in parasegment 8. One wonders whether wingless and Wnt-4 share the same enhancer (Bienz, 1994).

Further studies show that indeed Wnt-4 and wingless developmentally interact. wingless and DWnt4 are transcribed in overlapping embryonic territories under the control of the same regulatory molecules. Co-expression and co-regulation suggest first, that the close physical linkage results from the sharing of cis-control elements and second, that the two Wnt signals cooperate in developmental patterning events. Antisense RNA experiments reveal that signaling by DWnt4 is essential for cells from the anterior compartment of each parasegment to adopt a denticled fate. It is proposed that wingless and DWnt4 achieve opposite, but complementary functions in intrasegmental cell patterning of the embryonic ectoderm (Gieseler, 1996).

To address the question of the functional relationship of wingless and DWnt-4, an examination was made of how embryonic cells respond when they are exposed, whether simultaneously or not, to the encoded Wnt signals. DWnt-4 has the capacity to antagonize Wingless signaling both in the Drosophila ventral epidermis and in a heterologous system, the Xenopus embryo. Evidence indicates that DWnt-4 inhibits the Wingless/Wnt-1 signaling pathway upstream of the activation of transcriptional targets (Gieseler, 1999).

The functional relationship between the two Wnt products was examined by comparing their ability to induce embryonic axis duplication in Xenopus. To do this, different concentrations of mRNA, encoding either Wg or DWnt-4, were injected into the vegetal region of ventral blastomeres of four-cell stage Xenopus embryos. As expected, providing WG mRNA results in the development of a secondary axis. This effect occurs in a dose-dependent manner. Five picograms of the WG message are sufficient to induce in 65% of injected embryos a fully developed supernumerary axis including cement glands, a marker of most dorso-anterior structures. In contrast, injection of as much as 6 ng of DWnt4 mRNA fails to induce embryonic axis duplication. To determine whether DWnt-4 influences the axis-inducing activity of Wg, mRNAs encoding the two Wnt molecules were co-injected into the ventral marginal zone of cleavage-stage embryos. One ng of DWnt-4 message decreases the frequency at which Wg induces a complete ectopic axis, since embryos generally develop an incomplete axis devoid of cement gland. Embryos injected with 2.4 ng of DWnt-4 mRNA never show complete axis duplication. Further increases in the amount of DWnt-4 further inhibits the phenotype induced by Wg; with 6 ng of DWnt-4 mRNA, the formation of a secondary axis is blocked in most embryos. Taken together these results indicate that DWnt-4 antagonizes in a dose-dependent manner the ability of Wg to cause axis duplication. DWnt-4 also counteracts Wnt-1 signaling upstream of the activation of target genes in the Spemann Organizer (Gieseler, 1999).

In contrast with ventral injections, directing DWnt-4 mRNA into the dorsal marginal zone of a four-cell stage Xenopus embryo inhibits formation of the endogenous axis, resulting in the disappearance of the anterior-most dorsal structures. Of the embryos injected with 6 ng of DWnt-4 mRNA, 90% fail to develop head structures including eyes and cement glands. In Xenopus, the formation of axial structures, including the head, requires beta-catenin activity, which results in the activation of Spemann Organizer genes in the dorsal territories of the early gastrula. To investigate the mechanism by which DWnt-4 inhibits dorsal patterning, the expression of chd, gsc, otx2 and Xnr3 were examined. Whole mount in situ hybridization on embryos dorsally injected at the four-cell stage and left to develop until stage 10, show that DWnt-4 does not alter the transcription patterns of the Organizer genes, nor does injection into the ventral marginal zone induce expression at ectopic sites. Thus, unlike products of the Wnt-1 class of genes, DWnt-4 does not have the capacity to influence the transcription of Organizer genes. These results indicate that the repression of dorsal axis formation by DWnt-4 presumably does not result from deficient formation of the Spemann Organizer, but from changes occurring later in dorsal development (Gieseler, 1999).

In the process of patterning the embryonic ventral ectoderm, Wg has at least two temporally distinct roles. A paracrine function of Wg is first required for the maintenance of engrailed (en) transcription in the adjacent posterior two rows of cells. This activity results in the formation and stabilization of parasegmental boundaries (cell-stabilization phase). During this time, Wg is also required for establishment of the diverse array of denticle types within the anterior half of the parasegment. Later wg function is essential for maintaining its own transcription and for specifying the cell fate that secretes naked cuticle within the posterior half of the parasegment (cell-specification phase). To determine whether DWnt-4 can modulate Wg signaling in these processes, the ventral cuticle defects resulting from DWnt-4 overexpression using the UAS/Gal4 system were examined. Several UAS-DWnt-4 transgenic lines were established and crossed with the daughterless (da) Gal4 driver that provides ubiquitous expression of UAS transgenes beginning early in embryogenesis. Embryos bearing one copy of UAS-DWnt-4 transgene show mild but consistent cuticular defects consisting of ectopic denticles within the region of naked cuticle. Posterior segments are always more affected than anterior ones and segmentation and diversity in denticle morphology are not entirely lost. Stronger phenotypes were obtained by increasing DWnt-4 expression through two copies of the UAS-DWnt-4 transgene. Under these conditions, the naked cuticle is often completely lost and replaced with denticle-decorated cuticle, especially in the posterior-most segments. The phenotype produced by ubiquitous DWnt-4 expression is opposite that of the naked cuticle mutant phenotype, in that it is produced by ectopic Wg. Providing DWnt-4 ubiquitously thus results in a phenotype that resembles that induced by loss of wg activity. When wg function is removed throughout embryogenesis, naked cuticle does not form; compartment boundaries are lost and all denticles display similar morphology. However, when wg function is removed at stage 10 in development, naked cuticle is still affected, but not completely lost, while denticle diversity is maintained. The phenotypes observed in da-Gal4-UAS-DWnt-4 embryos resemble the latter, suggesting that DWnt-4 antagonizes Wg activity during the late cell-specification phase reproducibly observed (Gieseler, 1999).

Two series of experiments were carried out to determine whether the phenotypes produced by DWnt-4 in the ventral ectoderm result from a reduction in Wg signal transduction. First, the transcription patterns of Wg target genes were examined. The expression of en marks the activity of Wg during the early phase, while the expression of wg itself marks the late phase of Wg activity through its autoregulation. In stage 9 da-Gal4-UAS-DWnt-4 embryos, expression of the two target genes is not significantly altered. Thus, wg and en presumably still function in the definition and stabilization of parasegmental boundaries; this is consistent with the previous observation that ubiquitously provided DWnt-4 does not abolish the segmental organization of the embryo. In stage 11 embryos, since the expression of wg and en is no longer interdependent and Wg is required for the maintenance of its own transcription, wg expression in the ventral ectoderm becomes severely reduced. Consistent with the stronger defects observed in posterior cuticle, a more pronounced inhibition of wg in posteriormost segments has been reproducibly observed. At the same time, the En protein distribution remains unaffected. Second, it was asked whether DWnt-4 has the ability to influence intracellular accumulation of the Armadillo (Arm)/beta-catenin protein, a key mediator of the Wg/Wnt-1 signaling. Cytoplasmic Arm is stabilized in response to signaling by Wg, allowing its translocation into the nucleus where it acts in association with the Pangolin/dTCF factor, a transcriptional effector of the pathway. Loss of wg function results in a reduction of the amount of Arm, when compared to wild type. DWnt-4 overexpression also induces a clear, though less pronounced, reduction in the levels of Arm. Thus, signaling by DWnt-4 negatively interferes with the Wg signaling pathway upstream of the Arm stabilization step (Gieseler, 1999).

cDNA clone length - 3201

Exons - two

PROTEIN STRUCTURE

Amino Acids - 389

Structural Domains

WNT-4 has a signal sequence that functions in secretion and a putative N-linked glycosylation site. There are abundant, strongly conserved cysteine residues (Graba, 1995).

Evolutionary Homologies

Extensive information on Wingless homologs and their receptors is to be found at Roel Nusse's World Wide Web Wnt Window (WWWWW).

The WNT-4 sequence is 35% identical to Wingless, 32% to WNT-2 and 33% to WNT-3. WNT-4 is the most divergent of any WNT protein identified to date (Graba, 1995).

Transcriptional Regulation

Antennapedia, Ultrabithorax and Abdominal A repress Wnt-4 in parasegments 5-6, 7 and 9-12, respectively. Both UBX and ABD-A are required to activate Wnt-4 in parasegment 8. (Graba, 1995). UBX's action is indirect. It first activates dpp in parasegment 7, and this in turn stimulates Wnt-4 in parasegment 8. ABD-A's effect is direct. Its expression in segment 8 is required for Wnt-4 induction (Graba, 1995).

Is it possible to estimate the number of target genes of the homeoproteins Eve and Ftz? Eve and Ftz have been shown to bind with similar specificities to many genes, including four genes chosen because they were thought to be unlikely targets of Eve and Ftz. Eve and Ftz bind at the highest levels to DNA fragments throughout the length of three probable target genes: eve, ftz and Ubx. However, Eve and Ftz also bind at only two- to ten-fold lower levels to four genes chosen in an attempt to find non targets: Adh, hsp70, rosy and actin 5C, suggesting that Eve and Ftz bind at significant levels to a majority of genes. The expression of these four unexpected targets is controlled by Eve and probably by the other selector homeoproteins as well. A correlation is observed between the level of DNA binding and the degree to which gene expression is regulated by Eve (Liang, 1998).

At stages 10-14, 87% of cDNAs in the 8-12 hour library are likely to be directly or indirectly regulated by Eve, Ftz, Engrailed and all of the Hox proteins. These downstream genes are each expressed in unique, segmentally repeating patterns. Some are expressed at dramatically altered levels between segments. Most vary from segment to segment in the number and position of cells in which they are most prominently expressed. This is not simply because expression follows the distribution of a particular cell type. Between segments, the majority of genes are most highly expressed in differently positioned subsets of the same cell types, indicating that these patterns cannot result solely from the action of cell-type specific transcription factors. Eve, Ftz and Engrailed establish the segmentally repeating structure of the embryo. Therefore, all genes expressed in segmentally repeated patterns by stage 11 should be downstream of these three genes. This has been experimentally confirmed for eve and ftz. The expression of all 14 segmentally expressed genes tested is altered in eve and ftz mutant embryos at stage 11. These and other downstream genes can be divided into three classes: genes expressed in strong, moderate or weak segmentally repeated patterns. 33% of cDNAs fall into the strongly repeated class. For this class, staining levels vary five fold or more between cells across a transverse section of a segment along the anterior/posterior axis of the embryo. 24% of clones belong to the moderately regulated class. These genes show two- to five-fold variations in staining across the width of a segment. Finally, the weak segmentally repeated genes vary only 1.2 to 2 fold in staining between cells across a segment. Thus, most downstream genes are expressed in all cells, but each are still subject to specific and precise control by the selector homeoproteins. The more strongly regulated genes include many developmental control genes such as Enhancer of split [E(spl)] , tramtrack, division abnormally delayed (dally), and Dwnt4. A high proportion of the moderate and weakly regulated genes are involved in essential cellular functions such as splicing (e.g. RNA helicases), translation (e.g. met tRNA synthetase), general signal transduction (e.g. G-protein beta13F) and cytoskeletal structure (e.g. alpha tubulin 84B). This raises the question of whether or not modest changes in the expression of essential enzymes and structural proteins are important for morphogenesis. It is argued that they probably are. 11% of the genes picked from the 8-12 hour cDNA library do not appear to be downstream of the selector homeoproteins. Most of these genes are expressed relatively uniformly in all cells. But even these genes show some differences in expression pattern (Liang, 1998).

The wg-like ventral expression of DWnt-4 is dependent on hh, which may be due to shared regulatory elements between the two genes. However, since the dorsal expression of wg and DWnt-4 is nonoverlapping, this aspect of DWnt-4 expression appears to be regulated differently. Since the dorsal stripes of DWnt-4 lie in between the hh and wg stripes, the effect these genes have on dorsal DWnt-4 expression was examined. In wg mutants DWnt-4 is expressed normally, indicating that its expression in the epidermis is not dependent on the activity of wg. However, in hh mutants, both dorsal and ventral expression is eliminated. The regulation of DWnt-4 by hh within the anterior half of the dorsal parasegment suggests that it acts in concert with hh to pattern these cells (Buratovich, 2000).

DEVELOPMENTAL BIOLOGY

Embryonic

See the embryonic expression pattern of Wnt4 at the Berkeley Drosophila Genome Project Patterns of Gene Expression Site.

Transcripts of Wnt-4 are first detected by stage 8 in the ectoderm, at the anterior tip and in the trunk, where segmentally repeated stripes become visible ventrally. Shortly thereafter, expression in the head shows up in the foregut primordium and dorsal patches, presumably corresponding to cells of labral and ocular segment primordia. By germ band extension [Image], the segment polarity-like staining in the trunk intensifies, and 15 stripes are visible in parasegments 0-14. The stripes are one cell wide and overlap the ventral most wingless expressing cells.

Transcripts are present in the dorsolateral ectoderm of each parasegment in short rows of 3-4 cells, paralleling the Engrailed stripe, but located in the middle of the parasegment. An additional stripe becomes apparent at the posterior tip.

By germ band shortening, transcripts are apparent in the central nervous system and the developing gut. Transcripts are seen in foregut, hindgut and at two locations in the visceral mesoderm of the midgut. The two locations in the ventral mesoderm correspond to primordia for the gastric caeca and the second midgut constriction (Graba, 1995).

Wnt genes are often expressed in overlapping patterns, where they affect a wide array of developmental processes. To address the way in which various Wnt signals elicit distinct effects, the activities of two Wnt genes in Drosophila, DWnt-4, and wingless, were compared. These Wnt signals produce distinct responses in cells of the dorsal embryonic epidermis. Whereas wingless acts independently of hedgehog signaling in these cells, DWnt-4 requires Hh to elicit its effects. Expression of Wg signal transduction components does not mimic expression of DWnt-4, suggesting that DWnt-4 signaling proceeds through a distinct pathway. The dorsal epidermis may therefore be useful in the identification of novel Wnt signaling components (Buratovich, 2000).

DWnt-4 and wg are expressed in many of the same cells during Drosophila embryogenesis, including the ventral epidermis. However, in the cells of the dorsal epidermis each gene is expressed in distinct groups of cells. Whereas wg is expressed in the most posterior row of cells in each parasegment throughout most of embryogenesis, DWnt-4 is expressed in the anterior region of the parasegment. This expression is transient, beginning at stage 10 and fading by the end of stage 12. The wg-like ventral expression of DWnt-4 is dependent on hh, which may be due to shared regulatory elements between the two genes. However, since the dorsal expression of the two genes is nonoverlapping, this aspect of DWnt-4 expression appears to be regulated differently. Since the dorsal stripes of DWnt-4 lie in between the hh and wg stripes, the effect these genes have on dorsal DWnt-4 expression was examined. In wg mutants DWnt-4 is expressed normally, indicating that its expression in the epidermis is not dependent on the activity of wg. However, in hh mutants, both dorsal and ventral expression is eliminated. The regulation of DWnt-4 by hh within the anterior half of the dorsal parasegment suggests that it acts in concert with hh to pattern these cells (Buratovich, 2000).

Unlike the ventral epidermis, where hh and wg cooperate in the specification of naked cuticle, the dorsal epidermis is patterned through complementary activities of wg and hh. The anterior half of each parasegment consists of three cells types (1^o, 2^o, and 3^o) that are all dependent on the activity of hh. The position of the DWnt-4 expressing cells relative to the wg-expressing cells indicates that DWnt-4 is expressed in the presumptive 3^o cells. The posterior half of the parasegment consists of a single cell type, 4^o, and is dependent on wg activity. The dorsal epidermis is patterned beginning at 6 h of development, the time at which the expression of DWnt-4 is initiated in these cells. Prior to this point, between 3 h and 6 h, the expression of wg and hh are mutually dependent (Buratovich, 2000).

To determine whether DWnt-4 is able to modulate the patterning of the dorsal epidermis, and whether it mimics or otherwise regulates wg signaling in these cells, it was ubiquitously expressed using the GAL4 system. The results of ubiquitous expression of wg or DWnt-4 were compared. Ubiquitous expression of wg driven by a GAL4 insertion under the control of a daughterless enhancer (daGAL4) results in a uniform lawn of 4^o cells. Thus the hh-dependent cell types are deleted or transformed to 4^o fates. Ubiquitous expression of DWnt-4 elicits a distinct response in the hh-dependent cells, while having no effect on the wg-dependent cells. The phenotype produced by ectopic DWnt-4 is variable and dependent on levels of ectopic expression. With one copy of ectopic DWnt-4 expressed at 29^oC, 21% (23/108) of the segments exhibit a 2^o-3^o-4^o pattern, in which 1^o cells are missing and 3^o cells are expanded. In contrast, 62% of the segments exhibit either a 1^o-3^o-4^o or a 3^o-4^o pattern; it was found that 1^o and 3^o cells are difficult to distinguish. Lower levels of expression produced by rearing the flies at a lower temperature produces a higher percentage of embryos with a pattern that is more clearly 1^o-3^o-4^o along the dorsal midline, since the 2^o cell fate is still apparent laterally. Nevertheless, the 2^o-3^o-4^o phenotype shows that DWnt-4 can abolish 1^o cells, and indicates that the primary effect of DWnt-4 is to expand 3^o cells at the expense of the other two cell types (Buratovich, 2000).

These data show that cells in the anterior half of each parasegment have the ability to respond to both Wnt genes, but that each gene elicits a distinct response. Whereas Wg transforms these cells to 4^o cells or deletes them, DWnt-4 appears to modulate the specification of cell fate within the hh-dependent domain but has no effect on cell fate specification by wg. The phenotypes produced by ectopic DWnt-4 and wg therefore appear to be qualitatively distinct, in that each gene induces ectopic specification of different cell types (Buratovich, 2000).

The alteration in pattern by DWnt-4 suggests three possible interactions with hh. (1) DWnt-4 might affect the anterior half of the parasegment through modification of hh expression. However, analysis of hh transcripts following ectopic DWnt-4 expression has revealed that hh expression is not affected. (2) Since DWnt-4 expression requires hh activity, it could be a downstream effector of hh in pattern specification. (3) DWnt-4 could act in concert with hh to alter pattern. To address these possibilities, DWnt-4 was ectopically expressed in a hh temperature sensitive mutant shifted to the restrictive temperature at 6 h. Under these conditions the entire anterior half of the parasegment is missing in hh mutants. When DWnt-4 is ectopically produced in this background, anterior cell fates still fail to be specified, indicating that DWnt-4 does not simply act downstream of hh but requires hh for its activity after 6 h of development. If hh ts mutants are shifted to the restrictive temperature at 7 h, one row of 3^o denticles typically forms, while 1^o and 2^o fates are still missing. If DWnt-4 is ectopically expressed under these conditions, the number of 3^o rows increases, supporting the conclusion that DWnt-4 acts in concert with hh to specify 3^o cell fates (Buratovich, 2000).

Wnt genes encode evolutionarily conserved secreted proteins that provide critical functions during development. Although Wnt proteins share highly conserved features, they also show sequence divergence, which almost certainly contributes to the variety of their signaling activities. Wnt4 and wingless, two divergent clustered Wnt genes, can have either antagonist or distinct functions during Drosophila embryogenesis. Both genes can elicit similar cellular responses during imaginal development. Ectopic expression of Wnt4 along the anterior/posterior (A/P) boundary of imaginal discs alters morphogenesis of adult appendages. In the wing disc, Wnt4 phenocopies ectopic Wg activity by inducing notum to wing transformation, suggesting similar signaling capabilities of both molecules. In support of this, it has been demonstrated that Wnt4 can rescue wg loss-of-function phenotypes in the antenna and haltere and is able to substitute for Wg in wing field specification. In addition to effects on the axial pattern of haltere and wing, ectopic Wnt4 produces supernumerary bristles in specific regions. Wnt4, like Wg, induces extra sternopleural bristles. This may reflect an endogenous activity for Wnt4: unlike wg, it is transcribed in a dorsal domain of the leg disc that roughly corresponds to the sternopleura. Both genes are transcribed in overlapping domains in imaginal discs, suggesting that Wnt4 may cooperate with wg during limb patterning (Gieseler, 2001).

wg and Wnt4 are transcribed in overlapping patterns during embryogenesis. They also show largely similar expression profiles in third instar imaginal discs. In the wing disc, transcripts of the two genes are synthesized at the D/V compartment border, the future wing margin, and the presumptive wing hinge region. Wnt4 expression, however, is weak compared to that of wg and appears more spatially restricted. It proceeds following a cell stripe comparable in width to the wg stripe in the distal part of the future wing margin, but is more faintly detected in more proximal parts of anterior and posterior compartments. The patterns are different in the presumptive notum, where WG mRNA is found in a broad D/V stripe, whereas WNT4 transcripts faintly label a central cell domain. These cells are located beneath the columnar epithelium and, therefore, likely correspond to adepithelial cells. In antenna and leg discs, the wg domain corresponds to a ventral/anterior sector, whereas Wnt4 is expressed only in a subset of these cells. Whereas WG mRNA is detected along the entire proximal-distal axis in the leg disc, Wnt4 transcription is restricted to two segments that correspond to primordia of the tibia and a more distal segment and in a dorsally located cell cluster that presumably corresponds to the sternopleural region. The wg pattern in the haltere disc is reminiscent of that observed in the wing disc. Wnt4 is transcribed in two large stripes, which give rise to the pedicel and/or the scabellum. However, in contrast to wg, Wnt4 is not expressed in the notum and in the most distal part of the disc, which is homologous to the margin of the wing. A clear difference in expression of the two genes is also seen in the larval central nervous system, where wg is not expressed, while Wnt4 transcripts follow a segmentally repeated pattern of small clusters in the central cortex and in the optic anlagen. In summary, the transcription profile of Wnt4 in third instar imaginal discs partially overlaps that of wg but is not restricted to wg-positive cells. For example, in the wing blade, the two genes are expressed along the future margin and in the presumptive hinge region, suggesting that the encoded Wnt products may be required together for the differentiation of these structures. In the central nervous system or in the dorsal part of the leg disc, Wnt4 is expressed where wg is not, suggesting it may also function in cells that do not depend on wg (Gieseler, 2001).

The ability of Wnt4 to induce, as Wg does, additional wings indicates that the two molecules can elicit similar cellular responses. Strong support for this conclusion is provided by rescue experiments of wg loss-of-function phenotypes. Wnt4 can restore normal wing development in the absence of a functional Wg protein at the second instar. Eliminating Wg signaling during the third instar, allows wing field specification but not subsequent wing patterning and growth. Wnt4 therefore appears unable to substitute for Wg signaling during late wing development. This failure to replace Wg might be the result of ectopic expression in inappropriate domains or of distinct signaling abilities during wing growth and patterning. In support of the latter possibility, the two Wnt molecules produce different effects, depending on the developmental context. Ectopic expression of Wg, but not of DWnt4, perturbs leg, eye, or antenna development, which reveals competency to respond to Wg but not to Wnt4. In addition, Wnt4 is able to antagonize late embryonic Wg signaling in the Drosophila ventral epidermis and to block the Wg-induced body axis duplication in Xenopus. Furthermore, the two genes induce different phenotypes in the dorsal embryonic epidermis. Taken together these results strongly support a context dependence for Wnt4 activity (Gieseler, 2001).

The molecular basis underlying the ability of Wg and Wnt4 to perform antagonistic, distinct, or similar signaling activities remains to be explored. It has been proposed that the Wnt4/Wg antagonistic relationship in Drosophila embryonic ventral ectoderm and in Xenopus axis induction assay results from strong sequence divergences in the C-terminal parts of the two proteins. This has been supported by reports that C-terminal truncations in Wg and XWnt8 result in proteins with dominant negative or antagonistic functions. If this interpretation is correct, domains other than the divergent C-terminal ends in the Wnt proteins would be critical for function in other developmental contexts. Although interactions of the C-terminus with specific partners dictate activity in the ventral embryonic epidermis, the recruitment of imaginal disc-specific factors by other domains in Wg and Wnt4 would allow the proteins to exhibit similar activities during wing field specification. The existence of separated functional domains in Wnt proteins is also supported by the nature and effect of the wg^NEI mutation, where a single amino acid change affects a subset of Wg functions only (Gieseler, 2001).

An alternative model to explain antagonism versus similarity is that Wnt4 acts as a low-activity agonist of Wg. In this model, tissue-specific differences in receptor concentrations and/or differences in receptor-binding affinities would determine whether Wnt4 mimics or antagonizes Wg signaling. If, for example, receptor concentrations are limiting in the ventral epidermis, Wnt4 may act as a competitive inhibitor of Wg for receptor binding but would provide less stimulation of the pathway than would Wg itself, therefore lowering the level of Wg signaling. However, if receptor concentrations are high, Wnt4 may be able to increase signaling by binding receptors without competing with endogenous Wg. If Wnt4 interacts with Wg receptors with lower affinity than that of Wg, the tissue-specific differences in Wnt4's ability to engage the Wg signaling pathway might be explained. In particular, this may explain the inability of Wnt4 to stimulate the Wg pathway in wild type legs while stimulating the pathway, and rescuing leg defects, in the absence of Wg (Gieseler, 2001).

Wnt4 regulates the dorsoventral specificity of retinal projections in the Drosophila melanogaster visual system

In Drosophila, the axons of retinal photoreceptor cells extend to the first optic ganglion, the lamina, forming a topographic representation. DWnt4, a secreted protein of the Wnt family, is the ventral cue for the lamina. In DWnt4 mutants, ventral retinal axons misproject to the dorsal lamina. DWnt4 is normally expressed in the ventral half of the developing lamina and DWnt4 protein is detected along ventral retinal axons. Dfrizzled2 and dishevelled, respectively, encode a receptor and a signaling molecule required for Wnt signaling. Mutations in both genes caused DWnt4-like defects, and both genes are autonomously required in the retina, suggesting a direct role of DWnt4 in retinal axon guidance. In contrast, iroquois homeobox genes are the dorsal cues for the retina. Dorsal axons accumulate DWnt4 and misproject to the ventral lamina in iroquois mutants; the phenotype is suppressed in iroquois:Dfrizzled2 double mutants, suggesting that iroquois may attenuate the competence of Dfrizzled2 to respond to DWnt4 (Sato, 2005).

The spatial order of projecting neurons is preserved in the spatial order of their targets to establish the topographic maps in the nervous system. In the visual system, precise topographic mapping of photoreceptor neurons to their targets in the brain, termed retinotopic mapping, is necessary for the correct interpretation of visual information received in the retina. The Drosophila visual system includes the retina, the compound eye and the optic lobe, which is the visual center of the brain and is connected to the eye via the optic stalk. Each of the approximately 750 ommatidial units in the retina consists of eight unique types of photoreceptor neurons called R cells (R1-R8). During larval development, R cells sequentially differentiate behind the morphogenetic furrow, progressing in a posterior-to-anterior order in the third larval instar retina, and send their axons through the optic stalk to the most distal part of the optic lobe, the lamina. R1-R6 axons terminate in the lamina layer, whereas R7 and R8 axons project through the lamina to terminate in the medulla layer. Although all the retinal axons pass through the narrow optic stalk, they distribute evenly and project to their correct targets along the anteroposterior and the dorsoventral axes. Thus, R cell axon connections between the retina and the lamina (or the medulla) are precisely retinotopic in the adult. Similarly, R axon connections established during the third instar are anatomically retinotopic (Sato, 2005 and references therein).

Like the retina, the lamina must also be patterned along the dorsoventral axis so that retinal axons can project precisely to their targets. It is assumed that selector genes and genes encoding guidance molecules are asymmetrically expressed along the dorsoventral axis in the developing lamina and it was found that DWnt4, one of seven D. melanogaster Wnt family genes, is specifically expressed in the ventral half of the developing lamina during the third larval instar (Sato, 2005).

The present study investigated the involvement of DWnt4 in D. melanogaster retinotopic mapping along the dorsoventral axis. DWnt4 is normally expressed in the ventral half of the developing lamina and DWnt4 protein has been detected on the surface of ventral retinal axons. In DWnt4 mutant backgrounds, ventral axons misproject to the dorsal lamina. Conversely, ventral axons are redirected by an ectopic source of DWnt4, suggesting that DWnt4 is a ventral cue for retinal axon projections in the lamina. Furthermore, ventral axon projections are regulated by noncanonical Wnt signaling in R cells, which is most likely under the control of DWnt4. These genetic data also suggest the involvement of JNK signaling in this process. Finally, iro may attenuate the competence of Dfz2 in dorsal axons to respond to DWnt4, since dorsal-to-ventral misroutings in iro clones are significantly suppressed in iro:Dfz2 double mutant clones (Sato, 2005).

As a first step to investigating retinotopic mapping in D. melanogaster, focus was placed on the iroquois (iro) complex genes, three related homeobox genes that act as selector genes for the dorsal retina. To test if iro regulates retinotopic mapping, cells homozygous for an iro deficiency were generated and labeled with green fluorescent protein (GFP) to visualize axons using the MARCM system. Dorsal mutant R axons were occasionally observed projecting to the ventral lamina. A similar phenotype was observed by generating large iro clones using ey-flp. Both outer- and inner-photoreceptor axons visualized with ro-lacZ and ato-myc were affected by iro. Thus, iro genes seem to function as dorsal cues for the retina (Sato, 2005).

The lamina, to which the retinal axons project, must also be patterned along the dorsoventral axis. It is assumed that selector genes and genes encoding guidance molecules are asymmetrically expressed along the dorsoventral axis in the developing lamina. Attempts were made to identify genes specifically expressed in either the dorsal or the ventral lamina. It was found that DWnt4, one of seven Wnt family genes in D. melanogaster, is specifically expressed in the ventral half of the lamina during the late third instar. At this stage, the lamina expresses Dachshund (Dac) and forms a characteristic, crescent-like structure. In situ hybridization showed specific expression of DWnt4 in the ventral half of the lamina along the lamina furrow, restricted to the anterior-most two or three rows of Dac-positive cells (Sato, 2005).

Because DWnt4 is a secreted glycoprotein, the distribution of DWnt4 protein was compared with that of DWnt4 mRNA. DWnt4 localization along the lamina furrow of the ventral lamina is indistinguishable from that of DWnt4 mRNA, except for small dots found in the ventral lamina. Notably, DWnt4-positive dots are found on the surface of the anterior-most R axons, labeled by GMR-Gal4 UAS-GFP. Given that R axons sequentially project to the lamina from posterior to anterior, DWnt4 probably accumulates on the most recently arriving axons. In more apical optical sections, DWnt4 accumulation is observed between the optic stalk and the lamina. The source of DWnt4 on R axons is likely to be DWnt4 expression along the ventral lamina furrow, since DWnt4 localization is not detectable on R axons in the dorsal half of the lamina (Sato, 2005).

DWnt4 localization on ventral R axons implies its involvement in R axon guidance. This possibility was tested by investigating the DWnt4 loss-of-function phenotype. In wild-type flies, dorsal and ventral axons project to the dorsal and ventral lamina, respectively. In DWnt4 mutants, however, occasional misroutings of ventral axons toward the dorsal lamina were observed. In extreme cases, ventral axons looked as if they were about to project to ventral lamina but had abruptly reoriented toward the dorsal lamina. omb-lacZ was used to visualize the dorsal- and ventral-most axons and ventral axon misrouting within the optic stalk was noted. Together, these data suggest that DWnt4 influences ventral axon projection at various points along the path from the optic stalk to the lamina (Sato, 2005).

Frizzled family receptors and Dishevelled are required for a wide variety of Wnt signaling cascades7. Dfz2 and dsh mutant flies have ovarian defects similar to those of DWnt4, strongly suggesting that Dfz2 and Dsh are involved in DWnt4 signaling. A ventral-to-dorsal misrouting phenotype was observed in Dfz2 and dsh mutant backgrounds. The results suggest that Dfz2 and Dsh are involved in DWnt4 signaling for R axon guidance. In addition, the greater expressivity and penetrance of Dfz2 and dsh mutants suggests the involvement of other Wnt family ligands in this process. However, after examining mRNA expression of all the known D. melanogaster Wnt genes, no such genes were found acting as ventral cues in concert with DWnt4. DWnt2 is expressed just outside the lamina, but its expression is symmetric along the dorsoventral axis (Sato, 2005).

The accumulation of DWnt4 on the surface of ventral R axons implies reception of the ligand and subsequent signal activation in R cells. Consistently, Dfz2 is localized on the surface of the anterior-most R axons, but not on surrounding lamina cells. To test whether Wnt signaling autonomously regulates axon guidance, dsh homozygous clones were induced in the retina using ey-Gal4:UAS-flp18. Surprisingly, the axonal misrouting phenotype was rarely observed despite the presence of many dsh mutant clones in the retina. When the retina was entirely dsh homozygous, the same dsh allele showed axonal misrouting. It is suspected that mutant R axons project normally in the presence of surrounding wild-type axons due to axon fasciculation. If this is the case, R axons homozygous for dsh may show a misrouting phenotype in the absence of neighboring, wild-type R cells. To test this idea, GMR-hid was introduced in trans to the dsh mutant chromosome. Wild-type R cells eventually die by programmed cell death triggered by hid expression behind the furrow. In this context, a severe misrouting phenotype was observed that was ventral-to-dorsal. Notably, DWnt4 protein accumulation was observed along ventral axons that were mutant for dsh and had misprojected to the dorsal lamina. Retina-specific Dfz2 clones with GMR-hid also showing a ventral-to-dorsal misrouting phenotype. These observations are consistent with the idea that DWnt4 expressed in the lamina directly regulates R axon projections (Sato, 2005).

There are two Wnt signaling pathways: the canonical and noncanonical pathways. In the former, ß-catenin/Armadillo (Arm) and TCF/Pangolin (Pan) form a complex to activate target gene transcription. In the latter, Wnt signaling is transduced independently of Arm and Pan. Canonical Wnt signaling was manipulated using UAS-panN, which encodes a constitutive repressor form of Pan, and UAS-arm, which encodes a constitutively active form of Arm. Misrouting along the dorsoventral axis was hardly observed in either genotype. Thus it is concluded that canonical Wnt signaling plays a very minor role, if any, in dorsoventral specification of R axon guidance. The above results strongly suggest the involvement of noncanonical Wnt signaling in R cells. To confirm this idea, UAS-dshDEP, which acts as a dominant-negative mutant in noncanonical signaling, was expressed in the retina. Again a strong ventral-to-dorsal phenotype was observed (Sato, 2005).

Although the planar cell polarity (PCP) pathway is categorized as a noncanonical Wnt pathway transduced by the Fz receptor, no PCP defects were observed in DWnt4 and Dfz2 mutant retinae. In addition, the retinotopic phenotype was not observed in fz null mutant backgrounds. These results suggest that DWnt4 regulates R axon projections via a noncanonical Wnt signaling distinct from the PCP pathway. wingless (wg) is involved in the specification of the dorsal retina through the activation of iro expression. The distinct chiral forms of ommatidia in the dorsal and ventral retina reflect the dorsoventral specification of the retina and the PCP signaling. The normal iro expression and the normal ommatidial chirality suggest that axonal misroutings occur independently of the retinal dorsoventral specification in DWnt4 and Dfz2 backgrounds. Since dsh is required for PCP signaling and the specification of the dorsal retina, ommatidial chirality was disorganized and dorsal iro expression was eliminated in dsh retinae. However, the expression of Serrate (Ser), which is specific to the ventral retina in wild-type backgrounds, was not affected, suggesting that the ventral cell fate is correctly specified in dsh homozygotes. Additionally, UAS-dsh and UAS-dshDEP expression under the control of GMR-Gal4 did not affect the dorsoventral specification of the retina as visualized by iro and Ser expression. Note that GMR-Gal4 is expressed behind the morphogenetic furrow well after the dorsoventral specification at earlier stages. The data shown above suggest that dsh also regulates R axon projections independently of the dorsoventral patterning of the retina (Sato, 2005).

JNK signaling is known to act downstream of the noncanonical Wnt pathway in many developmental contexts. The involvement of JNK signaling was examined by expressing puckered (puc), which encodes a JNK phosphatase, and a dominant-negative form of JNK encoded by basket (bsk) to block JNK signaling in the retina. Defects were observed only rarely, and it was next asked whether genetic interactions exist between hemipterous (hep) encoding a JNK kinase and DWnt4 or Dfz2. In a strong hep mutant background, or in DWnt4, DWnt4 or Dfz2 heterozygous backgrounds, little or no ventral-to-dorsal misrouting was observed. However, a reduction in the dosage of DWnt4 or Dfz2 in the hep background resulted in a marked increase in the ventral-to-dorsal phenotype. These findings provide some support for the idea that JNK signaling is involved in the DWnt4/Dfz2 pathway in retinal axon guidance. Since iro expression and ommatidial chirality were normal in retinae expressing the dominant-negative form of bsk and in hep hemizygotes in combination with DWnt4/+ and Dfz2/+, the misrouting of ventral axons observed in the brain mutant for JNK signaling appears to be caused by a failure in axon guidance and independent of the dorsoventral cell specification or PCP signaling in the retina. Note that mutations in JNK pathway components alone have no PCP phenotype (Sato, 2005).

iro is thought to be the dorsal cue in the retina. The dorsal axons project to the ventral lamina in the absence of iro, perhaps because dorsal axons are attracted by ventral cues in the lamina, such as DWnt4. When iro mutant clones were generated under the control of ey-flp, dorsal axons projected to the ventral lamina in 32.4% of them, and ectopic accumulations were observed of DWnt4 on the surface of the dorsal axons misprojecting ventrally. Since Dfz2 was expressed in the dorsal axons, iro may attenuate the competence of Dfz2 in the dorsal axons to respond to DWnt4. If this is the case, simultaneous removal of Dfz2 in iro clones should suppress the iro phenotype, which was indeed observed. In iro:Dfz2 double mutant clones, 'dorsal-to-ventral' misroutings were observed in 3.5% of the cases, and the class III phenotype was no longer observed. Instead, abnormal bundles of dorsal axons were found in iro:Dfz2 clones. This might be because iro:Dfz2 axons do not respond to either dorsal or ventral cues in the lamina. Indeed, no DWnt4 accumulation was found in those abnormal bundles found in iro:Dfz2 clones. The absence of iro expression in differentiated R cells behind the morphogenetic furrow suggests indirect modulation of Dfz2-dependent Wnt signaling by iro (Sato, 2005).

It was hypothesized that three events are required for retinotopic mapping along the dorsoventral axis in D. melanogaster: (1) dorsoventral identity is specified by selector genes expressed in the retina; (2) dorsoventral identity is specified by selector genes in the lamina; (3) guidance molecules recruit R axons to their correct targets in the lamina. The results nicely fit the hypothesis. iro expressed in the dorsal retina specifies the dorsal axon identity, and DWnt4 expressed in the ventral lamina recruits ventral axon projections. The restricted expression of DWnt4 to the ventral lamina suggests there could be unidentified dorsoventral selectors in the lamina (Sato, 2005).

REFERENCES

Bienz, M. (1994). Homeotic genes and positional signalling in the Drosophila viscera. Trends in Genetics 10: 22-26

Buratovich, M. A., et al. (2000). DWnt-4 and Wingless have distinct activities in the Drosophila dorsal epidermis. Dev. Genes Evol. 210: 111-119. Medline abstract: 11180811

Gieseler, K., et al. (1996). wingless and DWnt4, 2 Drosophila Wnt genes, have related expression, regulation and function during the embryonic development. C. R. Acad. Sci. III, Sci. 318: 1101-1110 Medline abstract

Gieseler, K., et al. (1999). Antagonist activity of DWnt-4 and wingless in the Drosophila embryonic ventral ectoderm and in heterologous Xenopus assays. Mech. Dev. 85: 123-131. Medline abstract: 99345949

Gieseler, K., et al. (2001). DWnt4 and wingless elicit similar cellular responses during imaginal development. Dev. Bio. 232: 339-350

Graba, Y., et al. (1995). DWnt-4, a novel Drosophila Wnt gene acts downstream of homeotic complex genes in the visceral mesoderm. Development 121: 209-218 Medline abstract

Liang, Z. and Biggin, M. D. (1998). Eve and Ftz regulate a wide array of genes in blastoderm embryos: the selector homeoproteins directly or indirectly regulate most genes in Drosophila. Development 125(22): 4471-4482. Medline abstract: 9778506

Sato, M., et al. (2005). DWnt4 regulates the dorsoventral specificity of retinal projections in the Drosophila melanogaster visual system. Nat. Neurosci. 9: 67-75. Medline abstract: 16369482

date revised: 15 February 2006