Regulation of the Protein Kinase Activity of ShaggyZeste-white3 by Components of the Wingless Pathway in Drosophila Cells and Embryos*

The protein-serine kinase ShaggyZeste-white3 (SggZw3) is theDrosophila homolog of mammalian glycogen synthase kinase-3 and has been genetically implicated in signal transduction pathways necessary for the establishment of patterning. SggZw3 is a putative component of the Wingless (Wg) pathway, and epistasis analyses suggest that SggZw3 function is repressed by Wg signaling. Here, we have investigated the biochemical consequences of Wg signaling with respect to the SggZw3 protein kinase in two types ofDrosophila cell lines and in embryos. Our results demonstrate that SggZw3 activity is inhibited following exposure of cells to Wg protein and by expression of downstream components of Wg signaling, Drosophila frizzled 2 anddishevelled. Wg-dependent inactivation of SggZw3 is accompanied by serine phosphorylation. We also show that the level of SggZw3 activity regulates the stability of Armadillo protein and modulates the level of phosphorylation of D-Axin and Armadillo. Together, these results provide direct biochemical evidence in support of the genetic model of Wg signaling and provide a model for dissecting the molecular interactions between the signaling proteins.

The protein-serine kinase Shaggy Zeste-white3 (Sgg Zw3 ) is the Drosophila homolog of mammalian glycogen synthase kinase-3 and has been genetically implicated in signal transduction pathways necessary for the establishment of patterning. Sgg Zw3 is a putative component of the Wingless (Wg) pathway, and epistasis analyses suggest that Sgg Zw3 function is repressed by Wg signaling. Here, we have investigated the biochemical consequences of Wg signaling with respect to the Sgg Zw3 protein kinase in two types of Drosophila cell lines and in embryos. Our results demonstrate that Sgg Zw3 activity is inhibited following exposure of cells to Wg protein and by expression of downstream components of Wg signaling, Drosophila frizzled 2 and dishevelled. Wg-dependent inactivation of Sgg Zw3 is accompanied by serine phosphorylation. We also show that the level of Sgg Zw3 activity regulates the stability of Armadillo protein and modulates the level of phosphorylation of D-Axin and Armadillo. Together, these results provide direct biochemical evidence in support of the genetic model of Wg signaling and provide a model for dissecting the molecular interactions between the signaling proteins.
The product of the Drosophila wingless (wg) gene is a secreted protein homologous to vertebrate Wnts (1). Genetic analysis of wg has revealed roles in processes controlling embryonic segmentation, gut formation, and imaginal disc patterning (2)(3)(4). Additional genes have been implicated in the secretion, reception, or interpretation of the Wg 1 signal: dishevelled (dsh) (5) and armadillo (arm) (6). Dsh protein is a novel protein with a discs-large homology region, whereas the arm gene encodes the Drosophila homolog of ␤-catenin, a component of vertebrate adherens junctions. Drosophila frizzled 2 (Dfz2) was recently identified as a protein with an amino-terminal cysteine-rich extracellular domain followed by seven transmembrane domains (7). The Dfz2 protein functions as a Wg receptor in cultured cells, but as yet, there are no known Dfz2 mutants. Whereas the above-mentioned genes act positively in Wg signaling, an additional gene called shaggy or zeste-white3 (sgg zw3 ) plays an inhibitory role in this pathway (1,4,8). sgg zw3 encodes a protein-serine kinase that has been highly conserved throughout the eukaryotic kingdoms (4,9,10). The mammalian homolog of sgg zw3 is glycogen synthase kinase-3 (GSK-3), which is encoded by two independent genes, GSK-3␣ and GSK-3␤ (11).
By a combination of clonal analysis, genetic epistasis, and biochemical experiments, wg class genes have been ordered within the same pathway (12)(13)(14)(15). armadillo and dishevelled embryonic phenotypes are very similar to the wg embryonic phenotype (12)(13)(14), whereas sgg zw3 has a mutant phenotype very similar to that of embryos in which wg has been expressed in all cells (12,16,17). Genetic data in Drosophila suggest that the functions of sgg zw3 are antagonized by Wg signaling (4). In fact, mutations in wg and sgg zw3 have opposite effects on cell fate determination, and each mutation has an opposite effect on Arm protein levels (17,18). In embryos, the normal segmental accumulation of Arm protein is absent in wg, whereas sgg zw3 mutants have uniformly high levels of Arm protein.
These data have been assembled into a model in which Wg protein is secreted and received by neighboring cells, where a signal transduction cascade is initiated (1). The Wg signal, at least in embryos and cultured cells, is transduced through Dsh and induces hyperphosphorylation of Dsh protein, possibly via casein kinase-2 (15,23). Through an unknown mechanism, activation of Dsh blocks the function of Sgg Zw3 and D-Axin, resulting in decreased phosphorylation of Arm. Unphosphorylated Arm has increased stability and accumulates in the cytoplasm (15,24), where it interacts with an high mobility grouplike factor, LEF-1/pangolin (25,26).
Recently, the mammalian homolog of Sgg Zw3 , GSK-3, has been shown to be regulated by Drosophila Wg protein in fibroblasts (27), but direct biochemical evidence for inhibition of Sgg Zw3 by Wg signaling has yet to be demonstrated. To address the mechanism by which Wg signals via Sgg Zw3 , the effect of the known components of Drosophila Wg signaling (Wg, Dfz2, and Dsh) on Sgg Zw3 activity was investigated in cultured cells and embryos. We used an imaginal disc cell line (cl-8 (clone 8)) that responds to Wg signals and Schneider (S2) cells, which are unresponsive to Wg (15,24). Using Wg-conditioned medium, we show that the activity of Sgg Zw3 protein kinase is inhibited by Wg in cl-8 cells and that overexpression of Dfz2 or Dsh in cells reconstitutes Wg signaling in the absence of Wg as judged by inhibition of the kinase and accumulation of Arm protein.
We also demonstrate that the regulation of Sgg Zw3 activity, in turn, controls the stability of Arm protein by modulating the level of phosphorylation of D-Axin and Arm. These results provide direct biochemical evidence in support of previous genetic analyses.

EXPERIMENTAL PROCEDURES
Antisera-Rabbit antisera to Arm and Dsh were raised against glutathione S-transferase (GST) fusion proteins. GST-Dsh was constructed by cloning a 1256-base pair XhoI-NotI fragment of the dishevelled coding region, corresponding to amino acids 395-624, into XhoI-NotI sites in pGEX-4T-1 (Amersham Pharmacia Biotech). cDNA fragments encoding amino acids 1-367 of Arm protein and 1-514 of Sgg Zw3 protein were cloned into pGEX-4T-1 and pET15b (Novagen), respectively. Fusion proteins were produced in Escherichia coli strain BL21(DE3) and purified from bacterial lysates before immunization.
Transfections and Cell Culture-Drosophila Schneider line-2 and wing imaginal disc cl-8 cells were maintained as described (24). Wg protein assays were performed essentially as published (24,28). Selection of stably transformed cl-8 cell lines was performed using methotrexate (29). The expression vector pRmHa-1 is designed to express proteins under control of the metallothionein promoter. The 2.8-kilobase pair BamHI-HindIII fragment of dsh cDNA in pBluescript SK ϩ (Stratagene) corresponding to the entire coding region was cloned into the BamHI-HindIII sites of pRmHa-1. The dsh/pRmHa-1 or sgg zw3 / HApRmHa-1 vector was introduced into cl-8 cells by cotransfection with a second vector, pHGCO, carrying a selectable dhfr gene, which confers resistance to methotrexate (0.5 g/ml). Transformed cells were maintained between 1 ϫ 10 6 and 1 ϫ 10 7 cells/ml and examined for metalinducible gene expression (by addition of 0.5 mM CuSO 4 ) by immunoblotting.
For expression in cl-8 cells, the D-axin-(332-642) fragment (amplified by polymerase chain) was subcloned into the pAc5.1/V5-His 6 vector (Invitrogen) in frame with the His epitope. Transfected cells were washed with phosphate-buffered saline and lysed in 20 mM Tris-HCl (pH 8) and 100 mM NaCl. For purification of D-Axin-(330 -642)-His 6 , 10 l of nickel-Sepharose beads were added in lysates. The complexes were washed four times with 20 mM Tris-HCl (pH 8), 100 mM NaCl, and 10 mM imidazole and resolved by SDS-PAGE or incubated with [␥-32 P]ATP for 30 min.
Metabolic Labeling of S2 Cell Lines-Transfected Dsh S2 cells were treated with CuSO 4 to induce Dsh expression and labeled overnight with 1 mCi of [ 32 P]orthophosphate/ml of S2 phosphate-free medium ϩ 10% dialyzed fetal calf serum. Radioimmune precipitation assay buffer cell lysates were normalized for incorporation by Cerenkov counting (30). After immunoprecipitation of Sgg Zw3 protein and separation by SDS-PAGE, proteins were transferred to polyvinylidene difluoride membranes. 32 P-Labeled Sgg Zw3 was subjected to partial acid hydrolysis, and the phosphoamino acids were separated in two dimensions by thin-layer electrophoresis (31).
Preparation of Embryo Lysates-For overexpression of Sgg Zw3 , homozygous HS-Sgg Zw3 Drosophila eggs were collected 3 h after laying, heat-shocked for 8 min at 37°C, and allowed to recover for an additional 1.5 h at 25°C. To generate sgg zw3 M11-1 mutant embryos, germ line mosaics were produced using the yeast recombinase-base flippase-dominant female sterile system as described by Chou and Perrimon (32). Homozygous mutant embryos can be recognized morphologically by a lack of segmentation. For overexpression of Wg, Drosophila males homozygous for arm-Gal4 were crossed to virgin Drosophila females harboring pUAS-Wg, and their progeny embryos were collected at 3-6 h. Wild-type embryos of the same stage were used as controls. Embryos were lysed in Gentle Soft buffer (28) and were subjected to immunoprecipitation analysis as described below.
Immunoprecipitation and Sgg Zw3 Kinase Assays-Cells lines were washed with phosphate-buffered saline and lysed in Gentle Soft buffer (28). For Sgg Zw3 immunoprecipitation, 20 l of protein A-Sepharose or 20 l of protein G-Sepharose were pre-bound to rabbit polyclonal antiserum or to monoclonal antibodies (anti-Sgg Zw3 , 2G2C5), respectively, and were added to the clarified cell lysates at 4°C for 2 h. Immunocomplexes were washed four times with Gentle Soft buffer (28). In vitro Sgg Zw3 kinase assays were performed for 30 min as described previously (33,34). Phosphorylated peptide was separated from unincorporated [␥-32 P]ATP by Tricine/SDS-PAGE and quantified using a PhosphorImager.

Wingless Protein Represses Sgg Zw3
Activity and Induces Accumulation of Cytoplasmic Armadillo-To analyze the biochemical consequences of Wg signaling, we exploited an imaginal disc cell line (cl-8) that is responsive to Wg (24). To determine the biological effects of Wg, cl-8 cells were exposed to the serum-free conditioned medium from either heat-shocked Schneider HS-wg (Wg-conditioned medium) or Schneider control cells (S2 control medium), and cytoplasmic extracts were prepared and immunoblotted with antibodies to Wg, Arm, and Dsh ( Fig. 1A) (15). Wg-containing medium increased Arm levels within 2 h, reaching a maximum after 6 h. By contrast, cellular levels of Dsh did not change in this time period. How- . Equal amounts of cytoplasmic extracts from treated cl-8 cells were immunoblotted with polyclonal antibodies to Arm, Dsh, Sgg Zw3 , and Wg. Arm protein migrated as two bands of ϳ105 kDa, and the faster migrating form accumulated in cl-8 cells in response to extracellular Wg. Dsh protein migrated as multiple mass isoforms likely representing differences in phosphorylation state, the extent of which was increased by Wg. B, cytoplasmic extracts from cl-8 cells treated for different times (0 -6 h) with S2 control medium or Wg-conditioned medium were immunoprecipitated using rabbit polyclonal antibodies against Sgg Zw3 and assayed for Sgg Zw3 kinase activity. Activities are expressed as the percentage of those of the untreated controls (mean Ϯ S.E., three experiments). ever, Wg induced the formation of electrophoretically retarded forms of Dsh. These modifications have been previously observed by Yanagawa et al. (15) and Willert et al. (23) and correspond to hyperphosphorylation of Dsh protein. Exposure of cells to medium conditioned by control S2 cells affected neither Arm levels nor the Dsh electrophoretic pattern.
To determine whether Wg modulates Sgg Zw3 activity, Sgg Zw3 was immunoprecipitated from lysates of cl-8 cells treated with Wg-conditioned medium or S2 control medium. Protein kinase activity was measured using a peptide substrate specific for the GSK-3 family of protein kinases (GS-1 peptide (33)). Incubation of cl-8 cells with Wg-conditioned medium caused a time-dependent inhibition of Sgg Zw3 protein kinase activity (Fig. 1B). After 2-4 h of treatment with Wg-conditioned medium, total GS-1 peptide kinase activity was suppressed by 40 -50% compared with the activity observed in cells treated with S2 control medium. Wg did not affect the expression of Sgg Zw3 as judged by immunoblotting (Fig. 1A).
To confirm the effect of Wg protein on the activity of Sgg Zw3 , we investigated how Sgg Zw3 functions in Wg signaling during embryogenesis, analyzing Sgg Zw3 activity in embryos with a wild-type or sgg zw3 mutant genotype, embryos overexpressing sgg zw3 , and embryos expressing wg ubiquitously. sgg zw3 embryos were made homozygous for the sgg zw3 M11-1 allele, and Sgg Zw3 immunoprecipitates from these mutant embryos contained no detectable Sgg Zw3 activity, which verified the specificity of the assay (Fig. 2B). Furthermore, Sgg Zw3 immunoprecipitates from embryos overexpressing Sgg Zw3 from a heat shock-inducible transgene (HS-Sgg Zw3 ) exhibited 2.5-fold higher activity than immunoprecipitates from wild-type embryos (Fig. 2B).
To determine the effect of Wg overexpression on Sgg Zw3 activity, Wg was ectopically expressed in early embryos using a line that carries a GAL4-driven wg transgene (pUAS-Wg) crossed to a line that ubiquitously expresses GAL4 (arm-GAL4). The activity of Sgg Zw3 from these embryos was determined to be ϳ30% lower than that from wild-type control lysates (Fig. 2B). Immunoblotting of the embryonic extracts revealed equivalent Sgg Zw3 levels in the wild-type sgg zw3 M11-1 allele and in the pUAS-Wg-expressing embryos, as expected ( Fig. 2A). Armadillo immunoblots revealed accumulation of Arm protein in the Sgg Zw3 M11-1 and pUAS-Wg extracts.
Overexpression of Dsh Represses Sgg Zw3 Protein Kinase Activity-Overexpression of Dsh protein in cl-8 and S2 cells bypasses the need for Wg and mimics Wg signaling (15). To investigate the effect of overexpression of Dsh on Sgg Zw3 activity, we used S2 and cl-8 cell lines expressing Dsh under the control of an inducible metallothionein promoter. Treatment of these cell lines with CuSO 4 led to a time-dependent increase in Dsh protein levels, as well as induction of forms of the protein with reduced electrophoretic mobility similar to the forms observed in untransfected cl-8 cells exposed to Wg protein (Fig. 3,  A and C). Concomitant with the increase in Dsh protein levels was an increase in Arm levels (Fig. 3, A and C), indicating that overexpression of Dsh in S2 and cl-8 cells mimics Wg signaling.
To determine whether Dsh protein inhibits Sgg Zw3 activity, we examined Sgg Zw3 protein kinase activity in the Dsh-inducible cl-8 and S2 cell lines (Fig. 3, B and D). Dsh overexpression in cl-8 and S2 cells revealed similar inhibition curves in both lines and induced a rapid decrease in Sgg Zw3 activity that was detectable after 2 h and reached a maximum (70%) after 4 -6 h, whereas Sgg Zw3 expression levels were not affected (Fig. 3, A  and C). The decrease in Sgg Zw3 activity observed in the Dsh experiments in cl-8 cells coincided with the effects of Wg on Sgg Zw3 activity in cl-8 cells and supports the genetic model in which Wg repression of Sgg Zw3 is mediated via Dsh.

Overexpression of Drosophila Frizzled 2, a Putative Wg Receptor, Mimics Wg Signaling-Unlike cl-8 cells, S2 cells do not respond to extracellular Wg as judged by Dsh modification and
Arm stabilization (data not shown) (15,24). Transfection of the transmembrane protein Drosophila Frizzled 2 (Dfz2) into S2 cells enables the cells to accumulate Arm in response to Wg, suggesting that Dfz2 acts as a receptor for Wg and that the reason for the lack of responsiveness of these cells to Wg is simply due to lack of Dfz2 expression (7). To investigate whether Dfz2 expression affected Sgg Zw3 activity, we used S2 cell lines expressing Dfz2 under the control of an inducible metallothionein promoter. Addition of CuSO 4 to the medium of these cells induced an increase in the levels of Dfz2 RNA (Fig.  4A), leading to the appearance of slower migrating forms of Dsh and an increase in cytoplasmic Arm levels within 2 h, whereas Sgg Zw3 protein levels were unaffected (Fig. 4A). However, immunoprecipitates of Sgg Zw3 exhibited a time-dependent decrease in protein kinase activity upon induction of Dfz2 expression, similar to the effects of overexpression of Dsh in S2 cells (Fig. 4B). Together, these data demonstrate that overexpression of Dfz2 in S2 cells is sufficient to trigger the Wg pathway, including modification of Dsh, repression of Sgg Zw3 , and stabilization of Arm.
Dishevelled Induces Serine Phosphorylation of Sgg Zw3 -To probe the mechanism via which Wg, Dfz2, and Dsh inactivate

FIG. 2. Sgg Zw3 activity is inhibited by Wg in the Drosophila embryo.
A, equal amounts of embryonic extracts with the indicated genotype were separated by SDS-PAGE to detect levels of Arm and Sgg Zw3 . In two lanes, embryos homozygous for a heat shock-inducible transgene encoding Sgg Zw3 (HS-Sgg Zw3 ) were either heat-shocked (ϩ) or maintained at room temperature (Ϫ) prior to analysis. Immunoblotting showed an accumulation of Arm protein levels in the embryos expressing wg ubiquitously and sgg zw3 M11-1 mutant embryos. B, Sgg Zw3 proteins were immunoprecipitated from the extracts of embryos with the indicated genotype, and their activities were measured (percent of the wild type). Sgg Zw3 , S2 cell lines harboring inducible Dsh were metabolically labeled with [ 32 P]phosphate, and Sgg Zw3 was immunoprecipitated and resolved by SDS-PAGE. Induction of Dsh expression caused a 2-2.5-fold increase in [ 32 P]phosphate associated with Sgg Zw3 (Fig. 5A). Subsequent phosphoamino acid analysis revealed the presence of only phosphoserine in the S2 cell sample (Fig. 5B). These data suggest that Dsh induces a specific increase in serine phosphorylation of Sgg Zw3 , which may mediate the reduction in protein kinase activity. Surprisingly, Sgg Zw3 in S2 cells does not contain detectable phosphotyrosine (34). Sgg Zw3 contained both phosphotyrosine and phosphoserine in cl-8 cells. Since induction of the Wg pathway resulted in equal -fold inhibition in both S2 and cl-8 cells, we conclude that Wg-mediated regulation of Sgg Zw3 is independent of tyrosine phosphorylation.
Phosphorylation of Arm and D-Axin by Sgg Zw3 -We have shown that negative regulation of Sgg Zw3 activity leads to Arm accumulation in Drosophila embryos and cells. Biochemical analysis has indicated that D-Axin/Axin negatively regulates ␤-catenin/Arm by interacting with GSK-3␤/Sgg Zw3 (19). 2 D-Axin is structurally related to vertebrate Axins, with the regions of highest identity corresponding to previously defined binding domains of Axin. 2 Armadillo contains "consensus" phosphorylation site sequences for GSK-3/Sgg Zw3 (35). D-Axin also contains such sequences (19). 2 However, it has been reported that mammalian GSK-3 phosphorylates ␤-catenin significantly only in the presence of the Axin protein (19). Therefore, we examined whether Sgg Zw3 could phosphorylate Arm and D-Axin under conditions in which these proteins formed a complex. To determine whether D-Axin and Arm are substrates for Sgg Zw3 , we purified D-Axin or various deletion mutants of D-Axin and Arm from E. coli as histidine fusion proteins (Fig. 6). Baculovirus-expressed GST-Sgg Zw3 (36) phosphorylated D-Axin, D-Axin-(302-746), D-Axin-(356 -565), and D-Axin-(356 -746), but not D-Axin-(383-565) and D-Axin-(34 -356) (Fig. 6). In the absence of D-Axin, no significant phosphorylation of Armadillo was observed, whereas in its presence, the phosphorylation was greatly increased (Fig. 6). These data indicate that Sgg phosphorylation of Armadillo is directed via D-Axin.
Inhibition of Sgg Zw3 Activity by Wg Affects Its Phosphorylation and Interaction with D-Axin Protein-We found that D-Axin is phosphorylated by Sgg Zw3 and binds to both Sgg Zw3 and Arm. 2 We therefore examined whether the inhibition of Sgg Zw3 activity by Wg affects its interaction with D-Axin and monitored the level of phosphorylation of D-Axin. To test this possibility, in vitro binding and phosphorylation assays were carried out using a D-Axin-(330 -642) fusion protein containing Sgg Zw3 -binding sites and consensus sites of phosphorylation for Sgg Zw3 . D-Axin-(330 -642)-His 6 was transfected as a histidine fusion protein into cl-8 cells, cl-8 cells treated with Wg, and cl-8 cells expressing Sgg Zw3 . The histidine-tagged complexes from the cl-8 cell lysates were purified using nickel-Sepharose beads, and the amount of Sgg Zw3 captured on the beads was determined by immunoblotting. In addition, the phosphorylation of D-Axin-(330 -642)-His 6 by Sgg Zw3 was determined by addition of [␥-32 P]ATP.
In the lysates from cells treated with Wg, Sgg Zw3 was found in association with D-Axin-(330 -642)-His 6 . However, the degree of binding was reduced ϳ2-fold compared with the amount of Sgg Zw3 associated with Axin in lysates of untreated cl-8 cells (Fig. 7). The negative effect of Wg signal on the binding of Sgg Zw3 correlated with a decrease in phosphorylation of D-Axin. By contrast, Axin complexes within lysates expressing Sgg Zw3 contained more Sgg Zw3 protein as well as higher Axin kinase activity (Fig. 7). These results indicate that D-Axin physically interacts with Sgg Zw3 and that Wg signaling leads to a reduction of both Sgg Zw3 activity and its interaction with D-Axin. DISCUSSION Previous studies have shown that treatment of cl-8 cell lines with Wg leads to hyperphosphorylation of Dsh protein and to cytoplasmic accumulation of Armadillo (15,23,24). Here, we report that Wg signaling as initiated by Wg, Dfz2, or Dsh expression causes enzymatic inactivation of Sgg Zw3 activity in concert with stabilization of Arm. These data indicate that Wg or overexpression of "upstream" components of this pathway mimics Wingless signaling by specifically inhibiting the activity of Sgg Zw3 .
We have demonstrated that regulation of kinase activity, rather than protein levels, is the main determinant of the effects of Wg on Sgg Zw3 , suggesting post-translational modification of this protein kinase activity. In support of this, induction of Dsh expression increased the levels of Sgg Zw3 phosphorylation 2-fold (Fig. 6), and the presence of phosphoserine in Sgg Zw3 protein from S2 cells suggested that the mechanism of repression of Sgg Zw3 activity is mediated by serine phosphorylation. Previous studies have shown that members of the GSK-3 family are inhibited by phosphorylation at an amino-terminal serine residue (serine 9 in GSK-3␤ and serine 21 in GSK-3␣) (33,37). Phosphorylation of the Sgg Zw3 residue equivalent to serine 9 does not appear to be the mechanism via which the Wg pathway inhibits Sgg Zw3 for several reasons. In mammals, this site is targeted by agents acting via phosphatidylinositol 3Јkinase, and the residue can be phosphorylated in vitro and in transfected cells by protein kinase B/AKT (38). However, Wg inhibition of GSK-3 in 10-T1/2 cells is not sensitive to inhibitors of phosphatidylinositol 3Ј-kinase, nor is Drosophila protein kinase B activity stimulated by Wg (27). 3 Furthermore, Dshinduced tryptic phosphopeptides of Sgg Zw3 are inconsistent with phosphorylation of the site analogous to serine 9 in GSK-3␤. 3 Identification of the Wg/Dsh-inducible serine residue(s) on Sgg Zw3 is underway.
Although our data provide biochemical support for the genetically defined Wg pathway, at least three gaps remain in this signaling cascade: the mechanism via which Dsh is activated by Dfz2, the mechanism by which Dsh inhibits Sgg Zw3 , and the means by which Sgg Zw3 induces turnover of Arm. We found a correlation between the modification of the phosphorylation state of Dsh protein and an increase in Arm stability, in agreement with the studies of Yanagawa et al. (15) and Willert et al. (23). A similar correlation was observed between the decrease in Sgg Zw3 activity and accumulation of hypophosphorylated Arm protein. Willert et al. (23) found that whereas Dfz2 expression induced Dsh hyperphosphorylation, it did not induce stabilization of Arm. In our hands, Dfz2 expression was sufficient for both of these processes in S2 cells. The reason for the discrepancy is unclear, but may relate to the degree of overexpression of Dfz2.
Yost et al. (39) proposed that ␤-catenin is directly phosphorylated by GSK-3, consistent with the finding that phosphorylation of Arm protein is decreased with the inhibition of Sgg Zw3 activity. However, Arm is a poor in vitro target of Sgg Zw3 . Phosphorylation of Arm is enormously increased in the presence of D-Axin. We have demonstrated that D-Axin is phosphorylated by Sgg Zw3 and that the binding of Sgg Zw3 to D-Axin is dependent upon the level of Sgg Zw3 activity. Repression of Sgg Zw3 activity by Wg signaling induced dissociation of the Sgg Zw3 ⅐D-Axin⅐Arm complex, leading to an accumulation of Arm protein. Together, these data suggest that Sgg binding is dependent upon or stimulated by its phosphorylation of Axin. Once bound to Axin, it can access the Arm molecule that is associated with Axin and phosphorylate it. Inactivation of Sgg results in dephosphorylation of Axin and release of the kinase, compartmentalizing it away from Arm.
Mammalian studies have suggested that a more complex mechanism for the regulation of ␤-catenin levels by GSK-3 involved another player, APC. In this case, Axin forms a complex with GSK-3, ␤-catenin, and APC (19,20). APC is directly phosphorylated by GSK-3 via Axin, which increases binding of APC to ␤-catenin and its subsequent degradation (40,41). Mutation of a Drosophila APC homolog did not affect Wg function, suggesting either divergence of the molecular mechanisms of Arm stabilization or the existence of additional APC-like molecules in flies (21). Resolution of these mechanisms will require identification of the serine kinase acting to inhibit Sgg Zw3 and the means by which it is, in turn, controlled by Dsh.