Bridging Decapentaplegic and Wingless signaling in Drosophila wings through repression of naked cuticle by Brinker

During animal development, proper organogenesis is achieved by simultaneous actions of highly conserved signaling pathways. One of the best-studied examples is the Drosophila imaginal discs, in which multiple signaling molecules, including Wg and Dpp, function as organizational cues to guide patterning. Both Wg and Dpp are thought to act as morphogens that activate a set of pathway-specific target genes in a gradient-dependent manner (Cadigan, 2002; Tabata and Takei, 2004). For instance, Wg target genes are predominantly regulated by the Wnt response elements (WREs) occupied by the transcription factor T cell factor (TCF; Pan – FlyBase) (for a review, see Cadigan, 2012). Wg signal transduction is centered around the stability of the cytosolic pool of Armadillo (Arm, the Drosophila homolog of β-catenin). Wg stabilizes Arm, causing its cytosolic accumulation and translocation into the nucleus, where Arm displaces co-repressors such as Groucho (Gro) and binds to TCF (Cadigan, 2012). This switches the TCF complex from a repressor to a transcriptional activator and turns on Wg target gene expression. In the case of Dpp target gene activation, two mechanisms have been proposed (for reviews, see Affolter et al., 2001; Affolter and Basler, 2007). In the first, Dpp-dependent receptor activation induces phosphorylation of Mad, which then forms a heteromeric complex with Medea (Med) in the nucleus and functions as a sequence-specific activator (Wisotzkey et al., 1998). Alternatively, Dpp signaling can act by inhibiting Brk, a transcriptional repressor (Campbell and Tomlinson, 1999; Jaźwińska et al., 1999; Minami et al., 1999). This model has been well addressed in the developing fly wing, in which dpp is expressed in a narrow stripe across the anterior/posterior (A/P) boundary. brk is repressed in a dose-dependent manner by Dpp through a Schnurri-Mad-Med repressing complex and is expressed in complement with the Dpp gradient (Müller et al., 2003). Brk thus acts as a major interpreter of Dpp signaling in wing discs.

Developmental signaling pathways are often regulated by negative-feedback regulators, thus limiting their range of actions. Such regulation contributes to the refinement of target gene expression in cells receiving multiple signaling cues (Gerlitz and Basler, 2002; Chang et al., 2008a). nkd, expression of which is universally induced by Wg, is one such feedback inhibitor of Wg signaling in flies (Zeng et al., 2000). Nkd interacts directly with Dishevelled (Dsh) to antagonize Wnt signal transduction (Rousset et al., 2001), suggesting that nkd functions mainly to limit Wg activity in order to achieve proper activation of Wg target genes.

In general, crosstalk between signaling pathways ensures the proper balance of signaling during development. A notable example of this is in fly leg discs, in which Wg and Dpp signaling mutually repress each other to specify ventral and dorsal leg fates, respectively (Brook and Cohen, 1996; Jiang and Struhl, 1996; Johnston and Schubiger, 1996; Morimura et al., 1996; Penton and Hoffmann, 1996; Theisen et al., 1996; Morata, 2001). In addition, Wg and Dpp are both required for cell fate specification along the proximal/distal axis via direct activation of Dll, precise expression of which is crucial for leg development (Estella et al., 2008). In the wing disc, wg and dpp are expressed in perpendicular narrow stripes along the dorsal/ventral (D/V) and A/P boundaries, respectively. Unlike in leg discs, where wg and dpp are expressed in distinct domains, these two signaling molecules share a common expression pattern in the center of the wing pouch. Recent studies have suggested that Mad, one of the key transducers in Dpp signaling, might play a role in the Wg pathway in the wing. One study suggests that Mad and Arm compete for binding to TCF, with Mad serving as an intermediate through which Dpp signaling suppresses the Wg pathway (Zeng et al., 2008). Other studies, however, propose an alternative model in which Mad is required for both Wg and Dpp signal transduction depending on its phosphorylation status (Eivers et al., 2009; Eivers et al., 2011).

In this article, we uncover an additional level of Wg-Dpp crosstalk involving Nkd, a feedback inhibitor of Wg signaling, and Brk, the major transcriptional repressor for Dpp target genes. We show that Brk directly represses nkd in vitro and in vivo. The biological relevance of such regulation is exemplified in fly wing discs, where brk is required for proper expression of Wg targets in the presumptive hinge region, coincident with the fact that brk is expressed at the highest level in this area. In adult wing, altering brk levels gives rise to phenotypes similar to the gain and loss of Wg activities. Moreover, activation of Dpp signaling reduces Dll levels and increases nkd expression. Taken together, our data support a novel model in which Dpp signaling inhibits Wg outputs through a brk-nkd negative regulatory circuit.

All lines were cultured with standard medium at 25°C. brk somatic clones were generated exactly as described (Schwank et al., 2008). P[UAS-brk] (Jaźwińska et al., 1999), P[UAS-nkd] (Zeng et al., 2000), P[UAS-tkv^QD] (Nellen et al., 1996) and P[UAS-TCFΔN] (van de Wetering et al., 1997) were used in the overexpression experiments. P[UAS-brk^RNAi] (VDRC# GD2919, KK101887) and P[UAS-nkd^RNAi] (VDRC# GD3005) lines were obtained from the Vienna Drosophila RNAi Center. Experiments with En-Gal4 and Vg-Gal4 were carried out at 25°C, whereas those with Ptc-Gal4 were carried out at 20°C. For temporal ectopic expression of brk and tkv^QD driven by En-Gal4, P[UAS-brk] and P[UAS-tkv^QD] were crossed with P[En-Gal4], P[tubP-Gal80^ts], P[UAS-GFP] line, and the progenies were shifted to a restrictive temperature at 29°C from the second instar larval stage and thereafter (Schwank et al., 2008). For heat-shock induction of tkv^QD, P[Hsp70-Gal4]/P[UAS-tkv^QD] third instar larvae were heat shocked at 37°C for 1 hour, once a day for 2 days, and total RNAs were extracted from young adult flies 4 days after eclosion. For Nkd-lacZ reporter expression, we used a line containing both upstream and intron WREs of nkd as described (Chang et al., 2008a).

Rabbit polyclonal anti-Brk antiserum was generated against a Brk fragment (361-619 aa). For western blot analyses, anti-Brk (1:5000), anti-Tubulin (1:5000, Sigma) or anti-Flag (1:5000) antibodies were used. Immunostaining of imaginal discs was performed as described (Fang et al., 2006) using the following antibodies: anti-Dll (1:500) (Panganiban et al., 1995), anti-Sens (1:1000) (Nolo et al., 2000) and anti-Wg (1:200, Developmental Studies Hybridoma Bank at the University of Iowa, USA).

EMSA was performed with a Lightshift Chemiluminescent EMSA Kit (Thermo Scientific) according to the manufacturer’s instructions. For production of recombined GST-BrkDBD, a cDNA fragment encoding 44-99 amino acid residues of Brk (Saller et al., 2002; Cordier et al., 2006) was expressed in Escherichia coli BL21 and affinity purified. The probe sequences were 5′-acaaaacgctttgtcggtcattagctttgagtggacgccg-3′ for wild type and 5′-acaaaacgctttgtcggtcattagctttgagtgtcagccg-3′ for the mutant form.

Kc167 (Kc) cells were routinely cultured in the Schneider’s Drosophila media (Life Technologies) containing 5% FBS (Life Technologies) at 25°C. RNAi-mediated gene knockdown was performed as described (Worby et al., 2001). Double-stranded RNAs (dsRNAs) for control and gro were synthesized as described (Fang et al., 2006), and the dsRNA primer corresponding to brk were 5′-gaattaatacgactcactatagggagaaatttcaagcggcccaaggattcttcg-3′ and 5′-gaattaatacgactcactatagggagatgttgctccttctgatgctgcatttgagg-3′. A million cells were cultured in serum-free medium for 1 hour, 9 μg of a mixture of two dsRNAs (combination indicated in Fig. 1A, or the indicated one plus the control dsRNA) was added, and FBS was supplemented to reach a final concentration of 5%. Total RNAs were isolated 4 days later using the RNeasy Mini Kit (Qiagen), and 2 μg of each sample was reverse transcribed with a superscript III reverse transcriptase (Life Technologies). qPCR assays were performed with Applied Biosystems 7300 Real-Time PCR System (Life Technologies). qPCR detections for wg and brk were carried out with a FastStart Universal SYBR Green Master Mix (Roche Applied Science). The primer pairs were for wg, 5′-cgtcagggacgcaagcata-3′ and 5′-attgtgcgggttcagttgg-3′; for brk, 5′-tcgggagctttgaatttcaag-3′ and 5′-tggctgtttgtggcattctc-3′. nkd qPCR was performed with FastStart TaqMan Probe Master (Roche Applied Science). The primer pairs and probe used were 5′-cgcatccgactggaggaa-3′, 5′-ggtcgtagaacgtgaacgagaac-3′ and 5′-FAM-tcacctgcgacgtgtccgtgg-BHQ1-3′. For nkd qPCR analysis in wing discs, total RNAs were extracted from 30 wing discs of third instar larvae using the RNeasy Micro Kit (Qiagen).

The reporter gene assay was performed as previously described (Fang et al., 2006; Chang et al., 2008a), except that pArm-Arm^* and pArm-lacZ instead of pAc vectors were used. Reporter constructs were generated as previously described (Chang et al., 2008a). For IntE1200, primers 5′-agctcacgcgtgctctcgggccacttcttggaa-3′ and 5′-cgcatcccggggcctgccactttgcttcagtgaga-3′, with underlined bases indicating introduced restriction sites, were used to amplify the intronic region of nkd, and subcloned into a pGL3 (Promega) vector upstream to a hsp70 minimal promoter (Chang et al., 2008a). IntE255 was as described (Chang et al., 2008a). IntE255^Brkmut were generated by PCR-introduced base substitutions that were the same as in the mutant probe in the EMSA assay. pAc-Brk-2XFlag was made from insertion of a PCR-amplified full-length coding sequence of brk plus two Flag tags into pAc5.1 (Life Technologies).

Transient transfections were carried out with Fugene HD (Roche Applied Science) according to the manufacturer’s instructions. A plasmid mixture containing 100 ng reporter, 1 ng pArm-lacZ, 10 ng pArm-Arm^* and 10-80 ng pAc-Brk-2XFlag were co-transfected into a million cells. The empty pAc5.1 vector was used to normalize the DNA content or as a control. For experiments with RNAi followed by reporter assay, a million cells were incubated with 9 μg control or brk dsRNA for 2 days, washed with PBS twice and then transfected with pArm-lacZ (1 ng) and pArm-Arm^* (10 ng). Luciferase and β-galactosidase activities were assayed 2 days after transfection, using the Tropix Luc-Screen and Galacto-Star Kits (Life Technologies) and quantified with a Chameleon plate luminometer (Hidex Personal Life Science). Transfection efficiencies were normalized to the β-galactosidase activities.

ChIP assays were carried out according to a protocol from the Furlong lab (Sandmann et al., 2006). Briefly, about one gram of synchronized embryos were collected, fixed and sonicated with a Bioruptor sonicator (Diagenode) to generate DNA fragments of ∼500 bp in length. The immunoprecipitation were performed with anti-Brk serum (2 μl) or normal rabbit IgG (20 μg) and the subsequent steps were carried out as previously described (Fang et al., 2006). Primer pairs used in ChIP assays were 5′-accttctggctttggagcag-3′ and 5′-tgggctcctcataaactggc-3′ for CDS, 5′-tcaatcagacgtcagaggtaccg-3′ and 5′-ctgatggaagaaccgtgttgg-3′ for IntE^Brk.

In situ hybridizations for nkd in imaginal discs were performed as previously described (Chang et al., 2008a). For quantifying the in situ signals, images were processed in Adobe Photoshop, converted to gray scale and inverted (supplementary material Fig. S3). The mean gray values were measured by ImageJ (NIH, v1.46). Three nonoverlapping areas in the nkd expression domain (near D/V boundary) were sampled from both anterior or posterior compartments and, for each individual disc, the mean gray values were measured and averaged, and the mean background value was subtracted from the average. Statistical analysis of the results was performed using ANOVA followed by post-hoc Tukey’s test for comparison between groups. For fluorescence in situ hybridization (FISH), an nkd probe of the same sequence was labeled with biotin. After hybridization, TSA amplification (PerkinElmer) was applied and visualized by TRITC-strepavidin (1:300, Jackson Immunochemicals) and anti-GFP antibody (Roche) staining.

We have previously shown that C-terminal Binding Protein (CtBP) directly represses nkd, a direct target of the Wg signaling pathway, via an intronic region of nkd (Fang et al., 2006). In addition, the physical binding of CtBP to the nkd locus is independent of TCF (Fang et al., 2006). However, CtBP functions as a transcriptional co-repressor and is not likely to bind to DNA directly (for a review, see Chinnadurai, 2002). In exploring the potential transcription factor that would recruit CtBP to DNA, two pieces of data drew our attention to Brk, a CtBP-binding repressor in Dpp signaling (Hasson et al., 2001; Zhang et al., 2001). By RNAi-mediated gene knockdown in Drosophila Kc167 (Kc) cells, we found that Brk repressed nkd expression in parallel with Gro (Fig. 1A), consistent with the synergistic repression of nkd by CtBP and Gro (Fang et al., 2006). We took advantage of the aligned genome sequences from 12 fly species (Stark et al., 2007) and identified a putative Brk recognition site (Sivasankaran et al., 2000; Rushlow et al., 2001; Zhang et al., 2001) within a completely conserved 66-bp sequence in the first intron of nkd (Fig. 1B; supplementary material Fig. S1). In addition to the presumptive Brk site, we also noticed two perfect neighboring TCF sites (Fig. 1B) (Chang et al., 2008b). Interestingly, this 66-bp DNA region (referred to as IntE^Brk) falls into a region that was previously termed IntE and has been carefully characterized as one of the major WREs directing nkd expression in response to Wg signaling (Fang et al., 2006; Chang et al., 2008a).

To test whether Brk interacts directly with IntE^Brk, we performed an electrophoretic mobility shift assay (EMSA) using a recombined GST protein fused to the DNA-binding domain of Brk (GST-BrkDBD) (Jaźwińska et al., 1999; Saller and Bienz, 2001; Cordier et al., 2006). GST-BrkDBD caused a clear shift of a 40-bp IntE^Brk probe (Fig. 1B,C) in a dose-dependent manner, indicating a direct interaction between Brk and IntE^Brk. The interaction appeared to be specific: either an excessive amount of non-labeled probe or the addition of anti-GST antibody readily abolished the GST-BrkDBD-induced probe shift (Fig. 1C). Furthermore, we did not detect such a shift with a mutant probe in which three bases in the presumptive Brk site were substituted (Fig. 1C). Therefore, Brk is likely to interact with IntE^Brk via the Brk site that we have identified.

We next investigated whether Brk interacts with IntE^Brk in vivo and, if yes, whether such binding has any connection with Wg signaling? Therefore, we performed chromatin immunoprecipitation (ChIP) analysis on fly embryos. wg is initially expressed in embryos just prior to gastrulation, ∼3 hours after egg laying (AEL), and functions as one of the major factors directing the patterning processes in fly segmentation (Bejsovec and Martinez Arias, 1991; Dougan and DiNardo, 1992). Accordingly, as a negative-feedback regulator of Wg signaling, nkd transcript was absent 0-2 hours AEL and reached its highest level 4-8 hours AEL in embryos (Zeng et al., 2000). We therefore made extracts from 0-2 hours and 5-7 hours AEL embryos to test the binding of Brk to IntE^Brk. Our data show that the physical occupation of Brk on this region was significant in 0-2 hours AEL embryos (Fig. 1D). In the 5-7 hours AEL embryos, however, the occupation of nkd by Brk was dramatically lower despite brk being expressed at a much higher level than in earlier embryos (Fig. 1D,E). These results indicate that Brk binds to nkd in vivo and that the physical occupation of nkd by Brk is inversely correlated with the level of Wg signaling.

We have shown that Brk represses the Wg target nkd in Kc cells (Fig. 1A). To examine further whether Brk also influences Arm-dependent nkd activation, we carried out reporter assays in Kc cells using various fragments of the nkd intronic WRE (IntE), including IntE^Brk. As previously reported (Fang et al., 2006; Li et al., 2007; Chang et al., 2008a; Parker et al., 2008), IntE reporters are highly activated by a constitutively active form of Arm (Arm^*) (Freeman and Bienz, 2001) in Kc cells. We observed that a 1.2-kb IntE reporter (IntE1200, Fig. 1B) ∼4.5 kb downstream of the nkd transcription start site was activated ∼100-fold following a moderate dose of transfected Arm^* (Fig. 2A). Following co-transfection with a Brk expression vector, Arm-dependent activation of IntE1200 was inhibited by Brk in a dose-dependent fashion (Fig. 2A,E). Consistent with this observation, RNAi-mediated brk knockdown in Kc cells significantly increased the Arm^*-dependent activation of IntE1200 (Fig. 2B,F). Similar results have been observed in a shorter, 255-bp version of IntE (IntE255; Fig. 1B; Fig. 2C,D) (Chang et al., 2008a). To test whether the Brk site in IntE255 is responsible for the inhibitory effect of Brk, we constructed a mutant reporter in a similar manner to the mutant probe used in the EMSA (IntE255^Brkmut; Fig. 1B). Although it still responded to Arm^* stimulation, IntE255^Brkmut was no longer affected by the Brk protein (Fig. 2C,D). Consistent with our ChIP analysis in the embryos, these results indicate that Brk directly represses the intronic WRE activity upon Wg activation. The repressive role that we observed for Brk in these experiments cannot be explained by the interaction between Mad and TCF, as suggested by previous studies (Zeng et al., 2008; Eivers et al., 2011), because excess Brk did not affect the Arm-dependent activation of dTF12 (DasGupta et al., 2005), a Wg reporter composed of multimerized optimal TCF sites (supplementary material Fig. S2).

We have so far shown that Brk directly represses nkd in vitro. An obvious question is whether Brk also represses nkd in vivo. We made use of a lacZ reporter that is driven by a combination of WREs in the regulatory sequences in the nkd locus (Nkd-lacZ) (Chang et al., 2008a). When we knocked down brk using a brk RNAi line (UAS-brk^RNAi) driven by Dpp-Gal4, we detected an ectopic lacZ expression along the A/P boundary, as indicated by Engrailed expression (Fig. 3D-F), compared with control wing discs (Fig. 3A-C), suggesting that Brk represses nkd expression in wing discs. Consistent with this observation, by in situ hybridization we detected enhanced nkd transcript levels in the posterior half of the wing discs when brk^RNAi was driven by Engrailed-Gal4 (En-Gal4) (Fig. 3G,H). Conversely, En-Gal4-driven brk overexpression caused reduced nkd transcription (Fig. 3I). These changes of nkd expression are significant when mean gray values are compared and judged by statistical analysis (Fig. 3J; supplementary material Fig. S3).

To examine this effect in a more rigorous way, we generated mitotic clones of a null allele of brk, brk^M68 (Jaźwińska et al., 1999), and examined nkd transcripts in wing discs by fluorescence in situ hybridization (FISH). As expected, nkd FISH signals were dramatically increased in the brk clones (Fig. 3K-M). We noticed that nkd upregulations were more pronounced in brk clones positioned at A/P extremes, consistent with published studies that brk is expressed at higher levels in this region (Campbell and Tomlinson, 1999; Jaźwińska et al., 1999; Minami et al., 1999). In addition, we have used RT-qPCR analysis to confirm the increase of nkd transcripts in wing discs that underwent brk clones induction (Fig. 3N). Taken together, these data indicate that Brk represses nkd in wing discs.

To assess the role of brk in Wg signaling in vivo, we examined Wg targets, such as Dll and senseless (sens), in the wing imaginal disc. Ectopic brk driven by Patched-Gal4 (Ptc-Gal4), which is activated along the A/P compartment boundary, caused an expansion of Dll expression (Fig. 4A-F), whereas Wg expression was not affected. Based on our conclusion from Fig. 3 that Brk represses nkd, the increased Dll expression could be due to the lowered expression of Nkd, the negative regulator of Wg signaling. If so, then ectopic nkd should be able to abolish Dll expansions caused by the over production of brk. Indeed, we found that co-expression of nkd with brk gave rise to similar Dll patterns to those seen in control discs (Fig. 4G-I). These observations are consistent with the notion that brk enhances Wg signaling.

To look into the possible effects of endogenous brk on Wg signaling, we examined Dll and Sens expressions in the brk^M68 clones. Dll expression in these clones was not always the same. We did not observe a detectable change of Dll expression in brk clones positioned within wing pouch areas. However, in clones away from A/P boundary, both anteriorly and posteriorly in hinge areas, we saw dramatic Dll reductions (supplementary material Fig. S4). Fig. 4J-L shows a typical brk clone across the hinge and wing pouch, in which Dll was completely absent in the hinge region but was unaffected in the wing pouch area. Such a positional effect is consistent with the observation of increased nkd expression in brk clones (Fig. 3K-M) and might reflect a dosage dependence of brk. The hinge expression of Dll depends on canonical Wg signaling, as Dll was significantly reduced in pygo clones in this area (supplementary material Fig. S5) (Parker et al., 2002). Wg was not affected in brk clones (supplementary material Fig. S6), indicating that the reduction of Dll expression was not due to a direct change in Wg expression. Similar results were observed for Sens, another Wg target in wing discs (Fig. 4M-O). Again, if it stands true that the reduced Dll level in brk clones is due to de-repressed nkd expression, it should be reversed in a loss-of-nkd background. To test this, we used a UAS-RNAi allele (UAS-nkd^RNAi) generated according to nkd sequence. As shown in supplementary material Fig. S7, En-Gal4-driven UAS-nkd^RNAi dramatically knocked down nkd levels in the posterior compartment. Therefore, we examined Dll expression in brk clones in this nkd RNAi background. We did observe that Dll expression was lost in anterior brk clones but not in posterior ones in the hinge (Fig. 4P-R), indicating a need for nkd in the reduction of Wg signaling in brk clones. Together, these results demonstrate that brk is required for Wg signaling in hinge area of wing discs.

The data described thus far indicate that Brk represses nkd and is required for Wg target gene expression in the wing disc. Is brk also required for adult wing structures? It is well documented that the wing margin is defined and patterned from a stripe of cells expressing wg during the third instar larvae (Phillips and Whittle, 1993; Couso et al., 1994). Loss of Wg signaling leads to defects of the adult wing margin, characterized by loss of bristles and/or notches in the wing. Conversely, elevated Wg signaling typically causes extra bristles.

Consistent with a role of brk in wing margin formation, studies from independent laboratories have shown that loss of brk results in notched wing in addition to wing outgrowths (Campbell and Tomlinson, 1999; Jaźwińska et al., 1999). To examine this in more detail, we analyzed wing phenotypes in brk^M68 clones in a more quantitative way. In flies potentially bearing brk clones, we found that 19% of the flies gave rise to notches in the wing, a phenotype indicative of a loss of Wg signaling (Fig. 5C; n=163). In addition, we saw many swirled hairs in areas close to the wing notches (Fig. 5D), which are consistent with the previous finding that a timed over production of nkd gives rise to planar cell polarity (PCP) phenotypes (Rousset et al., 2001). By contrast, we only occasionally saw (5%, n=163) wing outgrowths in these flies. These results may indicate that aside from a repressive role of Brk in Dpp signaling, Brk might play an essential role in Wg signaling. In further support of this notion, in wings with ectopic brk driven by Vestigial-Gal4 (Vg-Gal4), we often observed stout bristles indicative of a gain of Wg signaling (Fig. 5E,F) (Couso et al., 1994; Zhang and Carthew, 1998; Zeng et al., 2008), although the wings were much smaller compared with wild-type flies, as previously reported (Martín et al., 2004). Furthermore, co-expression of TCFΔN, which inhibits Wg signaling (van de Wetering et al., 1997) as evidenced by causing notches (Fig. 5I,J), suppressed these stout bristles (Fig. 5G,H). We interpreted these results as evidence that Brk antagonizes Dpp signaling and promotes Wg signaling at the same time. From the experiments inducing the loss or gain of brk in fly wings, we have found phenotypes resembling those observed in the loss or gain of Wg signaling, respectively, which might have been overlooked in previous studies.

As mentioned earlier, brk is directly repressed by Dpp signaling (reviewed by Affolter and Basler, 2007). According to our model, increasing Dpp signaling should also increase nkd expression and inhibit Wg signaling in the wing. Indeed, Tkv^QD, a constitutively active form of the Dpp receptor, increased nkd expression in the wing disc and globally in adult flies (Fig. 6D,E). In wing discs, Dll is considered to be a long-range target of Wg (Zecca et al., 1996; Neumann and Cohen, 1997) and is normally expressed at a comparatively low level close to the A/P boundary, resulting in two expanded domains with one being slightly wider in the posterior region (Fig. 3A). When tkv^QD was overexpressed in the posterior half by En-Gal4, Dll was expressed in a domain that was apparently narrower posteriorly than anteriorly (Fig. 6A-C), indicating a significant decrease in Dll expression. These results clearly show that Dpp signaling promotes nkd expression and suppresses Wg signaling in wing discs.

In this study, we have shown that Brk directly represses nkd expression in vitro and in vivo. The direct repression of nkd by Brk is underscored by three of our observations. First, we have identified a Brk site in the intronic region of nkd, which Brk physically occupies in vitro (Fig. 1B,C). Second, ChIP analysis shows that Brk binds a DNA region near this Brk site in embryos in a manner inversely related to Wg activity (Fig. 1D,E). Third, our reporter analysis in Kc cells indicates that Brk represses Arm-dependent activation of an intronic WRE containing this Brk site, but only when the Brk site is intact (Fig. 2). In addition, our genetic analyses have shown that the repression of nkd by Brk is functionally significant. In the developing wing, we found that the loss of brk de-represses nkd (Fig. 3) and downregulates Wg target proteins, such as Dll and Sens (Fig. 4). Conversely, ectopic brk inhibits nkd expression (Fig. 3) and markedly enhances Dll expression (Fig. 4). Furthermore, removal of nkd prevents the loss of Dll in brk clones whereas co-expression of nkd abolishes the expanded Dll caused by ectopic brk (Fig. 4). In adult wing, the loss and gain of brk phenotypically resembles the loss and gain of Wg signaling, respectively (Fig. 5). Consistent with a repressive role of Dpp cascade on brk, we found that ectopic Dpp signaling enhances nkd and inhibits Wg signaling (Fig. 6). These results support a model in which Dpp signaling increases the expression of Nkd, a Wg inhibitor, by the downregulation of Brk, and thereby inhibits the Wg outputs. In another words, nkd might fall into a class of Dpp targets, which are de-repressed upon the activation of Dpp signaling. We have thus uncovered a previously unsuspected molecular mechanism underlying the interaction between Wg and Dpp signaling pathways in Drosophila wing development.

Until recently, little has been known about the cross-interaction between Wg and Dpp signaling in Drosophila wings, in spite of the fact that the fly wing has served as an excellent model system for the dissection of the molecular basis of these signaling transduction pathways. This is in contrast to Drosophila leg imaginal discs, in which mutual repression between Wg and Dpp signaling has long been suspected. However, several studies have indicated that manipulation of Dpp signaling levels in the wing sometimes leads to phenotypes resembling those caused by loss or gain of Wg activity. Notably, ectopic Dpp signaling increases notches in the wing (Marquez et al., 2001; Bennett and Alphey, 2002; Zeng et al., 2008), which is characteristic of reduced Wg signaling (Couso et al., 1994). However, the underlying mechanism for this effect of Dpp is not clear. Recently, independent research groups have suggested that Mad, the key effector of Dpp signaling, might play a role in the regulation of Wg target gene expression in fly wings (Zeng et al., 2008; Eivers et al., 2011). The molecular basis for their findings has mainly been the physical interaction between Mad and TCF, similar to the findings in mammals, in which several Smad proteins interact with members of the lymphoid enhancer binding factor 1/TCF family of DNA-binding HMG box transcription factors (Labbé et al., 2000; Hussein et al., 2003; Labbé et al., 2007; Minoo and Li, 2010). It remains to be determined whether the role of Mad is direct or indirect because the reporter assays in these studies were performed with TOPFlash (Korinek et al., 1997) or similar constructs in mammalian cell culture, which might not always accurately represent the complicated situation of the in vivo regulation of Wg target genes (Chang et al., 2008b). Furthermore, manipulation of Mad expression in wing discs influences Dll expression in different directions (Zeng et al., 2008; Eivers et al., 2011). Although these intriguing discrepancies can be explained by the physical interaction between Mad and TCF, our model offers an alternative interpretation based on the negative regulation of nkd by Brk, which might suggest an indirect role of Mad in Wg signaling. For example, our model could provide an explanation for the previous finding that ectopic Dpp signaling, caused by Mad, Medea, Tkv^QD, etc., results in notched wings (Zeng et al., 2008).

The role of Brk in Wg signaling has been previously documented in Drosophila. It has been suggested that brk is able to antagonize Wg signaling based on the activity of a midgut-specific Ubx reporter gene in which physical interactions among Brk, Teashirt and CtBP have been described (Saller et al., 2002). In leg discs, Wg signaling may directly repress Dpp morphogen expression via an Arm-TCF-Brk complex, offering a direct model for the cross-talk between Wg and Dpp (Theisen et al., 2007). However, our studies have indicated a positive role for Brk in Wg signaling through an indirect action. In addition to the repression of Dpp targets, the roles of Brk in Wg signaling described in these different models exemplify the pleiotropic actions of brk throughout development and might provide the molecular basis for tissue-specific consequences of developmental signaling pathways.

nkd was first identified as a Drosophila segment-polarity gene, mutation of which gives rise to major deficits in fly embryonic development (Jürgens et al., 1984). Its expression appears to be universally induced by Wg in fly embryos and larval imaginal discs (Zeng et al., 2000). It is interesting that although the loss of nkd in embryos has an effect similar to gain of wg, decreased nkd function in fly wings shows little impact (Zeng et al., 2000). However, none of the nkd alleles used in these studies has been well characterized at the molecular level (Zeng et al., 2000). Given the complexity of nkd transcriptional regulation (Chang et al., 2008a), it could be that these mutant forms of nkd still possess residual function in the wing. Alternately, overexpression of nkd blocks ectopic Wg signaling in the eyes and generates PCP phenotypes in the wing through a direct interaction with Dsh (Rousset et al., 2001). Consistent with these observations, we found that loss of brk can cause a dramatic increase of nkd expression in certain areas of the wing imaginal disc, leading to wing notches and PCP defects (Fig. 5D). Our findings suggest that nkd may indeed play roles, at a certain level, in both canonical and noncanonical Wg signaling in fly wings. However, a closer examination of nkd function in fly wings is needed.

In conclusion, we found that Brk influences Wg signaling by directly repressing nkd expression and could serve as a node for cross-talk between the Wg and Dpp signaling pathways. Wnt-BMP cross-interactions have been implicated in many developmental and disease processes (Itasaki and Hoppler, 2010). For example, a Wnt-BMP feedback circuit mechanism is important for inter-tissue signaling dynamics in tooth organogenesis in mouse (O’Connell et al., 2012). Our findings may therefore add new insights into cell differentiation and human cancer.

The authors thank Drs Ken Cadigan, Zhijun Shen and Haiyun Song for their critical reading and editorial help during the preparation of the manuscript; Drs Ken Cadigan, Hugo Bellen, Sean Carroll, Konrad Basler, Stuart Newfeld, Lei Xue, Helen Skaer for reagents; and Bloomington Stock Center, Vienna Drosophila RNAi Center and Developmental Studies Hybridoma Bank for flies and antibodies.