DAnkrd49 and Bdbt act via Casein kinase Iε to regulate planar polarity in Drosophila

Summary

The core planar polarity proteins are essential mediators of tissue morphogenesis, controlling both the polarised production of cellular structures and polarised tissue movements. During development the core proteins promote planar polarisation by becoming asymmetrically localised to opposite cell edges within epithelial tissues, forming intercellular protein complexes that coordinate polarity between adjacent cells. Here we describe a novel protein complex that regulates the asymmetric localisation of the core proteins in the Drosophila pupal wing. DAnkrd49 (an ankyrin repeat protein) and Bride of Doubletime (Bdbt, a non-canonical FK506 binding protein family member) physically interact, and regulate each other’s levels in vivo. Loss of either protein results in a reduction in core protein asymmetry and disruption of the placement of trichomes at the distal edge of pupal wing cells. Post-translational modifications are thought to be important for the regulation of core protein behaviour and their sorting to opposite cell edges. Consistent with this, we find that loss of DAnkrd49 or Bdbt leads to reduced phosphorylation of the core protein Dishevelled and to decreased Dishevelled levels both at cell junctions and in the cytoplasm. Bdbt has previously been shown to regulate activity of the kinase Discs Overgrown (Dco, also known as Doubletime or Casein Kinase Iε), and Dco itself has been implicated in regulating planar polarity by phosphorylating Dsh as well as the core protein Strabismus. We demonstrate that DAnkrd49 and Bdbt act as dominant suppressors of Dco activity. These findings support a model whereby Bdbt and DAnkrd49 act together to modulate the activity of Dco during planar polarity establishment.

Keywords:

Drosophila melanogaster – Phenotypes – Cloning – RNA interference – Phosphorylation – Trichomes – Cell polarity – Planar cell polarity

Introduction

Planar polarity describes the phenomenon whereby cells coordinate their polarity in the plane of a tissue: for example the hairs on the skin point in the same direction, cilia coordinate their beating, and cells coordinate their movements during tissue morphogenesis [1–3]. Understanding the mechanisms by which this coordinated polarisation occurs is of prime importance, as disruption of polarity can have diverse consequences, including neural tube closure defects, hydrocephalus and defects in neuronal migration [3, 4].

The fly wing is a well-characterised model system in which to study planar polarity. Each cell within the adult wing produces a single hair, or trichome, which points towards the distal end of the wing. Furthermore, viable mutations that cause characteristic swirling of the trichomes have been identified, and the genes associated with these mutations were subsequently found to be highly conserved, and to regulate planar polarity throughout the animal kingdom [5].

The core planar polarity proteins (hereafter known as the core proteins) are the best characterised group of proteins that regulate planar polarity. In the pupal wing, the core proteins adopt asymmetric subcellular localisations prior to trichome emergence, and in their absence trichomes emerge from the centre of the cell rather than at the distal cell edge [6]. The core proteins comprise the atypical cadherin Flamingo (Fmi, also known as Starry Night [Stan]), the transmembrane proteins Frizzled (Fz) and Strabismus (Stbm, also known as Van Gogh [Vang]), and three cytoplasmic proteins Dishevelled (Dsh), Prickle (Pk) and Diego (Dgo). Fmi localises to proximal and distal cell edges in the pupal wing, but is excluded from lateral cell edges, while Fz, Dsh and Dgo localise to distal cell edges and Stbm and Pk to proximal cell edges (Fig 1A). These proteins form intercellular complexes at cell junctions, that couple neighbouring cells and allow them to coordinate their polarity [1, 2].

The mechanisms by which the core proteins become asymmetrically localised are poorly understood. The overall direction of polarity is thought to be determined by tissue-specific ‘global’ cues: these may include gradients of morphogens or Fat/Dachsous cadherin activity, or other cues such as mechanical tension [7]. In the wing these global cues may directly regulate core protein localisation or act indirectly via effects on growth and tissue morphogenesis [7]. Global cues are thought to lead to subtle biases in core protein localisation within cells that are subsequently amplified by feedback between the core proteins, in which positive (stabilising) interactions between complexes of the same orientation are coupled with negative (destabilising) interactions between complexes of opposite orientation (Fig 1A). In mathematical models, such feedback interactions have been demonstrated to be sufficient to amplify weak biases in protein localisation, leading to sorting of complexes and robust asymmetry (e.g. [8–11]).

Experimental evidence for feedback is only beginning to be elucidated. Cell culture experiments have demonstrated competitive binding between several of the core proteins, which may be important for feedback [8, 12–14]. Furthermore, we have recently shown that the core protein Pk acts through Dsh to destabilise Fz in the same cell, while stabilising Fz across cell junctions via Stbm [15].

In order for feedback to operate, cells must utilise the general cellular machinery: for example active endocytosis is necessary for Pk to destabilise Fz [15]. Furthermore, post-translational modifications of the core proteins are likely to be key mediators of feedback. For example loss of ubiquitination pathway components and some protein kinases have been shown to disrupt planar polarity. In flies, a Cullin-3/Diablo/Kelch ubiquitin ligase complex regulates Dsh levels at cell junctions, while the de-ubiquitinase Fat Facets regulates Fmi levels [16]. Stbm also negatively regulates Pk levels, and ubiquitination of Pk by Cullin-1/SkpA/Slimb promotes internalisation of Fmi-Stbm-Pk complexes [17–19]. Similarly, in vertebrates, the Stbm homologue Vangl2 may promote local degradation of Pk via ubiquitination by Smurf E3 ubiquitin ligases [20, 21]. Furthermore, Drosophila Fz phosphorylation is mediated by atypical Protein Kinase C [22], and Dsh is a target of phosphorylation by the Discs Overgrown (Dco, also known as Doubletime [Dbt] or Casein Kinase Iε [CKIε]) and Abelson kinases [23–25]. Dco/CKIε has also been implicated in phosphorylation of Stbm in both flies and vertebrates [26–29].

Here we describe the identification of two new regulators of planar polarity in the Drosophila wing. We show that Bride of Doubletime (Bdbt) and DAnkrd49 are binding partners that regulate each other’s levels. Loss of either protein disrupts asymmetric localisation of the core proteins. Furthermore, they regulate overall levels of Dsh and Dsh phosphorylation in the pupal wing, and we provide evidence that they act by modulating the activity of the kinase Dco.

Results

Identification of two novel regulators of planar polarity

To identify novel regulators of planar polarity in Drosophila, we performed an unbiased RNAi screen, using the MS1096-GAL4 driver to express RNAi lines throughout the developing wing blade and examining the adult wings for evidence of planar polarity defects in trichome (wing hair) orientation (see also [18, 30]). We identified two RNAi lines which gave a similar phenotype, in which trichomes were misoriented in the proximal region of the wing (Fig 1B–1D). These RNAi lines targeted two different loci: CG4140 and CG17282. Additional RNAi lines corresponding to these loci were obtained, and when screened using MS1096-GAL4, or ptc-GAL4 in the presence of UAS-Dcr2 (S1 Table), gave qualitatively similar trichome planar polarity defects.

To confirm that these phenotypes were due to mislocalisation of the core polarity proteins, we examined pupal wings in which RNAi constructs were expressed along the anterior-posterior compartment boundary using ptc-GAL4. As in adult wings (S1A–S1C Fig), trichome polarity was disrupted, with trichomes emerging from incorrect cell edges and swirling towards the anterior-posterior boundary (Fig 1E–1G). Trichome initiation was also delayed within the ptc-GAL4 domain, and asymmetric localisation of the core protein Fmi was disrupted, such that it showed a more uniform distribution around the apical junctions (Fig 1E–1G and S1D–S1F Fig). This loss of asymmetry was not caused by a general defect in membrane organisation or apical-basal polarity, as E-cadherin localisation was not affected. However an increase in cell size was evident.

CG4140 (also known as l(2)35Be) encodes a 215 amino acid protein with ankyrin repeats (Fig 2A). It is a homologue of the poorly characterised human gene Ankyrin Repeat Domain-Containing Protein 49 (ANKRD49, S2A Fig), and we will hereafter refer to the fly gene as DAnkrd49. CG17282 is also known as Bride of Doubletime (Bdbt), and encodes a member of the FKBP (FK506 binding protein) family. Bbdt has 3 tetratricopeptide repeats (Fig 2A) that may mediate protein-protein interactions. Like other FKBP family members, its N-terminus has structural similarity to peptidyl prolyl isomerases (PPIases), but it lacks critical residues for catalytic activity [31]. Interestingly, its N-terminus binds to the kinase Dco: loss of Bdbt in adult flies promotes hyperphosphorylation of Dco, and reduces its ability to phosphorylate target proteins (Fig 2B) [31, 32].

Physical interaction between Bdbt and DAnkrd49

Bdbt and DAnkrd49 were identified as binding partners in an early release of the Drosophila Protein Interaction Mapping project [33], but were not considered to be high confidence interactors in the published analysis (DPIM-2) [34]. In order to confirm this interaction, EGFP-tagged Bdbt and Myc-tagged DAnkrd49 were co-transfected into Drosophila S2 cells. Immunoprecipitation with GFP antibody led to pulldown of Myc-DAnkrd49 (Fig 2C). This suggests that Bdbt and DAnkrD49 interact and act as a complex to regulate core protein asymmetry.

DAnkrd49 and Bdbt mutants cause loss of core protein asymmetry

To confirm that the RNAi phenotypes were due to knockdown of the expected target genes, we examined loss of function mutations of DAnkrd49 and Bdbt. EMS-induced alleles of DAnkrd49 have been identified, which are homozygous lethal, but otherwise uncharacterised [35]. Sequencing of one of these alleles, DAnkrd49^l(2)35Be4, revealed H87Y and A156T mutations (Fig 2A and S2A Fig). H87 is highly conserved in all species, and mutation to Y is predicted to be highly deleterious, while A156 is less well conserved, and mutation to T predicted to be tolerated (SIFT analysis) [36]. Mutant clones of DAnkrd49^l(2)35Be4 were generated in the pupal wing, which revealed reduced asymmetric localisation of Fmi (Fig 3A and 3C), accompanied by delayed trichome initiation, and formation of trichomes in the cell centre rather than at the distal cell edge (Fig 3E). Levels of the junctional protein Armadillo (Arm) were not affected in these clones (S3A Fig), suggesting there is no general defect in junctional integrity. When small patches of wild-type cells were surrounded by DAnkrd49^l(2)35Be4 mutant cells, polarity of the wild-type cells was often slightly disrupted. However, we did not see consistent non-autonomous effects on proximal or distal clone boundaries.

As DAnkrd49^l(2)35Be4 is unlikely to be a null mutation, and the chromosome also carries additional recessive markers, we made an independent allele by deleting the entire gene by homologous recombination (S2B Fig, see also Materials and Methods). Multiple independent alleles were generated, all of which were homozygous lethal, and lethal when transheterozygous with DAnkrd49^l(2)35Be4. We then made mutant clones of DAnkrd49^ΔD3.1; however few clones were recovered, and most clones that were seen consisted of only a few cells, suggesting a proliferation defect. However rare clones that were slightly larger gave a qualitatively similar polarity phenotype to DAnkrd49^l(2)35Be4 mutant clones, with reduced core protein asymmetry and delayed trichome formation (S3D and S3F Fig). Due to the poor proliferation of DAnkrd49^ΔD3.1 mutant clones, we used the DAnkrd49^l(2)35Be4 hypomorph for subsequent experiments.

No mutant alleles were available for Bdbt, so we again used homologous recombination to knock out the entire coding region (S2B Fig, see also Materials and Methods). A single mutant allele was generated, which was homozygous lethal and failed to complement deficiencies for the region. In pupal wing clones, Bdbt mutations gave a similar phenotype to DAnkrd49 mutations, notably reduced Fmi asymmetry and delayed trichome initiation (Fig 3B, 3D and 3F). In some clones we saw a disruption in tissue structure, as seen by reduced Armadillo (Arm) staining at junctions (S3C Fig). However core protein asymmetry was reduced even in clones with normal Arm staining (S3B Fig). No obvious effect on cell proliferation was seen, unlike in DAnkrd49 clones. This suggests that DAnkrd49 may have additional functions, independent of Bdbt, or that differential perdurance of the two gene products within mutant clones may cause different effects on cell growth and viability.

EGFP-tagged DAnkrd49 and Bdbt localise uniformly in the cytoplasm and DAnkrd49 and Bdbt regulate each other’s levels

Planar polarity proteins belonging to the ‘core’ localise asymmetrically within the pupal wing, and loss of any of these proteins disrupts the localisation of the others. In contrast, other regulatory proteins such as kinases regulate the asymmetry of the core proteins but have not been observed to themselves localise asymmetrically (e.g. [16, 24]). To determine to which class DAnkrd49 and Bdbt belong, we expressed EGFP-tagged DAnkrd49 and Bdbt in the pupal wing, under control of the ubiquitous Act5C promoter. In both cases we saw uniform expression in the cytoplasm, with some localisation to the plasma membrane; but no asymmetric localisation (Fig 4A and 4B). Notably, the transgenes rescued their respective mutant phenotypes (Fig 4A–4D, compare with Fig 3A–3D), confirming both that the transgenes are functional and that the mutant phenotype is associated with mutations in the expected gene locus. Cellular levels of DAnkrd49-EGFP were increased in the absence of endogenous DAnkrd49.

As our co-immunoprecipitation experiments suggest that DAnkrd49 and Bdbt act as a complex, we examined whether they regulated each other’s localisation or levels. Interestingly, loss of Bdbt causes a reduction in overall levels of DAnkrd49-EGFP in the pupal wing, and vice versa (Fig 4E and 4F). This is consistent with a model in which DAnkrd49 and Bdbt act in a complex in which they promote each other’s stability.

DAnkrd49 and Bdbt regulate overall levels of Dsh

To determine the basis for the loss of core protein asymmetry, we further investigated the effects of loss of DAnkrd49 and Bdbt on core protein localisation. For mutations in both genes, we found that levels of Dsh at junctions were severely reduced (Fig 5A and 5B and S3E Fig), while we did not observe changes in levels of any other core proteins we examined (S3A, S3B Fig and S4A–S4D Fig). In particular, levels of Fz, which recruits Dsh to junctions, were similar to wild-type (S4A and S4B Fig). As our Dsh antibody gives variable amounts of background staining, we confirmed this result using a P[acman]-EGFP-dsh rescue construct. This revealed a reduction in Dsh levels, not only at the junctions but also in the cytoplasm (Fig 5C and 5D). Levels of Dsh-ECFP expressed under a heterologous arm promoter were also reduced, consistent with the effects being post-transcriptional (S4E and S4F Fig). Furthermore Dsh levels were rescued in DAnkrd49 and Bdbt mutant clones when DAnkrd49-EGFP and Bdbt-EGFP were expressed, respectively (S4G and S4H Fig).

To confirm that overall cellular levels of Dsh are reduced in DAnkrd49 and Bdbt mutants, we carried out western blots on pupal wing extracts. As loss-of-function mutants were lethal, we expressed RNAi in the pupal wing using the MS1096-GAL4 driver. Expressing RNAi against Bdbt caused variable lethality (see S1 Table) and insufficient pupal wings were obtained for western analysis. However, expression of RNAi against DAnkrd49 did not affect pupal wing development, and a clear decrease in cellular levels of Dsh on western blots was observed (Fig 5E and 5F).

Bdbt and DAnkrd49 regulate levels of Dsh phosphorylation

Bdbt is a non-canonical member of the FKBP family, which lacks the critical residues for PPIase activity [31]. However, FKBP family members can also function as molecular chaperones, assisting in protein folding independent of PPIase activity [37, 38]. One possibility therefore is that Bdbt and DAnkrd49 act as molecular chaperones, directly promoting Dsh stability. To investigate this, we expressed Myc-tagged Bdbt or DAnkrd49 in S2 cells, and attempted to co-immunoprecipitate Dsh-ECFP. No co-immunoprecipitation was seen (S5 Fig): furthermore no interaction was seen in the genome-wide interaction screen that found the Bdbt-DAnkrd49 interaction [33, 34]. Thus we cannot find evidence to support a chaperone model.

An alternative model is suggested by the fact that Bdbt physically interacts with Dco [31, 32], and that Dco is known to regulate Dsh phosphorylation in both canonical Wnt signalling and in planar polarity [23, 24, 39–41]. Dco can autophosphorylate its own C-terminus, and this autophosphorylation reduces its activity [42]. Deletion of the C-terminus of Dco increases Dco activity in in vitro assays, consistent with it being an inhibitory domain [42]. Binding of Bdbt to Dco is dependent on the C-terminus of Dco, and loss of Bdbt causes hyperphosphorylation of Dco and reduced Dco activity in vivo (Fig 2B) [31]. This leads to an attractive model in which Bdbt and DAnkrd49 regulate Dsh indirectly, through regulation of Dco activity. Consistent with this, we see a reduction of relative levels of Dsh phosphorylation in pupal wings expressing RNAi against DAnkrd49 (Fig 5G), phenocopying the reduced Dsh phosphorylation that was previously shown in dco hypomorphs [24].

Previously we reported a loss of asymmetry in dco hypomorphic mutant clones [24], but we did not report any effect on overall Dsh levels. However, upon re-examination of dco mutant clones, we saw a reduction in levels of Dsh at junctions in mutant tissue, while no reduction in other core proteins such as Fmi was seen (Fig 6A). Furthermore, when P[acman]-EGFP-dsh was expressed in pupal wings, a reduction in both junctional and cytoplasmic levels of Dsh was seen in dco mutant tissue (Fig 6B), again phenocopying DAnkrd49 and Bdbt. Expression of dominant negative Dco has also been shown to reduce levels of P[acman]-EGFP-dsh [29].

Bdbt and DAnkrd49 interact genetically with Dco

Overexpression of dco causes defects in trichome orientation in the adult wing [23, 24, 29, 41]. As DAnkrd49 and Bdbt are expected to promote Dco activity, to seek support for our model we investigated whether a reduction in their activity would suppress the dco overexpression phenotype. dco was overexpressed in the posterior compartment of the wing using the en-GAL4 driver, which reproducibly caused a large trichome swirl close to the posterior cross vein, and a smaller swirl in the distal region of the wing (Fig 7A and 7D). Interestingly, the size of the proximal swirl was reduced, and the distal swirl was almost entirely abolished in flies heterozygous for either DAnkrd49 or Bdbt (Fig 7B, 7C and 7E–7G).

Overall the genetic interactions between dco and DAnkrd49 and Bdbt support them acting in a common pathway. Coupled with the phenotypic similarities, we propose that DAnkrd49 and Bdbt regulate Dsh levels via modulating Dco activity.

Discussion

Asymmetric localisation of the core proteins depends on the tight regulation of protein dynamics and stability. Current models suggest that in order for core protein complexes to become aligned in the same direction, complexes must destabilise each other when they are in opposite orientations. In contrast when complexes are in the same orientation, they become more stable. Understanding how core protein stability is regulated is therefore of key importance.

Here we present evidence for a novel protein complex regulating planar polarity, consisting of the ankyrin repeat protein DAnkrd49 and the non-canonical FKBP family member Bdbt. DAnkrd49 and Bdbt physically interact in vitro and regulate each other's levels in vivo. This suggests that they may act in a complex in which each is required to stabilise the other, a model supported by the phenotypic similarity in mutant clones. We show that their activity is required for core protein asymmetry in the pupal wing, and for normal levels and phosphorylation of the cytoplasmic core protein Dsh.

Previous work on circadian rhythms has established a physical interaction between Bdbt and the kinase Dco, and that Dco activity is regulated by Bdbt [31]. Notably, Dco has previously been implicated in promoting both Dsh and Stbm phosphorylation in planar polarity signalling [23, 24, 26–29]. Thus, our data are consistent with both DAnkrd49 and Bdbt acting to regulate Dco activity in planar polarity. Firstly, loss of function clones of dco and DAnkrd49/Bdbt share a common phenotype: in their absence we see reduced overall levels of Dsh and reduced levels of Dsh phosphorylation (this work and [24]). Secondly, we see strong genetic interactions between dco and DAnkrd49/Bdbt. Although a physical interaction has been observed between Dco and Bdbt [31], we do not see a direct interaction between Dco and DAnkrd49. Therefore, a simple model is that the role of DAnkrd49 is to stabilise Bdbt, while Bdbt directly regulates Dco activity. Thus we define a regulatory cascade, whereby DAnkrd49 and Bdbt promote Dco activity, which in turn regulates phosphorylation of core proteins and asymmetric localisation.

An alternative model is that DAnkrd49 and Bdbt act to stabilise Dsh, independently of Dco. Proteins of the FKBP family are known to regulate the stability of target proteins [37, 38]. In canonical FKBP family members this stabilisation is a result of PPIase activity, which assists protein folding; whereas other family members stabilise their target proteins by direct binding [37]. The catalytic sites for PPIase are not conserved in Bdbt [31]. We were also unable to see a binding interaction between Dsh and either DAnkrd49 or Bdbt, arguing that Bdbt is unlikely to directly stabilise Dsh.

In addition to defects in planar polarity, we also see other pleiotropic defects in DAnkrd49 clones and Bdbt clones, including cell size defects, reduced proliferation and poor viability. These could plausibly be explained by regulation of Dco activity by DAnkrd49 and Bdbt. Dco acts in multiple signalling pathways. It phosphorylates the tumour suppressor Fat, which regulates cell growth and survival via the Hippo signalling pathway [43–47]. Interestingly, a hypomorphic mutation in dco causes tissue overgrowth and increased activity of the caspase inhibitor DIAP1, phenocopying loss of Hippo signalling pathway components [43, 44], while cells completely lacking Dco activity have reduced expression of the caspase inhibitor DIAP1 and reduced proliferation [45]. Dco may also act in additional signalling pathways, and these are thought to include Hedgehog signalling and canonical Wnt signalling [48]. Hence, the multitude of signalling pathways regulated by Dco may explain the complex phenotypes seen in the absence of DAnkrd49 and Bdbt. Alternatively, it is possible that DAnkrd49 and Bdbt may regulate other downstream targets in addition to Dco.

As Dco has also been implicated in phosphorylating Stbm [26–28], loss of DAnkrd49 and Bdbt is also expected to reduce Stbm phosphorylation in pupal wings. However we have not been able to verify this directly, as Stbm mobility in SDS-PAGE is only marginally increased when Dco activity is reduced [27, 29], and is not noticeably altered in extracts from animals with reduced activity of DAnkrd49 or Bdbt. Nevertheless, our recent work has shown that phosphorylation of both Stbm and Dsh by Dco is functionally important in establishing correct planar polarity [29], making the regulation of Dco activity of great interest.

Interestingly, protein interactions studies in human cell lines have identified a STRING network between Ankrd49, CKIδ/CKIε and FKBP family members [49–51]. Although the FKBP proteins identified in these studies are not the closest orthologues to Drosophila Bdbt, this nevertheless suggests conservation of a regulatory cascade of Ankrd49/FKBP promoting Dco activity, which in turn might phosphorylate core planar polarity proteins. Furthermore, the DISEASES online resource, which integrates results from text mining, manually curated disease-gene associations and genome-wide association studies, has linked Ankrd49 with brachydactyly subtypes (diseases.jensenlab.org) [52]. Brachydactyly is a key feature of Robinow syndrome, a disease closely associated with core planar polarity mutations [53]; thus understanding this regulatory cascade may be of importance for human health.

Materials and methods

Fly stocks and genetics

RNAi lines are from the Vienna Drosophila Resource Centre or the National Institute of Genetics in Japan. Transgenic fly lines were UAS-dco [54] and P[acman]-EGFP-dsh [55]. dsh-ECFP was cloned into an pArm-polyA [56], and DAnkrd49-EGFP was cloned in pActP-FRT-polyA-FRT [56], followed by P-element mediated insertion into flies. Bdbt-EGFP was cloned in pAttB-ActP-FRT-polyA-FRT [18] and then inserted into the attP2 landing site.

Mutant alleles are described in FlyBase. dco^j3B9 is a hypomorphic loss of function P-element insertion in the 5'UTR [57, 58]. l(2)35Be¹ and l(2)35Be⁴ were isolated in an analysis of the Adh genomic region region [59] and subsequently found to carry mutations in CG4140/DAnkrd49 [35]. Both alleles were obtained from the Bloomington Drosophila Stock Centre, but l(2)35Be¹ was not used as it was found to complement Df(2L)ED3, Df(2L)ED800 and Df(2L)BSC254 that uncover the l(2)35Be locus. l(2)35Be⁴ however failed to complement all three deficiencies as expected. The l(2)35Be⁴ chromosome carries additional mutations (full genotype is Adh^UF l(2)35Be⁴ Pol32^rd-s pr¹ cn¹). Sequencing of l(2)35Be⁴ revealed H87Y and A156T mutations within the DAnkrd49 coding region. Homologues were identified using HomoloGene (https://www.ncbi.nlm.nih.gov/homologene), and protein alignments were carried out using Clustal Omega (http://www.ebi.ac.uk/Tools/msa/clustalo/). SIFT analysis on putative homologues was carried out to determine residues likely to be deleterious (http://sift.jcvi.org).

Null alleles of CG4140/DAnkrd49 and CG17282/Bdbt were made by deletion of the entire open reading frame by homologous recombination using the pRK2 vector (S2B Fig) [60]. 4 independent mutations of DAnkrd49 were obtained. All were homozygous lethal and lethal over DAnkrd49^l(2)35Be4, Df(2L)ED3 and Df(2L)ED800. A single mutation in Bdbt was obtained: Bdbt^Δ79.1 was homozygous lethal, and failed to complement Df(3R)BSC508 and Df(3R)ED10845 that uncover the Bdbt locus. The phenotypes of Ankrd49 and Bdbt null mutations were rescued by transgenes expressing the appropriate gene.

Mitotic clones were induced using the FLP/FRT system and Ubx-FLP. Overexpression of UAS lines or RNAi lines used the GAL4/UAS system with MS1096-GAL4 [61], ptc-GAL4 [62] or en-GAL4 [63].

Immunostaining and imaging

Adult wings were dehydrated in isopropanol and mounted in GMM (50% methyl salicylate, 50% Canada Balsam).

Unless otherwise indicated, pupal wings were dissected at 28 hr after puparium formation (APF) at 25°C, as previously described [56]. Primary antibodies for immunostaining were mouse monoclonal anti-Fmi (Flamingo #74, DSHB) [64], rat anti-Dsh [24], affinity purified rabbit anti-Fz [65], rabbit anti-Stbm [15], affinity purified rat anti-Pk [18], affinity purified rabbit anti-GFP (ab6556, Abcam), mouse monoclonal anti-β gal (DSHB), rabbit anti-β gal (#55976, Cappel), rat monoclonal anti-Ecad (DSHB) and mouse monoclonal anti-Arm N2 7A1 (DSHB). Actin was labelled with Alexa568-conjugated Phalloidin (A12380, Molecular Probes).

Pupal wings were imaged on a Nikon A1R GaAsP confocal microscope using a 60x NA1.4 apochromatic lens. 9 Z slices separated by 150 nm were imaged, and then the 3 brightest slices around cell junctions were selected and averaged for each channel in ImageJ. Membrane masks were generated in Packing Analyzer [66]. Polarity magnitude (maximum asymmetry ratio on a cell-by-cell basis) and the variation in polarity angle were calculated as previously described [55]. Values were compared using one-way ANOVA with Dunnett's multiple comparisons test, or using paired t-tests for clones, comparing control and mutant regions of the same wings.

Biochemistry

For pupal wing westerns, 28 hr APF pupal wings were dissected directly into sample buffer. One pupal wing equivalent was used per lane. Westerns were probed with affinity purified rabbit anti-Dsh [24] and Actin AC-40 mouse monoclonal (A4700, Sigma). Detection was using SuperSignal West Dura Extended Duration Substrate (Thermo Scientific). A BioRad ChemiDoc XRS+ was used for imaging, and band intensities from six biological replicates were quantified using ImageJ. Data were compared used unpaired t-tests.

For immunoprecipitations, cDNAs were tagged with EGFP or Myc, and cloned into the pAc5.1-V5/His vector (Invitrogen). Immunoprecipitations between Bdbt-EGFP and Myc-DAnkrd49 used GFP rabbit serum (ab290, Abcam), in 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1% TritonX-100, 1x protease inhibitor cocktail (#11697498001, Roche). Immunoprecipitations between Myc-DAnkrd49 or Myc-Bdt and Dsh-ECFP used Myc antibody resin (ab1253, Abcam), in RIPA buffer (50 mM Tris-HCl pH 7.5, 100 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 1x protease inhibitor cocktail). Westerns were probed with mouse monoclonal anti-GFP JL8 (#632381, Clontech) or mouse monoclonal anti-Myc 9E10 (DSHB).

Supporting information

S1 Fig [a]
RNAi knockdown of and in the expression domain.

S2 Fig [a]
DAnkrd49 conservation and generation of knockout alleles.

S3 Fig [2]
Protein localisation in and clones, and knockout mutations.

S4 Fig [2]
Fz, Pk and Dsh localisation in and mutant clones.

S5 Fig [a]
Immunoprecipitation experiments with DAnkrd49/Bdbt and Dsh.

S1 Table [docx]
DAnkrd49 and Bdbt adult wing phenotypes.

S2 Table [xlsx]
Numerical data for Figs , , and .