An Unconventional Nuclear Localization Motif Is Crucial for Function of the Drosophila Wnt/Wingless Antagonist Naked Cuticle
Sharon Waldrop, Chih-Chiang Chan, Tolga Cagatay, Shu Zhang, Raphaël Rousset, Judy Mack, Wenlin Zeng, Matt Fish, Mei Zhang, Manami Amanai, Keith A. Wharton, Jr.


Wnt/β-catenin signals orchestrate cell fate and behavior throughout the animal kingdom. Aberrant Wnt signaling impacts nearly the entire spectrum of human disease, including birth defects, cancer, and osteoporosis. If Wnt signaling is to be effectively manipulated for therapeutic advantage, we first must understand how Wnt signals are normally controlled. Naked cuticle (Nkd) is a novel and evolutionarily conserved inducible antagonist of Wnt/β-catenin signaling that is crucial for segmentation in the model genetic organism, the fruit fly Drosophila melanogaster. Nkd can bind and inhibit the Wnt signal transducer Dishevelled (Dsh), but the mechanism by which Nkd limits Wnt signaling in the fly embryo is not understood. Here we show that nkd mutants exhibit elevated levels of the β-catenin homolog Armadillo but no alteration in Dsh abundance or distribution. In the fly embryo, Nkd and Dsh are predominantly cytoplasmic, although a recent report suggests that vertebrate Dsh requires nuclear localization for activity in gain-of-function assays. While Dsh-binding regions of Nkd contribute to its activity, we identify a conserved 30-amino-acid motif, separable from Dsh-binding regions, that is essential for Nkd function and nuclear localization. Replacement of the 30-aa motif with a conventional nuclear localization sequence rescued a small fraction of nkd mutant animals to adulthood. Our studies suggest that Nkd targets Dsh-dependent signal transduction steps in both cytoplasmic and nuclear compartments of cells receiving the Wnt signal.

PATTERN formation in multicellular animals is governed by the intensity, duration, and combination of signals received by each developing cell. Wnts are a family of highly potent and potentially oncogenic protein signals that specify cell fate and behavior throughout the animal kingdom, and, in the vertebrate, renew stem cells (Logan and Nusse 2004; Reya and Clevers 2005). Abnormal Wnt signaling perturbs development and can cause human diseases (Moon et al. 2004). Feedback regulation—the signal-dependent induction of genes that control signal flux—is a prominent mechanism by which responses to Wnt and other signals are kept within a physiological range, thereby ensuring accurate patterning in the face of environmental perturbation or altered gene dosage (Freeman 2000).

Many Wnts manifest activity via accumulation of β-catenin, a bifunctional, intracellular adaptor protein that regulates cell adhesion at the plasma membrane and transmits Wnt signals into the nucleus (Bienz 2005; Harris and Peifer 2005). Indeed, in a variety of contexts and animals, loss or gain of β-catenin activity mimics absent or maximal Wnt signaling, respectively (e.g., Pai et al. 1997; Gat et al. 1998; Huelsken et al. 2001; Zechner et al. 2003). In the canonical “Wnt/β-catenin” pathway (recently reviewed by Cadigan and Liu 2006; Willert and Jones 2006), Wnt engages Fz/LRP receptors to activate Dishevelled (Dsh), which inactivates a β-catenin “destruction complex” composed of the tumor suppressors Axin/Apc and kinases CK1/GSK3β, leading to intracellular β-catenin accumulation and activation of Wnt target genes via binding to Lef/TCF and other transcriptional regulatory proteins (see∼rnusse/wntwindow.html). Dsh also relays a parallel, LRP-dependent signal that culminates in Axin/LRP association and Axin degradation (Cliffe et al. 2003; Tolwinski et al. 2003; Davidson et al. 2005; Zeng et al. 2005). Fz and Dsh, but not LRP or downstream proteins that regulate β-catenin turnover, participate in noncanonical pathways, the best understood of which executes planar cell polarity (PCP) (Veeman et al. 2003).

Although often dubbed a “scaffolding protein” by virtue of its ability to bind a multitude of proteins, Dsh has been likened to a network hub or node because it links distinct signaling inputs to pathway-specific effectors. However, Dsh's dynamic localization to several subcellular compartments, its tendency to aggregate, and its apparent lack of catalytic activity have rendered accurate comprehension of its molecular and cell biological mechanisms an unexpectedly daunting prospect (Torres and Nelson 2000; Capelluto et al. 2002; Schwarz-Romond et al. 2005; Smalley et al. 2005; reviewed by Boutros and Mlodzik 1999; Wharton 2003; Wallingford and Habas 2005; Malbon and Wang 2006). Complicating matters further, recent RNA interference screens indicate that nearly 250 genes—>1% of the fly genome—impinge upon Wnt/β-catenin signaling (DasGupta et al. 2005).

Key Wnt signal transducers and their epistatic relationships were discovered through genetic analysis of embryonic development in Drosophila melanogaster (Nüsslein-Volhard and Wieschaus 1980; Ingham 1988; Noordermeer et al. 1994; Siegfried et al. 1994). Following cellularization of the blastoderm embryo [embryonic stages 5–7, ∼2–3 hr after egg laying (AEL)], sequentially acting “gap” and “pair-rule” transcription factors initiate “segment-polarity” gene activity with a spatial periodicity that prefigures the segmented body plan (Nüsslein-Volhard and Wieschaus 1980; Ingham 1988). Segment-polarity genes largely encode transducers for Wnt and Hedgehog (Hh) signals. Pair-rule genes act in a dual capacity by conferring upon alternating cell territories a competence to produce either Hh or the Wnt protein Wingless (Wg) and by initiating wg and hh transcription in abutting single-cell-wide transverse stripes (Figure 1A) (Martinez Arias et al. 1988; Martinez Arias 1993; Cadigan et al. 1994). Wg has two temporally distinct activities in ectodermal patterning. First, during early to middle germband extension (stages 8–11, ∼3.5–6 hr AEL), Wg maintains the transcription of the target genes hh and engrailed (en) at close range in the two to three rows of cells posterior to Wg-producing cells (reviewed by DiNardo et al. 1994; Hatini and DiNardo 2001). Thus the width of the hh/en-expressing stripe—the posterior border of which marks the segment boundary, a guidepost for axonal pathfinding and muscle attachment—is a readout of the Wg-signaling gradient (Figure 1A) (Larsen et al. 2003). Second, ventral cells exposed to Wg after ∼6 hr AEL, upon differentiation, suppress the synthesis of cell protrusions termed denticles and appear “naked,” whereas cells out of range of Wg produce denticle bands that facilitate larval locomotion (Bejsovec and Martinez Arias 1991; Dougan and DiNardo 1992). Genetic evidence suggests that early Wg/Hh and later Notch and EGF signals influence denticle fates, the latter signal by activating the transcription factor Svb in denticle-bearing cells (Bejsovec and Wieschaus 1993; Alexandre et al. 1999; Payre et al. 1999; Hatini and DiNardo 2001; Price et al. 2006). Cuticle pattern thus is a sensitive indicator of Wg signaling: defective signaling results in a “lawn of denticles” phenotype, whereas enhanced signaling due to mutation of negative regulators such as axin, apc2, GSK3β, or the aptly named naked cuticle (nkd) results in secretion of “naked” cuticles (Nüsslein-Volhard and Wieschaus 1980; Siegfried et al. 1992; Hamada et al. 1999; McCartney et al. 1999; Zeng et al. 2000).

nkd is unique among known regulators of Wg signaling for its exclusively zygotic expression and genetic requirement (Jürgens et al. 1984; Zeng et al. 2000). [Hypomorphic mutations in the fly β-catenin homolog armadillo (arm) give rise to zygotic wg-like phenotypes due to the lability of maternal Arm protein (Riggleman et al. 1990)]. In nkd mutants, the expression of wg and hh/en by stage 8 appears normal, but by stages 9–10 apparently wild-type levels of Wg broaden the territory of hh/en transcription to include additional posterior cells that are distant from the source of Wg (Figure 1A) (Martinez Arias et al. 1988; Lee et al. 1992; Tabata et al. 1992; Bejsovec and Wieschaus 1993). Increased Wg protein has been observed in posterior signal-receiving cells of stage 10 nkd mutants, possibly due to defective Wg trafficking (Moline et al. 1999). By stage 11, depending on the severity of the nkd allele, elevated Wg and Hh signaling induces a new, thin stripe of Wg that abuts the widened hh/en stripe (Martinez Arias et al. 1988; Bejsovec and Wieschaus 1993). Ectopic Wg transforms the fate of nearby cells from denticle producing to naked and promotes cell death, resulting in a shortened cuticle (Bejsovec and Wieschaus 1993; Pazdera et al. 1998). Embryos homozygous for strong nkd alleles, such as nkd7H16 or nkd7E89, usually produce two or fewer (but frequently no) complete denticle bands due to robust ectopic Wg production, while weaker (yet still embryonic lethal) nkd alleles, such as nkd6J48 or nkd42J1, may exhibit a normal or nearly normal cuticle pattern due to rare ectopic Wg (Jürgens et al. 1984; Zeng et al. 2000; this study).

We previously showed that Drosophila nkd encodes a “pioneer” EF-hand protein that inhibits Wnt/β-catenin signaling by targeting Dsh (Zeng et al. 2000; Rousset et al. 2001, 2002). Mice and humans each have two Nkd-related proteins that can bind Dsh and inhibit Wnt signaling when overproduced in cultured cells (Katoh 2001; Wharton et al. 2001; Yan et al. 2001). Nkd family proteins are distinguished from an otherwise large number of EF-hand-containing proteins by an ∼60-amino-acid region of sequence homology within and immediately adjacent to the EF-hand motif that we have termed the EFX domain (Wharton et al. 2001). Surprisingly, EFX is the only region of extended sequence similarity shared between fly and mammalian Nkds, and it is sufficient to mediate cross-species binding to fly or mouse Dsh proteins (Wharton et al. 2001). Here we employ genetic, cell biological, and structure–function analyses to show that Drosophila Nkd uses separable, conserved motifs that target Dsh and allow nuclear entry, thereby limiting Arm/β-catenin accumulation and Wg target gene expression.


DNA constructs:

D. melanogaster Nkd (GenBank AF213376) constructs were built in Bluescript-II-KS+ (Stratagene, La Jolla, CA) with C-terminal enhanced GFP (CLONTECH, Palo Alto, CA) or myc tags. DNA fragments were synthesized by Pfu PCR, subcloned/sequenced, and cloned into pUAS-T (Brand and Perrimon 1993). NkdrEF/mycC replaced aa 179–225 with aa 87–133 of recoverin (M95858); NkdhEFX2/mycC replaced aa 179–248 with aa 113–178 of human Nkd2 (NM_033120). Residues deleted were the following: NkdΔRB/mycC 179–292; NkdΔEFX/GFPC 177–253; NkdΔYS/GFPC 248–370; NkdΔR1S/GFPC 179–370; NkdΔBBg/GFPC 295–824; NkdΔRBg/GFPC 178–824; NkdNBg/GFPC 827–928; NkdNGA/GFPC 713–928; NkdNIN4/GFPC 573–928; NkdNIN3/GFPC 448–928; NkdNBam/GFPC 296–928; NkdNR1/mycC 179–928; Nkd7H/mycC/Nkd7H/GFPC 60–928; NkdΔ30aa/GFPC 543–572; NkdR1S/mycC 1–176 and 373–928; NkdEFX/GFPC 1–176 and 249–928. NkdΔ30aaNLS/GFPC replaced aa 543–572 with the SV40 NLS (APKKKRKVGST) (Kalderon et al. 1984; Tolwinski and Wieschaus 2004). mNkd1f30aa/GFPC replaced aa 248–274 of mNkd1 (GenBank NM_027280) with Nkd aa 543–572. NkdD201A/mycC, NkdNIN3/GFPC, NkdNBam/GFPC, NkdR1S/mycC, NkdNR1/mycC, and NkdEFX/GFPC have been previously described (Rousset et al. 2002).

Fly stocks and genetics:

Fly transformation/culture were performed according to standard methods. Balanced UAS-Nkd lines (at least three/construct) were screened for activity by examining adult progeny when crossed to B119-Gal4 and GMR-Gal4,UAS-Dsh (Zeng et al. 2000; Rousset et al. 2001). Two (or more where specified) second chromosome lines with strong overexpression effects (when compared to all lines of each construct) were studied in nkd rescue, with the exception that a single second chromosome UAS-NkdNGA/GFPC line was obtained. For nkd rescue, all crosses were performed as described (Rousset et al. 2002) at 25°. All Dsh coexpression crosses were performed with UAS-Dsh or UAS-DshGFP (Axelrod et al. 1998; Axelrod 2001). wgIL114 (Gonzalez et al. 1991) was used at the nonpermissive temperature of 25° in the experiments in Figure 1D and Figure 12, A and B.


Cuticle preparations were performed as previously described, with 125–600 cuticles for each cross and genotype scored as wild type, mild, moderate, or strong according to described criteria (Zeng et al. 2000; Rousset et al. 2002). For UAS-Nkd constructs with substantial activity, rescued nkd cuticles selected for photography either were identified by nkd head involution or spiracle elongation defects not seen in control crosses or were individually selected by the absence of balancer-linked GFP fluorescence. All crosses to assess cuticle rescue, except where indicated, were performed in a nkd7H16/nkd7E89 background. Since nkd7E89 encodes a potentially truncated protein that retains Dsh-binding regions (Figure 9A), we confirmed that UAS-NkdΔR1S/GFPC, lacking the Dsh-binding regions, rescued nkd7H16 cuticles and En stripe width to an extent comparable to nkd7H16/nkd7E89 (not shown) to rule out the possibility that the rescuing activity of NkdΔR1S/GFPC was due to the presence of an intact Dsh-binding sequence encoded by the nkd7E89 chromosome. For scoring the phenotype of nkd alleles (Figure 9B), “complete” denticle bands were scored if six denticle rows could be unambiguously identified and trapezoidal morphology was evident, whereas “partial” bands had either fewer than six denticle rows (except A1, which has three rows) or discontinuous replacement of denticle rows by naked cuticle due to patchy ectopic Wg production. Although nkd47K1 and nkd7E89 have identical mutations and each has a wide En stripe during stage 11 (Figure 9A), their slight difference in cuticle phenotype (Figure 9B) could be due to the different genetic backgrounds in which each was generated (ru cu ca for nkd7E89 vs. red e for nkd47K1).

Computer programs and web tools:

Figures were constructed in Adobe Photoshop and Canvas (ACD Systems). Graphs were generated in Deltagraph (Red Rock Software). Secondary structure prediction was obtained at Helical wheel output (Figure 10B) was obtained at∼cmg/Demo/wheel/wheelApp.html. ClustalW analysis and linked Boxshade output was obtained at


Embryos were collected on yeast/grape juice agar, dechorionated in 50% bleach for 2 min, fixed in 50:50 PBS + 4% formaldehyde/heptane for 20 min, devitellinized in 50:50 heptane/methanol, rehydrated in PBS + 0.1% Triton X-100 (PBT), blocked in PBT + 5% goat serum (PBTN), and then incubated with antibodies/PBTN with rocking at 4°. Embryo heat fixation was performed as described (Zeng et al. 2000). Primary antibodies/dilutions were the following: rat-α-Nkd (Zeng et al. 2000) 1:100; rabbit-α-Dsh (a kind gift from R. Nusse) 1:200 and α-β-gal (Molecular Probes, Eugene, OR) 1:500; biotinylated-α-GFP (Molecular Probes) 1:200; monoclonal Abs: α-En (4D9, a kind gift from N. Patel) (Patel et al. 1989) 1:200; α-β-gal (Promega, Madison, WI) 1:1000; α-myc (Sigma, St. Louis) 1:500; α-Arm [N2-7A1, Developmental Studies Hybridoma Bank (DSHB)] (Riggleman et al. 1990) 1:500; α-Wg (4D4, DSHB) (Brook and Cohen 1996) 1:50. Secondary antibodies/dilutions were the following: biotinylated-α-mouse and α-rabbit IgG, FITC-α-rat, rhodamine-α-rabbit and rhodamine-α-mouse (Jackson ImmunoResearch, West Grove, PA) 1:200. En was visualized using Vectastain ABC Elite (Vector Labs, Burlingame, CA), 3,3′-diaminobenzidine (DAB), and 30 mm NiCl enhancement. NkdGFP's were visualized by streptavidin-FITC (Jackson ImmunoResearch) at 1:1000. DNA was visualized using DRAQ5 (Alexis Biochemicals) at 1:1000 following RNAse-A digestion to minimize cytoplasmic mRNA staining.


Adult flies were photographed as described (Rousset et al. 2002). DAB-stained embryos were cleared through graded ethanols into methyl salicylate and viewed under Nomarski optics on a Zeiss Axioplan IIie. Rescued nkd embryos were unambiguously identified by lack of head-specific β-gal encoded by the hb-lacZ transgene on the TM3 balancer chromosome. At least 20 rescued stage 11 embryos were examined in each collection, and representative embryos were photographed. Fluorescent-labeled embryos were mounted in Fluoromount G (Southern Biotechnology) and images were acquired on a Nikon C1 confocal microscope with 488-, 543-, and 633-nm lasers using 40× and 100× oil objectives and 8-pass Kalman averaging. FITC/DRAQ5 and rhodamine were sequentially scanned to prevent bleed through of FITC-emitted signals. Three or more embryos of representative stages (typically stages 9–11) were examined under multiple confocal fields for each construct/line tested. For SEM, adult flies were fixed in 100% ethanol, 2% glutaraldehyde, postfixed in 1% OsO4 in 0.1 m cacodylate buffer, and dehydrated through graded ethanols. Specimens were mounted, critical point dried, sputter coated, and viewed on a JEOL 820A electron microscope.

Image analysis:

Confocal images of Arm-stained embryos were oriented along the anterior–posterior axis, analyzed in ImageJ (National Institutes of Health), and presented graphically via Deltagraph.

nkd viability assays:

Hatching frequency (Figure 3O) was determined by selecting, 6–12 hr AEL, GFP− (or dim, for NkdGFP-expressing constructs) embryos from the cross UAS-Nkd;nkd7H16/TM3-GFP × nkd7E89da-Gal4/TM3-GFP and culturing the embryos on grape juice agar at 25° until hatching. Adult viability (Figure 3P) was determined by individually transferring GFP− or dim crawling first instar larvae to yeast paste/grape juice agar at 21° and scoring for viability to the indicated stage until death or eclosion. All rescued adults lacked Sb on the TM3 balancer, confirming the rescued nkd genotype. nkd7H16 animals survived to a comparable degree as nkd7H16/nkd7E89 animals (not shown).

Yeast two-hybrid:

Yeast two-hybrid and yeast extract preparation was performed as described (Rousset et al. 2001) in the yeast reporter strains Y190 (ONPG assay) and AH109 (growth assay) (CLONTECH). Figure 6C and Figure 8C show the mean ±SD β-gal units for four independent experiments performed in triplicate for each plasmid combination.

Leptomycin-B treatment of embryos:

Embryos expressing constructs indicated in Figure 4 were dechorionated and rocked in 50:50 octane/PBS ± 0.158 μg/ml leptomycin-B (LMB; Sigma) for 30 min at 21°. PBS/LMB was aspirated and the embryos were fixed, stained, and imaged as described.

nkd allele sequencing:

nkd alleles were generated in an isogenized red e background (W. Zeng and M. P. Scott, unpublished data). Genomic DNA was prepared from adult flies carrying each nkd allele over a balancer chromosome. Overlapping primers for each nkd coding exon were designed to amplify 400- to 500-bp fragments. PCR fragments were purified on an agarose gel, eluted using the QIAquick gel extraction kit (QIAGEN, Chatsworth, CA), and sequenced from the 5′- and 3′-ends.

Western blots:

Dechorionated 0- to 5-hr embryos were dounce homogenized in 5 vol of cold lysis buffer [50 mm Tris-HCl, pH 7.5, 150 mm NaCl, 1% NP40, 0.5% sodium deoxycholate, 1 mm DTT, 1 mm PMSF, supplemented with complete protease inhibitor (Roche)]. The homogenate was centrifuged at 14,000 × g for 15 min to pellet debris, and 100 μg crude extract was resolved on 10% SDS–PAGE and transferred to PVDF membrane. For yeast Western blots, 10 μg supernatant was loaded per lane. Antibodies/dilutions were the following: rabbit α-GFP1 (Santa Cruz Biotechnology) 1:3000; mouse mAbs α-myc (Sigma) 1:6000; α-DBD-Gal4 (Santa Cruz) 1:5000; α-porin (Molecular Probes) 1:1000; α-tubulin TU27 (Covance) 1:2500 (not shown). Signals were visualized with SuperSignal West Dura extended duration substrate (Pierce Biotechnology) followed by autoradiography.


Nkd antagonizes Arm/β-catenin accumulation:

Whether or not increased levels of Arm/β-catenin cause the expansion of Wg target gene expression in nkd mutants is not known; an early report, prior to the widespread use of confocal imaging, suggested that Arm levels are normal during early germ-band extension in nkd mutants (Riggleman et al. 1990). Confocal imaging revealed elevated Arm in nkd embryos by stage 10, prior to ectopic Wg synthesis (cf. Figure 1, B and C). Images of Arm-stained nkd embryos had peak pixel intensities increased approximately twofold over wild type (Figure 1D). To confirm this observation, Nkd was produced in alternate segments using prd-Gal4 (Yoffe et al. 1995), a Gal4 driver integrated into the pair-rule locus paired (prd), so that Arm distribution in adjacent rescued and nkd segments could be imaged simultaneously. In stage 10 nkd embryos, Nkd reduced Arm to wild-type levels (Figure 1E). In contrast, increased Nkd did not reduce peak Arm levels in nkd/+ or wild-type embryos (not shown).

Figure 1.—

Nkd antagonizes Arm/β-catenin accumulation. (A) Schematic of ectodermal gene expression in wild type (wt, left) and nkd (right) embryos. Stages 6–7: Wg (blue) and Hh/En (red) initiate in adjacent cell rows. Stage 9: In wild type, Wg protein distributes in a gradient (blue triangle) and maintains Hh/En in approximately two cell rows while, in nkd, Hh/En is expressed in all competent cells (red shading). Wg expression-competent cells are in blue shading. Stage 11: In wild type, the Wg gradient is biased to the anterior while, in nkd, Wg is ectopically produced by some anterior Wg-competent cells, causing boundary duplications with reversed polarity (arrows). Parasegmental (PS) and segmental (S) boundaries are designated. “hatch”: Denticle rows 1–6 are produced in a trapezoidal band, two belts of which are depicted. In nkd, ectopic Wg promotes cell death (X) and suppresses denticle synthesis. (B and C) Five segmental anlagen of stage 10 wild type (B) or nkd7H16 (C) embryos stained for Arm (green) above schematic. (D) Mean pixel intensity across five segmental anlagen from Arm-stained stage 10 wingless (blue), wild-type (green), and naked (red) mutants. (E) Stage 10 UAS-NkdGFPC;nkd7H16prd-Gal4/nkd7H16 embryo stained for Arm (green) and GFP (purple). Note reduced Arm in NkdGFPC-expressing segments (asterisks).

Distributions of endogenous Nkd and Dsh:

Nkd and Dsh localizations during segmentation may provide clues about how Nkd limits Wg signaling. Endogenous Nkd was not detected by existing antibodies in formaldehyde-fixed embryos, but could be visualized by heat-fixation-induced antigen retrieval (Zeng et al. 2000). Confocal microscopy of heat-fixed embryos stained with α-Dsh and α-Nkd revealed uniform Dsh throughout each segment, whereas Nkd accumulated in a complex striped pattern similar to nkd mRNA (Figure 2, A and B) (Yanagawa et al. 1995; Zeng et al. 2000). Within cells, each protein was observed in a diffuse and punctate distribution in both cytoplasm and nucleus, with rare sites of punctate cytoplasmic and/or plasma membrane colocalization (Figure 2, B–B″). No segmental variation of nuclear accumulation was observed for either protein. Heat fixation compromised embryo morphology, precluding a more accurate assessment of subcellular distributions.

Figure 2.—

Distributions of Nkd and Dsh in embryos. (A) Heat-fixed wild-type stage 11 embryo stained for endogenous Nkd. Arrowheads mark positions of parasegmental boundaries. Note segmental variation in Nkd abundance (brackets, also in B). (B–B″) Heat fixed wild-type embryo stained for endogenous Nkd (green, B) and Dsh (red, B′); B″ is a merged image. Insets in B–B″: each protein is cytoplasmic and weakly nuclear (large arrowhead) with some cytoplasmic and/or plasma membrane colocalization (yellow arrows). (C) Formalin-fixed stage 11 cas-Dshmyc/UAS-NkdGFPC;nkd7H16da-Gal4/nkd7H16 embryo stained with α-myc and α-GFP (only the latter is shown). Note uniform, predominantly cytoplasmic but low-level nuclear distribution of NkdGFPC. (D–D″) Medium power; same genotype as C. Note uniform distributions of both proteins in cytoplasm and nucleus (arrowhead in insets). Punctate areas with predominantly Nkd (green arrow), Dsh (red arrow), or both (yellow arrow) can be identified in magnified insets. (E) Western blot of extracts from 0- to 5-hr embryos expressing GFP (left) or NkdGFPC (right) and probed with α-GFP antibody highlighting specific GFP (arrow), NkdGFPC (large arrowhead), and nonspecific bands (small arrowheads). (F) Western blot of extracts from 0 to 5 hr. cas-Dshmyc embryos probed with α-myc. Note doublet in left lane that collapses to a single band following phosphatase (CIP) treatment. (G) High power; same genotype as C, with DNA (blue) marking nuclei in G′ and G″. Note punctate cytoplasmic Nkd (green arrow) and Dsh (red arrow) staining with rare cytoplasmic (yellow arrow) and nuclear (arrowhead) colocalization.

Epitope-tagged Nkd rescues nkd mutants to adulthood:

We sought to visualize Nkd and Dsh live or under gentle fixation, but first compared the activity of epitope-tagged Nkd to untagged Nkd in a rescue assay (Rousset et al. 2002). Nkd fused to either a C-terminal myc tag (NkdmycC) or GFP (NkdGFPC), driven by the ubiquitous da-Gal4 driver, narrowed the En stripe to two to three cells in stage 10–11 nkd embryos and rescued the cuticle phenotype as well as untagged Nkd (Figure 3, A–C, N; data not shown) (Zeng et al. 2000; Rousset et al. 2002).

Figure 3.—

Nkd rescues nkd mutants to adulthood. (A–C) Stage 11 embryos of indicated genotypes stained for En and viewed at low power (left) and high power showing two segments (middle). (Right) Representative cuticle. Wt (A) has a two- to three-cell En stripe (arrows and bar), while nkd7H16/nkd7E89 (B) has a wide En stripe and develops a strong nkd phenotype with a severe head involution defect (arrowhead) and widely split posterior spiracles (arrow). (C) UAS-NkdGFPC;nkd7E89da-Gal4/nkd7H16 has a narrow En stripe and a wild-type cuticle pattern. (D–M) Eye (D–H) or ventral abdomen (I–M) of wild type (wt) (D and I) or B119-Gal4/UAS-Nkd female (E–H and J–M). Weaker UAS-Nkd lines NkdmycC#9 and NkdGFPC#3 produce a teardrop-shaped ventral eye (arrowheads, E and F) relative to wild type (D), whereas the stronger lines NkdmycC#3-2 and NkdGFPC#1 cause ventral eye reduction (G and H). Wild-type female abdomen has ∼90 sternite bristles (arrowhead) in the ventral aspect of segments A2–A6 (I). Weaker UAS-Nkd lines (J and K) caused bristle loss and lateral bristle displacement with indicated mean ±SD bristle numbers, while stronger lines (L and M) resulted in near total bristle loss. (N) Distribution of wild-type and nkd cuticle phenotypes (weak, moderate, strong) in rescue cross (UAS-X; nkd7H16/TM3 × da-Gal4,nkd7E89/TM3, where X is indicated construct) for each Nkd construct. UAS-lacZ (light blue) is the negative control, giving rise to an ∼3:1 Mendelian ratio (75:25%) of wild-type-to-strong nkd mutants. w− is viability control, with 100% wild type. Magenta, untagged Nkd; blue, NkdmycC; green, NkdGFPC. Note that each UAS-Nkd line rescues nearly all cuticles to wild type. (O) Percentage of nkd embryos that hatched into crawling first instar larvae as a function of UAS-Nkd rescue construct. w− and UAS-lacZ are controls. Above each bar is the number of embryos cultured. (P) Percentage of rescued nkd crawling first instar larvae that survived to indicated stage. w− is the control for culture conditions. Key shows the number of larvae of each genotype that were cultured. Note that a lower percentage of NkdGFPC than NkdmycC-rescued embryos hatch and survive to later stages. (Q) UAS-NkdmycC;nkd7E89da-Gal4/nkd7H16 hatched adult with wing-to-notum transformation (arrowhead). (R) Wild-type anterior wing margin with evenly spaced stout bristles (blue arrowhead) and slender bristles (red arrowheads), each of the latter interspersed by four wing trichomes. (S and T) Wing margins from rescued nkd adult showing displaced stout bristle and decreased spacing between slender bristles (S) and ectopic sensory bristles near wing margin (T).

Overexpressing Nkd during larval development results in adult phenotypes that mimic loss of Wg signaling; B119-Gal4-driven Nkd expression caused ventral eye reduction and loss and lateral displacement of sternite bristles (Figure 3, D–M) (Zeng et al. 2000; Wharton et al. 2001). From a panel of transgenic UAS-Nkd fly lines, we selected comparably “weak” and “strong” UAS-NkdmycC and UAS-NkdGFPC lines as judged by their expressivity when produced by B119-Gal4 (Figure 3, D–M). Although all four tagged UAS-Nkd lines rescued nkd cuticles as well as untagged Nkd (Figure 3N), NkdmycC lines rescued to later postembryonic stages than NkdGFPC lines (Figure 3, O and P). Although size and/or type of epitope tag, particularly a bulky GFP moiety, may subtly alter Nkd activity later in development, our data show that C-terminal epitope tags do not affect Nkd activity in the embryo.

Depending on the construct, transgenic line, and culture conditions, up to 60% of rescued nkd mutant crawling first instar larvae survived to adulthood (Figure 3P; data not shown). Although no NkdGFPC-expressing nkd mutant larvae survived to adulthood when cultured at 21° (Figure 3P), rare survivors eclosed when cultured at 18° (not shown). Dissection of pharate adults that failed to eclose revealed defects associated with altered Wg signaling, including reduced or missing wings and legs, altered sternite bristle pattern, and segmentation defects (not shown), while adults often lacked wings and/or halteres (Figure 3Q) (Baker 1988; Couso et al. 1993; Zeng et al. 2000). Although most of the defects in rescued mutants are associated with reduced wg activity (possibly due to Gal4-driven Nkd overproduction), we also observed wing-margin bristle patterns and ectopic wing bristles indicative of locally increased wg activity (Figure 3, R–T) (Couso et al. 1994).

Distributions of tagged Nkd and Dsh:

Having established the functional equivalence of tagged and untagged Nkd in a rigorous rescue assay, we proceeded to investigate the distributions of tagged Nkd and Dsh. Unfortunately, when expressed in embryos, we were unable to detect NkdmycC with α-myc by Western blot or immunofluorescence (not shown). Likewise, NkdGFPC was not visible with live fluorescence microscopy but was detectable with α-GFP, the specificity of which was verified by expressing NkdGFPC with striped Gal4 drivers (Figure 1E; data not shown). To monitor Dsh, we used a myc-tagged Dsh transgene driven by the endogenous dsh promoter (cas-Dshmyc) that can rescue dsh mutants to adulthood (J. Axelrod, personal communication). Formalin-fixed nkd embryos ubiquitously expressing NkdGFPC carrying cas-Dshmyc revealed α-myc- and α-GFP-staining patterns similar to endogenous Dsh and Nkd (cf. Figure 2, A and B with C and D). da-Gal4-driven NkdGFPC accumulated during stage 7 in a diffuse and punctate cytoplasmic pattern that persisted through stages 10–11 to include diffuse nuclear staining (Figure 2, C and D; data not shown). Dshmyc and NkdGFPC accumulated uniformly across the segment, consistent with a lack of post-transcriptional control of steady-state Nkd or Dsh levels or distributions by physiologic levels of Wg signaling, an observation that we confirmed by producing NkdGFPC in alternate segments of nkd mutants using prd-Gal4 (not shown). Western blots of embryo extracts revealed predominant bands with slightly retarded mobilities relative to the expected Mr of each fusion protein (NkdGFPC = 129 kDa; DshmycC = 79 kDa) (Figure 2, E and F), indicating that the distributions observed by microscopy likely correspond to that of full-length proteins as opposed to degradation products.

In contrast to the segment-polarity protein Lines, which undergoes rapid nucleo-cytoplasmic shuttling in response to Wg or Hh signaling, the observed embryonic distributions of NkdGFPC and Dshmyc were unaffected by pharmacologic inhibition of CRM1-dependent nuclear export following a 30-min exposure to leptomycin-B (Figure 4) (Fornerod et al. 1997; Ossareh-Nazari et al. 1997; Hatini et al. 2000).

Figure 4.—

NkdGFPC and Dshmyc do not undergo rapid CRM-1-dependent nuclear export. Each panel shows low- and high-power views of representative stage 11 embryos expressing the indicated construct via da-Gal4 (except cas-Dshmyc, which expresses Dshmyc at appropriate levels via the dsh promoter) stained with the appropriate antibody to reveal subcellular distributions ±30 min treatment with the CRM-1-dependent nuclear export inhibitor LMB. The positive control (A and B) is Linesmyc, which is nuclear in Wg-receiving cells and cytoplasmic in Hh-receiving cells of the dorsal ectoderm (Hatini et al. 2000). LMB treatment drives Linesmyc into the nucleus of most dorsal ectoderm and head (arrow) cells. Asterisks designate positions of nuclei. The negative control (C and D) is the myristoylated, N-terminal 89 amino acids of D. melanogaster Src (Simon et al. 1985) fused to GFP. No change in the membrane-associated SrcGFP distribution is seen after LMB treatment. NkdGFPC (E) is cytoplasmic and nuclear, whereas Dshmyc (G) is predominantly cytoplasmic in the absence of LMB, and these distributions remain unchanged in LMB-treated embryos (F and H).

With either fixation or staining method, Nkd and Dsh exhibited largely nonoverlapping subcellular distributions, with only rare Nkd/Dsh punctate cytoplasmic or nuclear colocalization (Figure 2, G–G″), observations consistent with our previous unsuccessful attempts to co-immunoprecipitate endogenous Nkd and Dsh from embryos or S2 cells (Rousset et al. 2001).

In vivo requirements for Nkd/Dsh-interacting regions:

Misexpressed Nkd can inhibit endogenous dsh during PCP signaling as well as Dsh-overexpression-induced gain-of-Wg signaling phenotypes (Axelrod 2001; Rousset et al. 2001). If Nkd/Dsh association is necessary for Nkd to antagonize Wnt signaling, then Nkd should not alter a gain-of-Wnt signaling phenotype induced by a Dsh protein that lacks Nkd-interacting sequences. Dsh consists of DIX, basic/PDZ, and DEP domains, of which Nkd binds to the central basic/PDZ region (Rousset et al. 2002). Misexpression of full-length Dsh in the larval eye imaginal disc by GMR-Gal4 resulted in adult eye phenotypes similar to those caused by a dominant wg allele (wgGlazed) in which ectopic Wg causes expanded head cuticle and pigment cell fates, ommatidial apoptosis, and reduced or absent interommatidial bristles (Figure 5, A–C) (Cadigan and Nusse 1996; Ahmed et al. 1998; Brunner et al. 1999; Rousset et al. 2001; Cadigan et al. 2002; Tomlinson 2003; Lin et al. 2004). Nkd coexpression inhibited the Dsh-induced phenotype but had no effect on a similar phenotype induced by a mutant Dsh protein, DshΔbPDZ, that lacked Nkd-interacting regions (cf. Figure 5, D and E) (Axelrod et al. 1998; Rousset et al. 2001). Nkd also inhibited Dsh-induced, but not DshΔbPDZ-induced, ectopic wing-margin bristles (not shown).

Figure 5.—

Nkd reverses Dsh- but not DshΔbPDZ-induced eye phenotype. (A) Wild-type eye. (B) wgGlazed eye. (C) GMR-Gal4/UAS-Dsh;UAS-lacZ eye. (D) GMR-Gal4/UAS-Dsh;UAS-Nkd eye showing strong suppression of Dsh-induced phenotype. (E) GMR-Gal4/UAS-DshΔbPDZ;UAS-lacZ eye (left) is identical to GMR-Gal4/UAS-DshΔbPDZ;UAS-Nkd eye (right), with no ommatidia or bristles.

Next we made Nkd proteins with mutant Dsh-binding regions. NkdrecEF/mycC and NkdhEFX2/mycC replaced the Nkd EF hand with either the high-affinity Ca2+-binding EF hand of bovine recoverin (Flaherty et al. 1993) or the EFX of human Nkd2 (hNkd2) (Wharton et al. 2001). NkdD201A/mycC substituted an alanine for a conserved aspartate in the EF hand loop, a mutation that reduced the extent of Nkd/Dsh association by yeast two-hybrid (Y2H) (Rousset et al. 2002). All three constructs rescued the nkd cuticle phenotype and adult viability, although NkdrEF/mycC and NkdhEFX2/mycC cuticle rescue was less efficient than wild type (Figure 6B; data not shown).

Figure 6.—

Properties of Nkd proteins with mutant Dsh-binding regions. (A) Dsh (purple) has DIX, PDZ, and DEP domains. Nkd (yellow) has EFX domain. Red bars indicate regions that mediate Nkd/Dsh association (Rousset et al. 2002). Nkd constructs lacking Dsh-binding regions (designated by spliced-out regions) are schematized at left, with representative images from activity assay (stage 11 α-En and larval cuticle) and α-GFP localization in stage 11 embryo, depicted from left to right. The top two constructs, lacking the EFX domain, each narrowed the En stripe to two to three cells and restored cuticle pattern, while the bottom two constructs, lacking the Zn2+-binding domain, partially narrowed the En stripe but also restored most of the cuticle pattern. Note that EFX deletion increased the ratio of nuclear-to-cytoplasmic GFP, with the exception of the two NkdΔR1S/GFPC-expressing cells in the center of the field with condensed chromatin indicative of late telophase. ND, not done. (B) Distribution of wild-type and nkd cuticles in rescue cross for indicated Nkd construct. For crosses in which the UAS-Nkd insertion is lethal, the presence of the CyO balancer in the rescue cross results in ∼12.5% of cuticles remaining unrescued and hence “strong” (indicated by the black-topped bars). Note that constructs lacking Dsh-binding regions have a reduced ability to rescue relative to wild-type Nkd, as evidenced by a reduced proportion of wild-type cuticles and increased weak and moderate cuticles. (C) Y2H of strains expressing Nkd bait constructs (left) and Dsh (top) or DshbPDZ (middle) prey assayed for growth under double-dropout (2D) or quadruple-dropout (4D) media. Bar graphs on right show ONPG units for strains harboring indicated constructs in quantitative Y2H assay. (Bottom) Western blot of yeast extracts expressing indicated Nkd-Gal4 DNA-binding domain (DBD) fusion proteins. Porin is loading control. (D) High power of stage 11 UAS-NkdΔR1S/GFPC/cas-Dshmyc;nkd7H16da-Gal4/nkd7H16 epidermis stained for α-GFP (D, green), α-myc (D′, red), and DNA (D″, blue); D‴ is a merged image. Although the majority of NkdΔR1S/GFPC is nuclear, rare cytoplasmic Nkd/Dsh colocalization (yellow arrows) can be identified.

Surprisingly, NkdΔRB/mycC, which lacks the EFX, rescued nkd mutants to adulthood as well as NkdmycC (Figure 6B; data not shown), but the mutant protein's residual activity could be due to Dsh-association via adjacent Zn2+-binding sequence in fly Nkd (Rousset et al. 2002). We made Nkd–GFP constructs lacking the EFX (NkdΔEFX/GFPC) or the Zn2+-binding region (NkdΔYS/GFPC) or both Dsh-binding regions (NkdΔR1S/GFPC). NkdΔEFX/GFPC rescued nkd cuticles as well as NkdGFPC or NkdΔRB/mycC, but NkdΔYS/GFPC or NkdΔR1S/GFPC had a reduced but a still significant rescue activity (Figure 6, A and B). The Nkd proteins lacking Dsh-binding regions had reduced associations with Dsh or DshbPDZ by quantitative Y2H (Figure 6C). During stages 10–11 NkdΔEFX/GFPC and NkdΔR1S/GFPC, in contrast to NkdGFPC, accumulated predominantly in nuclei while retaining punctate cytoplasmic distribution and rare Dsh colocalization (Figure 6, A and D–D″), suggesting that the EFX domain and/or Dsh association opposes nuclear localization signals in Nkd. Consequently, NkdEFX/GFPC, consisting of only the EFX domain fused to GFP, accumulated in the cytoplasm during germ-band extension but did not inhibit Wnt signaling or rescue the nkd cuticle (Rousset et al. 2002; data not shown).

Dsh-binding sequences in Nkd are important to reverse the consequences of Dsh overproduction, because NkdΔR1S/GFPC restored few bristles or ommatidia to GMR-Gal4/UAS-Dsh eyes (cf. Figure 7, B and C). Conversely, NkdR1S/mycC, composed of only sequences sufficient to bind Dsh, suppressed the Dsh-induced eye phenotype (Figure 7D). Nkd's Zn2+-binding sequences are important for inhibiting Dsh, because NkdEFX/GFPC, lacking the Zn2+-binding sequences, did not alter the Dsh-induced eye phenotype (not shown). In summary, Dsh-binding regions of Nkd are necessary for full nkd rescue activity and are largely sufficient to inhibit Dsh.

Figure 7.—

Nkd's Dsh-binding regions are important for blocking Dsh-induced eye phenotype. (A–D) SEMs of representative wild type (A), GMR-Gal4/UAS-Dsh;UAS-lacZ (B), GMR-Gal4/UAS-Dsh;UAS-NkdΔR1S/GFPC (C), or GMR-Gal4/UAS-Dsh;UAS-NkdR1S/mycC (D) eyes. Note that NkdΔR1S only weakly rescues the Dsh-induced phenotype, while NkdR1S, consisting of only Dsh-binding sequences, restored ommatidia and bristles.

A 30-amino-acid motif is necessary for Nkd activity and nuclear localization:

Despite the ability of Nkd's Dsh-binding sequences to block Dsh in vivo, da-Gal4-expressed NkdR1S/mycC had no nkd rescue activity (Rousset et al. 2002). To identify regions of Nkd important for function, we made additional transgenes and sequenced nkd alleles. NkdΔBBg/GFPC and NkdΔRBg/GFPC, lacking amino acids 295–824 and 178–824, respectively, did not rescue, indicating deletion of critical region(s) (Figure 8A). The C-terminal truncations NkdNBg/GFPC (retaining aa 1–826), NkdNGA/GFPC (1–712), and NkdNIN4/GFPC (1–572) restored cuticle pattern and narrowed the En stripe of nkd mutants (Figure 8, A and B). NkdNIN3/GFPC (1–447), which includes Dsh-association sequences, did not narrow the En stripe and only partially rescued the nkd cuticle (Figure 8A) (Rousset et al. 2002). The shorter constructs NkdNBam/GFPC (1–295), NkdNR1/mycC (1–178), or Nkd7H/GFPC/Nkd7H/mycC (1–59) did not affect the nkd phenotype (Figure 8A) (Rousset et al. 2002). The distributions of NkdNIN4/GFPC and longer constructs during stages 10–11 resembled NkdGFPC, whereas the shorter constructs NkdNIN3/GFPC, NkdNBam/GFPC, NkdΔRBg/GFPC, and NkdΔBBg/GFPC were excluded from nuclei during stages 10–11 (Figure 8A). Nkd7H/GFPC was detectable in nucleus and cytoplasm (Figure 8A), suggesting that the EFX retains Nkd in the cytoplasm, consistent with deletion of the EFX domain (as in constructs NkdΔEFX/GFPC or NkdΔR1S/GFPC) resulting in enhanced nuclear localization relative to NkdGFPC (cf. Figure 2G with Figure 6, A and D). Alternatively, the small size of Nkd7H/GFPC may allow free diffusion through nuclear pores. NkdNIN3/GFPC, NkdNIN4/GFPC, and longer constructs fully reversed the Dsh-induced eye phenotype, while NkdNBam/GFPC, retaining the EFX domain but lacking the Zn2+-binding region, restored ommatidia but few bristles (not shown). Thus Nkd aa 448–572 (the sequence in NkdNIN4/GFPC but not in NkdNIN3/GFPC) are crucial for both Nkd activity and nuclear localization but not for inhibiting overexpressed Dsh.

Figure 8.—

A 30-aa motif is necessary for Nkd activity and nuclear localization. (A) Data from indicated UAS-Nkd constructs as described in Figure 6A. Note that constructs with the 30-aa motif (light blue box) narrowed the En stripe, restored cuticle pattern, and were detected in nuclei, whereas constructs lacking the 30-aa motif did not narrow the En stripe, did not restore or only partially restored cuticle pattern, and (with the exception of Nkd7H/GFPC) were excluded from nuclei. (B) Quantitation of cuticle phenotypes. Note that constructs with a 30-aa motif (blue bars) efficiently rescued, while those lacking the 30-aa motif (white bars) partially rescued or did not rescue. (C) Y2H of Nkd or NkdΔ30aa binding to Dsh or DshbPDZ. (Right) Western blot of yeast extracts expressing Nkd or NkdΔ30aa. Note that deletion of the 30-aa motif did not affect Dsh binding.

Sequencing DNA from flies heterozygous for a panel of lethal EMS-induced nkd alleles (Jürgens et al. 1984; W. Zeng and M. P. Scott, unpublished data) revealed six nonsense mutations predicted to encode truncated proteins ranging from 399 to 516 aa (Figure 9A and Figure 10A). Each mutant develops a variable nkd cuticle phenotype—possibly due to variation in Wg-dependent nkd expression, nonsense-mediated mRNA decay, or stability or activity of each truncated protein—but stage 11 embryos exhibit a wide En stripe indicative of elevated Wg signaling (Figure 9A). As shown in Figure 9B, each allele exhibits a weaker nkd cuticle phenotype than the first allele that we molecularly characterized, nkd7H16, which harbors a nonsense codon at residue no. 60 (Zeng et al. 2000). Since Nkd7H/GFPC and Nkd7H/mycC lacked rescue activity (Figure 8, A and B), we presume that nkd7H16 represents a null allele but a definitive determination awaits examination of nkd deficiency chromosomes.

Figure 9.—

Phenotypes and molecular lesions of nkd alleles. (A) Predicted truncated Nkd protein based on sequencing of indicated nkd allele, relative to full-length Nkd (top). Note that each predicted protein is truncated N-terminally of the 30-aa motif. Stage 11 mutant embryos have a widened En stripe and a variable nkd phenotype. Right column shows DNA sequence trace from a heterozygote, indicating the codon number with mutation (arrow). (B) Mean number of complete (black bars) and partial (shaded bars) abdominal denticle belts (A1–A8) in each nkd allele (n = 25 mutant cuticles scored for each allele). The molecular lesion of nkd7H16 was reported in Zeng et al. (2000).

Figure 10.—

Nkd alignment and protein family. (A) ClustalW alignment of predicted An. gambiae (Ag, GenBank BK005845) and D. melanogaster (Dm, GenBank AF213376) Nkd proteins. Identities are in a black background, and similarities are shaded. EFX is designated by a red line and the 30-aa motif by light blue shading. Arrowheads designate the position of nonsense mutations in each nkd allele shown in Figure 9A as well as the C-terminal extent of constructs NkdNIN3/GFPC and NkdNIN4/GFPC (each named for the position of nkd intron 3 and 4, respectively). (B) Nkd protein family showing region of EF-hand similarity between Recoverin (Rec), consisting of four EF hands, and Nkd proteins. Vertical lines indicate sequence homology, whereas dashed lines indicate preserved motif order, but not sequence conservation, between recoverin, fly and mosquito Nkd, and Mus musculus (Mm) and Homo sapiens (Hs) Nkd1 (orange) and Nkd2 (green). Thirty-amino-acid motifs (light blue in insect, dark blue in mammal) are each predicted to be an amphipathic α-helix, part of which is depicted in helical wheel format above the fly sequence (beginning with Nkd residue 552). Diagonal line separates charged (+ or −) residues on one face of the predicted helix from hydrophobic residues (H) that mostly lie on the opposite face. Residues: pink, acidic; blue, basic; orange, nonpolar; green, polar uncharged.

Next we assembled a putative nkd cDNA sequence from the Anopheles gambiae genome (GenBank BK005845) (Holt et al. 2002). Mosquito Nkd is predicted to be 886 aa and Mr = 96.7 kDa (Figure 10A). Fly and mosquito Nkd share 29% aa identity, with homology clustered in four blocks interspersed by mostly dissimilar sequence (Figure 10A). Fly and mosquito Nkd share 22/30 identical amino acids (fly aa 543–572) that lie precisely within the interval critical for nkd activity defined by transgenes and alleles. Mammalian nkd paralogs also share four sequence blocks interspersed by variable sequence (Figure 10B) (Katoh 2001; Wharton et al. 2001). Strikingly, the third conserved sequence block in mammalian Nkds, like those of the two insect Nkds, is exactly 30 aa, but no homology exists between insect and mammalian 30-aa motifs (Wharton et al. 2001; data not shown). However, a secondary structure prediction algorithm (see materials and methods) indicates that insect and mammalian motifs may adopt an amphipathic α-helical secondary structure (Figure 10B; data not shown). NkdΔ30aa/GFPC, lacking the 30-aa motif, only slightly narrowed the nkd mutant En stripe and partially rescued the nkd cuticle (Figure 8A). NkdΔ30aa/GFPC and NkdNIN3/GFPC were also compromised in their ability to reduce Arm levels in stage 10 nkd mutants relative to NkdGFPC and NkdNIN4/GFPC (Figure 11, A–E). By Y2H, Nkd and NkdΔ30aa each interacted to a comparable extent with Dsh or DshbPDZ (Figure 8C), and the fly 30-aa motif bound neither Dsh nor DshbPDZ (not shown). Like NkdNIN3/GFPC, NkdΔ30aa/GFPC exhibited reduced nuclear staining in stage 11 embryos relative to NkdGFPC (Figure 8A; also cf. Figure 13, A and B), indicating that the 30-aa motif is necessary for Nkd function and nuclear localization during segmentation.

Figure 11.—

The 30-aa motif is required for Nkd to reduce Arm/β-catenin levels. Each panel shows four segmental anlagen of a representative stage 10 UAS-X;nkd7H16prd-Gal4/nkd7H16 embryo, where X is each indicated Nkd construct, stained for Arm (green) and GFP (purple). Colored arrowheads indicate position of nkd mutant (green) and rescued (purple) segments. Right column shows mean grayscale intensity (ordinate range 0–200) of Arm staining as a function of position along the anterior–posterior axis. To quantitate the extent of Arm reduction by each Nkd construct, we defined γ as the ratio of peak grayscale intensity in nkd mutant segments to that in adjacent segments expressing each construct. When GFP is used in the rescue assay, γ = 1.0 (i.e., no rescue), while for full-length Nkd, γ > 1.8. Both constructs that lacked the 30-aa motif had γ = 1.2, indicating a reduced ability to decrease Arm levels.

The fly 30-aa motif increases activity and confers nuclear localization to mouse Nkd1:

The distribution and activity of the fly 30-aa motif fused to GFP did not differ from that of GFP alone (not shown). As a test of sufficiency, we replaced the mouse Nkd1 30-aa motif with the fly motif (mNkd1f30aa). Previously, we reported that misexpression of mNkd1 during fly larval development induced weak wg loss-of-function adult phenotypes whose penetrance was sensitive to a reduced dose of wg (Wharton et al. 2001). When expressed by da-Gal4, neither mNkd1GFPC nor mNkd1f30aa/GFPC affected the nkd phenotype (not shown). However, mNkd1f30aa/GFPC caused a twofold greater reduction in sternite bristle number than did mNkd1GPFC when driven by B119-Gal4 in a wg/+ background (Figure 12, A and B). In addition, the fly 30-aa motif conferred low-level nuclear localization to mNkd1 in stage 10 embryos but robust nuclear accumulation in third instar larval salivary glands (Figure 12, C–F).

Figure 12.—

Fly 30-aa motif increases mouse Nkd1 activity and confers nuclear localization. Schematic of mouse Nkd1GFPC (left) and chimeric construct with substitution of fly 30-aa domain (light blue, right). (A and B) Representative B119-Gal4 wg/UAS-mNkd1GFPC (A) or B119-Gal4 wg/UAS-mNkd1f30aa/GFPC (B) female abdomens showing segments with reduced numbers of sternite bristles (arrows). mNkd1f30aa/GFPC reduced sternite bristle counts approximately twofold relative to mNkd1GFPC (the mean ±SD of bristles for 10 flies from the indicated number of lines for each construct is shown). (C and D) Epidermal cells of stage 11 UAS-mNkd1GFPC;nkd7H16da-Gal4/nkd7H16 (C) or UAS-mNkd1f30aa/GFPC;nkd7H16da-Gal4/nkd7H16 embryos (D) stained with α-GFP (green). Magnified insets show low-level nuclear GFP in mNkd1f30aa/GFPC but not mNkd1GFPC (asterisks mark center of nuclei). C and D were stained in parallel and imaged under identical confocal settings. (E and F) Salivary gland expressing mNkd1 constructs driven by da-Gal4. Note absence of nuclear GFP in mNkd1GFPC-producing salivary gland cells (E) but abundant nuclear GFP (arrow) in mNkd1f30aa/GFPC-producing cells (F). DNA is purple.

A heterologous nuclear localization sequence partially restores activity to NkdΔ30aa/GFPC:

If Nkd acts in the nucleus to antagonize Wg signaling, then a heterologous nuclear localization sequence (NLS) may restore the ability of NkdΔ30aa/GFPC to enter the nucleus and rescue a nkd mutant. NkdΔ30aaNLS/GFPC, in which the SV40 NLS replaced the 30-aa motif, narrowed the En stripe and restored cuticle pattern to nkd mutants to an extent intermediate between NkdΔ30aa/GFPC and full-length Nkd (Figure 13, A–D). After stage 10, NkdΔ30aaNLS/GFPC was detectable in the nucleus and cytoplasm (Figure 13C), with rare punctate cytoplasmic Dsh colocalization but with no apparent effect on cas-Dshmyc distribution (not shown). One of three UAS-NkdΔ30aaNLS/GFPC transgenic lines (no. 7A) produced rescued crawling first instar larvae, 1.6% (of 252) of which survived to adulthood (Figure 13E). In contrast, no rescued crawling first instar larvae emerged from rescue crosses with two independent UAS-NkdΔ30aa/GFP transgenic lines (n > 1000 progeny examined/line). Adults rescued by NkdΔ30aaNLS/GFPC exhibited more severe wing-margin bristle patterning abnormalities than those rescued by Nkd (cf. Figure 13F with 3S), indicating that enhanced nuclear localization alters Nkd activity.

Figure 13.—

A heterologous NLS partially restores NkdΔ30aa activity. (A–C) Representative stage 11 nkd7H16 mutant embryos expressing NkdGFPC (A), NkdΔ30aa/GFPC (B), or NkdΔ30aaNLS/GFPC (C) stained for GFP (green) and En (red). The NLS promotes Nkd nuclear localization and narrows the En stripe of most segments (arrowheads). (D) Quantitation of nkd cuticle rescue. Note increased proportions of wild type and reduced moderate and weak nkd cuticles with rescue by NkdΔ30aaNLS/GFPC (red bars) as compared to NkdΔ30aa/GFPC (purple). (E) UAS-Nkd1Δ30aaNLS/GFPC;nkd7H16da-Gal4/nkd7H16 adult. (F) Anterior wing margin of animal in E with marked bristle disarray, including absent (x) and ectopic (blue arrowhead) stout bristles and mispatterned and ectopic slender bristles (red arrowhead).


Here we report a structure–function analysis of D. melanogaster Nkd, a novel intracellular protein that acts in a feedback loop to limit Wnt/β-catenin signaling during embryogenesis. Our finding that nkd mutants have elevated Arm/β-catenin levels concomitant with broadened domains of Wg target gene expression is consistent with prior reports of Nkd targeting Dsh, an enigmatic Wnt signal transducer that acts upstream of β-catenin degradation. Although Wnt-signal-induced Dsh accumulation has been observed in cultured cells (Yanagawa et al. 1995), transgenic mice (Millar et al. 1999), and some cancers (Uematsu et al. 2003), and recent studies indicate that Dsh, like β-catenin, can be degraded by the ubiquitin–proteasome pathway (Simons et al. 2005; Angers et al. 2006; Zhang et al. 2006), our data show that Nkd does not attenuate Wnt signaling in the embryo by significantly altering steady-state Dsh levels or distribution. If Nkd promotes Dsh degradation in the fly embryo, as has recently been proposed on the basis of overexpression of mammalian Nkd in cultured cells (Creyghton et al. 2005), it must act only on a subset of Dsh, perhaps the fraction engaged in signaling. Consistent with this idea, we observed rare, punctate Nkd/Dsh colocalization in embryonic ectodermal cells.

Several Nkd constructs with mutant or deleted Dsh-binding regions possessed a reduced but still substantial nkd rescue activity. Perhaps NkdΔR1S/GFPC, lacking both Dsh-binding regions, is able to target Dsh in vivo (and hence rescue a nkd mutant) by virtue of overexpression, through other low-affinity Nkd/Dsh-binding regions, or by as yet uncharacterized proteins that bridge Nkd to Dsh. Consistent with these possibilities, we also observed some NkdΔR1S/GFPC/Dsh colocalization.

Three independent lines of investigation—evolutionary sequence comparisons, sequencing of lethal nkd alleles, and transgenic nkd rescue assays—pinpointed a 30-aa motif, separable from Dsh-binding regions, that is crucial for fly Nkd activity and nuclear localization. The comparable positions, identical sequence length, and similar predicted structure of insect and mammalian 30-aa motifs suggests that the family of Nkd proteins may inhibit Wnt signaling through a common mechanism. Given the small size and presumably simple α-helical structure of the 30-aa motif, it is unlikely to possess intrinsic catalytic activity but, in addition to its weak NLS activity, it could serve as a protein-docking motif.

In addition to several reports that have documented nucleo-cytoplasmic shuttling of β-catenin, Axin, and APC (Schneider et al. 1996; Arbesfeld et al. 2000; Cong and Varmus 2004; Rosin-Wiechens et al. 2004; Xiong and Kotake 2006), it is noteworthy that two recent reports revealed a potential role for Fz and Dsh in the nucleus (Itoh et al. 2005; Mathew et al. 2005). In response to Wg signaling at the fly neuromuscular synapse, the Fz2 C terminus was detected in puncta of postsynaptic muscle nuclei although not in ectodermal nuclei (Mathew et al. 2005), so this report's significance to Nkd's action in ectoderm is unclear. Xenopus Dsh has a separable NLS and nuclear export sequence (NES), with the former required for “signaling activity” in gain-of-function assays (Itoh et al. 2005). However, a vertebrate Dsh construct with a mutant NES exhibited increased nuclear accumulation but no activity increase relative to that of wild-type Dsh, arguing against nuclear Dsh concentration—at least when it is overexpressed—being rate limiting for activity (Itoh et al. 2005). Intriguingly, Dsh NES and NLS motifs seem to be conserved in D. melanogaster Dsh, but their significance remains to be investigated.

Our data extend our still rudimentary knowledge of Nkd action in the fly embryo. The epistatic relationship between wg and nkd suggests that, in the absence of Wg ligand, the low levels of Nkd in a wg mutant (because Wg normally upregulates nkd transcription) inhibit spontaneous ligand-independent signaling through the Wnt receptor complex (Bejsovec and Wieschaus 1993; Zeng et al. 2000). Wg exposure promotes Arm accumulation and induction of target genes, including en, hh, and nkd. Nkd, synthesized in the cytoplasm, accumulates and targets an uncharacterized fraction of cytoplasmic Dsh. However, Nkd/Dsh binding alone is apparently insufficient to limit Wg signaling during stages 10–11, as Nkd uses its 30-aa motif to inhibit Arm accumulation, restrict Wg-dependent gene expression, and access the nucleus. Although it is possible that the 30-aa motif is required in the cytoplasm, and that the ability of the 30-aa motif to confer nuclear access is a consequence rather than a cause of activity, three lines of evidence support a nuclear role for Nkd: (1) a subpool of Nkd normally accumulates in the embryonic nuclei after stage 10; (2) the 30-aa motif, distinct from the Dsh-binding sequence, was necessary for both nuclear localization and activity and was sufficient to increase the activity of mouse Nkd1 when expressed in the fly; and (3) a heterologous NLS increased nuclear localization and nkd rescue activity of NkdΔ30aa.

While our experiments strongly suggest a role for Nkd in the nucleus, they do not reveal the nature of that role. Likewise, lacking insight into how Dsh transmits Wnt signals into the nucleus, our experiments thus far reveal neither the relevant subcellular location(s) of Nkd action nor a molecular mechanism by which Nkd inhibits Dsh activity. The punctate Nkd/Dsh colocalization that we observe in embryonic cytoplasm, and rarely, in nuclei, is consistent with Nkd either affecting Dsh nucleo-cytoplasmic transport or impinging directly on the chromatin of Wnt-responsive genes. Our inability to observe increased nuclear Nkd or Dsh after treatment with a nuclear export inhibitor suggests that nuclear export of Nkd (and possibly Dsh) in the fly embryo (1) does not occur (e.g., if each protein were degraded in the nucleus following import); (2) occurs over a longer time period relative to proteins such as Lines that can rapidly shuttle between nucleus and cytoplasm; (3) is independent of CRM-1; or (4) like the presumed Nkd/Dsh interaction, involves only a fraction of the total pool of each protein. Future experiments will be required to distinguish among these possibilities.

The four-domain structure of both insect and vertebrate Nkd's argues that there once existed an ancient “core” mechanism by which Nkd engaged Dsh to inhibit Wnt signaling. However, given the sequence divergence between insect and mammalian Nkds, their current mechanisms may share little similarity beyond Dsh binding. Recently, PR72 and PR130, two alternatively spliced B″ subunits of the multi-subunit enzyme protein phosphatase 2A (PP2A), were shown to associate with mammalian Nkd and to modulate its inhibitory effect on ectopic Wnt signaling (Creyghton et al. 2005, 2006). As Dsh is phosphorylated by kinases such as CK1, CK2, and Par1 following Wnt stimulation (Yanagawa et al. 1995; Willert et al. 1997; Peters et al. 1999; Sun et al. 2001), recruitment of phosphatases to Dsh by Nkd represents an attractive hypothesis to explain the inhibitory action of Nkd on Wnt signaling via Dsh. Consistent with this possibility, Nkd, PP2A, and Dsh kinases co-immunoprecipitated with vertebrate Dsh (Angers et al. 2006). However, unlike the vertebrate Nkd/PR72 interaction, thus far we do not detect direct interactions by Y2H between the fly PR72 homolog (CG4733) and full-length fly Nkd or any of the regions in Nkd, in particular the 30-aa motif, that are crucial for activity (our unpublished observations). Thus, regulation of Nkd activity by PR72/PR130 may be a derived, vertebrate-specific phenomenon—analogous in some ways to the effect of mammalian Nkd2 but not Nkd1 on intracellular TGF-α trafficking (Li et al. 2004)—that may be distinct from the mechanism by which Nkd regulates Wg signaling in Drosophila.

In Drosophila, nkd is crucial for shaping gradients of Wnt activity, but is this role conserved in vertebrates? Mouse nkd genes are expressed during embryogenesis in dynamic patterns reminiscent of known Wnt gradients (Wharton et al. 2001; Ishikawa et al. 2004; Nakaya et al. 2005). A recent report described nkd1 mutant mice with a targeted deletion of exons 6 and 7 (encoding the EFX domain) but allowing in-frame splicing between exons 5 and 8, resulting in expression of a residual Nkd1 protein very much analogous to our NkdΔR1S/GFPC construct that lacks Dsh-binding sequences but retains three conserved motifs (Li et al. 2005). Given that nkd1 is more broadly expressed than nkd2 during mouse development (Wharton et al. 2001), it was surprising that nkd1−/− mice were viable and fertile, even though mutant mouse embryo fibroblasts showed elevated Wnt reporter activity and homozygous male mice exhibited a sperm maturation defect (Li et al. 2005). Although genetic redundancy between nkd1 and nkd2 could account for these observations, our results suggest an alternative hypothesis, namely that the residual protein produced in the reported nkd1 mutant mice, like our EFX-deleted NkdΔEFX/GFPC and NkdΔR1S/GFPC constructs, has significant activity in vivo, despite the observation that a mutant mNkd1 protein lacking the EF hand is defective at blocking Wnt signaling in cultured cells (Yan et al. 2001). Resolution of this quandary awaits an investigation of strong loss-of-function mutations in each mammalian nkd gene. Given the broad involvement of Wnt/β-catenin signaling in mammalian development and cancer, coupled with the similar loss-of-function phenotypes of fly nkd, axin, and apc homologs, we hope that our studies guide future investigations of vertebrate Nkd proteins as regulators of Wnt signaling and candidate tumor suppressor genes.


Thanks go to Matthew Scott and his lab for support, generosity, and encouragement throughout the early stages of this project and for the UAS-Nkd, UAS-NkdmycC, and UAS-NkdΔRB/mycC flies and new nkd alleles; Roel Nusse and his lab for α-Dsh and continued interest; Victor Hatini for UAS-Linesmyc flies and the LMB protocol; Gregor Zimmermann for cloning UAS-NkdGFPC; Jeff Axelrod for cas-Dshmyc, UAS-Dsh, UAS-DshΔbPDZ, and UAS-DshGFP flies; Nipam Patel for α-En; Heather Stevens for help with sequencing nkd alleles; George Lawton for electron microscopy; Jon Graff and Scott Cameron for use of their fluorescent dissection microscopes; Steve DiNardo, Victor Hatini, Marcel Wehrli, and Eric Wieschaus for insightful discussion; and Jin Jiang, Rueyling Lin, Jeff Axelrod, and anonymous reviewers for critical review of the manuscript. R.R. was supported by the Human Frontier Science Program and the Howard Hughes Medical Institute (HHMI) and J.M. and W.Z. were supported by the National Institutes of Health (NIH) and HHMI. K.W. is a W.W. Caruth, Jr., Scholar in Biomedical Research and has received a Beginning Grant-In-Aid Award from the Texas Affiliate of the American Heart Association, a Junior Faculty Institutional Research Grant from the American Cancer Society, and NIH grants K08-HD01164 and R01-GM65404.


  • Wnts were discovered through their ability to cause breast cancer in mice (Nusse and Varmus 1982). This article—an incremental step closer to understanding how Wnt signaling is normally regulated—is dedicated to the memory of Colleen Werner, one of too many women who have prematurely succumbed to breast cancer.

  • Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession no. BK005845.

  • 1 Present address: Institute of Signaling, Developmental Biology and Cancer, Centre de Biochimie, University of Nice, Parc Valrose, Nice 06108, France.

  • 2 Present address: Cleveland Clinic Foundation, Lerner Research Institute, Department of Biomedical Engineering, Cleveland, OH 44195.

  • 3 Present address: MedImmune, Department of Process Development, Santa Clara, CA 95054.

  • 4 Present address: Division of Cardiology, Department of Medicine, Duke University Medical Center, Durham, NC 27710.

  • 5 Present address: Laboratory of Mammalian Molecular Embryology, RIKEN Center for Developmental Biology, Kobe 650-0047, Japan.

  • Communicating editor: K. V. Anderson

  • Received June 11, 2006.
  • Accepted July 13, 2006.


View Abstract