|Home | About | Journals | Submit | Contact Us | Français|
Robust development, tissue homeostasis, and stem cell renewal in animals requires precise control of the Wnt/β-catenin signaling axis. In the embryo of the fruit fly Drosophila melanogaster, the naked cuticle (nkd) gene attenuates signaling by the Wnt ligand Wingless (Wg) during segmentation. nkd mutants have been reported to exhibit abnormalities in wg transcription, Wg protein distribution and/or transport, and the intracellular response to Wg, but the relationship between each alteration and the molecular mechanism of Nkd action remains unclear. In addition, whether Nkd acts in a cell-autonomous or nonautonomous fashion in the embryo is not known. Mammalian Nkd homologs have N-terminal consensus sequences that direct the post-translational addition of a lipophilic myristoyl moiety, but fly and mosquito Nkd, while sharing N-terminal sequence homology, lack a myristoylation consensus sequence. Here we provide evidence that fly Nkd acts cell-autonomously in the embryo, with its N-terminus able to confer unique functional properties and membrane association that cannot be mimicked in vivo by heterologous myristoylation consensus sequences. In conjunction with our recent observation that Nkd requires nuclear localization for function, our data suggest that Nkd acts at more than one subcellular location within signal-receiving cells to attenuate Wg signaling.
Secreted Wnt proteins regulate development, tissue homeostasis, and stem cell renewal throughout the animal kingdom. The ability of Wnts to regulate gene expression in a concentration-dependent fashion across fields of cells and/or layers of tissue embodies their designation as “morphogens” (Hoppler and Bienz, 1995; Immergluck et al., 1990; Zecca et al., 1996). The importance of precisely controlling Wnt signal transmission – from ligand production in signal-producing cells to signal reception and interpretation signal-receiving cells – is evidenced by the plethora of developmental defects and diseases - including cancer, osteoporosis, and diabetes - that arise as a consequence of altered signaling (Clevers, 2006). A thorough understanding of mechanisms that maintain Wnt signaling cascades within a physiological range will be a prerequisite to the effective manipulation of signaling for therapeutic purposes.
One common consequence of Wnt receptor engagement in widely diverged animals is the accumulation of β-catenin, a bifunctional adaptor protein that regulates cell adhesion and acts as a transcriptional coactivator to regulate Wnt-dependent target gene expression (Brembeck et al., 2006; Molenaar et al., 1996; Peifer et al., 1994; Riggleman et al., 1990). Graded activity of the so-called “canonical Wnt/β-catenin” signaling axis, elicited by the Wnt protein Wingless (Wg), is crucial for segmentation in the embryo of the fruit fly Drosophila melanogaster (Baker, 1988; Bejsovec and Martinez Arias, 1991; DiNardo et al., 1994; Noordermeer et al., 1992). Discrete phases of differential gene expression, cell-cell interactions, and cell morphogenesis are evident during segmentation: First, a hierarchy of gap, pair-rule, and segment-polarity genes specifies segmental primordia (Nüsslein-Volhard and Wieschaus, 1980). Next, Wg and Hedgehog (Hh) signals, expressed within each segmental anlage in adjacent, narrow stripes of cells, reinforce each other’s expression through a ~2½ hour period during germ band extension (~3½–6 hrs. after egg laying, stages 9–11) (DiNardo et al., 1994). Thereafter, a dynamic and unique combination of Wg, Hh, Notch, and EGF signals influences each cell’s fate (Bejsovec and Martinez Arias, 1991; Bejsovec and Wieschaus, 1993; Colosimo and Tolwinski, 2006; Hatini and DiNardo, 2001; Payre et al., 1999; Price et al., 2006; Urban et al., 2004; Wiellette and McGinnis, 1999). Upon differentiation, ventral epidermal cells distant from Wg-producing cells secrete a belt-like array of protrusions termed denticles, while cells close to the source of Wg - except those that will make the most anterior row of denticles - suppress denticle synthesis and secrete “naked” cuticle (Bejsovec and Martinez Arias, 1991; Dougan and DiNardo, 1992).
How Wnts and other morphogens transit through tissues to regulate target gene expression remains the subject of intense study and conjecture (Lander, 2007; Martinez Arias, 2003; Sampedro et al., 1993), with important, early contributions made possible through the study of Wg signaling during Drosophila segmentation (Klingensmith and Nusse, 1994). While morphogens such as Wg may in principle freely diffuse to distant cells (Lander et al., 2002), the facts that Wnts are glycosylated and lipid-modified (Smolich et al., 1993; Willert et al., 2003), that there exists an extracellular jungle of proteoglycans that can bind Wnts and limit free diffusion (Baeg et al., 2004; Franch-Marro et al., 2005; Han et al., 2005; Kirkpatrick et al., 2004), that Wg-producing and receiving cells may migrate away from the signal source prior to degrading Wg (Pfeiffer et al., 2000), and that Wg may be actively transported - “transcytosed” – through a field of cells (Greco et al., 2001; Moline et al., 1999), indicates the existence of multiple regulatory mechanisms. In the stage 9 embryo, Wg is found in the endoplasmic reticulum and Golgi apparatus of signal-sending cells, and is symmetrically distributed in the extracelluar space and in endolysosomal vesicles in a few rows of adjacent signal-receiving cells; but, by stage 11, the gradient of Wg distribution and activity is asymmetric with an anterior bias due to Hh and EGF signal-dependent Wg degradation in posterior signal-receiving cells (Dubois et al., 2001; Gonzalez et al., 1991; Sanson et al., 1999; van den Heuvel et al., 1989). Despite enhanced degradation of Wg in posterior cells during stages 10–12, genetic studies suggest that Wg influences the diversity of denticle morphologies synthesized by those cells (Bejsovec and Martinez Arias, 1991; Bejsovec and Wieschaus, 1993), raising the question of how Wg reaches these cells to influence their fate.
Embryos mutant in the naked cuticle (nkd) gene have been reported to develop sequential abnormalities in Wg target gene transcription, Wg protein distribution, and wg mRNA expression (Bejsovec and Wieschaus, 1993; Dougan and DiNardo, 1992; Jürgens et al., 1984; Lee et al., 1992; Martinez Arias et al., 1988; Moline et al., 1999; Tabata et al., 1992; Waldrop et al., 2006; Zeng et al., 2000). In nkd mutants, wg and hh expression initiates normally, but during stage 9, the Wg target genes hh and engrailed (en) are transcribed in additional, posterior cells distant from Wg-producing cells, which though stage 10–11 leads to the induction of a new stripe of wg just posterior to the expanded en/hh domain (Fig. 1A) (Martinez Arias et al., 1988). Since later Wg signaling directs “naked” cuticle differentiation, the ectopic Wg stripe creates mirror-image pattern duplications, extra naked cuticle, and increased cell death in embryos homozygous for strong nkd alleles, while weaker nkd alleles show replacement of denticle belts by naked cuticle in proportion to the number of cells that ectopically produce Wg (Bejsovec and Martinez Arias, 1991; Dougan and DiNardo, 1992; Jürgens et al., 1984; Pazdera et al., 1998; Waldrop et al., 2006; Zeng et al., 2000).
Although the sequence of gene expression abnormalities in nkd mutants is well recognized, the causal relationships between the early and later abnormalities are not understood. Intact Wg and Hh signaling is required for induction of the ectopic wg stripe in nkd mutants (Bejsovec and Wieschaus, 1993; Dougan and DiNardo, 1992), likely due in part to a requirement for wg in autoregulating its own transcription (Yoffe et al., 1995), but the source of the initiating Wg signal – anterior vs. posterior to the incipient ectopic wg stripe - is not clear. Moreover, although Nkd can block responses to ectopic Wg in a cell-autonomous manner in eye imaginal disc (Rousset et al., 2001), whether Nkd acts cell-autonomously in the embryo has not been investigated. Previously, the observation that misexpression of shibireK44A (shiD), a dominant-negative version of the motor protein dynamin whose expression disrupts endocytosis, in en-expressing cells of nkd embryos narrowed the En stripe and prevented ectopic wg transcription was argued as evidence in support of the hypothesis that Nkd restricts the cell-to-cell transport of Wg posterior to wg-expressing cells (Damke et al., 1994; Moline et al., 1999). Consistent with this possibility was an apparent increase in Wg immunoreactivity posterior to the endogenous wg stripe in stage 10 nkd mutants (Moline et al., 1999). However, experiments showing that Wg signaling is not required for normal denticle synthesis in cells posterior to en-expressing cells (Sanson et al., 1999), coupled with the recent finding that misexpression of shiD may cause toxic and/or non-specific effects on Wg signaling or cell viability (Rives et al., 2006), demand a careful reappraisal of any conclusions derived from shiD misexpression.
By what mechanism does Nkd inhibit Wnt signaling? As a first step to answer this question, our laboratory has developed in vivo assays - including a stringent nkd rescue assay - to study Nkd activity in Drosophila, and has tested the activities of mutant and chimeric Nkd proteins in these assays (Rousset et al., 2002; Waldrop et al., 2006). Nkd is a novel intracellular protein that inhibits Wg signaling, in part, by binding to Dishevelled (Dsh), a scaffold protein that links – in a still mysterious fashion - Wnt receptor activation to β-catenin accumulation (Rousset et al., 2001; Wallingford and Habas, 2005; Wharton, 2003; Zeng et al., 2000). The only sequence homology shared between fly Nkd and its two mammalian homologs - Nkd1 and Nkd2 - is an EF-hand-containing domain (which we have termed “EFX,” but is a.k.a. “NH2” (Katoh, 2001) or “NHR1” (Li et al., 2004)) that binds Dsh or its mammalian counterparts, the Dvl proteins (Rousset et al., 2002; Wharton et al., 2001). Nkd also has an unconventional nuclear localization motif, separate from Dsh-binding regions, that is essential for activity (Waldrop et al., 2006). A consensus sequence (MGXXXS) that directs the post-translational addition of a lipophilic myristoyl moiety is found at the N-terminus of mammalian Nkd1 and Nkd2, but no such sequence is found at the N-terminus of fly Nkd (Li et al., 2004; Resh, 2004; Wharton et al., 2001; Zeng et al., 2000), raising the question of whether the fly and mammalian Nkd N-termini have homologous functions. Although possibly a coincidence, EF-hands in known Nkd proteins are most similar to EF-hand proteins of the recoverin family of “myristoyl-switch” proteins that undergo Ca2+-sensitive alteration in tertiary structure to expose a buried N-terminal myristate, thereby allowing reversible, stimulus-dependent membrane association (Flaherty et al., 1993; Zeng et al., 2000). The absence of a N-terminal myristoylation sequence in fly Nkd suggests either that Nkd does not associate with membranes, or that if it does engage membranes it may do so through protein-protein interactions as a peripheral membrane protein.
In this paper, we show that Nkd acts cell-autonomously during segmentation, independently limiting Wg signaling in cells anterior and posterior to Wg-producing cells, a result inconsistent with models that invoke posterior, cell-to-cell transport of Wg as necessary for induction of ectopic wg stripes in nkd mutants (Moline et al., 1999). Additionally, we show that the N-terminus of Drosophila Nkd confers unique functional and membrane-association properties that cannot be mimicked by N-terminal myristoylation. Our results suggest that Nkd acts at multiple subcellular locations in signal-receiving cells to attenuate the response to Wg signaling.
All Nkd constructs were built in Bluescript-II-KS+ (Stratagene) and cloned into the P-element vector pUAS-T for P-element transformation and misexpression using the Gal4/UAS system (Brand and Perrimon, 1993). NkdGFPC, NkdNBg/GFPC, Nkd7H/GFPC, mNkd1GFPC, mNkd1f30aa/GFPC, Src(1-89)GFP have been previously described (Waldrop et al., 2006). DNA fragments were synthesized by Pfu PCR and then subcloned and sequenced. NkdGFPX/mycC has linker sequence plus enhanced green fluorescent protein (EGFP) (Clontech) amino acids (aa) 1-244 inserted after Nkd aa 294 (Nkd residues underlined: …GSIATMV… (EGFP) …GLRSGG…), with a C-terminal myc tag. NkdGFPN/mycC has EGFP residues 1-239 fused to Nkd aa 2-928, with a C-terminal myc tag. NkdSrcNBg/GFPC has Drosophila Src aa 1-89 (Simon et al., 1985) fused to NkdNBg/GFPC; NkdSrcG2ANBg/GFPC is identical to NkdSrcNBg/GFPC except Gly2 was mutated to Ala. NkdSrcR1Bg/GFCP fused Src aa 1-89 to NkdNBg/GFPC aa 177-928. Nkdm1N N/GFPC fused mouse Nkd1 aa 1-27 to Nkd aa 79-928 to GFP. mNkd1fNf30aa/GFPC fused fly Nkd aa 1-78 to mNkd1f30aa/GFPC aa 28-end.
Fly husbandry and P-element transformation were performed by standard methods. All crosses were performed at 25°C. nkd7H16, which harbors a nonsense mutation at codon #60 and is the strongest known nkd allele (Jürgens et al., 1984; Waldrop et al., 2006; Zeng et al., 2000), was used in all rescue assays. At least three independent transgenic lines were obtained for each UAS-Nkd construct, and two or more independent chromosome II inserts for each line were tested in each nkd rescue cross: UAS-Nkd/UAS-Nkd or CyO ; nkd7H16/TM3-Hb-lacZ X nkd7H16 da-Gal4/TM3-Hb-lacZ. For cases where the UAS-Nkd line used was a lethal insertion, 12.5% of cuticles are predicted to remain unrescued nkd mutants due to the persistence of the CyO balancer in the rescue cross. Fly stocks/chromosome: UAS-GFP (II) (Yeh et al., 1995), UAS-lacZ (II) (Brand and Perrimon, 1993), UAS-AxinGFP (II) (Cliffe et al., 2003), UAS-dTCFΔN (II) (van de Wetering et al., 1997), da-Gal4 (III) (Wodarz et al., 1995), prd-Gal4 (III) (Yoffe et al., 1995), en-Gal4 (II) (Fietz et al., 1995), ptc-Gal4 (II) (Hinz et al., 1994), B119-Gal4 (II) (Zeng et al., 2000), 71B-Gal4 (III) (Brand and Perrimon, 1993).
Cuticle preparations were performed, scored as wild type or weak, moderate or strong nkd, and selected for photography as previously described (Waldrop et al., 2006; Zeng et al., 2000). Criteria for scoring nkd cuticles were based on the existing nkd allelic series as well as the spectrum of nkd cuticle rescue observed when Nkd is expressed in a strong nkd mutant: “strong” class cuticles, exemplified by nkd7H16, nkd7E89, and nkd47K1, are <75% of wild-type length, have two or fewer complete denticle bands and have a fully exteriorized head skeleton and non-everted, widely spaced posterior spiracles (Jürgens et al., 1984; Waldrop et al., 2006; Zeng et al., 2000); “moderate” class cuticles, exemplified by nkd9G33, or nkd7H16 rescued by ubiquitous expression of a UAS-Nkd construct that lacks the 30 aa motif (Jürgens et al., 1984; Rousset et al., 2002; Waldrop et al., 2006), have three or more complete denticle bands (usually in odd numbered abdominal segments), a partially internalized head skeleton, and partially to fully everted spiracles. “Weak” class cuticles, exemplified by nkd42J1, nkd71-72K1, and l(3)4869 (a lethal P element inserted in the 5’ end of nkd), have normal denticle bands in most segments but focal denticle band loss, and typically subtle or no head or tail defects (Jürgens et al., 1984; Waldrop et al., 2006; Zeng et al., 2000). Rescue of the nkd head and tail defects was not as efficient with expression of the Wg antagonists dTCFΔN, AxinGFP, or mNkd1fNf30aa as compared to fly Nkd, so only denticle belt criteria were used to score nkd phenotypic classes in Figs. 3J and and6B6B.
Figures were prepared in Photoshop (Adobe), drawings in Canvas (ACD Systems) and Powerpoint (Microsoft), and graphs in Deltagraph (Red Rock Software). Channel spectra were maximized below saturation for selected confocal images, with the exception that anti-Arm channels were not altered. ImageJ (NIH) was used to quantitate mean greyscale pixel intensity of raw confocal images of Arm-stained embryos as previously described (Waldrop et al., 2006). Statistical calculations were performed using Prism (GraphPad Software) and Microsoft Excel.
All embryo collection, fixation, staining procedures including antibody dilutions, and confocal microscopy were performed as previously described (Waldrop et al., 2006). Mean +/− s.d. of En stripe width was calculated by counting the number of En-positive cell diameters along the anterior-posterior axis in dorsomedian and ventral-median positions in parasegments 2–14 of five stage 11 nkd da>Nkd embryos for each construct (130 data points per construct).
pArm-GFP, or pArm-Gal4 with either pUAS-Src(1-89)GFP or pUAS-Nkd7H/GFPC plasmids were transfected into Drosophila Kc cells grown at 25°C in Schneider medium with 10% fetal bovine serum. Lysates as follows were prepared 36 hrs. post-transfection: Whole cell lysates were obtained by scraping and dispersion of cells into Drosophila cell lysis buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% NP40, 0.5% sodium deoxycholate, 1 mM PMSF and 1×protease inhibitor cocktail (Roche)) for 1 hr on ice, then centrifuging at 12,000 × g for 10 min. Triton-soluble fractions from cultures transfected in parallel were obtained by gently adding MES buffer, pH 6.8 (2.5 mM EDTA, 5 mM MgCl2 and 0.5 % Triton X-100) × 3 min. and then aspirating the liquid phase into a fresh tube. The remaining, triton-insoluble lysates were obtained by adding cell lysis buffer to the remaining material, and processing above as for whole cell lysate. Samples were concentrated with Amicon filters, and equal amounts of lysate were resolved by 10% SDS-PAGE and transferred to Hybond membrane. Western blot was probed with rabbit polyclonal anti-GFP (Santa Cruz) at 1:3,000 followed by HRP-conjugated anti-rabbit (Pierce) at 1:1,000. Equal loading was confirmed by probing all blots with anti-β-tubulin (Covance) (not shown). Signals were visualized by the SuperSignal West Chemiluminscent Substrate kit (Pierce). Confocal images of Kc cells pre- and post-Triton-X100 treatment were collected from transfected cells grown on coverslips and then subject to the extraction procedure described above, followed by fixation in 4% paraformaldehyde and counterstaining with DRAQ5 (Biostatus Limited).
The abundance of the fly β-catenin homolog Armadillo (Arm) and the width of the Engrailed (En) stripe are readouts of Wg activity during segmentation of the germ band-extended Drosophila embryo (DiNardo et al., 1988; Martinez Arias et al., 1988; Peifer et al., 1994; Riggleman et al., 1990). By stage 10, the markedly elevated Arm levels in nkd embryos are restored to wild-type levels in alternate segments when NkdGFPC (Nkd fused to a C-terminal GFP tag) is produced via prd-Gal4, a Gal4 driver whose expression in a ~5–6 cell wide domain in alternate segmental anlagen is centered on the Wg stripe (Fig. 1A) (Waldrop et al., 2006; Yoffe et al., 1995). As a consequence of elevated Wg signaling in nkd mutants, the En stripe approximately doubles in width along the anterior-posterior axis (2.7+/−0.7 cell diameters for wild type vs. 5.9+/−1.4 cell diameters for nkd7H16 mutants) (cf. Fig. 1B, C). If nkd has cell-nonautonomous effects in the embryo, perhaps due to its ability to limit posterior, cell-to-cell transport of Wg (Moline et al., 1999), then we would expect to observe reduced or absent production of En in cells posterior to prd-expressing cells of nkd embryos in which NkdGFPC is driven by prd-Gal4 (nkd prd>NkdGFPC) due to reduced delivery of Wg to those cells. Occasional En-negative, prd-Gal4-expressing cells posterior to En-producing cells allowed us to test this hypothesis (Fig. 1B). As shown in Fig. 1C, En-positive cells are readily identified posterior to GFP-positive cells of nkd prd>GFP embryos. As expected, in nkd prd>NkdGFPC embryos, reduced to absent En was observed in ventral NkdGFPC-producing cells more than 2–3 cells posterior to the source of Wg, due to the normal role of nkd in repressing en in these cells (Martinez Arias et al., 1988). However, En-producing cells were consistently observed just posterior to the NkdGFPC-producing cells (Fig. 1D), indicating that those cells have received Wg input. The distribution of Wg protein around the endogenous wg stripe did not appear different between mutant and rescued segments of stage 10–11 nkd prd>NkdGFPC embryos (Fig. 1E, F–F′), suggesting that the reduced Arm in NkdGFPC-rescued cells (Waldrop et al., 2006) is due to a restored ability of those cells to appropriately respond to Wg. Cuticles secreted by nkd prd>NkdGFPC embryos exhibited a partial restoration in ventral denticle bands, consistent with a slight reduction in the number of Wg stripes in stage 11 nkd prd>NkdGFPC vs. nkd prd>GFP embryos (Fig. 1E, G, and not shown). Complete denticle bands in nkd prd>NkdGFPC cuticles had a wild-type assortment of denticle morphologies but subtle patterning abnormalities (cf. Supplementary Fig. 1A, B), consistent with En production in epidermal cells fated to give rise to denticles having no effect on denticle morphology (Bejsovec and Martinez Arias, 1991). These observations, coupled with a cell-autonomous inverse relationship between NkdGFPC levels and Arm levels in nkd prd>NkdGFPC embryos (Waldrop et al., 2006), refute the hypothesis that nkd has essential cell non-autonomous roles in the embryo.
Through stages 9–11, en transcription, whether in the presence or absence of nkd, requires Wg signaling (Bejsovec and Wieschaus, 1993; DiNardo et al., 1988; Dougan and DiNardo, 1992; Martinez Arias et al., 1988), raising the question of whether Wg ligand that maintains En production in the posterior cells of nkd prd>NkdGFPC embryos originates from the endogenous wg stripe that is several cell diameters anterior to those cells, or – the more likely scenario - from an ectopic wg stripe just posterior to those cells. First, we confirmed that nkd limits Wg-dependent Arm accumulation in cells anterior to en-expressing cells by inhibiting Wg signaling (and Arm accumulation) within en-expressing cells through expression of either a GFP-tagged version of the Wnt antagonist Axin (AxinGFP) (Cliffe et al., 2003) or a dominant-negative version of Pangolin, the transcriptional effector of Wg signaling (dTCFΔN) (van de Wetering et al., 1997) via en-Gal4 (Fietz et al., 1995); both gave similar results. (Because of the Wg-dependent nature of the expanded En domain in a nkd mutant, en-Gal4 mediated expression of a Wg antagonist creates a negative feedback loop that variably reduces the width of the En stripe as well as that of en-Gal4 itself, leading by stage 11 to patchy narrowing of the En domain.) As shown in Fig. 2A and B, in nkd en>AxinGFP embryos, cells anterior to the En domain retain high levels of Arm 4–5 cell diameters anterior to the wg/en boundary - evidence that nkd regulates Arm levels in cells anterior to En-producing cells.
During stage 10, the segment-polarity gene patched (ptc) is expressed in epidermal cells that do not express en, with the normally broad domain of ptc expression narrowed in a nkd mutant due to expanded en/hh expression and the repressive role of en on ptc transcription (Fig. 1A) (Hidalgo and Ingham, 1990; Hooper and Scott, 1989; Nakano et al., 1989). Since en-Gal4 and ptc-Gal4 are expressed in complementary domains (Fietz et al., 1995; Hinz et al., 1994), we investigated which abnormalities in nkd embryos could be rescued by expressing NkdGFPC using each Gal4 driver. As with prd-Gal4, En is consistently reduced or absent in ventral epidermal cells in the middle of the expanded en-Gal4 domain of nkd en>NkdGFPC embryos, but not of nkd en>GFP embryos (Fig. 3A, B). However, En continues to be produced in the most posterior en-Gal4-producing cells of nkd en>NkdGFPC embryos, again suggesting that those cells are close to a source of Wg, which we confirmed by staining with an antibody against Wg (Fig. 3C, D). We observed little to no difference in the number of parasegments that produce Wg ectopically in nkd en>NkdGFPC embryos as compared to nkd en>GFP or nkd ptc>GFP embryos (Fig. 3C, D, G), indicating that providing Nkd to en-expressing cells in nkd mutants does not prevent the appearance of ectopic Wg stripes. Conversely, nkd ptc>NkdGFPC embryos show a dramatic reduction in the number of ectopic ventral Wg stripes but retain wide En stripes (Fig. 3E–G). Reduced to absent En in cells in the middle of the expanded en domain in nkd en>NkdGFPC embryos is due to Nkd’s activity as a Wg antagonist, because a similar but more severe reduction in the number of posterior cells that produce En can be observed in nkd en>AxinGFP (Fig. 3H) or nkd en>dTCFΔN embryos (not shown). Accordingly, we observed a more complete restoration of ventral denticle belts and body length in nkd mutant cuticles when Nkd or dTCFΔN was expressed by ptc-Gal4 than by en-Gal4 (Fig. 3I, J). As with nkd prd>NkdGFPC cuticles, ventral denticle bands in nkd ptc>NkdGFPC and nkd ptc>dTCFΔN cuticles displayed mostly normal denticle morphology, but denticle bands of nkd ptc>dTCFΔN cuticles had marked disarray (Supplementary Fig. 1C, D), likely due to the distinct mechanisms by which Nkd and dTCFΔN inhibit Wg signaling. As suggested by the near complete loss of ventral En in nkd en>AxinGFP embryos (Fig. 3H), otherwise wild type en>AxinGFP embryos showed complete loss of ventral En in stage 11 embryos and developed cuticles with fused denticle belts and complete loss of ventral naked cuticle (not shown), likely due to the greater strength of misexpressed AxinGFP as a Wnt antagonist as compared to dTCFΔN or Nkd.
Previously we showed that fusion of a myc epitope tag or GFP to the Nkd C-terminus did not alter rescue activity during embryogenesis (Waldrop et al., 2006). In contrast, fusion of GFP to the Nkd N-terminus (NkdGFPN/mycC), but not internally just C-terminal of the EFX (NkdGFPX/mycC), eliminated Nkd activity in three assays: First, when ubiquitously expressed in nkd mutants by da-Gal4, NkdGFPX/mycC limited the En stripe to a width comparable to that of wild type embryos (2.9+/−0.8 cells) and rescued the nkd cuticle phenotype like NkdGFPC, but NkdGFPN/mycC had little to no rescue activity (En stripe = 5.3+/−1.4 cells) (Fig. 4A, B) (Waldrop et al., 2006). Second, when expressed by prd-Gal4, NkdGFPX/mycC but not NkdGFPN/mycC reduced Arm levels of stage 10 nkd mutants (Fig. 4C). Third, B119-Gal4-driven expression of NkdGFPX/mycC in pupal abdomen prevented the appearance of nearly all of the sternite bristles whose induction requires Wg signaling (Shirras and Couso, 1996), but NkdGFPN/mycC had no effect on sternite bristle quantity or pattern (Fig. 4D–G). Like endogenous Nkd and misexpressed NkdGFPC, NkdGFPX/mycC localized in a punctate, predominantly cytoplasmic and diffuse nuclear pattern in embryonic epidermal cells when detected by anti-GFP immunocytochemistry (Waldrop et al., 2006), but NkdGFPN/mycC was predominantly nuclear by stage 10 (Fig. 4A), indicating that Nkd N-terminal sequence is required to reside at the N-terminus in order for Nkd to function and to prevent Nkd from accumulating in the nucleus. Similarly, NkdΔN/GFPC, which lacks N-terminal sequence conserved in mosquito Nkd (Waldrop et al., 2006), was unable to rescue nkd mutant phenotypes (En stripe = 5.0+/−1.2 cells) and showed increased nuclear localization relative to NkdGFPC when examined either in embryos or in third instar larval salivary glands (Fig. 4A–C; cf. Supplementary Fig. 2A, B) (Waldrop et al., 2006).
Although fly Nkd does not have a N-terminal myristoylation consensus sequence, like the mammalian Nkd proteins it could act, in part, by associating with membranes. If so, then increased association of fly Nkd with the plasma membrane may enhance its ability to rescue a nkd mutant. To test this possibility, we fused the N-terminal 89 amino acids of Drosophila Src - which encodes a myristoylation consensus sequence and is sufficient to direct GFP to the plasma membrane - to a partially truncated but fully functional Nkd construct, NkdNBg/GFPC, to form NkdSrcNBg/GFPC (Fig. 5A) (Simon et al., 1985; Waldrop et al., 2006). As shown in Fig. 5A, nkd da>NkdNBg/GFPC embryos have En stripes of 1.9+/−0.8 cells wide, with rare, focal loss of En production, consistent with a slight enhancement of NkdNBg/GFPC potency relative to NkdGFPC or NkdGFPX/mycC (Waldrop et al., 2006). Consequently, 3.7% of cuticles in the NkdNBg/GFPC rescue cross had one or more fused denticle belts due to loss of the parasegmental boundary caused by focal En repression (Martinez Arias et al., 1988). However, nkd da>NkdSrcNBg/GFPC embryos have En stripes 3.7+/−1.0 cells wide and give rise to weak to moderate nkd cuticles, while nkd prd>NkdSrcNBg/GFPC embryos retain higher Arm levels relative to nkd prd>NkdNBg/GFPC (Fig. 5A–C). Similarly, B119>NkdNBg/GFPC adults had no sternite bristles, but B119>NkdSrcNBg/GFPC adults retained an intermediate number of sternite bristles (Fig. 5E, H). While the localization of NkdNBg/GFPC, like NkdGFPC, was predominantly punctate and cytoplasmic, NkdSrcNBg/GFPC also localized in intracellular aggregates (Fig. 5A), indicating that Nkd sequences counteract Src(1-89)GFP’s ability to localize at the plasma membrane.
NkdSrcG2ANBg/GFPC, in which myristoylation was inhibited by mutation of the myristoylated glycine to an alanine (G2A) (Kamps et al., 1985), showed very weak to absent nkd rescue (En stripe = 4.4+/−1.7 cells wide) and misexpression activity, but the fusion protein was undetectable by anti-GFP antibody in embryos or in transfected cells (Fig. 5A–C, F, H, and not shown), indicating that myristoylation also facilitates NkdSrcNBg/GFPC protein stability. Like NkdΔN/GPFC, NkdSrcG2ANBg/GFPC was localized predominantly in the nucleus when produced in larval salivary gland (Supplementary Fig. 2C), consistent with the N-terminal region of Nkd promoting protein stability and opposing nuclear localization.
We attribute the residual nkd rescue activity of NkdSrcNBg/GFPC and NkdSrcG2ANBg/GFPC to endogenous Nkd N-terminal sequences in the two constructs, because replacement of sequence N-terminal of the EFX motif with the Src(1-89) tag (NkdSrcRBg/GFPC) further reduced nkd rescue activity relative to NkdSrcNBg/GFPC (En stripe = 4.9+/−1.2 cells) but restored predominant localization to the plasma membrane (Fig. 5A–C). Replacement of the fly Nkd N-terminus with the myristoylation consensus sequence of mouse Nkd1 (to make Nkdm1NΔN/GFPC) also reduced nkd rescue activity relative to NkdGFPC (En stripe = 4.7+/−1.3 cells) and conferred localization to the plasma membrane (Fig 5A–C). NkdSrcRBg/GFPC may have dominant-negative properties, because B119>NkdSrcRBg/GFPC adults had increased numbers of sternite bristles relative to B119>GFP adults (Fig. 5G, H) (Rousset et al., 2002). These experiments demonstrate that fly Nkd and mouse Nkd1 N-termini, as well as a heterologous myristoyl tag, oppose Nkd nuclear localization, but that the fly Nkd N-terminus possesses unique activity that cannot be mimicked by myristoylation.
Mouse Nkd1, when misexpressed in otherwise wild-type flies, can induce weak loss-of-Wg signaling adult phenotypes (Wharton et al., 2001). Fly Nkd has a 30 amino acid (aa) motif, conserved in mosquito Nkd, that is crucial for function, and is sufficient, when substituted for the 30 aa motif in mouse Nkd1 (to form mNkd1f30aa/GFPC), to confer increased Wg antagonist activity and nuclear localization (Waldrop et al., 2006; Wharton et al., 2001). However, neither mNkd1GFPC nor mNkd1f30aa/GFPC had significant activity in nkd embryo rescue assays (En stripe = 5.1+/−1.2 and 4.7+/−1.2 cells, respectively) (Fig. 6A–C) (Waldrop et al., 2006). Importantly, replacement of mouse Nkd1 N-terminal sequence in mNkd1f30aa/GFPC with the fly Nkd N-terminus (to form mNkd1fNf30aa/GFPC) conferred partial embryonic nkd rescue activity when the double-chimeric protein was expressed by da-Gal4 or prd-Gal4 (Fig 6A–C). mNkd1fNf30aa/GFPC effectively reduced Arm levels in nkd mutants, but it only partially narrowed the En stripe to 3.6+/−1.0 cells, predominantly in the ventral region of most segments (Fig. 6A–C), indicating that sequences in fly Nkd that are not in mNkd1fNf30aa/GFPC are required for full En repression.
Given the importance of Nkd’s N-terminal region, we compared the subcellular localization and detergent solubility of the N-terminal 59 aa – a conserved block of sequence that is 75% identical and 86% similar to the mosquito Nkd N-terminus - fused to GFP (Nkd7H/GFPC) with GFP and Src(1-89)GFP (Waldrop et al., 2006). When expressed in third instar salivary glands via 71B-Gal4, GFP was diffusely cytoplasmic with some enrichment in the nucleus, while Src(1-89)GFP was localized exclusively to the plasma membrane (Fig. 7A, B). In contrast to GFP, Nkd7H/GFPC was enriched at plasma and nuclear membranes, with the remainder diffusely distributed (Fig 7C). Next, we expressed each protein in cultured Drosophila Kc cells and inferred their lipid solubility by determining the extent to which each protein could be extracted by brief, gentle exposure to the non-ionic detergent Triton X-100 (see Materials and Methods). As shown in Fig. 7D, under the extraction conditions employed, all of the GFP but little to no Src(1-89)GFP was Triton-extractable. Consistent with the observed localizations in salivary gland cells, most Nkd7H/GFPC was Triton X-100-extractable and - despite the presence of protease inhibitor cocktail during extraction - degradation-sensitive, but we consistently detected some Nkd7H/GFPC that was insoluble and degradation-resistant (Fig. 7D). The subcellular distributions of each protein in Kc cells was similar to that observed in salivary glands (Fig. 7E–G). Examination of Kc cells post-Triton X-100 treatment revealed residual Nkd7H/GFPC and Src(1-89)GFP in both plasma and internal membranes, whereas nearly all GFP was extracted from GFP-expressing cells (cf. Fig. 7H vs. I, J). Thus, like the mammalian Nkd proteins, the N-terminus of fly Nkd confers membrane association.
In this paper, we investigate the requirements for Nkd activity in each developing segment, as well as demonstrate critical functional and membrane-anchoring roles for fly Nkd’s unique N-terminus. Our studies provide further support for the hypothesis that Nkd limits intracellular responses to Wg signaling. In contrast to a prior report (Moline et al., 1999), we observe neither consistent nor dramatic differences in the distribution of Wg protein that originates from the endogenous wg stripe in stage 10–11 nkd-mutant vs. rescued segments. In this regard, it should be noted that in nkd embryos by stage 11 the endogenous stripe of Wg protein exhibits an anterior bias just as in wild type, whereas the ectopic Wg stripe adopts a mirror-image posterior bias (see Fig. 3C), with the diminution of Wg immunoreactivity in the wide en-domain likely due to En and/or Hh signal-dependent enhancement of Wg degradation as observed in wild type embryos (Sanson et al., 1999).
During stages 10–11, the expression pattern of nkd mRNA and Nkd protein resolves into a complex, Wg-dependent striped pattern repeated across each segmental anlagen (schematized in Fig. 8A) (Zeng et al., 2000). There, nkd acts via a negative feedback mechanism to limit signaling across each segment (Fang et al., 2006; Zeng et al., 2000). Consequently, in a wg mutant, nkd mRNA decays, while in a nkd mutant, the defective nkd transcript accumulates to higher levels and more broadly across each segment (Zeng et al., 2000). That ubiquitously produced Nkd can rescue a nkd mutant to adulthood suggested that striped nkd expression is not essential for activity, although the efficiency of rescue past embryogenesis was reduced relative to wild type (Waldrop et al., 2006). In this paper we demonstrate, by driving Nkd production in defined domains of each segmental anlagen of nkd mutants, that the early expansion of en and the later induction of the ectopic wg stripe are separable events. Because we observe neither any cell-nonautonomous effects on Wg-dependent Arm accumulation or En production in nkd mutants, nor any obvious differences in Wg distribution in nkd mutant vs. rescued segments, we conclude that the action of Nkd can be explained by its cell-autonomous ability to attenuate responses to Wg. The observed patterns, summarized in Fig. 8A, are as follows:
In nkd prd>NkdGFPC and nkd en>NkdGFPC embryos, we observed En-positive cells at the posterior of each “en-competent” domain (defined as the population of epidermal cells that express en in a nkd mutant (Martinez Arias et al., 1988)), which in the latter genotype were in close proximity to an ectopic Wg stripe a few cells anterior to the endogenous Wg stripe. With prd-Gal4, Nkd was not expressed in the most posterior cells of alternate en-competent domains, so En production in posterior cells was likely maintained by Wg produced by an ectopic wg stripe just posterior to those cells. The production of Nkd in posterior cells of each en-competent domain in nkd mutants, as in nkd en>NkdGFPC embryos, was not sufficient to reduce En levels in those posterior cells, indicating - as is observed with the endogenous 2–3 cell wide En stripe - that the relatively weak Wg-antagonist activity of Nkd is not sufficient to overcome the levels of signaling induced by the high levels of Wg ligand to which those cells are exposed. Expression of Axin, a more potent antagonist, via en-Gal4 in nkd mutants led to a more extensive yet still incomplete suppression of endogenous and ectopic En, particularly in ventral epidermal cells.
In contrast, nkd ptc>NkdGFPC embryos retained wide En stripes but developed far fewer ectopic ventral Wg stripes than nkd embryos. In contrast to nkd prd>NkdGFPC and nkd en>NkdGFPC, in nkd ptc>NkdGFPC embryos the most posterior En-positive cells are likely receiving Wg input, at least initially, from the endogenous Wg stripe that is just to the anterior, although our experiments do not allow us to rule out the possibility that later on those cells are also sensing Wg from the further posterior source, as previously hypothesized to occur in similarly staged nkd mutants (Moline et al., 1999). Embryos homozygous for some weaker nkd alleles show a phenotype similar to nkd ptc>Nkd embryos: posterior expansion of En but rare to absent ectopic Wg, a further indication that nkd-mutant cells at the posterior of each en-competent domain receive Wg input from the anterior, endogenous source of Wg. In addition to wg expression requiring input from Hh signaling, wg also autoregulates its own expression during germ band extension (Bejsovec and Martinez Arias, 1991; Yoffe et al., 1995). Perhaps in nkd ptc>NkdGFPC embryos the presence of Nkd in cells anterior to the endogenous wg stripe that are competent to express wg but only do so when exposed to high levels of Hh signaling, raises the threshold for activation of wg transcription by Wg and Hh, even though those cells are exposed to high levels of Hh. Nkd expression in the anterior ptc-expressing cells of nkd mutants was not 100% effective at preventing ectopic wg transcription, as 1–3 ecopic wg stripes per embryo were observed.
The distinct N-terminal sequences of fly and mammalian Nkd have raised the question of whether the Nkd N-termini have homologous functions in vivo. Previously, it was reported that mammalian Nkd2 is myristoylated and localizes to cell membranes, whereas Drosophila Nkd, lacking a myristoyl modification, was diffusely localized in the cytoplasm when misexpressed in cultured mammalian cells (Li et al., 2004). In this work, we show that the fly Nkd N-terminus confers membrane localization and in vivo activity that cannot be mimicked by heterologous membrane-targeting sequences. Taken together with our previous studies that have probed relationships between structure and function of Drosophila Nkd (Rousset et al., 2002; Waldrop et al., 2006), here we show that four regions of Nkd – N-terminus, EFX, 30 aa nuclear localization sequence, and histidine-rich C-terminus, conserved between fly and mosquito Nkd through ~250 million years – are sufficient for significant in vivo Nkd function. These data suggest that the four conserved regions constitute a minimal platform upon which additional domains and/or activities may have been acquired during evolution. Consistent with this idea, Nkd2 but not Nkd1 binds to the intracellular tail of TGF-α precursor, using a motif that is between the 30 aa motif and histidine-rich regions of Nkd2 (Li et al., 2004); and, by yeast two hybrid assay the B” protein phosphatase 2A (PP2A) subunit PR72 binds to sequences in Nkd1 between the conserved EFX and 30 aa motifs, that are not conserved in Nkd2 (C.-C.C. and K.A.W., unpublished observation) (Creyghton et al., 2005). Since the EFX – the only motif conserved in primary sequence between fly and mammalian Nkd proteins - binds Dsh or its three mammalian homologs (Wharton et al., 2001), we surmise that regulation of Dsh activity constitutes Nkd’s most ancient function.
Our data suggest that the action of Nkd at the plasma membrane in flies may be very similar to the presumed mechanism by which mammalian Nkd proteins inhibit Wnt signaling, but the means by which membrane association is achieved appears distinct in flies and mammals; for mammalian Nkd, a myristoyl anchor facilitates membrane association, while in the fly it is mediated by a N-terminal ~60 aa motif, possibly via protein-protein interactions. That each functionally important region of fly Nkd promotes localization to different subcellular regions – the N-terminus promoting membrane association, the EFX retaining Nkd in the cytoplasm, possibly via associations with Dsh, and the 30 aa motif promoting nuclear localization (Waldrop et al., 2006) – supports the hypothesis that Nkd acts at multiple subcellular locations to attenuate Wg signaling. However, our experiments do not distinguish between models whereby separate pools of Nkd act in concert to inhibit signaling (Fig. 8B, Model 1), or whether Nkd action requires intracellular transport between cytoplasm, intracellular membranes, and the nucleus (Fig. 8B, Model 2). Yet a third possibility, suggested by recent studies, is that Nkd impinges on the ability of Dsh to oligomerize and promote Wnt-receptor complex “signalosomes” that have been proposed to trigger intracellular signaling (Bilic et al., 2007; Schwarz-Romond et al., 2007), but such a model does not readily account for Nkd’s apparent activity in the nucleus. Future investigations that probe the subcellular dynamics of Nkd trafficking and identify additional Nkd-associated proteins will be required to generate additional clues about the mechanism by which Nkd attenuates Wnt/β-catenin signaling.
Special thanks to Matthew Scott for support and encouragement during the early stages of the Nkd structure-function analysis. Thanks to John Manak for the GFP plasmid used to make NkdGFPX/mycC; Matt Fish for injection of Src-Nkd fusion constructs; Jin Jiang and anonymous reviewers for critical reading of the manuscript; Marilyn Resh and Sandra Hofmann for advice regarding lipid-modified proteins; and DSHB and Nipam Patel for antibodies. K. W. is a W.W. Caruth, Jr. Scholar in Biomedical Research at UT Southwestern and has received support from the Texas Affiliate of the American Heart Association; institutional support from the American Cancer Society, UT Southwestern Medical Center and the Department of Pathology; and NIH R-01 GM65404.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.