Dachs: an unconventional myosin that functions downstream of Fat to regulate growth, affinity and gene expression in Drosophila

Adult structures of the head, thorax and genitalia in Drosophiladevelop from clusters of undifferentiated cells in the larva called imaginal discs. The discs grow throughout the three larval instars, and then evert and differentiate during pupal development to form the adult cuticle. The growth and patterning of imaginal discs is governed by highly conserved intercellular signaling pathways (reviewed by Irvine and Rauskolb, 2001; Klein,2001; Lawrence and Struhl,1996), including Notch, Wnt, Egf, Hedgehog and Bmp pathways. Imaginal discs have proven to be an outstanding model for identifying and characterizing components of these pathways. Recent work has led to the suggestion that disc growth and patterning are also dependent upon a new signaling pathway, the Fat pathway.

fat encodes a large protocadherin(Mahoney et al., 1991). Genetic studies in Drosophila have identified three crucial requirements for fat during imaginal disc development. First, fat is required to limit wing growth, as mutation of fatcauses disc overgrowth (Bryant et al.,1988). Second, fat acts cell-autonomously to influence an intercellular signaling process between distal and proximal wing cells that establishes a ring of Wingless (Wg) expression in the proximal wing(Cho and Irvine, 2004). Third, fat plays a crucial role in the establishment of tissue polarity(Casal et al., 2002; Rawls et al., 2002; Strutt and Strutt, 2002; Yang et al., 2002).

The molecular basis for the influence of fat on these processes has not been determined, but genetic studies have identified genes that function together with fat (Adler et al., 1998; Casal et al.,2002; Cho and Irvine,2004; Rawls et al.,2002; Strutt and Strutt,2002; Yang et al.,2002). Two, four-jointed (fj) and dachsous (ds), act genetically upstream of fat in the regulation of tissue polarity (Yang et al., 2002), and act non-autonomously to influence the expression of Wg in the proximal wing (Cho and Irvine, 2004). ds encodes a protocadherin(Clark et al., 1995) that appears to participate in heterophilic interactions with Fat(Ma et al., 2003; Matakatsu and Blair, 2004; Strutt and Strutt, 2002). These observations suggest that Ds and Fat might bind each other to mediate intercellular signaling, with Ds acting as a ligand and Fat as a receptor. fj encodes a protein found in both secreted(Villano and Katz, 1995) and Golgi-resident (Strutt et al.,2004) forms, and might influence Fat-Dachsous interactions.

A third gene that has been genetically linked to fat is dachs, which was first described by Bridges and Morgan(Bridges and Morgan, 1919). The original dachs mutant allele, d¹, results in reduction of wing and leg growth similar to that observed in alleles of fj and ds (Waddington,1940; Waddington,1943). dachs also interacts genetically with fjto influence leg segmentation and growth(Buckles et al., 2001),suggesting that its function is related to these Fat pathway components. More recently, dachs was shown to be epistatic to fat both for the regulation of wg expression in the proximal wing, and for imaginal disc growth (Cho and Irvine,2004). These observations suggest that dachs might act as a downstream component of Fat signaling. However, the nature and extent of the requirement for dachs has not been well understood because only one allele was characterized, and the molecular nature of Dachs had not been described.

Here, we present the first phenotypic characterization of strong mutations in dachs. These mutations define requirements for dachs in different fat-dependent processes. We also show that dachsencodes an unconventional myosin, and characterize its subcellular localization. Finally, we show that the localization or stability of Dachs at the membrane is influenced by Fat signaling, thus providing molecular evidence that Dachs is a downstream component of a Fat signaling pathway.

The d²¹⁰, d^GC2 and d^GC13alleles were the kind gift of F. Michael Hoffmann. For male recombination(Chen et al., 1998), a stock was constructed to contain two flanking markers for the area of interest, black (b) and clot (cl), as well as a source of P transposase (Robertson et al.,1988). w; cl¹ d¹ b¹/Cyo; SbΔ2-3/TM6B flies were crossed to a set of P element insertion stocks (Fig. 5A). Recombination in males is confined to these P insertion sites. For each P-element, 22-28 crosses were set up and 1500-3000 progeny were screened. dachs phenotypes were analyzed using the d¹,d²¹⁰, d^GC2 and d^GC13alleles, in homozygous, transheterozygous, and hemizygous combinations, using Df(2L)N22-5 and Df(2L)ED623. We also examined d^GC13 ck¹³, fat⁸, fat^G-rv,d^GC13 fat⁸, d¹ fat⁸,d^GC13 fat^G-rv, fj^d1, ds^UA071,and ds^36D mutant animals.

UAS-d insertions were isolated on the second (UAS-d[2D],UAS-d[2A]) and third (UAS-d[2B]) chromosomes. UAS-d:V5insertions were isolated on the second (UAS-d:V5[50-5],UAS-d:V5[18-4]) and third (UAS-d:V5[9-F], UAS-d:V5[18-2],UAS-d:V5[8-3]) chromosomes. Rescue and overexpression experiments were conducted by crossing to tub-Gal4, arm-Gal4, C765Gal4, ptc-Gal4 or da-Gal4. Ectopic expression clones were created by Flp-out using the UAS-d and UAS-d:V5 lines, as well as the following stocks: y w hs-Flp[122]; act>y⁺>Gal-4 UAS-GFP(AyGal4); UAS-fj; GS-ds; UAS-fat; UAS-fat; UAS-d:V5[9F]; UAS-fj; UAS-d:V5[9F]; and GS-ds; UAS-d:V5[9F].

Simple mutant clones were created using FLP-FRT mediated recombination,with the following stocks: y w; d^GC13 FRT40A/CyO-GFP; y w; d¹ FRT40A/CyO-GFP; y w; fat⁸FRT40A/CyO-GFP; y w; fat^G-rv FRT40A/L14; y w;d^GC13 fat⁸ FRT40A/CyO-GFP; d^GC13ck¹³ FRT40A/CyO-GFP; y w; d¹ fat⁸FRT40A/CyO-GFP; y w hs-flp[122]; Ubi-GFP FRT40A; w hs-flp[122] f; M(2)25A P[f⁺30B] FRT40A; y w hs-flp[122];y⁺ FRT40A/CyO; y w hs-flp[122]; M(2)25A Ubi-GFP FRT40A/CyO.

To examine Dachs:V5 expression in fat mutant clones we used y w;ft^G-rv FRT40A; UAS-d:V5[9F]/L14 × y w hs-FLP;Ubi-GFP FRT40A; C765Gal4 and related stocks, substituting UAS-d:V5[18-2], tub-Gal4 or arm-Gal4.

For RFLP analysis of dachs mutant DNA, probes were prepared by XL PCR (Perkin-Elmer) based on sequence in GadFly across the 70 kb interval indicated in Fig. 5. Candidate genes were then amplified by PCR from wild-type and mutant DNA.

The identification of dachs was communicated to GenBank by F.K. prior to publication (Accession Number AF405293). FlyBase(Drysdale and Crosby, 2005)currently lists three transcripts for dachs, d-RA, d-RB and d-RC. d-RA (Accession Number NM_175991) differs slightly from the original submission of AF405293, but is consistent with BDGP EST sequences and our own more recent sequence analysis of cDNAs and RT-PCR products, and encodes the product depicted in Fig. 5; the entry for AF405293 has thus been corrected to match d-RA. d-RB and d-RC correspond to Myo29D transcripts reported by Tzolovsky et al.(Tzolovsky et al., 2002), but we were unable to detect them by RT-PCR from larval total RNA. dachscDNAs were recovered from a third instar cDNA library enriched for imaginal discs (Brown and Kafatos,1988). We also prepared cDNA by RT-PCR from total RNA using either Superscript II reverse transcription (GibcoBRL) or OneStep RT-PCR kit(Qiagen). For construction of UAS-dachs, cDNA fragments were prepared by RT-PCR and then cloned into pGEM-T (Promega) to make a cDNA corresponding to nucleotides 52-3915 of d-RA, which was then verified by DNA sequencing. Plasmid pUAST-d was constructed by cloning nucleotides 71-3915 of dachs into pUAST using EcoRI and NotI; the start codon begins at nucleotide 71. A CCACC Kozak sequence was introduced in front of the ATG. V5-tagged Dachs was created by replacing the stop codon and 3′ UTR of dachs with V5 and His₆ sequences from pMTBip/V5-His (Invitrogen).

Imaginal discs were fixed and stained as described previously(Cho and Irvine, 2004), using as primary antibodies rabbit anti-V5 (1:1000, Novus), mouse anti-V5 (1:200,Invitrogen), rat anti-Fat intracellular domain (1:100. H. McNeill), mouse anti-Wg (1:800, 4D4, DSHB), rat anti-Ser (1:1000)(Papayannopoulos et al.,1998), goat anti-β-gal (1:1000, Biogenesis), rat anti-Elav(1:40, 7E8A10, DSHB), mouse anti-Dac (1:40, DSHB), rat anti-DE-Cad (1:40,DSHB), mouse anti-Dll (1:400) (Duncan et al., 1998) and mouse anti-Prospero (1:50, DSHB).

In situ hybridization was carried out as described previously(Rauskolb and Irvine, 1999),using a labeled antisense RNA probe corresponding to nucleotides 1710-3915 of the predicted dachs transcript.

Western blotting was performed on larval tissue (wing, leg and haltere discs, attached to fragments of cuticle) boiled in SDS-PAGE loading buffer. The Dachs:V5 was detected with mouse anti-V5 (Invitrogen).

Mutant clones in adult flies were generated and analyzed as described by Hao et al. (Hao et al., 2003),except that flies were heat-shocked 0-72 hours AEL. Pupal legs were dissected in PBS, fixed for 45 minutes in PLP (McClean and Nakane, 1974), and mounted in fluorescent mounting medium. Polarity was scored in male abdominal segments 2-4. Mutant clones were marked with yellow, which is only scorable in bristles; thus, we did not distinguish between autonomous and non-autonomous effects in these experiments. We scored only clones that included more than one bristle; the severity of the polarity phenotype did not correlate with clone size among clones analyzed (2-11 bristles).

The original dachs allele, d¹, is viable, with shortened legs and wings (Bridges and Morgan, 1919; Waddington,1940; Waddington,1943). The reduction in wing length is most obvious in the middle of the wing, as evidenced by the reduced distance between the cross veins(Fig. 1B, compare with 1A). As each cell in the wing normally secretes a single hair, the number and spacing between hairs serves as a measure of cell number and size. Counting hairs between the crossveins in d¹ mutants revealed that the reduction in inter-crossvein distance is associated with a corresponding decrease in cell number (not shown).

To better define requirements for dachs, additional alleles were characterized. The strength of these alleles was then evaluated by examining the phenotypes of homozygous, hemizygous and transheterozygous animals. This analysis identified two alleles, d²¹⁰ and d^GC13, that are stronger than d¹, and another allele, d^GC2, that is weak like d¹ (Figs 1, 2; data not shown). Animals homozygous for d²¹⁰ die by early larval stages, but this likely results from other mutations on the d²¹⁰chromosome, as d^GC13 homozygotes survive until the end of pupal development, yet d²¹⁰ behaves similarly to d^GC13 in combinations with d^GC13 and d¹ (Figs 1, 2; data not shown), and molecular analysis (see below) predicts that d^GC13 encodes a more severely truncated protein than d²¹⁰. d^GC13 animals make adult tissue, but usually fail to eclose from the pupal case (less than 1% of animals eclose), forming pharate adults that can be dissected out and examined. The d^GC13phenotype is similar in homozygotes and in transheterozygous combinations with dachs deficiencies (Figs 1, 2; data not shown), suggesting that it could be a null mutation.

The legs of d^GC13 mutants are more severely affected than d¹ mutants (Fig. 2). All of the intermediate and distal segments of the leg (femur through tarsus) are noticeably shortened, and some tarsal segments are fused,typically resulting in the formation of only three tarsal segments instead of the normal five. Although each of the genes in the Fat pathway influence leg development, their phenotypes are different, presumably reflecting their distinct roles within the pathway (see Fig. S1 in the supplementary material). In addition to a general reduction in length, in some cases no external leg tissue is evident in d^GC13 mutants, or the leg appears to form a single mass of tissue. Pupal legs, however, are consistently shortened but never absent or severely deformed. Examination of pupae suggested that the legs absent phenotype actually derives from defects in the morphological changes that occur during leg disc eversion(Fig. 2L; data not shown). Indeed, direct examination of d^GC13 pharate adults revealed that when external legs are absent, masses of leg tissue could be found differentiating within the body cavity(Fig. 2M). The disc eversion phenotype was also observed in d^GC2, but not in d¹. To further define the dachs leg phenotype, we also characterized d^GC13 mutants clones. These caused reduced growth and fusions of tarsal segments similar to those observed in mutant animals, but even large mutant clones were not associated with a failure of disc eversion (not shown).

d^GC13 wings are smaller than d¹ or wild-type wings (Fig. 1). They also exhibit abnormal wing patterning, as evidenced by a variable phenotype that includes extra, missing and mis-positioned crossveins(Fig. 1). Extra or missing crossveins are also often observed in adult flies with d^GC13 mutant clones, and in d¹/d^GC13,d¹/d²¹⁰, or d¹/Df animals(Fig. 1; data not shown).

In wing imaginal discs, clones of cells mutant for fat are associated with upregulation of Wg expression in the proximal wing(Cho and Irvine, 2004). Conversely, clones of cells mutant for d¹ exhibit a severe loss of Wg expression in the proximal wing, but this loss of Wg is transient(Cho and Irvine, 2004). To investigate whether the transience of Wg loss in d¹results from remaining dachs activity, we examined d^GC13 mutant clones, but their affect on Wg expression was similar - Wg is severely reduced at early third instar, but essentially normal by late third instar (not shown).

Clones of cells mutant for fat have been reported to be associated with upregulation of fj expression in eye imaginal discs(Yang et al., 2002), and we have observed that fat mutant clones are also associated with induction of fj expression in leg (not shown) and wing(Fig. 3D) imaginal discs, as assayed by a fj-lacZ reporter. Importantly then, clones of cells mutant for d^GC13 are associated with decreased fjexpression in eyes and wings (Fig. 3A,C). However, the influence of dachs on fjexpression was detected during early and mid-third instar, but not late third instar.

Loss of leg joints and reduced leg growth is characteristic of mutations in genes in the Notch signaling pathway. To evaluate the potential relationship between dachs phenotypes and Notch signaling, we investigated the expression of the Notch ligand Serrate (Ser) in discs containing d^GC13 mutant clones. During third instar, loss of Ser could not be detected (Fig. 3F). Some loss of Ser expression could be detected in pupal legs in regions where dachs mutation is associated with leg segment fusions (Fig. 3I), but it is not clear in this context whether loss of Ser is a cause or a consequence of the loss of leg tissue. Nonetheless, Fj(Buckles et al., 2001) and Ds(data not shown) can induce the expression of Ser and Delta in neighboring cells when ectopically expressed in the leg. These observations suggest that the Fat pathway does have a role in regulating Notch ligand expression during leg development. To further evaluate this, we examined Ser expression in clones of cells mutant for fat. fat mutant clones could be associated with ectopic Ser expression, although this was preferentially observed in the proximal leg (Fig. 3G). Within the proximal leg (defined for these experiments as proximal to the dachshund expression domain), 36/47 fat clones exhibited an obvious upregulation of Ser expression. Thus, fat has a normal role in repressing Ser expression during leg development. An influence of fat on Notch ligand expression probably accounts for the occasional outgrowths of leg tissue observed in fat mutants and in association with fat mutant clones (Bryant et al., 1988; Mahoney et al.,1991) (Fig. 2I), as these outgrowths appear similar to those observed upon ectopic Notch activation (Rauskolb and Irvine,1999).

Prior work has identified a set of broadly expressed genes, the leg gap genes, that are responsible for the initiation of Notch ligand expression in the leg (Rauskolb, 2001). To position the action of Fat within the leg segmentation hierarchy, we examined the expression of two key leg gap genes, dachshund and Distal-less. Both of these genes were expressed normally within fat mutant clones (not shown), suggesting that the action of Fat in leg segmentation is downstream of these broadly expressed genes.

dachs suppresses the consequences of fat mutation on wing disc growth (Cho and Irvine,2004). Moreover, even though dachs mutation does not lead to permanent loss of Wg expression in the proximal wing, it nonetheless completely suppresses the ability of fat mutation to induce ectopic Wg expression (Cho and Irvine,2004) (Fig. 4). To further evaluate the hypothesis that dachs is a general downstream component of a Fat pathway, we evaluated the ability of dachs to suppress additional fat mutant phenotypes, and to suppress fat phenotypes in other tissues.

Even though mutation of dachs has little effect on Ser expression in the legs of otherwise wild-type animals, dachs completely suppresses the Ser expression induced by fat mutation(Fig. 3H, 0/37 fat dachs double mutant clones in the proximal leg exhibited upregulation of Ser). Additionally, mutation of dachs suppresses the ability of fat mutant clones to induce outgrowths in adult legs(Fig. 2J: 12/50 legs with fat mutant clones had outgrowths, but 0/62 legs with fat dachs clones had outgrowths). dachs also suppresses the induction of fj expression associated with mutation of fat(Fig. 3E).

Another striking feature of fat mutant clones is their roundness(Fig. 4A). Normally, clones of cells adopt irregular, elongated shapes. However, genetic manipulations that influence cell affinity cause clones to be rounder and smoother. Strikingly,mutation of dachs also suppresses the roundness of fatclones (Fig. 4C, see Table S1 in the supplementary material), suggesting that dachs is required for an altered affinity of fat mutant clones. A difference in cell affinity probably also accounts for the appearance of internal vesicles of cuticular tissue within the legs of animals containing fat mutant clones (Fig. 2I: 40/50 legs with fat clones had internal vesicles). The appearance of these vesicles is suppressed by mutation of dachs(Fig. 2J: 0/62 legs with fat dachs clones had internal vesicles). Examination of clone size also confirmed that the suppression of fat mutant overgrowth by dachs is observed not only at the level of the whole disc(Cho and Irvine, 2004), but also within individual clones (Fig. 4C, see Table S1 in the supplementary material). Altogether, these results indicate that dachs is epistatic to fat for multiple phenotypes, including gene expression, growth and cell affinity, and in multiple tissues.

Another crucial function of fat is to regulate tissue polarity(Casal et al., 2002; Rawls et al., 2002; Strutt and Strutt, 2002; Yang et al., 2002). To investigate the possibility that dachs influences polarity, we examined d^GC13 mutant animals and d^GC13 mutant clones. Planar polarity is evident throughout most of the adult cuticle in the polarized orientation of hairs and bristles. In fat mutants, the normal orientation is disturbed, and swirling patterns of hairs and bristles occur in many tissues (e.g. Fig. 5B). It has been reported that d¹ can have mild polarity phenotypes in the leg(Held et al., 1986). We examined d^GC13 for polarity phenotypes in wings, legs,abdomens and eyes. Only mild polarity phenotypes were observed. As in wild-type animals, hairs in the abdomen point posteriorly(Fig. 5D), most leg bristles and wing hairs point distally (Fig. 2, Fig. 5I), and most ommatidia are correctly oriented (Fig. 5K).

The relatively mild polarity phenotypes of dachs contrast with the strong polarity phenotypes of fat, and thus we investigated double mutant animals to determine whether dachs is also required for the influence of fat on polarity. For these studies, we focused on eyes and abdomens. In the abdomen, fat mutant animals exhibit extensive swirling (Fig. 5B)(Casal et al., 2002), dachs mutants have no obvious polarity phenotype(Fig. 5D) and fat dachs double mutants exhibit an intermediate phenotype(Fig. 5E), in which there is some irregularity in the orientation of abdominal hairs, but they are less disturbed than in fat mutants. To provide a quantitative comparison,we scored disturbances in polarity associated with clones of mutant cells in abdominal segments. All (39/39) fat clones had polarity phenotypes(Fig. 5C), whereas only 3%(2/60) of dachs clones (Fig. 5G) and 54% (37/68) of fat dachs clones(Fig. 5F) exhibited polarity phenotypes. There was also a difference in the severity of the polarity phenotypes, as all of the fat clones were associated with some hairs at an angle greater than 90° relative to the normal orientation, whereas 26% (18/68) of fat dachs clones and 0% of dachs clones were associated with hairs at an angle of 90° or greater. In the remaining fat dachs clones (19/68), the degree of change in polarity was less severe (Fig. 5F). Thus,mutation of dachs partially suppresses polarity phenotypes associated with mutation of fat in the abdomen.

Polarity in the eye is manifest in the specification of chiral forms of ommatidia, which differ in their orientation and placement of photoreceptor cells. Staining with a neural marker (Elav) and an R7 marker (Prospero)illustrates the regular polarized orientation of ommatidia in the eye(Fig. 5J). In d^GC13, some disorganization in ommatidial orientation is observed, indicative of a mild polarity phenotype, but overall polarity is again largely normal (Fig. 5K). To examine the relationship of dachs to fat, we again focused on animals containing clones of mutant cells. fat mutant clones are consistently associated with strong polarity phenotypes(Rawls et al., 2002; Strutt and Strutt, 2002; Yang et al., 2002), which include completely reversed ommatidia (Fig. 5M, 32/33 fat clones included ommatidia rotated more than 90° away from normal), whereas dachs mutant clones exhibited mild polarity phenotypes, with only slightly mis-rotated ommatidia(Fig. 5L, 1/32 dachsclones included ommatidia rotated more than 90° away from normal). fat dachs double mutant clones appear similar to fat(Fig. 5N, 28/33 fat dachs clones included ommatidia rotated more than 90° away from normal).

Deficiency mapping and male recombination localized dachs to 29D1-2, between the most distal P element used for male recombination mapping(7704) and the distal end of a small non-complementing deficiency[Df(2L)N22-5; Fig. 6A]. Correlation with the genomic map(FlyBase, 1999) gave an interval of ∼70 kb, containing 14 potential transcripts. We used RFLP analysis throughout this interval to identify a single predicted gene, CG10595, which contained an internal deletion in d^GC2, and a large insertion in d¹ (not shown). Subsequent analysis showed that this same candidate gene contained mutations in two other d alleles (see below), providing strong evidence that this locus encodes dachs.

Analysis of dachs cDNAs and genomic sequences defined seven exons,encoding a predicted protein of 1232 amino acids(Fig. 6). The predicted protein is a member of the Myosin family, identified independently as Myo29D(Tzolovsky et al., 2002). The central region of Dachs includes all of the defining features of a myosin head domain, including a well-conserved ATP binding domain, actin-binding domain and active thiol region (Fig. 6). However, in the middle of the head domain, just before the actin-binding domain, is an unusual insertion of 44 amino acids that is not found in any other Myosin. By blast analysis, the closest mammalian myosins are members of the myosin V, X, and VII families. However, Dachs is not clearly orthologous to any of these mammalian myosins, as the Drosophila genome encodes other genes that appear to be more closely related to mammalian myosins V, X, and VII(Tzolovsky et al., 2002).

In addition to the head domain, unconventional myosins sometimes have an N-terminal extension preceding the head domain, which is characteristic for each class (Korn, 2000). Dachs has an N-terminal extension of 235 amino acids that does not have significant similarity to other proteins. As in other myosins, the head domain is followed by neck and tail domains. In Dachs, the neck domain contains a single calmodulin-type IQ-like motif (Fig. 6), the binding site for regulatory light chains. Dachs also encodes a tail domain of 187 amino acids, which does not show extensive similarity to any other proteins. Surprisingly, sequence analysis identifies a potential transmembrane domain C-terminal to the IQ motif, with a predicted type II orientation. This computer prediction has to be treated cautiously, as there is no precedent for transmembrane myosins.

We sequenced the four dachs mutant alleles(Fig. 6). d^GC2 contains a deletion that removes part of the N-terminal extension while retaining the reading frame. d¹contains an insertion of the blood retrotransposon(Bingham and Chapman, 1986) in the tail domain. blood insertions have been reported to affect transcript stability in other genes(Bingham and Chapman, 1986). d^GC13 and d²¹⁰ are both predicted to encode proteins truncated within the head domain. d^GC13contains an 11 bp deletion near the N terminus of the head domain, causing a frame shift followed closely by a stop codon, while d²¹⁰contains a point mutation that results in a stop codon in the active thiol region (Fig. 6). If the myosin head domain is required for dachs function, then both of these mutations would be expected to encode non-functional alleles.

To gain further insight into potential functions of dachs, dachsexpression was examined by in situ hybridization to embryos and imaginal discs. dachs mRNA is expressed broadly throughout embryonic and imaginal development, although at certain stages some local upregulation of dachs expression was observed(Fig. 7C). Early embryonic expression of dachs was near background until stage 9(Fig. 7), which suggests that there is no significant maternal contribution of dachs mRNA.

Based on the identification of Dachs as a myosin, we considered the possibility that dachs might be partially redundant with other Drosophila myosins. In particular, crinkled encodes a Drosophila Myosin VII family member(Kiehart et al., 2004), and exhibits a multiple wing hair phenotype, which is also sometimes observed in tissue polarity mutants. To evaluate the possibility of redundancy between dachs and crinkled, we examined double mutant clones. These displayed both dachs and crinkled phenotypes, but the phenotypes were not obviously more severe than in the respective single mutants. Thus, loss of Wg expression in the proximal wing was still transient(not shown). Moreover, clones of cells mutant for strong tissue polarity mutants, like fat, can influence the polarity of surrounding wild-type cells (Casal et al.,2002), but the polarity of hairs surrounding dachs crinkled mutant clones in the abdomen appeared normal(Fig. 5H).

Both to confirm the identification of the dachs gene, and to evaluate the consequences of Dachs overexpression, we created transgenic animals in which a dachs cDNA was expressed under UAS-Gal4 control. When dachs is expressed at moderate levels using a tub-Gal4driver, apparently normal flies develop. Importantly, expression of dachs under tub-Gal4 control can result in substantial rescue of d^GC13 mutants, confirming the identification of dachs. Both untagged and V5-tagged forms of Dachs were constructed;V5-tagged Dachs can also rescue dachs mutations(Fig. 1I, Fig. 2G), indicating that the V5 tag does not significantly impair Dachs function.

High level expression of dachs, achieved using different UAS-dachs insertions, or by raising animals at higher temperature,can result in disruptions of normal wing and leg development. In the wing,Dachs overexpression increased wing size and resulted in vein abnormalities(Fig. 1J), while in the leg occasional outgrowths of leg tissue and formation of internal vesicles were observed (Fig. 2H). These phenotypes resemble those observed in weak, viable alleles of fat(Bryant et al., 1988),consistent with the possibility that downregulation of dachs is a crucial fat function. Examination of Dachs-expressing clones in adult abdomens did not reveal any effects on polarity (not shown).

As attempts to raise anti-sera against Dachs were unsuccessful, we used V5 epitope-tagged Dachs (Dachs:V5) to characterize the localization of Dachs protein. Immunostaining of tub-Gal4 UAS-dachs:V5 or Ay-Gal4 UAS-dachs:V5 animals revealed low-level, diffuse and punctate staining for Dachs in the cytoplasm, and higher levels near the adherens junctions(Fig. 8A,H; data not shown).

Genetic studies suggest that Dachs acts as a downstream component of a Fat signaling pathway. However, it was conceivable that Dachs could instead act in parallel to Fat. To further explore the functional connection between Dachs and Fat signaling, we examined Dachs and Fat protein localization under conditions where Fat pathway activity was altered, either by mutation of fat, or by overexpression of Fj, Ds or Fat.

Over-expression of either Ds or Fj in clones of cells can result in elevated Fat protein staining at the edge of the clone, with the strongest effects observed where the endogenous expression levels are low (i.e. where there is the greatest contrast between endogenous and ectopic expression)(Cho and Irvine, 2004; Ma et al., 2003; Matakatsu and Blair, 2004)(Fig. 8D,E). However, when Ds is overexpressed, Fat appears depleted from other membranes of neighboring cells (Fig. 8E,F), suggesting that the elevated Fat staining at the edge of Ds-expressing clones represents Fat protein in neighboring cells recruited to cellular interfaces where it can bind higher levels of Ds (Fig. 9D) (Cho and Irvine,2004; Matakatsu and Blair,2004). Conversely, when Four-jointed is overexpressed, Fat staining can appear partially depleted from other membranes of cells just inside the clone border (Fig. 8C,D). This suggests that the elevated Fat staining at the edge of Fj-expressing clones represents Fat protein recruited from cells inside the clone (Fig. 9C).

Because Dachs:V5 protein was expressed under UAS-Gal4 control, Dachs:V5 staining in these experiments was specifically detected within Ds- or Fj-expressing clones. Strikingly, these manipulations had opposite affects on Dachs. Overexpression of Ds was associated with elevated Dachs:V5 staining at the edges of the clone, while overexpression of Fj was associated with loss of Dachs:V5 staining at the edges of the clone(Fig. 8C,F). Given the Fat staining profile, these observations suggest that accumulation of Fat protein at the membrane inhibits the localization or stability of Dachs. This inference is consistent with the genetic epistasis of dachs to fat, and is further supported by the observation that removal of fat within clones of cells was associated with an increase in Dachs staining at the membrane (Fig. 8G), whereas overexpression of Fat was associated with a decrease in Dachs staining at the membrane (Fig. 8B). When Dachs:V5 levels in wild-type and fat mutant wing discs were compared by western blotting, there was no detectable difference (Fig. 8I). Thus, if Dachs stability is affected by Fat, Dachs at the membrane must represent only a small fraction of the total protein. As this does not appear to be the case(Fig. 8A,H), we suggest that the influence of Fat on Dachs protein staining most probably reflects protein re-localization.

Fj is expressed most strongly by distal wing cells, while Ds is expressed most strongly by proximal wing cells(Brodsky and Steller, 1996; Clark et al., 1995; Villano and Katz, 1995). Thus,if Fj and Ds normally influence Dachs protein localization, there might be some asymmetry in Dachs localization even in wild-type animals. Within the middle of clones expressing Dachs:V5, polarization of staining can not be assessed because the resolution of confocal microscopy can not distinguish which of two neighboring cells is contributing membrane staining. However, at the edges of Dachs:V5-expressing clones, all staining observed comes from a single cell. Importantly then, examination of clones of Dachs:V5-expressing cells in the wing pouch revealed that Dachs:V5 staining is generally stronger along the distal sides of the clones, and weaker along the proximal sides(Fig. 8A; 64/79 clones were scored as having elevated Dachs on the distal side, 15/79 were scored as having no significant difference, and 0/79 were scored as having stronger Dachs:V5 on the proximal side). This observation implies that asymmetric localization of Dachs in response to regulation of Fat occurs during normal development, and at endogenous expression levels of Fj and Ds.

The observation that a hypomorphic mutation of dachs could suppress the effects of fat mutations on wing growth and Wg expression in the proximal wing led to the suggestion that dachsmight act as a downstream component of a Fat signaling pathway(Cho and Irvine, 2004). Here,we have provided two types of evidence that confirm this suggestion. First, dachs is epistatic to fat for multiple phenotypes in multiple tissues, including gene expression, growth and cell affinity. Indeed,with the notable exception of the influence of fat on tissue polarity, all known fat mutant phenotypes are completely suppressed by mutation of dachs. Second, we found that expression of regulators of Fat, Fj and Ds, or of Fat itself, influence the localization or stability of Dachs protein at the membrane, thus providing a molecular link from Fat to Dachs.

The predicted structure of Dachs is unique within the myosin superfamily,and places Dachs in a new class of unconventional myosins. It has most similarity to myosins V, VII, and X. This is intriguing, as a mammalian protocadherin, Cdh23, has been functionally linked to myosin VIIa during the development of sensory hair cells in the inner ear(Boeda et al., 2002).

Within the myosin head region, the major conserved domains are all present,suggesting that Dachs functions as a motor protein. However, it is also possible that Dachs serves a structural or scaffolding role. For example, in the Hedgehog pathway, a kinesin-related protein, Costal2, is thought to function largely as a scaffold that brings together crucial kinases with their substrates (reviewed by Ogden et al.,2004).

The d^GC2 mutation deletes part of the N terminal extension. As d^GC2 mutants have relatively weak phenotypes, the N terminal extension might not be not essential for Dachs activity. Conversely, the severe phenotypes of alleles that truncate Dachs in the myosin head region imply that the myosin domain is essential. d^GC13 in particular is predicted to eliminate almost all of the myosin head domain, and genetically it appears to act as a null allele.

Characterization of new dachs alleles has provided an opportunity to define more clearly the requirements for dachs. dachs is required for normal wing and leg growth, although some appendage growth is dachs independent. Importantly, the identification of dachsas a downstream component of a Fat signaling pathway that influences growth implies that the reduced growth in dachs mutants is reflective of a normal role for a Fat pathway in growth promotion. That is, while fatis a gene whose normal role can be thought of as to restrain growth, as mutant tissue overgrows, we suggest that inhibition of Fat occurs during normal development, and that this inhibition contributes to normal appendage growth,as defined by the reduced growth of dachs mutants. Normal inhibition of Fat activity would presumably be effected by the two known regulators of Fat, Fj and Ds.

Whether available dachs mutations fully define the normal involvement of the Fat pathway in growth promotion is not yet clear. We cannot exclude the possibility that dachs is partially redundant with other proteins (e.g. other myosins), although this seems unlikely given the complete suppression of all non-polarity phenotypes of fat by dachs. It is also possible that dachs is required only for peak Fat signaling. This explanation is suggested by the observation that expression of the Fat target genes wg, Ser and fj is only partially or transiently lost in dachs mutants, yet the elevated or ectopic expression of these genes in fat mutants is completely eliminated by mutation of dachs.

The relatively mild tissue polarity phenotypes of dachs mutants,and the inability of dachs mutation to completely suppress the influence of fat on tissue polarity, contrast with the absolute dependence of fat gene expression, growth and affinity phenotypes on dachs. These observations suggest that there are two distinct Fat pathways. One, crucially dependent on Dachs, influences gene expression,growth and cell affinity, and another, partially independent of Dachs,influences tissue polarity. Studies of the atrophin protein Grunge also support the suggestion that there is a distinct Fat polarity pathway, as Grunge interacts with Fat and influences tissue polarity(Fanto et al., 2003), but does not exhibit other phenotypes observed in fat mutants(Cho and Irvine, 2004; Fanto et al., 2003). Thus,Dachs might act redundantly with another protein in a polarity pathway, but non-redundantly in a pathway that influences gene expression. It should also be noted that effects of dachs on gene expression might contribute to the polarity phenotypes of dachs mutants. For example, fj is regulated by dachs (Fig. 4), and fj has polarity phenotypes(Casal et al., 2002; Zeidler et al., 1999; Zeidler et al., 2000).

The asymmetric localization of Dachs observed in wild-type wings, and the influence of Fj and Ds on Dachs localization, have important implications for tissue polarity. First, the asymmetric localization of Dachs is itself a form of polarity, and its detection in third instar imaginal discs emphasizes that these cells are polarized well before core polarity proteins such as Frizzled and Dishevelled become asymmetrically localization in pupal wings (reviewed by Eaton, 2003). A similar conclusion can be drawn from the recent observation that fat and ds influence the orientation of cell divisions in third instar discs(Baena-Lopez et al., 2005). Second, our observations identify an ability to induce asymmetric protein localization as a mechanism through which the Fat pathway might influence tissue polarity. Dachs is one target, but the Fat polarity pathway might similarly involve asymmetric localization of other myosins, or of other types of proteins, to affect tissue polarity.

Mutation of fat is associated with elevated Dachs staining at the membrane, and overexpression of Fat decreases Dachs staining at the membrane. Although this negative effect of Fat on Dachs is subject to the caveat that we can only detect tagged overexpressed Dachs:V5, this tagged protein rescues dachs mutants, and the effects of Fat on Dachs staining are consistent with their opposite phenotypes and the epistasis of dachsto fat. Manipulations of the expression of Fat regulators provide further evidence that Fat regulates Dachs levels at the membrane, and altogether our observations implicate Dachs as a crucial intracellular component of a Fat signaling pathway (Fig. 9A,B).

The concomitant elevation of Fat staining and loss of Dachs staining observed at the perimeter of Fj-expressing clones is consistent with the conclusion that Fat can antagonize the localization or stability of Dachs at the membrane. Because the elevation of Fat is limited to the periphery of Fj-expressing clones, we hypothesize that it results from an influence of Fj on Fat-Ds interactions, rather than the expression of Fj per se. Tissue polarity studies have implied that Fj and Ds have opposite affects on Fat. Although it has not yet been determined whether Fj can directly modify Fat or Ds, the simplest explanation for the elevated Fat staining at the edge of Fj-expressing cells would be to propose that Fj modifies Ds to inhibit its interactions with Fat (Fig. 9B). In this case, Fat protein within Fj-expressing clones would be predicted to prefer to bind to Ds outside of the clone, and hence to accumulate at the clone perimeter, where it would then downregulate Dachs(Fig. 9C).

The interpretation of the elevated Dachs staining at the perimeter of Ds-expressing clones is more complex. Although Fat is elevated at the clone perimeter, the depletion of Fat from neighboring cells suggests that the elevated Fat staining largely reflects Fat outside of the clone, rather than in Ds-expressing cells (Fig. 8E,F; Fig. 9D). Given that dachs and fat influence transcriptional targets cell autonomously, and dachs acts genetically downstream of fat, the link between elevated Fat in one cell and elevated Dachs in a neighboring cells must be indirect. It might be that Ds can also influence Dachs localization, and does so in opposite fashion to Fat. According to this scenario, the elevated Fat staining in cells neighboring the clone would be reflective of high levels of Ds engaged by Fat at the clone perimeter, which would then recruit or stabilize Dachs at the membrane. However, mutation of ds did not result in any noticeable decrease of Dachs:V5 staining(data not shown). Alternatively, it might be that Fat antagonizes the accumulation of Ds within the same cell. High Fat accumulation at the edge of one cell could then result in low Fat accumulation at the edge of its neighbor through this hypothesized downregulation of Ds. In this case, the elevated Dachs accumulation at the edge of Ds-expressing clones would be a consequence of low levels of Fat. This model would also imply that asymmetric localization of Fat could be propagated from cell to cell, which could have important consequences for Fat pathway regulation. However, there is as yet no evidence that Fat is asymmetrically localized at wild-type levels of Fj and Ds expression.

We thank F. M. Hoffmann, T. Schupbach, H. McNeill, M. Simon, S. Blair, the Developmental Studies Hybridoma Bank and the Bloomington stock center for antibodies and Drosophila stocks. We are especially grateful to F. Michael Hoffmann for providing us with unpublished mutant alleles. We thank Gerri Buckles and Olga Dunaevsky for excellent technical assistance. This work was supported by NSF grant IBN 9724930 (F.N.K.), NIH grant GM63057 (C.R.) and by the Howard Hughes Medical Institute (K.D.I.).