The wingless signalling pathway and the patterning of the ......622 Mahowald, 1987; Simpson et al., 1988; Perrimon and Smouse, 1989). Therefore, it is not clear at present if the proteins

INTRODUCTION

The

wingless (wg) gene of Drosophila is a member of the Wntgene family and encodes a secreted protein homologous to theproduct of the Wnt-1 vertebrate proto-oncogene (Rijsewick etal., 1987). Studies of the absence of the gene (reviewed byIngham, 1991; Hooper and Scott, 1992; Peifer and Bejsovec,1992; Nusse and Varmus, 1992, and Martinez Arias, 1993) orof its misexpression in flies (Nordemeer et al., 1992; Struhl andBasler, 1993), mice (Tsukamoto et al., 1988) and frogs(reviewed by Moon, 1993), suggest that wg encodes a signalthat is involved in cell interactions. The product of wg plays aprominent role in many of the events that pattern the larvalepidermis of Drosophila, for example in the maintenance ofthe expresssion of the engrailed (en) gene (reviewed byIngham and Martinez Arias, 1992; Hooper and Scott, 1992;Peifer and Bejsovec, 1992). In addition, wg is also requiredduring the development and patterning of the nervous system(Patel et al., 1989 and unpublished obs.), Malpighian tubules(Skaer and Martinez Arias, 1992) and imaginal epidermis(Baker, 1988a; Couso et al., 1993). Experiments with a tem-perature sensitive allele of wg show that many of these require-ments reflect spatially and temporally discrete functions for wg(Bejsovec and Martinez Arias, 1991). Thus, during the devel-opment of the imaginal discs, wg is initially required for theestablishment of their primordia from the larval epidermis(Cohen, 1990) and later for their patterning during larval life(Couso et al., 1993).

The multiple functions of wg, revealed by the use of tem-perature sensitive mutants, raise the question of whether theyreflect multiple responses of the cells to a single signallingevent, dependent on a single signalling pathway, or whether

they are related to different signalling events mediated bydifferent signalling systems.

Studies of mutants that might be involved in the same sig-nalling event as wg during Drosophila embryogenesis haveidentified a small group of genes whose activity appears to benecessary for wingless signalling (reviewed by Peifer andBejsovec, 1992; Klingensmith and Perrimon, 1991; MartinezArias, 1993). Mutations in two of them, armadillo (arm) anddishevelled (dsh), produce a wg mutant phenotype in a cellautonomous manner in the embryo, suggesting that their wild-type products are indeed required for proper wg function(Riggleman et al., 1990; Klingensmith and Perrimon, 1991).Mutations in another gene, shaggy/zeste white 3 (sgg), lead toa larval cuticular phenotype (Perrimon and Smouse, 1989)similar to that produced by overexpression of wg (Noordemeeret al., 1992). Molecular studies indicate that arm encodes amember of the plakoglobin gene family (Peifer and Wieschaus,1990) and sgg a non-receptor serine threonine kinase withhomology to the vertebrate GSK3 (Bourouis et al., 1990;Siegfried et al., 1990, 1992). Studies of the interrelationshipsbetween wg and sgg and wg and arm during embryogenesishave led to the suggestion that the wingless signal antagonizesthe activity of the sgg gene in the maintenance of theexpression of en (Siegfried et al., 1992) and that arm isrequired for this event (Riggleman et al., 1990; Peifer et al.,1991). Although no molecular information is yet availableabout dsh, genetic studies indicate that it is also involved in theimplementation of the wingless signal (Klingensmith andPerrimon, 1991). Despite such specific effects, mutations ineither sgg, arm or dsh are highly pleiotropic i.e. absence of anyone gene results in a variety of apparently unrelated pheno-types in embryos and adults (Peifer et al., 1991; Perrimon and

621Development 120, 621-636 (1994)Printed in Great Britain © The Company of Biologists Limited 1994

The margin of the wing of

Drosophila is defined andpatterned from a stripe of cells expressing the wingless (wg)gene that is established during the third larval instar in thedeveloping wing blade. The expression of the genes cut andachaete in a small domain in the prospective wing marginregion reflects the activity of wg and probably mediate itsfunction. Our results indicate that, in the wing margin, thewingless signal requires the activity of at least three genes:armadillo (arm), dishevelled (dsh) and shaggy (sgg) and that

the functional relationship between these genes and wg isthe same as that which exist during the patterning of thelarval epidermis. These observations indicate that arm, dshand sgg encode elements of a unique ‘wingless signallingpathway’ that is used several times throughout develop-ment.

Key words: wingless, wingless signalling pathway, wing margin, cellsignalling

SUMMARY

The wingless signalling pathway and the patterning of the wing margin in

Drosophila

Juan Pablo Couso*, Sarah A. Bishop and Alfonso Martinez Arias

Department of Zoology, University of Cambridge, Downing Street, Cambridge CB2 3EJ, UK

*Author for correspondence

622

Mahowald, 1987; Simpson et al., 1988; Perrimon and Smouse,1989). Therefore, it is not clear at present if the proteinsencoded in these genes are common elements of a ‘winglesssignalling pathway’, or they are only functionally related to wgin the maintenance of en.

Here we describe a function of wg in the patterning of thewing margin of Drosophila during the third larval instar. Weshow that the Wingless protein (Wg) acts as a signal for thedevelopment of the pattern elements along the wing marginand that these functions are dependent on the same elementsand functional relationships that control en expression duringearly embryogenesis. Our results indicate that in Drosophila,the wingless signal uses a conserved group of elements that acton different transcription factors at different times duringdevelopment.

MATERIALS AND METHODS

Genetic variations and experimental crossesThe mutations and allelic combinations used in these studies are asfollows.

wgIL

At 17°C, embryos homozygous for the temperature sensitive wgIL

allele display a wild-type cuticular pattern and hatch normally.However, the original stock wgIL cn bw sp/CyO (Nüsslein-Volhard etal., 1984) never yields homozygous wgIL pupae or adults. Variousallelic combinations (unpublished observations) suggested that thispupal lethality was not associated with the wg allele but with a second-site mutation in the original wgIL cn bw sp chromosome. For thisreason, we generated a new recombinant chromosome wgIL ck pr cnthat retains the wgIL allele but has lost any pupal lethals existing inthe original chromosome. Although the newly obtained recombinantchromosome is homozygous lethal because of the presence of ck,adult flies that are homozygous for the wgIL allele can be obtained asheterozygous wgIL cn bw sp/wgIL ck pr cn individuals. When rearedat 25°C these individuals die as embryos with a characteristic wg nullphenotype but, at 17°C, wgIL homozygous adults emerge, survive anddisplay a visible mutant phenotype (see results and Fig. 5B). Both inthe temperature shift experiments and in the antibody stainings,mutant larvae or adults were identified as the Tb+ Hu+ progeny of across between flies from wgIL stocks balanced with the SM6aTM6btranslocated balancer (in which the markers Tb and Hu are linked tothe second chromosome). In the temperature shift experiments, theappropiate crosses were set up at 17°C and layings were collected at5 hours intervals. The vials were transferred to 25°C at the appropri-ate times, either until the end of development (shifts) or for a periodof time and then transferred back to 17°C (pulses). The age of thelarvae at the time of shift was estimated after studying the develop-ment of the homozygous wgIL control animals continuously main-tained at 17°C. Pharate adults were dissected, their wings inflated byheating in 10% NaOH, and mounted in Hoyer’s for microscope exam-ination.

spdflg

This mutation is homozygous viable and fertile and behaves as a weakspecific regulatory allele of wg in the third instar. Homozygous spdflg

flies display small wings with a severely reduced alula and defects atthe wing margin (Tiong and Nash, 1990; see also results and Fig. 4).When spdflg is placed in trans with lethal alleles of wg, some defectsin the wing margin can be observed and the alula is still very reduced(unpublished observations). For antibody stainings, mutant larvaewere taken from a homozygous spdflg stock.

A temperature sensitive condition for armIn the embryo, the phenotype of the temperature sensitive allelearmH.6 ranges from strong at 25°C to wild type at 17°C (Klingensmithet al., 1989), but the larvae raised at 17°C have very small discs anddie as undifferentiated pupae. To study the adult phenotypes of lossof function of arm, we have used the synthetic temperature sensitiveallelic combination armH.6; armBCD7/+, which produces wild-typeflies at 17°C and mutant phenotypes at 25°C (see results and Fig. 3).armBCD7 is a duplication for arm that can provide partial arm functionand was generated as a transformant with a transposon containing anarm minigene (Peifer et al., 1991). The temperature shifts were doneas described above for wgIL.

dshv26

Mutant dshv26 animals die as undifferentiated pupae (Perrimon andMahowald, 1987). However, we recovered imaginal discs forantibody stainings from the mutant larvae. These were recognized byvirtue of the y and w markers in a y w dshv26/Binsn stock.

cutThe ‘ct-wing mutants’ are a class of viable regulatory mutants with amutation at the ct locus that affects a specific enhancer element thatdrives ct expression during the third larval instar, in the stripe alongthe prospective wing margin (Jack et al., 1991), thus reducing or elim-inating only that expression of ct, which coincides with the expressionof wg (unpublished observations). In our experiments we have usedthe mutants FM6, ct/ct145 and ct6. For antibody staining of imaginaldiscs, mutant larvae were taken from the homozygous stock ct6 orfrom a stock y w ct145/FM6, y w ct/Y ct+y+, where mutant heterozy-gous females were recognized by their gonads and y mutantphenotype. Such females have no expression of Ct in the stripe alongthe wing margin. Control experiments with FM6 alone indicate thatthe results presented in this work are not due to any mutation on thebalancer chromosome.

Clonal analysisFlies of the appropiate genotypes were crossed and their eggscollected every 12 hours in food plates or split bottles. The vials con-taining larvae were irradiated with 1,000 rads using an Al filter 48-96hours after they were laid. In all cases mutant and control clones weregenerated, and their sizes (number of cells) were studied andcompared. Because the layings were 12 hours long and the experi-ments were carried out in batches, the average size of the controlclones was used to match comparable experiments and eliminatespurious age dispersion.

To study the phenotype of arm mutant clones we have used twodifferent mutant conditions. In the first instance we used the genotypey armXK f36a/M(1)OSp; mwh jv/+. After X-rays induced mitoticrecombination, mutant arm M+ clones were generated and marked yf36a, whereas mwh jv clones served as controls. We have alsogenerated clones of the temperature sensitive allele armH.6 (seeabove). To do this, animals of the genotype y armH.6 f36a/+; mwhjv/+ were cultured at 25°C until irradiation, and then transferred to170C. In the case of dsh, the genotypes used were y dshv26 sn3/+;mwh jv/+, which produce dsh mutant clones marked y sn and controlmwh jv clones. Animals of the genotype y dshv26 sn3/M(1)OSp; mwhjv/+, were used to generate dsh M+ clones. To study sgg mutantclones, twin clones were generated in animals of the genotypesggM11-1/y f36a. The same recombination event produces a control yf36a clone and an accompaning sgg mutant clone, identifiable by itsphenotype (Ripoll et al., 1988; see results and Fig. 5 ). To generateclones doubly mutant for sgg and dsh, a recombinant chromosomesggM11-1 dshv26 was generated. The experimental genotype wassggM11 dshv26/y f 36a. In this situation, because the f locus is proximalto dsh, control f clones are necessarily accompanied by a mutant sggdsh twin clone.

J. P. Couso, S. Bishop and A. Martinez Arias

623wingless signalling in the Drosophila wing margin

ImmunocytochemistryImaginal discs were fixed for 10-15 minutes in cold 4%paraformaldehyde in PBS and stained, using standard procedures,with DAB. Polyclonal anti-Wg antibody was provided by M. vanden Heuvel and used diluted 1:250. Monoclonal anti-Ac and anti-Ct antibodies were provided by S. Carroll and Y. N. Jan respectivelyand used diluted 1:200 and 1:500 respectively. In some experiments,

gene expression was monitored with lacZ reporter gene fusions, andin those cases a standard enzymatic assay coloured the

β-gal-expressing cells blue. For ct, the transformant ctwHZ, which carriesthe specific ct enhancer that drives expression in the edge cells only(Jack et al., 1991) was used. The transformant line neu-A101 wasused to detect expression of β-galactosidase in the bristle precur-sors.

Fig. 1. Fate map, developmentand differentiation of the wingmargin of Drosophila. (A) Themargin of the wing ofDrosophila contains threedomains, each with astereotyped array of patternelements. In both the anteriortriple row, and the distal doublerow the sensory elements - stoutbristles (open circles), slenderbristles (triangles) andchemosensory bristles (solidcircles) - are inervated, whereasin the posterior row the bristles(hatched circles) are notinnervated. Epidermal hairs areindicated as dots.(B) Correlation of geneexpression and cellulardifferentiation during thedevelopment of the margin. Themain surface of the drawingrepresents a wing margin thathas been opened out to allow thedorsal and the ventral surfacesto lie on the same plane. Thisrepresentation permits atopological correlation betweenthe final pattern elements on thewing and the patterns of geneexpression in the third instarwing disc. A section of the latteris represented by a row of cellswith the pseudostratifiedappearance characteristic ofthird larval instar disc epitheliawhich, after wing evagination inthe pupa, adopt a squamous

morphology. In the anterior margin of the adult wing (main surface), it is possible to discern several rows of pattern elements from ventral (left)to dorsal (right). Ventrally there are two rows of epidermal hairs which form the ventral side of the marginal vein (not indicated) and a row ofbristles containing a sequence of one recurved chemosensory bristle every four slender mechanosensory bristles. Medially there is a row ofhairs that defines the edge of the margin and a row of thickly packed stout mechanosensory bristles. Finally, on the dorsal side of the marginalvein there is another row of hairs followed by a row of chemosensory bristles each interspersed by four hairs. This pattern of bristles representswhat is commonly known as the ‘triple row’ and is modified as it progresses into the posterior region of the wing, first in the double and then inthe posterior row (see A above and Fig. 2). This pattern of cellular differentiation is derived from the presumptive margin of the third instarimaginal discs, where it is prefigured by the patterns of expression of wingless (wg), cut (ct) and achaete (ac) (shown in fig. 3). As indicated inthe figure, the expression of Wingless protein (Wg, dark stippling) and Cut protein (Ct, black nuclei) outline the edge proper, and the marginalelements are defined by the expression of Achaete protein (Ac, hatched nuclei) under the influence of Wg (apical stippling). The differentlocation of the stippling reflects differences in the distribution of Wg, characteristic of wild-type discs. The differences between dorsal andventral, as well as those between the different regions of the margin along the anteroposterior axis, are probably due to the activity of othergenes (see text). (C) Chronology of events during the development of the margin (O’Brochta and Bryant 1985; Hartenstein and Posakony 1989;Cubas et al., 1991; Blochlinger et al., 1993; Jack et al., 1991; Blair 1992). All times refer to hours after egg laying (AEL). The margin isindicated by the double line. The top line indicates the cells of the edge i.e. those that express ct and will give rise to the stripe of hairs in themiddle of the margin, while the bottom line indicates the cells of the margin. The shading indicates the periods of mitotic activity in thedifferent regions of the margin. L3e and L3l indicate third instar larval stages, early (e) and late (l). PF indicates the time of puparium formationand the symbols at the bottom, the time at which the precursors of the different sensory organs are visible. For symbols, see above, and forfurther details, see text.

624

RESULTS

The development of the wing margin inDrosophilaThe wings of Drosophila develop during larvallife within the dorsal imaginal discs of the secondthoracic segment, the ‘wing discs’. While theestablishment of the basic system of polar coor-dinates of positional information in these discstakes place during the first and second larvalinstars (Couso et al., 1993), the fine patterning ofthe tissue, as reflected in the determination of thesense organs and hairs that decorate the marginof the wing, is achieved during the third larvalinstar and the early stages of pupal development(Figs 1, 2; García-Bellido and Santamaría, 1978;Hartenstein and Posakony, 1989, 1990; Cubas etal., 1991; Rodriguez et al., 1990).

Studies of wg expression in wing discs duringthe third larval instar have revealed a complexpattern of wg RNA (Baker, 1988b) and proteinexpression (Couso et al., 1993; Williams et al.,1993). A striking feature of this pattern is a stripeof wg expression along the presumptive margin


Fig. 2. Detail of the triple row showing the edgecells. Two planes of focus through the wing margin,as seen from the ventral side of the wing, showingthe row of edge cells (arrows) between the ventralrow of slender (large arrow) and the dorsal row ofstout (arrowheads) mechanosensory receptors. Theblue color reflects β-galactosidase activity which hasbeen expressed in the stripe of ct-expressing cellsalong the wing margin in the third larval instar andearly pupa (Jack et al., 1991, see methods and fig. 3c)and therefore shows that the edge cells of the adultwing arise from the stripe of ct-expressing cellsshown in Fig. 3A,C.

Fig. 3. Patterns of gene expression in the late third instar larval disc(approx. 120 hours AEL) along the presumptive wing margin. Allimages show details of the presumptive anterior wing margin; theorientation of the discs is always such that posterior is to the bottomand dorsal to the left. At this stage the imaginal cells form apseudostratified epithelium and at different planes the width of thestripes of stained cells can vary, although the relationships betweenthe patterns of gene expression shown here are constant (see fig.1B). (A) Apical plane of focus through a disc stained with anti-Ctantibody. The Ct protein is expressed in a stripe of about three cellswide along the wing margin, which will give rise to the edge cells.(B) Apical plane of focus through a disc stained with anti-Acantibody. Ac is expressed in two parallel bands on the presumptiveanterior wing margin. Within the bands, the precursors of thechemoreceptors can be discerned by the bigger size of their nuclei,which accumulate high levels of Ac (arrows). Note that theseprecursors are surrounded by cells with lower levels of protein andthat they arise close to the edge cells (see below). (C) Prospectivewing margin stained to reveal simultaneously ac and ct expresssion.The blue colour shows β-galactosidase activity under the control ofa ct enhancer element specific to the edge of the wing margin (Jacket al., 1991, see Methods and Fig. 2), while the red one shows Acexpression. The ct gene is expressed between the two stripes of Ac.

Although different planes of focus show, for the most part, nucleiexpressing one or the other gene, a degree of coexpression exists incertain domains (arrows). (D) Detail of a wgIL/+ disc grown at 25°Cstained with anti-Wg antibody. The staining highlights two kinds ofcells, those darkly stained at the center of the stripe, which expressthe wg gene and therefore have accumulated the protein from thewgIL allele (see González et al., 1991), and those on either side,lightly stained, that reveal the movement of Wg. It can be seen thatthe stripe of wg-expressing cells is about three cells wide(unpublished observation; see also E and F). (E) Apical plane offocus through a wild-type disc stained with anti-Wg antibody. Highlevels of stain can be observed in a stripe of three cells which areexpressing the wg gene (unpublished observation), but lower levelscan be detected up to two or three cells away from this stripe.Comparison of this domain with A, B and C indicates that Wg canbe found over the whole field of ct- and Ac-expressing cells.(F) Medial plane of focus through a disc stained to show β-galactosidase expression in the chemoprecursors, through a reportergene insertion in the neuralized gene (blue color; see Methods) andWg (brown; as above). Notice that the cells with high levels of Wgare located between the rows of chemoprecursors. Therefore, thewg-expressing cells coincide with the ct-expressing cells (see B andC).

Fig. 2


of the wing established around the mid-third instar larva (Fig.4A) and which in the late third instar (ca. 120 hours after egglaying; AEL) is about three cells wide (Figs 3D, 4B). Thesecells, which continue to express wg after puparation (Fig. 4C),have high levels of Wingless protein (Wg) and coincide withcells that express the cut (ct) gene (Jack et al., 1991) all along

the presumptive wing margin (Fig. 3 and unpublished obser-vation). Lower levels of Wg can be detected up to three cellsaway on either side of the stripe of wg expression. It is likelythat this stain reveals binding of Wg by cells that do not syn-thesize it (Fig. 3D,E). These parallel stripes overlap, in theanterior compartment, with two bands of cells that express the

Fig. 3

626

Achaete protein (Ac) (Fig. 3 and for a review of the achaetescute complex (AS-C) see Campuzano and Modolell, 1992).

Towards the end of the third larval instar, the precursors ofthe sensory elements of the wing margin begin to be deter-mined, when the chemosensory precursors can be discernedfrom amidst the Ac-expressing cells that are nearer to thosethat express wg (Fig. 3). Later, during the early pupal stages,mechanosensory precursors arise from the interface betweenct- and Ac-expressing cells (Blair, 1993) and finally, the cellsof the presumptive wing margin rearrange and differentiate togive rise to the array of pattern elements characteristic of thewing margin (Fig. 1; Hartenstein and Posakony, 1989). Oneelement of this pattern not previously described is a row ofepidermal cells located between the medial and the ventralbristles of the anterior triple row that can be traced back to theregion that expresses ct and wg in the third instar disc (Fig. 2).We call these cells the ‘edge cells’ and those on each side ofthem, the ‘marginal cells’. The spatial relationships betweenthese pattern elements and the different patterns of geneexpression superimposed on a fate map of the wing marginduring the third larval instar are shown and discussed in Fig.1B.

Wingless is required for the patterning of the wingmargin At 17°C, embryos homozygous for the temperature sensitivewgIL allele (for details see methods) display a wild-typecuticular pattern, hatch normally and develop into adults whichsurvive but display some defects in the epidermis: the sternitesare for the most part missing (not shown) and the margin ofthe wing lacks some bristles (Fig. 5B). Stronger variations ofthese phenotypes are observed in other combinations of wgIL

with hypomorphic alleles of wg (Phillips and Whittle, 1993 andunpublished observations; see Methods). This requirement forwg in the patterning of the wing margin is emphasized by theobservation that flies homozygous for the mutation spdflg, anallele of wg which specifically affects the wing margin (Tiongand Nash, 1990, Methods and Fig. 5C), show reduced levelsof Wg in the corresponding presumptive region of the disc(Fig. 4D). Finally, very large clones of wg null alleles occa-sionally produce notches when they reach the wing margin(Baker, 1988b).

Altogether, these results indicate that the pattern of wgexpression during the third larval instar might be related to afunction of wg in the patterning of the wing margin. To test


Fig. 4. Expression of Wg inthe imaginal discs during thethird larval instar and pupalstages. All images at thesame magnification, posteriorto the bottom. (A) Mid-thirdinstar larval wing and legdiscs. The final pattern of Wgexpression is established inthe wing disc: a broad stripein the notum (n), a thinnerone in the prospective wingmargin (arrow) and anotherone encircling theprospective wing blade (for adetailed fate map of the thirdinstar wing disc showing thecorrespondence with adultstructures see Campuzanoand Modolell, 1992).(B) Late third instar wingdisc (shortly before 120hours AEL). The stripe ofWg expression in the margin(see Fig. 3) is the same as inA, but, at this stage there aretwo stripes over theprospective wing hingeregions (h), around the wingblade region (v, prospectiveventral wing blade).(C) Dorsal view of a pupalwing 4-8 hours APF, afterevagination of the wing

regions shown in B to the right of the figure so that the ventral wing blade (labelled v in B) is now underneath this plane of focus. Wg is stillpresent along the wing margin, in a stripe of about five cells wide. Note that while in the late third instar disc (B) the protein levels decreasefrom a central high point of expression (see also fig. 3), in the pupal wing the protein appears constrained within a bounded domain. Thisbounded distribution is reminiscent of that described for the parasegmental boundary during late stages of embryogenesis (González et al.,1991). (D) Late third instar disc from a spdflg mutant. Note the narrower width of the stripe of Wg in the presumptive posterior margin (bottompart of the stripe), where spdflg mutant flies display the stronger phenotype (see Fig. 5C).


this further and to define the precise requirements for Wg inthis process, we have eliminated its activity by shifting wgIL

homozygous animals from 17°C to 25°C at defined timesduring the third larval instar (see Methods). Such shifts resultin the complete inactivation of Wg (Baker, 1988a; Gonzálezet al., 1991; Bejsovec and Martinez Arias, 1991) and producedifferent effects at the wing margin depending on their timingand length. Shifts started after mid-third instar (approx. 96hours AEL) or at the time of pupation (120-124 hours AEL,i.e. 0-4 hours after puparium formation, APF) produce pharateadults with a ‘narrow margin’ phenotype: most bristles aremissing, and the remaining ones are intermingled in two oreven one row, instead of the three rows interspaced by

tricomes, characteristic of the wild-type (compare Figs 5Ewith 1B and 2). A 24 hour-long pulse, from 96 to 120 hoursAEL, produces weaker narrow margin phenotypes, verysimilar to those of the viable allelic combinations describedabove. However, shifts shortly before 96 hours AEL result inpharate adults in which the wing lacks all the features char-acteristic of a wild-type wing margin. In these wings, everybristle, whether non-innervated, chemo or mechanoreceptors,as well as the marginal vein, is missing; otherwise such wingsbear a fair pattern of veins and sensilla (Fig. 5D). Unlike thephenotype of other mutants in Drosophila that affect the wingmargin, this phenotype is not associated with notching, andthe number of cells of the wing blade does not appear to be

Fig. 5. Phenotypes of adult wings, oriented as Fig. 4C, distal tothe right and posterior to the bottom. (A) Wild-type wing;(B) wing from a wgIL homozygous animal raised at 17°C. Bristlesare lost along the entire wing margin. In the triple row, thepositions of lost stout mechanoreceptors are occupied by recurvedchemoreceptors. In slightly stronger mutant conditions, thisresults in the transformation of the the triple row into a narrow‘double row’. (C) spdflg mutant wing. Some bristles are missingalong the anterior margin and the posterior margin is completelylost. Defects in wing hinge regions that express Wg produce anabnormally shaped wing. (D) Wing from a wgIL animal shifted to25°C shortly before 96 hours AEL. These wings have beendissected out from the pharate adults and inflated (see methods),hence the apparent defects asociated with veins. (E) Detail of theanterior wing margin of a wgIL animal shifted to 25°C at 0-4

hours APF. The wing was mounted to allow a frontal view of the streched margin. The mechanoreceptors (and posterior bristles, not shown),which should have been determined at about 12 hours APF (see Fig. 1C), have been lost and the remaining chemoreceptors are arranged in a‘narrow margin’ (compare with Figs 1B and 2).

628

significantly reduced. The absence of these two features,which are usually associated with extensive cell death(Fristrom, 1969; Spreij, 1971), suggests that the primary causeof the total absence of wing margin is the lack of marginalcell fates.

The products of dsh and arm are required forwingless signalling during the development of thewing margin to antagonize the activity of sggGenetic studies suggest that the products of arm, dsh and sggare involved in wingless signalling during embryogenesis



(reviewed by Nusse and Varmus, 1992; Martinez Arias, 1993).To test whether the function of wg in the development of thewing margin is mediated by the same signalling pathway thatfunctions in the early patterning events in embryos, we haveanalyzed the function of arm, dsh and sgg during the thirdlarval instar (Fig. 6; Table 1).

armadilloStudies with arm mutants suggest that its gene product mightbe required for wingless signalling not only in the embryo butalso during the establishment of polar coordinates in theimaginal discs: hypomorphic mutant conditions for arm canproduce pharate adults with wg-like mutant phenotypes ofabnormal polar patterning in legs and thorax (Peifer et al.,1991). Unfortunately, the nature of these alleles precludes astudy of arm function during the third larval instar (seeMethods). To study this function, we have used the armH.6

allele as the basis for a synthetic temperature sensitive alleliccombination that produces wild-type adults at 17°C (for detailssee Methods). When these flies are shifted from 17°C to 25°Cduring the third larval instar, they produce adults with missingpattern elements along the wing margin and the alula, pheno-types which are reminiscent of weak wg hypomorphic con-ditions (Fig. 6A). These arm mutant adults also resemble wgmutants in that their abdomens lack sternites (not shown).

To study the effects of stronger loss of arm function, wehave performed a clonal analysis of the arm gene (seeMethods; Table 1). Mosaic flies carrying clones of cellshomozygous for the armXK allele, which behaves as a strongbut not null allele in the embryo (Peifer and Wieschaus, 1990),were generated. Our findings agree with those reported byPeifer et al. (1991) in that clones of arm mutant cells are found

with lower frequency and smaller sizes than expected,revealing a requirement for arm in cell viability and/or celldifferentiation (Table 1). In line with their results, we also finda high frequency of phenocopies of wg phenotypes among theirradiated flies. However, in addition we find that when theclones are generated late in development, surviving armMinute+ clones display visible phenotypes: abnormal celldifferentiation and alterations of the patterning around the wingmargin (Table 1 and Fig. 6B,C) similar to those produced bydsh mutant clones (see below). Finally, to overcome the lowviability of arm mutant cells, we studied clones of the mildmutant condition armH.6 at 17°C (see methods). In accordancewith their weaker hypomorphy, these clones appear withhigher frequency and larger sizes than armXK M+ clones, andproduce the same wing margin phenotypes, although withlower penetrance and expressivity (Table 1).

dishevelledIn the absence of conditional or strong viable mutant alleles ofdsh that would allow us to study the imaginal requirements fordsh in the patterning of the wing margin (Perrimon andMahowald, 1987), we have generated clones of cells homozy-gous for the strongest available lethal allele, dshv26 (Table 1).The resulting mosaic flies display different phenotypesdepending on the position of the clone within the wing blade(Fig. 6E). If the dsh mutant cells lie within the wing blade they

Fig. 6. Phenotypes of clones of cells mutant for arm, dsh and sgg.(A) Detail of the alula and proximal posterior region of an armmutant wing (see Methods). Notice the loss of bristles (arrows)reminiscent of the weak wg mutant phenotypes shown in Fig. 5B,C.(B,C) Small surviving arm M+ clones in the anterior and posteriormargin, respectively. The arm mutant cells are small and abnormal(arrowheads). Although the clones result in the loss of bristles(arrows), neighbouring wild-type cells located in the wing blade canoccassionally form bristles (stars). (D,E) Clones of cells mutant fordsh in the anterior ventral and posterior dorsal wing margin,respectively. Every bristle is lost in the surface of the wing marginoccupied by the mutant clone (to the right of the arrows), but wild-type neighbours can occasionaly differentiate bristles (stars). Inaddition, the dsh cells differentiate multiple tricomes with abnormalorientation and display abnormally highlighted contours. (F) Extrabristles in the anterior wing margin produced by a sgg mutant clonein the dorsal surface of the wing. (G) Ectopic bristles produced by asgg mutant clone in the wing blade. (H) Clone doubly mutant for sggand dsh (from the arrow to the right) in the dorsal anterior wingmargin, showing the epistatic sgg mutant phenotype (compare with Fand D). Notice the presence of y f bristles produced by the controltwin clone (arrowhead, and to the left). (I) Clone doubly mutant forsgg and dsh stradling the double row. Cells in the clone differentiateextra bristles either on the margin or on the internal blade, as do sggmutant clones and unlike the dsh mutant clones. Some cells of the ftwin can be seen to the left. Also, some cells differentiate multipletricomes. y f bristles were never seen to be recruited into the tufts ofbristles produced by sgg dsh clones, thus showing that the non-autonomous effect of dsh mutant cells on their wild-type neighboursis also supressed.

Table 1. Clonal analysis of the genes involved in winglesssignalling in the adult fly

Clones Age Wing margin studied

Genotype (h. AEL) phenotype* Penetrance† n

y armXK f36a/ 72-84 non-viable 1.0 (38)‡M(1)OSp 90-96 lack of WM 0.88 111

y armH.6 f36a (170)§ 48-66 ” 0.33 15

y dshv26 sn3 48-66 ” 1.0¶ 1572-84 ” 0.76 89

sggM11-1/y f36a 48-66 ectopic WM 0.65 1772-84 ” 0.32** 53

sggM11-1 dshv26/ 48-66 ” 0.68 22y f36a 72-84 ” 0.29†† 79

Age range estimation is based on the hours from egg laying to irradiationand on the mean size of the control clones (see methods). *See text and Fig. 6for a more detailed description of phenotypes. †Fraction of clones thatproduce a wing margin phenotype. The data for arm and dsh is based on theclones that reached the margin. For the sgg and sgg dsh experiments, thepenetrance is calculated as the percentage of f clones accompanied by a twinclone of sgg bristles. In these two cases, the maximum value is lower thanone because a fraction of sgg clones in the wing blade is lost (see also Ripollet al., 1988). In all cases, however, the penetrance decreases in the latestgenerated clones. This can be explained by the residual presence of wild-typeprotein in the cells until the requirement for the gene function ends. ‡Numberof control clones. No f (arm) clone was found at this age. §The armH.6 cloneswere produced at 17°C. Because their weak hypomorphy (see methods) theyshow higher viability and lower penetrance and expressivity than the armXK

M+ clones. ¶These dsh clones have half the average size and frequency ofappearance than controls. **These sgg clones produce additional phenotypeswhen not at the wing margin, like ectopic vein differentiation (see also Ripollet al, 1988). ††The sgg dsh clones generated at this age that do not show asgg phenotype begin to show dsh-like phenotypes of abnormal celldifferentiation (see Fig. 6).

630

show several abnormalities in their differentiation related to afunction of dsh in tissue polarity (Adler, 1992 and J. P. C.unpublished observations). However, when the clone reachesthe margin of the wing, dsh mutant cells display an additionalautonomous phenotype: they do not differentiate as patternelements of the margin and instead behave as if they were inthe middle of the wing blade i.e. they develop tricomes andnon-pigmented cuticle. Interestingly, in wings bearing suchmarginal clones of dsh mutant cells, wild-type cells located afew cell diameters away from the wing margin but which lieadjacent to dsh mutant cells are induced to produce ectopicbristles characteristic of the margin (Fig. 6D,E). Such pheno-types are also produced by clones of other alleles of dsh (N.Perrimon, personal communication) and, depending on theirsize, by arm mutant clones (Fig. 6B,C).

The loss of pattern elements of the wing margin in clones ofcells mutant for arm and dsh are identical to those observedwhen wg function is inactivated from the middle of the thirdlarval instar, suggesting that, just as in the patterning of thelarval epidermis, dsh and arm are necessary for proper wgfunction. This suggestion is reinforced when the mutant phe-notypes are studied at the level of gene expression (see below).Furthermore, the autonomy of the arm and dsh phenotypes ofloss of marginal pattern elements indicate that their wild-typefunctions are required for the implementation of the winglesssignal in the marginal cells.

shaggyClones of sgg mutant cells in the wing blade autonomously dif-ferentiate tufts of bristles characteristic of the wing margin(Ripoll et al., 1988; Simpson et al., 1988, Fig. 6F,G and Table1), suggesting that the function of sgg is antagonistic to that ofwg, arm and dsh. The marginal phenotype of sgg mutant cellsis associated with the ectopic expression (Blair, 1992) andfunction (Simpson and Carteret, 1989) of products of the AS-C, e.g. ac. However, in our experiments, staining of imaginaldiscs bearing clones of sgg failed to show ectopic expressionof wg (not shown; Blair, 1994). Because sgg mutant clonesproduce the ectopic appearance of wing margin withoutrecourse to wg expression, it can be concluded that sgg doesnot act upstream of wg.

To understand the relationship between the wingless signaland the activity of sgg, we have generated clones of cellssimultaneously mutant for both sgg and dshs, which displayviable and readily visible cell autonomous phenotypes (seeMethods). These clones do not produce any dsh mutantphenotype at the wing margin, either autonomously or non-autonomously. Moreover, cells mutant for both sgg and dshdisplay the same phenotype as cells mutant for sgg alone:autonomous ectopic appearance of wing margin anywhere inthe wing (Fig. 6 and Table 1). These results show that sgg isepistatic over dsh and therefore allow us to conclude that allthe function of dsh in the patterning of the wing margin ismediated through sgg, and more precisely that, in the wild type,the wingless signal represses the function of sgg in the targetcells.

The products of ct and ac implement wg function atthe wing marginThe most conspicuous phenotype in the wing margin, resultingfrom the absence of wg function, is the loss of sensory

elements. It is known that the expression and function of theproducts of the AS-C are required in clusters of cells fromwhich the precursors of the peripheral nervous system of thefly, including both the chemo- and mechanoreceptors of thewing margin, develop (Campuzano and Modollel, 1992;Dominguez and Campuzano, 1993). In the hypothesis that wgis directing the patterning of the wing margin, it would beexpected that one aspect of wg function would be to controlthe expression of the AS-C products, e.g. Ac, along the pre-sumptive margin in the wing discs. When we examine the dis-tribution of Ac in wg mutant conditions, we find a strong cor-relation between the mutant phenotype and the distribution ofAc in the marginal cells (Figs 5, 7). Conditions that lead to thecomplete absence of the margin completely abolish Acexpression (Fig. 7B), whereas conditions that produce a narrowmargin lead to a sparse distribution of Ac in narrower stripes(Fig. 7C). A similar observation has been made recently withanother hypomorphic allelic combination of wg (Phillips andWhittle, 1993). In addition, in mutant dshv26 discs (seeMethods), the expression of Ac is eliminated in the regionswhere Wg is normally expressed (Fig. 7). This corroborates theconclusions of the study of mutant phenotypes (see above), thatthe dsh product is involved in the transduction of the winglesssignal.

The integrity of the wing margin is also affected bymutations at the ct locus that eliminate or reduce only theexpression of Cut protein (Ct) in the stripe that outlines theedge cells along the prospective wing margin (Jack, 1985; Jacket al., 1991; Blochlinger et al., 1993). The correlations betweenthe phenotypes and patterns of expression of ct and wg mightsuggest that the activity of Ct in the edge cells of the wingmargin is involved in the control of wg expression. However,antibody stainings of ct mutant discs show a wild-type distri-bution and abundance of Wg, despite the total absence of Ctexpression in the edge cells. Moreover, the pattern ofexpression of wg monitored by a β-galactosidase reporterinsertion at the wg locus also remains unaltered in a ct mutantbackground (see Methods).

These observations have led us to test whether ct expressionin the edge cells is dependent on the activity of wg. To do thiswe have monitored the expression of Ct in wing discs withvarious allelic combinations of wg. In all cases, the expressionof Ct is reduced in correlation with the strength or spatial speci-ficity of the wg mutant condition used. Thus, when wgIL

animals are shifted to 25°C before mid-third instar, they showcomplete absence of Ct in the edge cells (Fig. 8). Also, in spdflg

mutant discs, which display reduced levels of Wg (Fig. 4D),there are correspondingly low levels of Ct in those cells (notshown).

Our observations suggest that low levels of Wg may activateAc expression at the margin and this raises the question of whyit is that the higher levels of Wg present in the edge cells donot direct Ac expression in this region. One possibility is thatgenes that are expressed between the two main bands of Acprevent this expression (see Cubas and Modollel, 1992). Totest this we have studied the expression of Ac in ct mutantdiscs. In these discs, there is Ac expression in many of thosecells of the edge that in the wild type never express it (compareFigs 3B and 7D). This result indicates that Wg can direct theexpression of Ac at the margin but that this expression isprevented, at least partially, by the activity of Ct.



DISCUSSION

The wingless signalling pathway in the wingmarginIn the absence of wg function during the third larval instar inDrosophila, wings develop normally but the cells on themargin do not make the bristles and veins characteristic of thewild type. Clones of wg mutant cells generated at this time donot show a mutant phenotype probably because their small sizeallows rescue by diffusion of the wild-type protein from neigh-bouring wild-type cells (Baker, 1988a). However, at the wingmargin, clones of cells mutant for arm or dsh display pheno-

types similar to those caused by the loss of wg during the thirdlarval instar (Figs 5 and 6). Because these mutant phenotypesreflect the absence of wg function in a cell autonomous manner,it can be concluded that arm and dsh functions are required inthe cells that receive and interpret the wingless signal. Theadditional non-autonomous ‘domineering’ influence thatmutant arm or dsh cells exert on their wild-type neighbourssuggest that, in the wild type, once cells within the marginalfield have become determined as wing margin elements by thewingless signal, a process of lateral inhibition prevents theirneighbours from doing so (Couso and Martínez Arias, unpub-lished data). Thus, the loss of bristles in such clones seems to

Fig. 7. Ac expression in imaginal discs 110-120 hours AEL. (A) Wild-type wing disc. A complex pattern of Ac expression can be appreciatedover a low background in the anterior compartment. Note the twoparallel bands along the anterior prospective wing margin (arrowhead),the clusters in the prospective notum (star), and the cluster of the veinsensilla (arrow). (B) Disc from a wgIL homozygous animal shifted to25°C before 96 hours AEL. Ac expression is abolished in the wingmargin and in some clusters in the notum (star). Overall expression isreduced. The arrow points to the remnants of the cluster of the veinsensilla. (C) Disc from a dshv26 mutant. Symbols as in A and B. Acexpression is lost in those regions that require Wg, like the wingmargin and some notal clusters (compare with B). (D) Ac expresion ina wgIL mutant exposed to a pulse of restrictive temperature that givesrise to narrow margin phenotypes as in Fig. 5E. Notice that the bandsof Ac expression are very reduced (compare with Fig. 2B) but that

some cells (probably the precursors of the chemoreceptors) still express high levels of Ac. (E) Ac expression in a ct145/FM6, ct6 mutant disc.The bands of Ac are disorganized, and there are more cells expressing Ac in the region between the original stripes than there are in wild-typediscs (compare with Fig. 3B).

632

be due to a failure in a process of determination rather than inone of differentiation.

In contrast with the behaviour of cells mutant for arm or dsh,clones of sgg mutant cells in the wing autonomously differen-tiate ectopic marginal pattern elements (Ripoll et al., 1988;Simpson et al., 1988 and Fig. 6). This phenotype suggests thatin the wild type, where the function of wg, arm and dsh is topattern the wing margin, that of sgg is to antagonize thisprocess. A functional relationship between both activities isdemonstrated by clones of cells simultaneously mutant for sggand dsh. These clones display an epistatic sgg phenotype in allregions of the wing, including the margin, indicating that thewild-type function of dsh in the wing margin is to repress theactivity of sgg. These results lead to a model in which the Wg-dependent patterning of the wing margin is instructed by a sig-nalling pathway that receives and transduces the winglesssignal through the dsh gene product, in order to repress theactivity of sgg. This model is consistent with two observations,our finding that the expression of Ac is eliminated in wg anddsh mutants, and the ectopic expression of Ac in the absenceof sgg function (Blair, 1992).

Our results also suggest that the arm gene product isnecessary for this process. Although it has been shown that theArmadillo protein is regulated by Wg both in the embryo(Riggleman et al., 1990) and the wing margin (Peifer et al.,1991), at present it is not clear whether arm acts upstream ordownstream of sgg.

The functional relationships that we have described betweenwg, arm, dsh and sgg at the wing margin resemble those thathave been found during early embryogenesis, when thewingless signal is used to establish and maintain the expressionof en in the neighbouring cells of the anterior region of eachparasegment. In the embryo, mutations in arm and dsh lead toa loss of en expression (Perrimon and Mahowald, 1987; vanden Heuvel et al., 1993; Peifer et al., 1991), as do mutationsin wg (Bejovec and Martinez Arias, 1991), whereas absence ofsgg function results in the ectopic expression of en in a wg-independent manner (Perrimon and Smouse, 1989; Siegfried etal., 1992). Thus, in the wild type, it seems that the localizedexpression of en in the embryo and the patterning of the wingmargin during the third larval instar rely on the wg-dependentinactivation of sgg activity within a small domain around thesource of Wg protein.

The patterning of the wing marginOur results together with those of others (García-Bellido andSantamaría, 1978; Hartenstein and Posakony, 1989; Rodriguezet al., 1990; Cubas et al., 1991; Blochlinger et al., 1993; Jack etal., 1991; Blair, 1992) demonstrate that the patterning of thewing margin is a progressive event. It is initiated during the thirdlarval instar with the deployment of the wg and ac gene productsand continues until the differentiation of hairs and sense organsin the pupa. Using temperature sensitive wg mutants we haveshown that Wg is required continuously for the patterning of thewing margin: initially to establish a marginal field within whichpatterning of cells will take place, and later for the progressiveand spatially restricted commitment of cells to the particularfates that make up the final pattern.

The establishment of a marginal field requires the expressionof Wg in a stripe across the prospective wing blade half waythrough the third larval instar. Later, cells that express high


Fig. 8. Ct protein expression in wing imaginal discs. (A) Wild-typelate third instar disc. Ct is found in the adult myoblasts (not shown),in developing sensory organs (arrow) and in the stripe along the wingmargin (Bloechinger et al., 1993). (B) Wing disc from a homozygouswgIL animal shifted to 25°C before 96 hours AEL. Ct expression isabsent from the wing margin but remains in the myoblasts(arrowhead) and in developing sensory organs, both of which are notaffected by wg mutant conditions (arrow). (C) Wing from a wgIL

homozygous animal, 4-8 hours APF, treated as in B. Ct expressioncannot be detected on the edge cells, nor at either side of it,correlating with the absence of sensory organ development over thewing margin. However, Ct expression can still be detected in thedeveloping sensilla of the third vein (arrow), which are present in thedifferentiated mutant wing (see Fig. 4E).


levels of Wg within this stripe also express Ct in a wg-dependent manner. In addition, in the anterior region of thewing disc and on either side of the stripe of Ct expression thereare two bands of Ac-expressing cells. These bands lie withina region of low levels of Wg and are eliminated if its activityor that of dsh are abolished. Despite this dependence of Acexpression on Wg (see also Phillips and Whittle, 1993), ac isonly expressed in the anterior region of the wing disc (Cubaset al., 1991; Skeath and Carroll, 1991 and Figs 3, 7). Interest-ingly, a pattern of low levels of Ac expression can sometimesbe seen in the posterior compartment of discs stained with ananti-Ac antibody (unpublished observation). We believe thatthe expression of ac is confined to the anterior compartmentdue to the activity of en which, in this manner would create adifferent final pattern of differentiation in the posterior com-partment of the wing (Morata and Lawrence, 1975) simply bycontributing to the repression of ac expression in this region.Mutations in en lead to the appearance of elements of the triplerow in the posterior margin of the wing (García Bellido andSantamaría, 1972) and while clones of cells mutant for sgg inthis region produce elements of the posterior row, clones ofcells doubly mutant for en and sgg produce elements of thetriple and double row (Ripoll et al., 1988). Thus, in the margin,as is likely to be the case in dorso-ventral differences (Fig. 1),the identity of pattern elements differentiated will depend onfactors other than Wg.

The stripe of wg expression over the prospective wingmargin persists throughout the late third larval instar and theearly pupal stages. Removal of wg function during this timeeliminates the mechanoreceptors and non-innervated bristles(both singled-out around 12 hours APF, Hartenstein andPosakony, 1990; see Fig. 1) and leads to a narrow margin inwhich the chemoreceptors are decimated and mixed inderanged rows. Once more, these phenotypes correlate withalterations in the pattern of Ac expression. In addition,mutations in ct or in the AS-C affect the integrity of the margin(Jack et al., 1991; García-Bellido and Santamaría 1978;Domínguez and Campuzano 1993). These phenotypes,together with the dependence of the patterns of Ac and Ctexpression on wingless signalling suggest that the function ofWg at the wing margin is mediated, in part, through the activityof Ct and the products of the AS-C, which are putative tran-scription factors (Bloechinger et al., 1988; Murre et al., 1989).

The role of ac and other members of the AS-C in the pat-terning of the wing margin appears to be similar to theirfunction elsewhere in the epidermis: they initiate the develop-ment of sensory organs by contributing to their determinationand differentiation. Initially, their expression defines a smallfield of competent cells from which later neural precursors willappear at defined positions among those cells expressing higherlevels of AS-C products (Cubas et al., 1991; Campuzano andModolell, 1992; Skeath and Carroll, 1991). By up-regulatingthe levels of AS-C products during this process, wg can playa role in both the definition of the competent region and in theselection and positioning of the precursors that arise from it,both in the wing margin and in the notum (Fig. 3 and 7; seealso Phillips and Whittle, 1993; Couso and Martinez Arias,unpublished data).

However, the function of the stripe of ct expression in theedge cells of the wing margin has been the subject of debate.Although Jack et al. (1991) identified the cells in this stripe as

the precursors of the mechanoreceptors, further studies indicatea more complex situation. Blochlinger et al. (1993) have shownthat the total number of mechanoreceptors does not account forthe total number of cells in the stripe of Ct expression. Inaddition, when the mechanoreceptors can be first identified,they appear on the edge of this stripe and they do not expressCt themselves (Blair, 1993). Nonetheless, it is still possible thatthey arise from amongst cells that were originally in the stripeof ct expression. Indeed, in the anterior region of the wing at4 hours APF, we observe Ct-expressing cells outside the stripeof Ct expression that do not correspond to chemosensory pre-cursors (unpublished observations). It is not impossible thatthey represent cells that are being recruited as elements ofsensory organs of the wing margin since it has been observedthat, at the wing margin, sensory elements can sometimes bederived from different cell lineages (Hartenstein and Posakony,1989).

Notwithstanding these observations, a significant number ofthe cells in the stripe of ct expression must contribute to theepidermal cells of the wing margin, including the row of edgecells that we have described between the medial and ventralrows of sensory elements of the triple row. Interestingly, inwings mutant for ct, these cells are largely absent and thesensory elements that remain mix themselves in a single dis-organized row (unpublished observation). Finally, in theabsence of Ct expression, there is a partial derepression of Acin those edge cells that usually do not expresss it. Takentogether these observations do suggest that the expression ofCt in the edge cells is essential to develop the pattern of rowsof bristles interspaced by epidermal cells characteristic of thewing margin. Ct probably plays multiple roles in this processbut, an important one is likely to be the control of theexpression of the members of the AS-C, in concert with othertranscription factors like the product of the emc gene (Cubasand Modollel, 1992). In this framework, the wing margin nicksof ct and similar mutants could be traced back to an imbalancebetween bristle precursors and edge cells which might disturbthe morphogenesis of the wing margin at the end of develop-ment (Jack et al., 1991).

Pattern formation by the wingless signal inDrosophilaThe involvement of wg in the patterning of the wing margin isreminiscent of its involvement in the patterning of larvalsegments during embryogenesis. In both instances Wg acts asa molecular source of positional information for a group ofcells and then contributes to the development of patternelements within the group (Bejsovec and Martinez Arias, 1991;Dougan and Di Nardo, 1992; this work). These two functionsappear to be mediated through the modulation of the activityof various transcription factors, like Engrailed and Gooseberryduring embryogenesis, and Ct and Ac in the patterning of thewing margin. Here we have shown that the products of arm,dsh and sgg are integrated in a wg signalling pathway duringthe patterning of the wing margin (Fig. 9). Since this is alsothe case during the patterning of parasegmental units duringembryogenesis (Siegfried et al., 1992 and reviewed by Klin-gensmith and Perrimon, 1991 and Martinez Arias, 1993), ourresults suggests that Wg controls the expression of differentgenes through a common signalling pathway. It has beensuggested that different concentrations of Wg might elicit

634

different responses from the receiving cells (Struhl and Basler,1993). Our results provide some evidence for this because, atthe wing margin, cells containing high levels of Wg preferen-tially express ct, whereas lower levels of Wg correlate with theexpression of Ac. Despite this observation, even consideringthat cross regulatory interactions between putative trancriptionfactors might modulate these differences, concentration aloneis not likely to explain the different patterns of transcriptionthat depend on Wg during development.

The molecular structure of the arm gene product, aDrosophila homologue of plakoglobin (Peifer and Wieschaus,1990), has led to the suggestion that it is asociated with the Wgreceptor (Peifer and Bejsovec, 1992) and the same has beenproposed for the product of dsh (Klingensmith and Perrimon,1991) even though its molecular structure remains to be eluci-dated. Because these two genes are involved in all functions ofwg that have been analyzed so far, unless there is some as yetundisclosed complexity in the molecular nature of the receptoritself, there are no simple possibilities to explain the multi-plicity of responses elicited by the activity of these molecules.One way to understand the versatile regulatory activity of the

wingless signalling pathway is derived from an appreciation ofthe central role that the product of sgg plays in the function ofwg (see above and Siegfried et al., 1992).

The sgg gene encodes a non receptor serine threonine kinasehomologous to vertebrate GSK3 (Bourouis et al., 1990;Siegfried et al., 1990, 1992; Ruel et al., 1993). Both have beenshown in vitro to phosphorylate serine and threonine residuesat the C terminus of c-jun and in this manner render inactiveits DNA binding ability (reviewed by Woodget, 1991). Thisobservation has led to the proposal that, in the early stages ofdevelopment, the sgg gene product could play a similarfunction with respect to Engrailed and thereby prevent itsnormal self-activation (Siegfried et al., 1992). A generalizationof this idea would imply that Wg controls gene expression byrepressing the inhibitory activity of sgg, thus allowing thefunctioning of nuclear autoregulatory networks that lead to theaccumulation of higher levels of specific transcription factors.In this hypothesis, the products of ac and ct would be in asimilar position with respect to sgg as it has been suggested forthe product of en.

In this model, there are two variables whose control andcombination might yield different patterns of gene expressionupon wingless signalling. The first one must be what iscommonly understood as the ‘developmental history’ or ‘com-petence’ of the cells, which might be related to the availabil-ity of specific transcription factors at different times and placesin development. The second variable is implicit in the possi-bility that the regulation of different transcription factors couldrequire different thresholds of kinase activity. From this itwould follow that different concentrations of Wg couldmodulate the activity of sgg and in this manner elicit qualita-tively different patterns of protein activity and gene expression.

Altogether our results suggest that a conserved core ofproteins and a similar mechanism underlie most functions of wgduring the patterning of the larval and adult epidermis. However,many more elements implicated in wingless signalling remainto be elucidated, in particular the receptor. Moreover, theexistence of specific components for different functions of wghas not been ruled out. Our observations on the role of wg in thepatterning of the wing and of its conserved relationship to sgg,dsh and arm, suggest that the wing margin provides an excellentsystem in which to explore these questions further. Preliminaryexperiments along these lines have identified some genesrequired for wg function, among them the Notch protein, areceptor molecule with multiple roles during development(Couso and Martinez Arias, unpublished data).

We thank S. Blair and N. Perrimon for discussing results prior topublication. We also want to thank M. Taylor and H. Skaer forcomments on the manuscript and M. Bate, J.F. De Celis, M. Ruiz-Gómez and other members of our lab for constructive discussions. Weare grateful to S. Carroll, Y. N. Jan and L. Jan for antibodies, J. Jack,E. Wieschaus and P. Lawrence for fly stocks and J. Rodford for hisrendition of the wing margin in Fig. 1. J. P. C. is supported by a FPUpostdoctoral fellowship from the Spanish Ministerio de Educación yCiencia and the research of A. M. A. and S. A. B. is funded by TheWellcome Trust.

REFERENCES

Adler, P. (1992). The genetic control of tissue polarity in Drosophila.Bioessays 14, 735-741.


Fig. 9. Summary diagram of regulatory interactions between genesthat mediate wingless function at the wing margin. The bell shapedistribution on the top represents presumed levels of Wg protein andthe circles underneath represent cells at the wing margin. The innercircles represent nuclei which express ct (black) or ac (hatched). Asindicated in Fig. 3 and in the text, the ct-expressing nuclei respond tothe higher levels of Wg. The arrow highlights two adjacent cells, oneexpressing ct and the other expressing ac, to illustrate the regulatoryinteractions between the genes. The wg gene is expressed only in theedge cells - those that express ct - but its product reaches themarginal cells (see Fig. 1B and text). In either case, it is likely that,upon binding to a receptor, Wg activates the product of dsh, which ispart of a mechanism that antagonizes (blunt arrows) the activity ofsgg. In turn the wild-type function of sgg is to repress the functionsof ct and ac. We postulate that the sgg product has a higher activityover cut than over ac -indicated by the bolder line in the blunt arrow.As discussed in the text, this would mean that higher levels of Wgare needed to maintain cut expression. In addition, at the edge, Ctcontributes to the repression of ac, which would otherwise beexpressed in this region. The arm gene is required for the function ofWg both in the edge cells and the marginal cells but its preciserelationship to the other genes is not yet known.


Baker, N. E. (1988a). Embryonic and imaginal requirements for wingless, asegment-polarity gene in Drosophila. Dev. Biol. 125, 96-108.

Baker, N. E. (1988b). Transcription of the segment polarity gene wingless inthe imaginal discs of Drosophila, and the phenotype of a pupal-lethal wgmutation. Development 102, 489-497.

Bejsovec, A. and Martinez Arias, A. (1991). Roles of wingless in patterningthe larval epidermis of Drosophila. Development 113, 471-485.

Blair, S. S. (1992). shaggy (zeste-white 3) and the Formation of supernumerarybristle precursors in the developing wing blade of Drosophila. Dev. Biol.152, 263-278.

Blair, S. S. (1993). Mechanisms of compartment formation: evidence that non-proliferating cells do not play a critical role in defining the D/V lineagerestriction in the developing wing of Drosophila. Development 119, 339-351.

Blair, S. S. (1994). A role for the segment polarity gene shaggy-zeste white 3 inthe specification of regional identity in the developing wing of Drosophila.Dev. Biol., in press.

Blochlinger, K., Jan, L. Y. and Jan, Y. N. (1993). Postembryonic patterns ofexpression of cut, a locus regulating sensory organ identity in Drosophila.Development 117, 441-450.

Bloechinger, K., Bodmer, R., Jack, J., Jan, L. Y. and Jan, Y. N. (1988).Primary structure and expression of a product from cut, a locus involved inspecifying sensory organ identity in Drosophila. Nature 333, 629-635

Bourouis, M., Moore, P., Ruel, L., Grau, Y., Heitzler, P. and Simpson, P.(1990). An early embryonic product of the gene shaggy encodes aserine/threonine protein kinase related to the CDC28/cdc2 subfamily. EMBOJ. 9, 2877-2884.

Campuzano, S. and Modolell, J. (1992). Patterning of the Drosophila nervoussystem: the achaete-scute gene complex. Trends Genet. 8, 202-208.

Cohen, S. (1990). Specification of limb development in the Drosophila embryoby positional cues from segmentation genes. Nature 343, 173-177.

Couso, J. P., Bate, M. and Martinez Arias, A. (1993). A wingless-dependentpolar coordinate system in Drosophila imaginal discs. Science 259, 484-489.

Cubas, P., de Celis, J. F., Campuzano, S. and Modolell, J. (1991). Proneuralclusters of achaete-scute expression and the generation of sensory organs inthe Drosophila imaginal wing disc. Genes Dev. 5, 996-1008.

Cubas, P. and Modolell, J. (1992). The extramacrochaetae gene providesinformation for sensory organ patterning. EMBO J. 11, 3385-3393.

Dominguez, M. and Campuzano, S. (1993). asense, a member of theDrosophila achaete-scute complex, is a proneural and neural differentiationgene. EMBO J. 12, 2049-2060.

Dougan, S. and DiNardo, S. (1992). Drosophila wingless generates celldiversity using engrailed expressing cells. Nature 360, 347-349.

Fristrom, D. (1969). Cellular degeneration in the production of some mutantphenotypes in Drosophila melanogaster. Molec. Gen. Genet. 103: 363-379.

González, F., Swales, L., Bejsovec, A., Skaer, H. and Martinez Arias, A.(1991). Secretion and movement of the wingless protein in the Drosophilaembryo. Mech. Dev. 35, 43-54.

García-Bellido, A. and Santamaría, P. (1972). Developmental analysis of theengrailed gene of Drosophila. Genetics 72: 87-107

García-Bellido, A. and Santamaría, P. (1978). Developmental analysis of theachaete-scute system of Drosophila melanogaster. Genetics 88, 469-486.

Hartenstein, V. and Posakony, J. W. (1989). Development of adult sensillaon the wing and notum of Drosophila melanogaster. Development 107, 389-405.

Hartenstein, V. and Posakony, J. W. (1990). A dual function of the Notchgene in Drosophila sensillum development. Dev. Biol. 142; 13-30.

Hooper, J. and Scott, M. P. (1992). The molecular genetic basis of positionalinformation in insect segments. In Results and problems in celldifferentiation: Early embryonic development of animals (ed. W. Henig) vol.18, pp 1-48. Berlin: Springer Verlag..

Ingham, P. (1991). Segment polarity genes and cell patterning within theDrosophila body segment. Curr. Op. Genet. Dev. 1, 261-267.

Ingham, P. W. and Martinez Arias, A. (1992). Boundaries and fields in earlyembryos. Cell 68, 221-235.

Jack, J. W. (1985). Molecular organization of the cut locus of Drosophilamelanogaster. Cell 42, 869-876.

Jack, J. and DeLotto, Y. (1992). Effect of wing scalloping mutations on cutexpression and sense organ differentiation in the Drosophila wing margin.Genetics 131, 353-363.

Jack, J., Dorsett, D., DeLotto, Y. and Liu, S. (1991). Expression of the cutlocus in the Drosophila wing margin is required for cell type specificationand is regulated by a distant enhancer. Development 113, 735-747.

Klingensmith, J. and Perrimon, N. (1991). Segment polarity genes and

intercellular communication. In Cell Activation : Genetic Approaches. (ed. J.J. Mond et al. ) pp. 251-274. New York: Plenum Press

Klingensmith, J., Noll, E. and Perrimon, N. (1989). The segment polarityphenotype of Drosophila involves differential tendencies towardtransformation and cell death. Dev. Biol. 134, 130-145

Martinez Arias, A. (1993). Development and patterning of the larvalepidermis of Drosophila. In The Development of Drosophila. (eds. M. Bateand A. Martinez Arias). pp 517-608. New York: Cold Spring Harbor Press.

Moon, R. (1993). In pursuit of the functions of the Wnt family of developmentalregulators: insights from Xenopus laevis. BioEssays 15, 91-97.

Morata, G. and Lawrence, L. (1975). Control of compartment developmentby the engrailed gene in Drosophila. Nature 255, 614-617

Murre, C., McCaw, P. S., Vassin, H., Caudy, M., Jan, L. Y., Jan, Y. L.,Cabrera, C. V., Lassar, A. B., Weintraub, H. and Baltimore, D. (1989).Interactions between heterologous helix-loop-helix proteins generatecomplexes that bind specifically to a common DNA sequence. Cell 58, 537-544

Noordermeer, J., Johnston, P. Rijsewijk, F. Nusse, R. and Lawrence, P. A.(1992). The consequences of ubiquitous expression of the wingless gene inthe Drosophila embryo. Development 116, 711-719.

Nusse, R. and Varmus, H. (1992). Wnt genes. Cell 69, 1073-1087. Nüsslein-Volhard, C., Wieschaus, E. and Kluding, H. (1984). Mutations

affecting the pattern of the larval cuticle in Drosophila. I. Zygotic loci on thesecond chromosome. Roux’s Arch. Dev. Biol. 193, 267-282.

O’Brochta and Bryant, P. (1985). A zone of non-proliferating cells at alineage restriction boundary in Drosophila. Nature 313, 138-141.

Patel, N., Schafer, B. Goodman, C. and Holmgren, R. (1989). The role ofsegment polarity genes during Drosophila neurogenesis. Genes Dev. 3, 890-904.

Peifer, M. and Bejsovec, A. (1992). Knowing your neighbours : cellinteractions determine intrasegmental patterning in Drosophila. TrendsGenet. 8, 243-249.

Peifer, M., Rauskolb, C. Williams, M. Riggleman, R. and Wieschaus, E.(1991). The segment polaity gene armadillo interacts with the winglesssignalling pathway in both embryonic and adult pattern formation.Development 11, 1029-1045.

Peifer, M., McCrea, P. D. Green, K. J. Wieschaus, E. and Gumbiner, B. M.(1992). The vertebrate adhesive junction proteins β-catenin and plakoglobinand the Drosophila segment polarity gene armadillo form a multigene familywith similar properties. J. Cell Biol. 118, 681-691.

Peifer, M. and Wieschaus, E. (1990). The segment polarity gene armadilloencodes a functionally modular protein that is the Drosophila homolog ofhuman plakoglobin. Cell 63, 1167-1178.

Perrimon, N. and Mahowald, A. P. (1987). Multiple functions of segmentpolarity genes in Drosophila. Dev. Biol. 119, 587-600.

Perrimon, N. and Smouse, D. (1989). Multiple functions of a Drosophilahomeotic gene zeste white 3 during segmentation and neurogenesis. Dev.Biol. 135, 287-305.

Phillips, R. G. and Whittle, J. R. S. (1993). wingless expression mediatesdetermination of peripheral nervous system elements in late stages ofDrosophila wing disc development. Development 118, 427-438.

Riggleman, R., Schedl, P. and Wieschaus, E. (1990). Expression of theDrosophila segment polarity gene armadillo is posttranscriptionallyregulated by wingless. Cell 63, 549-560.

Rijsewijk, F., Schuermann, M., Wagenaar, E., Parren, P., Weigel, D. andNusse, R. (1987). The Drosophila homologue of the mouse mammaryoncogene int-1 is identical to the segment polarity gene wingless. Cell 50,649-657.

Ripoll, P., El Messal, M., Laran, E. and Simpson, P. (1988). A gradient ofaffinities for sensory bristles across the wing blade of Drosophilamelanogaster. Development 103, 757-767.

Rodriguez, I., Hernandez, R., Modolell, J. and Ruiz-Gomez, M. (1990).Competence to develop sensory organs is temporally and spatially regulatedin Drosophila epidermal primordia. EMBO J. 9, 3583-3592.

Ruel, L., Bourouis, M., Heitzler, P., Pantesco, V. and Simpson, P. (1993).Drosophila shaggy kinase and rat glycogen synthase kinase-3 haveconserved activities and act downstream of Notch. Nature 362, 557-559.

Siegfried, E., Perkins, L. A., Capaci, T. M and Perrimon, N. (1990). Putativeprotein kinase product of the Drosophila segment-polarity gene zeste-white3. Nature 345, 825-829.

Siegfried, E., Chou, T. and Perrimon, N. (1992). wingless signalling actsthrough zeste-white 3, the Drosophila homologue of glycogen synthasekinase 3, to regulate engrailed and establish cell fate. Cell 71, 1167-1179.

636

Simpson, P. and Carteret, C. (1989). A study of shaggy reveals spatialdomains of expression of achaete-scute alleles on the thorax of Drosophila.Development 106, 57-66.

Simpson, P., El Messal, M., Moscoso Del Prado, J. and Ripoll, P. (1988).Stripes of positional homologies across the wing blade of Drosophilamelanogaster. Development 103, 391-401.

Skaer, H. and Martinez Arias, A. (1992). The wingless product is required forcell proliferation in the Malpighian tubule anlage of Drosophilamelanogaster. Development 116, 745-754.

Skeath, J. B. and Carroll, S. B. (1991). Regulation of achaete-scute geneexpression and sensory organ pattern formation in the Drosophila wing.Genes Dev. 5, 984-995.

Spreij, T. H. (1971). Cell death during the development of the imaginal disks ofCalliphora erythrocephala. Neth. J. Zool. 21, 221-264.

Struhl, G. and Basler, K. (1993). Organizing activity of the Wingless proteinin Drosophila. Cell 72, 527-540.

Tiong, S. Y. K. and Nash, D. (1990). Genetic analysis of the adenosine3 (Gart)

region of the second chromosome of Drosophila melanogaster. Genetics124, 889-897.

Tsukamoto, A., Grosschedl, R., Guzman, R., Parslow, T. and Varmus, H.(1988). Expression of the int-1 gene in transgenic mice is associated withmammary gland hyperplasia and adenocarcinomas in male and female mice.Cell 55, 619-625.

van den Heuvel, M., Klingensmith, J., Perrimon, N. and Nusse, R. (1993).Cell patterning in the Drosophila segment: engrailed and wingless antigendistributions in segment polarity mutant embryos. Development 1993Supplement, 105-144.

Williams, J. A., Paddock, S. and Carroll, S. B. (1993). Pattern formation in asecondary field: a hierarchy of regulatory genes subdivides the developingDrosophila wing disc into discrete subregions. Development 117, 571-584.

Woodget, J. R. (1991). A common denominator linking glycogen metabolism ,nuclear oncogenes and development. Trends Biochem. Sci. 26, 177-187.

(Accepted 30 November 1993)


Documents

The wingless signalling pathway and the patterning of the ......622 Mahowald, 1987; Simpson et al., 1988; Perrimon and Smouse, 1989). Therefore, it is not clear at present if the proteins