Dissecting the mechanism of torso receptor activation

Marc Furriols, Andreu Casali, Jordi Casanova*

Centre d’Investigacio´ i Desenvolupament (CSIC), C/Jordi Girona 18-26, 08034 Barcelona, Spain

Received 5 September 1997; revised version received 27 October 1997; accepted 27 October 1997

Abstract

Regulated activation of receptor tyrosine kinases depends both on the presence of the receptors at the cell surface and on the availabilityof their ligands. InDrosophilathe torso (tor) tyrosine kinase receptor is distributed along the surface of the embryo but it is only activated atthe poles by a diffusible extracellular ligand generated at each pole which is trapped by the receptor, thereby impeding further diffusion.However, it is not well understood how this signal is generated, although it is known to depend on the activity of many genes such astorso-like (tsl) andtrunk (trk). To further investigate the mechanism involved in the local activation of the tor receptor we have altered the normalexpression of the tsl protein by generating females in which thetsl gene is expressed in the oocyte under the control of the tor promoterrather than in the ovarian follicle cells. Analysis of the phenotypes generated by this hybrid gene and its interactions with mutations in othergenes in the pathway has enabled us to further dissect the mechanism of tor receptor activation and to define more precisely the role of thedifferent genes acting in this process. 1998 Elsevier Science Ireland Ltd.

Keywords: Torso; Torso-like; Trunk; fs(1)pole hole; fs(1)Nasrat; Signal transduction

1. Introduction

Receptor tyrosine kinases are involved in many signallingpathways in cell proliferation and cell differentiation. It isupon ligand binding that these receptors are able to inducetheir responses; thus, regulated activation of these pathwaysdepends both on the presence of the receptors at the cellsurface and on the availability of their ligands. Indeed, theextent of receptor activation can be set up by the limitedproduction or by the restricted diffusion of their ligands (seeJessell and Melton, 1992 for a review). The torso pathway inDrosophila is a particularly appropriate model to addressthis issue because the torso (tor) receptor is distributed alongall the surface of the Drosophila embryo but it is only acti-vated at the poles, where it specifies development of itsterminal structures (Schu¨pbach and Wieschaus, 1986a;Klingler et al., 1988; Casanova and Struhl, 1989; Sprengeret al., 1989). Activation of the tor receptor has been shownto depend on a diffusible extracellular ligand generated ateach pole which is trapped by the receptor, thereby imped-

ing further diffusion (Sprenger and Nu¨sslein-Volhard, 1992;Casanova and Struhl, 1993).

The situation for generating the terminal signal is not wellunderstood. Two genes,torso-like(tsl) and trunk (trk), havebeen identified which are normally required for activation ofthe tor receptor and have been implicated in the local pro-duction or action of the tor ligand (Casanova and Struhl,1989; Stevens et al., 1990).tsl expression is restricted tospecialised follicle cells at each end of the maturing oocyte(Savant-Bhonsale and Montell, 1993; Martin et al., 1994)and its product appears to be found at both embryonic polesafter fertilisation (Martin et al., 1994). Moreover, itsrestricted expression in these follicle cells appears to becritical for generation of the localised terminal signal inthe embryo because ectopictsl expression during oogenesiscauses central portions of the early embryo to develop term-inal structures (Savant-Bhonsale and Montell, 1993; Martinet al., 1994). By contrast,trk is required in the female germ-line (Schupbach and Wieschaus, 1986b) and it codes for aprotein with similarities to several types of extracellulargrowth factors and likely to be secreted and cleaved (Casa-nova et al., 1995).

Two other genes,fs(1)pole hole(fs(1)ph) andfs(1)Nasra

Mechanisms of Development 70 (1998) 111–118

0925-4773/98/$19.00 1998 Elsevier Science Ireland Ltd. All rights reservedPII S0925-4773(97)00181-0

* Corresponding author. Tel.: +34 34006135; fax: +34 32045904.

(fs(1)Nas), seem also to be required for local activation ofthe tor receptor (Casanova and Struhl, 1989). These genesare unusual because the majority of mutant alleles at bothloci lead to a collapsed egg phenotype, probably due toabnormalities in the vitelline membrane, and only a parti-cular hypomorphic mutation for each gene gives rise to aterminal phenotype (Degelmann et al., 1990). Thus, it hasbeen proposed that these two genes may have a dual func-tion: they may serve to stabilise the vitelline membrane andthey may also be involved in retention or stabilisation of thetsl product in the perivitelline space at the egg pole (Degel-mann et al., 1990).

To further investigate the mechanism involved in thelocal activation of the tor receptor we have altered the nor-mal expression of the tsl protein by generating females inwhich the tsl gene is expressed in the oocyte under thecontrol of thetor promoter rather than in the ovarian folliclecells. We show thattsl expressed in the germ-line cells isfully functional in the mechanism of tor activation, enablingmore precise definition of the spatial and temporal require-ments of the tsl product in the generation of the terminalpattern. Moreover, analysis of the interactions between thishybrid gene and mutations in other genes in the terminalsystem has enabled us to further dissect the mechanism oftor receptor activation. In particular, we discard the possi-bility that trk could act as an oocyte-to-follicle cell signalinstructing tsl expression and we show thatfs(1)ph and

fs(1)Nas actively link tor receptor activation to a mem-brane-bound event.

2. Results

2.1. Ubiquitous expression of tsl in the oocyte

To investigate the requirement oftsl function for tor acti-vation we have generated transgenic flies in which thetslcDNA is under the control of thetor promoter (referred tofrom now on as the tor-tsl construct; see Section 4). Thetorgene is expressed in the germ-line; it is first detected in thenurse cells and once these cells transport their contents intothe oocyte thetor mRNA becomes evenly distributed withinthe oocyte; after egg deposition, the transcript is uniformlydistributed in the embryo (Sprenger et al., 1989). Accord-ingly, we can detect a uniform low level oftsl expression inearly embryos derived from females carrying thetsl geneunder the control of thetor promoter (Fig. 1B).

We have generated 12 lines of transgenic flies that haveindependently integrated the tor-tsl construct. Five of theselines do not show any particular phenotype but the otherseven lines show a variable degree of what has been namedthe splice phenotype, i.e. deletions of the middle segmentsof the embryo (Schu¨pbach and Wieschaus, 1989). Thesplice phenotype originated by the expression of this con-

Fig. 1. Expression oftsl andtll in embryos derived from mothers that expresstsl in the oocyte under the control of the tor promoter. (A) Notsl expression isdetected in wild-type early embryos astsl is expressed during oogenesis only in the follicle cells. (B) Conversely, a uniform level oftsl expression is detectedin early embryos derived from females carrying thetsl gene under the control of thetor promoter. (C) In wild type embryostll expression is restricted to thepoles. (D) However, we observe expandedtll expression in embryos derived from females that expresstsl ubiquitously in the oocyte, an indication ofexpansion of the terminal portions of the body at the expense of the middle body segments.

112 M. Furriols et al. / Mechanisms of Development 70 (1998) 111–118

struct is very variable and also very sensitive to its copynumber (Fig. 2); even the weaker lines produce splice phe-notypes as the copy number of the tor-tsl constructsincreases. This is a phenotype associated with thetor domi-nant mutations that cause an expansion of the terminal por-tions of the body at the expense of the embryonic centralregion (Klingler et al., 1988; Strecker et al., 1989). There-fore, these phenotypes offer a first indication that generalexpression oftsl in the oocyte could result in ubiquitousactivation of the tor receptor. Furthermore, in the mostextreme cases, we can even detect the appearance of ectopicterminal structures, such as filzko¨rpers (Fig. 2D). Three ofthe stronger lines have been chosen for further analysis.

2.2. Uniform expression of tsl in the oocyte is able totrigger unrestricted activation of the tor receptor

The phenotypes associated with the tor-tsl construct sug-gest that uniform expression oftsl in the oocyte is able totrigger activation of the tor receptor all over the embryo. Wehave performed two additional tests of this hypothesis. First,we have corroborated that in these embryos there is indeedan expansion of the terminal portions of the body at theexpense of the middle body segments. Specifically, wehave monitored in these embryos the expression pattern oftailless(tll ), a gene whose expression is regulated by thetortransduction pathway; while in wild type embryostllexpression is restricted to the embryonic poles (Pignoni etal., 1990) and it expands in mutant embryos where there isspatially-indiscriminate activation of the tor pathway(Steingrımsson et al., 1991). Indeed, we observed expandedtll expression in embryos derived from females that express

tsl ubiquitously in the oocyte (Fig. 1D). Second, the pheno-type generated bytsl expression in the germ-line is sup-pressed in embryos derived from females lacking thetorgene (Fig. 3). In addition, those embryos also display thetor− mutant phenotype, as expected from the upstreamrole of tsl in tor receptor activation (Casanova and Struhl,1989). Hence, we can attribute the phenotype associatedwith the tor-tsl construct to unrestricted activation of thetor receptor.

2.3. Germ-line expression of tsl overcomes the absense ofendogenous tsl activity

As mentioned above, a distinct feature oftsl is its somaticexpression in some follicle cells flanking the oocyte(Savant-Bhonsale and Montell, 1993; Martin et al., 1994).In addition, mosaic analysis and transplantation experi-ments have demonstrated that it is in the somatic folliclecells thattsl function is precisely required (Stevens et al.,1990). However, the phenotypes associated with the tor-tslconstruct indicate that the tsl product has the capacity tofunction when supplied from the germ-line. To test whetherthe germ-line expression oftsl can substitute for the endo-genoustsl function we have analysed the progeny fromtslmutant females carrying the tor-tsl construct and mutant fordifferent tsl alleles. In particular, we have used thetsl5allele, an allele of the most severe class oftsl mutants(Savant-Bhonsale and Montell, 1993).

First, we have analysed their cuticle pattern and foundthat they show a variable rescue of thetsl phenotype whichdepends on the copy number of tor-tsl constructs. Weobserved embryos derived fromtsl females carrying one

Fig. 2. Cuticle phenotype of embryos derived from mothers that expresstsl in the oocyte under the control of the tor promoter. (A) Wild-type embryoniccuticle. (B,C,D) cuticle phenotypes originated by the expression of the tor-tsl construct; the phenotypes are variable and very sensible to the number of tor-tslconstructs inserted in the genome (the embryo in B is derived from a female carrying two copies of the tor-tsl construct; the embryos in C and D are derivedfrom females carrying four copies of the tor-tsl construct). Note the deletions of the middle embryonic segments typically associated with thetor dominantmutations and the appearance of ectopic terminal structures, such as filzko¨rpers (fk in D), in the most extreme cases.

113M. Furriols et al. / Mechanisms of Development 70 (1998) 111–118

copy of the tor-tsl construct that display a partial rescue ofthe tsl phenotype without any sign of ectopic activation ofthe tor receptor. Embryos derived fromtsl females carryingtwo copies of the tor-tsl constructs can even show a com-plete rescue of thetsl phenotype as judged by the appear-ance at the posterior cuticle of all the terminal structures(denticle belt of the eighth abdominal segment, filzko¨rperand posterior spiracles, anal pads and tuft) (Fig. 3B,C). Inaddition, some of these embryos also show the abdominaldefects associated with inappropriate activation of the torreceptor in the middle body segments (Fig. 3B).

To further corroborate that germ-line expression oftsl

can substitute for the lack of endogenoustsl function wehave also analysed the pattern ofhuckebein(hkb) expres-sion in embryos derived fromtsl females carrying the tor-tslconstructs.hkb expression is thought to respond to higherlevels of terminal signalling (Furriols et al., 1996) and inwild type embryos it is restricted to the more terminalregions (Bronner and Ja¨ckle, 1991). While there is noexpression ofhkb in the posterior pole of embryos derivedfrom tsl females (Fig. 3D),hkbexpression is restored at theposterior pole by the presence of the tor-tsl construct (Fig.3E), indicating that even high levels of terminal signallingcan be driven bytsl germ-line expression.

Fig. 3. tsl expression in the oocyte rescues the phenotype oftsl mutants but not of mutants of the other terminal genes. (A) Cuticle phenotype of an embryoderived from atsl7 female. Note the absence of the posterior terminal structures (denticle belt of the eighth abdominal segment, filzko¨rpers, anal pads). (B)Cuticle of an embryo from atsl7 female carrying two copies of the tor-tsl construct. A full rescue of the posterior terminal structures can be observed, as wellas the deletions of the middle segments associated with ectopic activation of the tor receptor. (C) The same full rescue can be observed in more detail in anembryo from a mutant female for a strongertsl allele (tsl5) (A8, eight abdominal denticle belt; ps, posterior spiracles; ap, anal pads. t, tuft). (D) In embryosfrom tsl604 femaleshkbexpression is not detected at the posterior pole; expression remains in the anterior domain due to the activity of the anterior system.(E) Rescue of thetsl phenotype can also be observed at the level ofhkbexpression as its posterior domain is present in embryos derived fromtsl5 females thatcarry two copies of the tor-tsl construct. (F) Conversely, no rescue of terminal structures or deletions of the middle segments can be observed in embryosfrom trk1 females that carry two copies of the tor-tsl construct. The same phenotype as in F is observed in embryos from mutant females fortorXR1, fs(l)ph1901

and fs(1)Nas211 that also carry two copies of the tor-tsl construct.

114 M. Furriols et al. / Mechanisms of Development 70 (1998) 111–118

2.4. Triggering of tor activation by tsl expressed in thegerm-line requires the product of the trk gene

trk is another gene that is essential for normal activationof the tor receptor (Casanova and Struhl, 1989). In contrastto tsl, trk is expressed in the germ-line and we have shownthat it codes for a protein with similarities to several types ofextracellular growth factors and is likely to be the tor ligand(Casanova et al., 1995). However, it has also been proposedthat tsl might itself be the tor ligand (Martin et al., 1994).According to this view,trk could have a role upstream of thelocalised expression oftsl, e.g. as an oocyte-to-follicle cellsignal involved in specifying the differentiation of specia-lised follicle cells at each end of the oocyte. Similar induc-tion events have already been documented from the oocyteto specify different fates among follicle cells (Schu¨pbach,1987; Gonza´lez-Reyes and St. Johnston, 1994). To ascertainthe role oftrk we have investigated whethertsl expression inthe germ-line still requirestrk function in triggering torreceptor activation. Indeed, we have found that embryosfrom trk mutant females carrying the tor-tsl constructshow neither the splice phenotype nor any rescue of thetrk phenotype (Fig. 3), indicating thattrk function is stillrequired for the activation of tor bytsl even when it isexpressed in the germ-line under the control of a heterolo-gous promoter. Therefore,trk is required in a process otherthan for induction oftsl expression. Accordingly,tsl isexpressed in the border cells and in a subset of the posteriorfollicle cells in trk mutant ovaries (Fig. 4).

2.5. Triggering of tor activation by tsl expressed in thegerm-line requires the product of the fs(1)ph and fs1)Nasgenes

Two other genes of the terminal system,fs(1)ph andfs(1)Nas, are special because the majority of mutant allelesat both loci lead to a collapsed egg phenotype, probably dueto abnormalities in the vitelline membrane; only a particularhypomorphic mutation for each gene produces a specificterminal phenotype (Degelmann et al., 1990). We haveshown that they seem to be required for ligand-inducedactivation of the tor receptor sincetor constitutive mutationssuppress their terminal phenotypes (Casanova and Struhl,1989). Accordingly, it has been proposed that these twogenes may have a dual function; they may serve to stabilisethe vitelline membrane and they may also be involved inretention or stabilisation of the tsl product in the perivitel-line space at the egg pole (Degelmann et al., 1990). Inparticular, as it has been proposed that thetsl gene productis secreted from the follicle cells to the vitelline membraneor perivitelline space (Savant-Bhonsale and Montell, 1993;Martin et al., 1994), thefs(1)ph andfs(1)Nasgene productscould be involved in the process of transference of the tslproduct from the follicle cells to the germ-line. Thus, theirrole in the process of tor receptor activation could be dis-pensable if the tsl product is supplied from the germ-line.

However we have found the opposite, as the tor-tsl constructhas no effect in embryos from eitherfs(l)ph or fs(l)Nasmutant females (Fig. 3), indicating that thefs(1)ph and thefs(1)Nasfunctions are still required for tor receptor activa-tion even whentsl is expressed in the germ-line and itsproduct is likely to be secreted from the oocyte ratherthan from the follicle cells.

3. Discussion

tsl is the unique gene in the tor pathway that hasbeen found to have restricted expression. Given the uni-form distribution of the other known components of thepathway, the restricted expression oftsl in the folliclecells at each end of the oocyte during oogenesis has beensingled out as the determinant for local activation of the torreceptor at both poles of the embryo. In agreement with thisrole, ectopic expression oftsl by means of a heat-shockinducible promoter is able to activate the tor receptor allover the embryo (Savant-Bhonsale and Montell, 1993; Mar-tin et al., 1994). Previous reports have also been publishedindicating thattsl RNA injected at the posterior pole ineggs from tsl mutant females could correct thetsl pos-terior phenotype (Martin et al., 1994). We have developeda new system of general expression oftsl in the germ-linecells which has allowed us to study more precisely the spa-tial and temporal requirements fortsl and the relationsbetweentsl and other genes also required for tor receptoractivation.

3.1. Spatial and temporal requirements of tsl in inducingtor receptor activation

The results reported indicate that the somatic expressionof tsl in a subset of follicle cells can serve to ensure that itsproduct is strictly confined to a restricted domain and thus,that the tor receptor is only activated at the embryonic poles.However, expression oftsl in the somatic cells is not criticalfor its functional role in tor activation since the tsl productsupplied from the germ-line is equally functional. The nat-ure of thetsl gene product as a protein likely to be secretedcould account for this ability of the tsl protein to be func-

Fig. 4. tsl is normally expressed intrk mutant ovaries as visualised in anegg chamber around stage 10.tsl RNA can be detected in the border cells(bc), the posterior follicle cells (pfc) and in the anterior centripetal cells(acc) (only some of the anterior centripetal cells are in focus in this image).

115M. Furriols et al. / Mechanisms of Development 70 (1998) 111–118

tional when expressed either in the somatic or in the germ-line cells (Martin et al., 1994; and see below).

The finding that the tor protein is expressed during arestricted period of time just before and during the syncytialblastoderm indicates that activation of the receptor occursduring this period (Casanova and Struhl, 1989), a conclu-sion that agrees with the results of cytoplasmic rescueexperiments (Klingler et al., 1988) and with the time courseof tor phosphorylation (Sprenger et al., 1993). The timelapse betweentsl expression in the follicle cells and toractivation in the oocyte when the follicle cells are no longerpresent, has prompted the suggestion that a mechanism mustexist that ensures that the spatial asymmetry in the somaticcells is transmitted during oogenesis and early embryogen-esis until activation of the tor receptor takes place (Casa-nova and Struhl, 1989). However it is not clear whethertslitself is required during embryogenesis at the time of acti-vation of the tor receptor. Our experiments indicate thatexpression oftsl at the time of tor expression is sufficientto fulfil its role in tor activation and suggest that the tslproduct (and not a downstream function of the gene) isfunctional around the period of time when the tor receptoris present at the blastoderm membrane.

3.2. tsl expression and terminal pattern

Two additional remarks can be drawn from the uniformexpression oftsl in the germ-line. First, uniform low levelsof tsl elicit the tor pathway response just at the embryonicpoles; only increasingly high levels oftsl are able to alterpattern in the central regions of the embryo. This observa-tion is in agreement with all the previous reports that haveunderlined the distinct outcome of the tor pathway whenequally activated along all over the embryo (Casanovaand Struhl, 1989; Lu et al., 1993; Tsuda et al., 1993; Casa-nova et al., 1994). These results have been interpreted as anindication that centrally-located nuclei might respond dif-ferently to activity of the tor pathway from nuclei located atthe poles, probably due to interactions between terminal andcentral gap genes (Casanova et al., 1994). Second, highlevels of uniform tsl expression not only disturb centralsegmentation but are also able to generate ectopic terminalstructures, such as filzko¨rpers, indicating that ectopictslexpression is able to induce enough levels of tor activityto elicit the morphogenetic responses of the terminal path-way. In a previous report, expression oftsl driven by a heat-shock promoter was found not to be able to generate ectopicterminal structures (Savant-Bhonsale and Montell, 1993).The critical levels achieved by the two systems drivingtslectopic expression could account for these different results.

3.3. trk does not work as an oocyte-to-follicle cell signalinstructing tsl expression

Local activation of the tor receptor requires thetrk gene,which codes for a protein with similarities to several extra-

cellular growth factors. However, it is the localised expres-sion of tsl in the follicle cells that is responsible for therestricted activation of the tor receptor in the embryo. Asimilar mechanism operates in the dorsoventral systemwhere spatial information from the follicle cells is respon-sible for the formation of the embryonic dorsoventral axis(Schupbach, 1987). In the dorsoventral system the asymme-try in the follicle cells is induced by a signal from theoocyte: it is the germ-line expression ofgurken (grk), agene coding for a growth factor-like protein, which signalsto dorsal follicle cells leading ultimately to the ventral pro-duction of an active form of spa¨tzle (spz), the putativeligand of the Toll receptor (Neuman-Silberberg andSchupbach, 1993; Morisato and Anderson, 1994). Basedon the parallelisms between the two systems,trk could beplaying a similar role togrk as a germ-line inducer ofrestricted tsl expression in the follicle cells. Our resultsclearly argue against that model, in thattsl expression isnormal in trk mutant ovaries andtsl still requires thetrkgene product to be effective in the mechanism of tor recep-tor activation even when expressed out of the follicle cellsand under a heterologous promoter. Conversely, theseresults suggest a more direct role of trk in the activationof the tor receptor and strengthen the suggestion that trkitself could act as a ligand for the tor receptor (Casanovaet al., 1995).

3.4. fs(1)ph and fs(1)Nas actively link tor activation to amembrane-bound event

Our results assign an active role to the vitelline mem-brane in tor activation. Tsl secreted from the oocyte isable to trigger tor activation but still requiresfs(1)ph andfs(1)Nas, two genes associated with the integrity of thevitelline membrane. Hence, the products of thefs(1)phandfs(1)Nasgenes are not involved in the mere transferenceor holding of tsl from the follicle cells. Instead, our resultsargue for a membrane-bound activity in the mechanism oftor receptor activation. Our results also argue against anypolar specificity of thefs(1)ph and fs(1)Nasgene productssince they are also required for activation of the tor receptorin the middle region of the embryo by ectopic expression oftsl. Although a direct interaction of tsl with either fs(1)ph orfs(1)Nas could account for these observations, this interac-tion could occur with another component of the mechanismof tor receptor activation. However, since the tsl productcontains leucine-rich repeats implicated in mediating pro-tein-protein and protein-lipid interactions (Savant-Bhonsaleand Montell, 1993), it is tempting to postulate that the tslproduct itself could require to be membrane-bound in orderto be active. Two additional observations are worth consid-ering. First, it has been proposed that thefs(1)ph andfs(1)Nas products might be assembled into the oocyteplasma membrane or transported into the perivitellinespace where they could stabilise the vitelline membrane(Spradling, 1993). Second, R. Ollo and co-workers (Martin

116 M. Furriols et al. / Mechanisms of Development 70 (1998) 111–118

et al., 1994) have made the unique observation that tsl pro-tein is detected at the surface of the syncytial blastoderm inthe form of symmetrical caps at both poles. Although it hasnot been possible to reproduce this last result, both observa-tions make it possible to speculate that localised membraneanchoring of tsl, mediated directly or indirectly by thefs(l)ph or fs(l)N as gene products could be essential fortslfunction and hence for the process of tor activation.

Previous characterisation of thetrk gene raised the pos-sibility that its product might be secreted intact into theperivitelline fluid layer and locally activated by proteolyticcleavage to generate the ligand for the tor receptor (Casa-nova et al., 1995). Our present results are consistent withthis prospect. The existence of a membrane-bound event inthe mechanism of tor activation fits very well with the puta-tive proteolysis of thetrk gene product; the cleavage site inthe trk gene product is similar to the one present in the thirdcomponent of the complement system and cleavage in thecomplement is an event that occurs in membrane-boundcomplexes (see Mu¨ller-Eberhad, 1988 for a review).Furthermore, the local confinement of this membrane-bound activity could be the determinant of the spatialrestriction of tor activation. Indeed, the tsl product couldconstitute the link between the membrane-bound eventand local restriction in tor activation, perhaps by means ofproteolysis of the trk product. Alternatively, it is also pos-sible that tsl itself could directly bind to the tor receptor inthe embryonic poles. Unfortunately, the present tools do notmake it feasible to ascertain whether the possible binding oftsl at the embryonic poles is dependent on any of the othergene functions of the terminal system. Further experimentswill be required to discriminate between these possibilities.

4. Experimental procedures

4.1. Fly strains

trk1 is a mutation that results from a premature stop codonand behaves as a null allele (Casanova et al., 1995).torXR1 isa small deficiency that deletes thetor gene (Sprenger et al.,1989). tsl5, tsl3, tsl2 are amino acid substitution mutationsordered according to their decreasing severity (Savant-Bhonsale and Montell, 1993),tsl7 is a spontaneous mutationand tsl604 is a P element induced mutation (Martin et al.,1994). fs(1)ph1901 and fs(1)Nas211 are mutations in thosegenes displaying the specific terminal phenotype (Degel-mann et al., 1990).

4.2. tor-tsl construct

An XbaI-EcoRI fragment from thetsl cDNA comprisingall the coding sequence (Savant-Bhonsale and Montell,1993) was fused to an EcoRI-XbaI genomic fragment of3.5 Kb from thetor promoter (Sprenger et al., 1989) andthe resulting fusion was inserted into the P(ry+) transforma-

tion vector C20.

4.3. In situ hybridisation

tsl, tll andhkbprobes were generated from cDNA clonesprovided by D. Montell, J. Lengyel and H. Ja¨ckle, respec-tively. Whole-mount in situ hybridisations were done fol-lowing the method of Tautz and Pfeifle (1989) with minormodifications.

Acknowledgements

We thank H. Ja¨ckle, J. Lengyel, A. Mahowald and D.Montell for providing flies and materials, and A.Gonzalez-Reyes and our colleagues in the lab for manydiscussions and critical comments on the manuscript. Wealso thank Nicola´s Martın for his technical assistance. A.C.is supported by a fellowship from the Ministerio de Educa-cion y Ciencia and M.F. is supported by a fellowship fromthe Maratode TV3 and the Fundacio´n RICH. This work wassupported by the DGICYT and the CIRIT.

References

Bronner, G., Ja¨ckle, H., 1991. Control and function of terminal gap geneactivity in the posterior pole region of theDrosophilaembryo. Mech.Dev. 35, 205–211.

Casanova, J., Struhl, G., 1989. Localized surface activity oftorso, a recep-tor tyrosine kinase, specifies terminal body pattern inDrosophila. GenesDev. 3, 2025–2038.

Casanova, J., Struhl, G., 1993. Thetorso receptor localizes as well astransduces the spatial signal specifying terminal body pattern inDrosophila. Nature 362, 152–155.

Casanova, J., Llimargas, M., Greenwood, S., Struhl, G., 1994. An onco-genic form of humanraf can specify terminal body pattern inDrosophila. Mech. Dev. 48, 59–64.

Casanova, J., Furriols, M., McCormick, C.A., Struhl, G., 1995. Similaritiesbetween trunk and spa¨tzle, putative extracellular ligands specifyingbody pattern inDrosophila. Genes Dev. 9, 2539–2544.

Degelmann, A., Hardy, P.A., Mahowald, A.P., 1990. Genetic analysisof two female-sterile loci affecting eggshell integrity and embryonicpattern formation inDrosophila melanogaster. Genetics 126, 427–434.

Furriols, M., Sprenger, F., Casanova, J., 1996. Variation in the number ofactivated torso receptors correlates with differential gene expression.Development 122, 2313–2317.

Gonzalez-Reyes, A., St. Johnston, D., 1994. Role of oocyte position inestablishment of anterior-posterior polarity inDrosophila. Science 266,639–642.

Jessell, T.M., Melton, D.A., 1992. Diffusible factors in vertebrate embryo-nic induction. Cell 68, 257–270.

Klingler, M., Erdelyi, M., Szabad, J., Nu¨sslein-Volhard, C., 1988. Functionof torso in determining the terminal anlagen of theDrosophilaembryo.Nature 335, 275–277.

Lu, X., Chou, T.B., Williams, N.G., Roberts, T., Perrimon, N., 1993.Control of cell fate determination by p21ras/Ras1, an essential compo-nent of torso signaling inDrosophila. Genes Dev. 7, 621–632.

Martin, J.-R., Raibaud, A., Ollo, R., 1994. Terminal pattern elements inDrosophilaembryo induced by thetorso-likeprotein. Nature 367, 741–745.

117M. Furriols et al. / Mechanisms of Development 70 (1998) 111–118

Morisato, D., Anderson, K.V., 1994. Thespatzle gene encodes a compo-nent of the extracellular signaling pathway establishing the dorso ventralpattern of theDrosophilaembryo. Cell 76, 677–688.

Muller-Eberhad, H.J., 1988. Molecular organization and function of thecomplement system. Ann. Rev. Biochem. 57, 321–347.

Neuman-Silberberg, F.S., Schu¨pbach, T., 1993. TheDrosophiladorsoven-tral patterning genegurken produces a dorsally localized RNA andencodes a TGFalfa-like protein. Cell 75, 165–174.

Pignoni, F., Baldarelli, R.M., Steingrı´msson, E., Diaz, R.J., Patapoutian,A., Merriam, J.R., Lengyel, J.A., 1990. TheDrosophilagenetailless isexpressed at the embryonic termini and is a member of the steroidreceptor superfamily. Cell 62, 151–163.

Savant-Bhonsale, S., Montell, D.J., 1993.torso-likeencodes the localizeddeterminant ofDrosophila terminal pattern formation. Genes Dev. 7,2548–2555.

Schupbach, T., 1987. Germ line and soma cooperate during oogenesis toestablish the dorsoventral pattern of egg shell and embryo inDrosophilamelanogaster. Cell 49, 699–707.

Schupbach, T., Wieschaus, E., 1986a. Maternal-effect mutations alteringthe anterior-posterior pattern of theDrosophila embryo. Roux’s Arch.Dev. Biol. 195, 302–317.

Schupbach, T., Wieschaus, E., 1986b. Germ-line autonomy of maternaleffect mutations altering the embryonic body pattern ofDrosophila.Dev. Biol. 113, 443–338.

Schupbach, T., Wieschaus, E., 1989. Female sterile mutations on the sec-ond chromosome ofDrosophila melanogaster. Maternal effect muta-tions. Genetics 121, 101–117.

Spradling, A., 1993. Developmental genetics of oogenesis. In: Bate, A.,Martınez-Arias, A. (Eds.), The Development ofDrosophila melanoga-

ster. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY, pp.1–70.

Sprenger, F., Stevens, L.M., Nu¨sslein-Volhard, C., 1989. TheDrosophilagene torso encodes a putative receptor tyrosine kinase. Nature 338,478–483.

Sprenger, F., Nu¨sslein-Volhard, C., 1992. Torso receptor activity is regu-lated by a diffusible ligand produced at the extracellular terminalregions of theDrosophilaegg. Cell 71, 987–1001.

Sprenger, F., Trosclair, M.M., Morrison, D.K., 1993. Biochemical analysisof torso and D-raf duringDrosophilaembryogenesis: implications forterminal signal transduction. Mol. Cell. Biol. 13, 1163–1172.

Stevens, L.M., Fronho¨fer, H.G., Klingler, M., Nusslein-Volhard, C., 1990.Localized requirement fortorso-like expression in follicle cells fordevelopment of the terminal anlagen of theDrosophila embryo.Nature 346, 660–663.

Steingrı´msson, E., Pignoni, F., Liaw, G.-J., Lengyel, J.A., 1991. Dual roleof the Drosophila pattern genetailless in embryonic termini. Science254, 418–421.

Strecker, T.R., Halsell, S.R., Fisher, W.W., Lipshitz, H.D., 1989. Recipro-cal effects of hyper- and hypoactivity mutations in theDrosophilapat-tern genetorso. Science 243, 1062–1066.

Tautz, D., Pfeifle, C., 1989. A non-radioactive in situ hybridization methodfor the localization of specific RNAs inDrosophila embryos revealstranslational control of the segmentation gene hunchback.Chromosoma 98, 81–85.

Tsuda, L., Inoue, Y.H., Yoo, M.-A., Mizuno, M., Hata, M., Lim, Y.-M.,Adachi-Yamada, T., Ryo, H., Masamune, Y., Nishida, Y., 1993. Aprotein kinase similar to MAP kinase activator acts downstream ofthe raf kinase inDrosophila. Cell 72, 407–414.

118 M. Furriols et al. / Mechanisms of Development 70 (1998) 111–118