Developmental control by the Drosophila EGF receptor homolog DER

Itt V E S 30 Jermyn, K.A. and Williams, J.G. (1991) Development 111, 779-787 31 Ceccarelli, A., Mahbubani, H. and Williams, J.G. (1991) Cell 65, 983-990 32 Richardson, D.L., Hong, C.B. and Loomis, W.F. (1991) Dev. Biol. 144, 269-280 33 Kay, R.R. (1989) Development 107, 753-759 34 Kopachik, W. et al. (1983) Cell 33, 397-403 35 Reymond, C.D. et al. (1986) Nature 323, 340-343 36 Mehdy, M., Ratner, D. and Firtel, R.A. (1983) Cell 32, 761-771 37 Jermyn, K.A., Berks, M., Kay, R.R. and Williams, J.G. (1987) Development 100, 745-755 38 Morris, H.R. et al. (1987) Nature 328, 811-814 39 Berks, M. and Kay, R.R. (1990) Development 110, 977-984 40 Ginsburg, G. and Kimmel, A.R. (1989) Proc. NatIAcad. Sci. USA 86, 9332-9336 41 Wang, M., van Driel, R. and Schaap, P. (1988) Development 103, 611-618

T h e DER protein has the canonical structure of a receptor tyrosine kinase belonging to the subclass that includes the vertebrate EGF receptor 1,2. While three members of this class have been identified in vertebrates, in Drosophila DER appears to be unique. The protein has a single transmembrane domain separating the intracellular and extracellular domains. The highest degree of sequence conservation is found in the intracellular kinase domain. The DER protein has also been shown to have tyrosine kinase activity, and can phosphorylate itself3. The carboxy-terminal region, beyond the kinase domain, contains the sites for autophosphorylation and shows the lowest degree of structural conservation, although this region plays a pivotal role in signal transduction (see below). The extracellular portion comprises four subdomains. Two are cysteinerich and are likely to generate the scaffold of the ligand-binding domain, but not its recognition specificity. The cysteine-rich subdomain closer to the membrane is the longer of the two. Interestingly, this structural feature is also found in the Caenorhabditis elegans EGF receptor homolog, encoded by the let-23 gene ~, but not in the vertebrate counterparts, indicating that the ancestral form of these receptors was probably similar to DER. The structure of DER makes it a likely candidate for transmitting regulatory signals into the cell. What is the function of these signals during Drosophila development? This article describes the genetic analysis of the locus, pointing to a diverse set of roles for the DER protein during embryonic and postembryonic development.

Mutations in the DER locus Screens for mutations in the DER locus led to the isolation of embryonic-lethal alleles whose phenotype turned out to be identical to that of a previously described set of mutations designated f a i n t little ball ~flb)3,5. In parallel, mutations isolated on the basis of their interesting postembryonic phenotype in the eye TIG NOVEMBER/DECEMBER ~'t991 Etse~icr Science Publishers Ltd ( I K I 01(~

9479,91:$0200

42 Peters, D.J.M. etal. (1989) J. CellSci. 93, 205-210 43 Blumberg, D.D., Comer, J.F. and Higinbothem, K.G. (1988) Dev. Genet. 9, 359-369 44 Haberstroh, L. and Firtel, R.A. Development (in press) 45 Hjorth, A.L., Pears, C., Williams, J.G. and Firtel, R.A. (1990) Genes Dev. 4, 419-432 46 Felt, I.N., Bonner, J.T. and Suthers, H.B. (1990) Dev. Genet. 11,442--446 47 Insall, R. and Kay, R.R. (1990) ~ B O J . 9, 3323-3328 . 48 Gomer, R.H. and Firtel, R.A. (1987) Science 237, 758-762 49 McDonald, S.A. and Dumston, A.J. (1984)J. Cell Sci. 66, 195-204 50 Weijer, C.J., Duschl, G. and David, C.N. (1984)J. Cell Sci. 70, 133-145 R.A. FIRTEL IS IN THE DEPARTMENT OF BIOLOGY, CENTER Fog MOLECULAR G£NETIC~ UNIVERSITY OF CALIFORNIA, SAN DIEG~

LaJoLL4, CA 92093-0634, USA.

Developmental control by the DrosophilaEGF receptor h0m010gDER BEN-ZION SHILOAND EREZ RAZ

The identification of receptor tyrosine kinases in Drosophila has provided an opportunity to study the requirement for these proteins during the development of a ~ e U u l a r organism. Genetic attalysis of the function of the Drosophila epidermal growth factor (EGF) receptor bomolog (DER) has revealed an extremetF diverse set of roles for this protein throughout the life cycle of the orgaaism, for example in eye development and in the establishme~ of dorsoventral polarity in the oocyte. We discuss the possible basis for the pleiotropic activity of DER, and the similarities and differences in the function of the homo4ogoas proteins in other invertebrates and vertebrates. or in the ovary were shown to reside at the same genetic locus ~ . Severe mutation: f a i n t little ball The embryonic phenotype of null or severe alleles of the DER locus was termed f a i n t little ball (fib), for good reason3,5,s: the cuticle of mutant embryos has a rounded shape because the germ band fails to retract and head structures are absent. The cuticle also lacks the ventral denticle belts (Fig. la, b). Other characteristics of the phenotype include severe collapse of the central nervous system (CNS), discontinuities of the longitudinal axon tracts and fusion of commissures 3 (Fig. lc, d). Both head and CNS structures appear to develop normally and collapse only at a later stage of embryogenesis3,9. The description of the fib phenotype raises a new set of questions. Is the receptor involved in determining

1991 vot. 7 NO. 11/12

[]~EVIEWS J~

J "~'

• • •,

~,~ ,~

• ,~,,,c

~

'

~

d

]

C FIGE!

The cuticle and central nervous system (CNS) phenotypes of fib embryos. (a) Cuticle of a wild-type embryo. (b) Cuticle of a null fib embryo. Note the absence of head structures, failure of the germ band to retract, and the absence of ventral denticle belts. (c) CNS of a wild-type embryo stained with BPI02 monoclonal antibody, which recognizes a carbohydrate epitope present exclusively on cells of the CNS. (d) CNS of a null fib embryo. Note the discontinuities in longitudinal axon tracts and the nonretracted germ band. The anterior end of the embryo is at the left side, and ventral is at the bottom. Abbreviations: db, denticle belts; c, commissures; 1, longitudinal axon tracts. cell fate in the affected tissues, or is it required later in embryogenesis for survival and maintenance of these tissues? The availability of a temperature-sensitive allele (fib 1F26) has allowed the complex embryonic fib phenotype to be dissected, and the temporal requirements for DER activity to be determined (R. Clifford and T. Schtipbach, pers. commun.; E. Raz and B. Shilo, submitted). Although the disintegration of tissues in fib embryos occurs late in embryogenesis, the actual function of the receptor is required very early in embryonic development. For example, the collapse of the CNS can be prevented by providing DER activity early in development, at the time when the neuroblasts are delaminating from the neuroectoderm. It thus appears that in embryogenesis DER is mediating processes that determine cell fate.

Determining the number of photoreceptor clusters A special class of dominant gain-of-function mutations in DER (Ellipse, Elp) leads to a specific defect in the development of the c o m p o u n d eye. Elp flies have smaller eyes with fewer ommatidia (in homozygous Elp flies, as few as 10% of the normal number). However, the structure of the ommatidia that do develop appears normal 6. To extrapolate from the Ellipse phenotype to the normal role of DER in eye development, it is important to define the nature of the Elp

mutations. Elp was shown to be a semidominant mutation that is sensitive to the activity of the DER allele on the homologous chromosome. The severe eye phenotype is seen only in homozygous flies, while heterozygous flies have rough eyes that are only slightly smaller than wild type. In addition, when an Elp allele is present in trans to a null allele of DER, a wild-type eye phenotype is observed. Elp therefore appears to represent only a subtle deregulation or hyperactivation of the normal pathway, leading to a reduced number of photoreceptor clusters. In wildtype eye disks, the number and spacing of photoreceptor clusters may be established by a process of lateral inhibition, in which the differentiated photoreceptor cells inhibit their neighbors from assuming a similar fate. It is possible that DER transmits these inhibitory signals to the nondifferentiated cells. Tile Elp mutation provides a useful genetic tool for the isolation of genetic loci that interact with DER or with genes encoding components of its signal transduction pathway. The dependency of the Elp phenotype on the other DER allele suggests that the DER signal transduction pathway in Elp heterozygous flies is just above the threshold. The Elp phenotype is thus likely to be sensitive to the gene dosage of other elements in its signalling pathway. This feature of Elp has been used to ask if there are common steps in the

llG NOVEMBER/DECEMBER1991 VOL. 7 NO. 11/12

H~EVIEWS

~ )i ~! i~¸¸3i¸ i ~! ~ ? i i i~i!~ ~ i > !

signalling pathways of different receptor tyrosine kinases in Drosophila, such as sevenless and DER. Indeed, a gain-of-function mutation in a gene termed Son ofsevenless (Sos), isolated on the basis of its ability to suppress the sevenless eye phenotype, could also increase the severity of the Elp phenotype 1°. A loss-offunction mutation in the Sos locus resulted in the opposite set of phenotypes: an increase in the severity of sevenless and a suppression of the klp phenotype (Ref. 10 and M. Simon and G. Rubin, pers. commun.).

Establishment of dorsoventral polarity in the egg A third class of mutations in the DER locus is torpedo (top). These mutations are recessive, and lead to female sterility due to the production of ventralized eggs in ventralized egg shells 7. Extensive analyses, including pole cell transplantation and germ-line mosaics, have shown that in the ovary DER is required in the follicle cells surrounding the oocyte, but not in the oocyte itself 7. Immunolocalization of DER has indeed shown that it is expressed in the follicle cells but not in the oocyte, and is found predominantly in the membrane of the follicle cells facing the oocyte (R. Schweitzer, N.B. Zak and B-Z. Shilo, unpublished). Epistatic relationships with other mutations that give rise either to a similar phenotype, or alternatively to an opposite phenotype, have suggested a model in which the signal for dorsoventral polarity is initiated in the oocyte n (perhaps due to the asymmetric localization of the oocyte nucleus at the dorsal side). The cue for dorsoventral polarity is likely to be the asymmetric distribution of the active form of the DER ligand within the oocyte. The signal is transmitted to the follicle

cells, where it is received by DER. DER is thus activated only at the dorsal side and its activation leads to development of dorsal follicle cells. If DER is not activated, follicle cells follow the default pathway, ventralization 11. In wild-type ovaries, one could envisage a scenario in which variations in the level of DER activation in each of the follicle cells, depending upon its dorsoventral position, could elicit a graded set of responses. Thus, by transmitting the coarse dorsoventral information from the oocyte, which is a single large cell, to the grid of about 1000 follicle cells surrounding it, the information becomes more refined and compartmentalized. The function of DER is not restricted to the three phenotypes described above. A systematic screen for additional mutant alleles has shown that it is also important for patterning and development of certain imaginal disks la. Table 1 summarizes the known functions of DER surmised from the genetic analysis of the different alleles.

Multiple pathways for signal transductionby DER? The pleiotropic role of DER during development raises the question of whether the diverse set of processes it regulates are driven by the same signal transduction pathway, which can lead to different consequences depending upon the cellular context in which it is activated. Mternatively, some of the specificity may be generated by the use of tissue-specific components (e.g. ligands or substrates) in the signalling pathway. The availability of multiple alleles in the DER locus allows this issue to be addressed genetically. If

TIG N O V E M B E R / D E C E M B E R 1 9 9 1

VOL. 7 N O . 1 1 / 1 2

~]~EVIEWS the same signal transduction pathway is used in 1 2 3 all cases, the basis for differences in the tissues and stages affected by the various alleles must lie in the different sensitivity of these stages to the level of DER activity. The implication is that one should observe a nested set of responses in a phenotypic series of DER alleles. The tissue most sensitive to DER activi~/should always be affected, while Trans-phosphorylation Substrate phosphorylation the less sensitive tissues would be affected only in alleles that reduce the level of DER activity more drastically. Alternatively, if there are differences in the components of the DER pathway in the various tissues, it should be possible to identify DER alleles that are defective in only ............ ~ ~ P~ a single or a limited set of processes, for each of the functions of the Class A Class B protein. It has not been possible FTGM to separate genetically the A model for interallelic complementation of fib mutations. Signal transduction by receptor different roles of DER tyrosine kinases is initiated by ligand binding, leading to aggregation and trans-phosphorylation (a2). The activated receptors are then capable of recognizing and phosphorylating cellular during embryonic developsubstrates, thus transmitting the signal into the cell (a3). In the case of complementation ment. The phenotypic between two classes of fib mutations, one group (class A) could be defective in its ability to series of fib alleles shows a recognize and phosphorylate exogenous substrates. Conversely, the second group (class B) graded severity of all ascould be defective in the trans-phosphorylation process. In embryos carrying one Jib allele of pects of the embryonic each class, the heterodimers formed after ligand binding are partially functional. The class A phenotype, suggesting that protein is capable of trans-phosphorylating the class B protein (b2). Once phosphorylated, the a single signal transduction class B protein can then associate with and phosphorylate exogenous substrates (b3), thus pathway is involved. A difrestoring biological activity. The sites of missense mutations in the two class B alleles were ferent conclusion was shown to be close to each other, and are indicated by the white circles within the coding reached when the role region for the kinase domain (dark rectangle). Substrates are indicated by the tinted circles. of DER was examined throughout the life cycle of the fly. Complementation tests between various DER signal transduction mechanism by this receptor alleles suggest that there may indeed be specific sig- family 14. Activation of the kinase by the ligand appears nalling elements in the DER pathway at different to be indirect, that is, binding of the ligand affects stages. Clifford and Schtipbach 12 have identified alleles only the extracellular domain and promotes aggrethat preferentially affect embryogenesis, while others gation of receptors. Once two kinase domains are in have been shown to have selective effects on particuclose proximity, they transduce the signal by a twolar imaginal disks. The differential mutability of DER phase mechanism, requiring catalytic activity at each functions can thus result in positive complementation, of the two steps. First, autophosphorylation sites on w h e n two such alleles are crossed. It is interesting to the carboxy-terminal tail are phosphorylated in trans note that a similar observation was made in C. elegans, by the kinase domain of the partner receptor. The where the EGF receptor homolog was also dissected phosphotyrosine adducts provide a 'magnet' for asgenetically through the let-23 set of mutations 13. In this sociation with substrate proteins. A diverse array of system the effects of the receptor on vulval induction, substrate proteins shares a c o m m o n structural motif formation of male spicules or hermaphrodite fertility termed SH2, which has been shown to be responsible for the recognition of the phosphorylated tyrosine appear to be differentially mutable. residues on the kinase molecule 15. Once the substrate Signal transductionby receptortyrosinekinases proteins are associated with the receptor, they are An elaborate set of studies on the vertebrate recepphosphorylated by the kinase on tyrosine residues, and thus activated (Fig. 2a). tor tyrosine kinases has led to the notion of a universal

(a)

,t,t

(bl

"nc, NOVEMBF~R/DECEMSER1991 VOL. 7 NO. 11/12

[BEVIEWS

Since the signalling pathway of receptor tyrosine kinases appears to work by dimerization, it was logical to expect that among the various f i b alleles induced by treatment with ethylmethane sulfonate (EMS), several heteroallelic combinations would show positive complementation, since each mutant receptor in the dimer may be defective in a different phase of signalling. For the assay, it was important to concentrate on a single developmental stage (embryogenesis), to avoid complications of genetic complementation due to differentially mutable, stage-specific functions of the receptor mentioned earlier. Indeed, several such positive combinations were identified, and two classes of complementing alleles were defined 16. Interestingly, the product of one of the alleles involved in positive complementation has an intact kinase domain, but lacks the entire carboxy-terminal tail, including the autophosphorylation sites, and presumably cannot associate with its substrates. This allele may represent a class o f f i b mutations that give rise to receptors that can recognize the tail of the partner as a substrate, but do not recognize or associate with exogenous substrates. This class can be complemented by a second class of alleles whose products show the opposite properties. Thus, when the two types of mutant DER protein are expressed in t h e same embryo, the pathway is restored. A scheme for the complementation by the two classes is shown in Fig. 2b.

General conclusions and open questions Although the DER protein has been shown to be involved in multiple, seemingly unrelated processes, in cases where a detailed phenotypic description exists a common unifying principle emerges. Analyses of the various aspects of the f i b embryonic mutant phenotype, or of the postembryonic oogenesis and eye phenotypes, suggest that DER is crucial for the transmission of signals for determination of cell fate. In C. elegans, the let-23 gene product also appears to play a pivotal role in developmental decisions, demonstrated most dramatically by its involvement in the induction of vulval fate 17. It is striking that DER functions are not involved in transmission of mitogenic signals, traditionally associated with growth factor receptors of the EGF receptor class in vertebrates. Since genetic mutations in the vertebrate members of the EGF receptor class are not available, it is not clear if this discrepancy is a result of the different conditions used to assay the function of the receptor (in vivo in Drosophila, tissue culture and tumor cells in vertebrates). Alternatively, it may indicate that despite their structural similarity, these receptors play very different roles in vertebrates and invertebrates. Two other Drosophila receptor tyrosine kinase genes that have been studied in detail, torso and sevenless, were also shown to determine cell fate, in the embryonic terminal regions and in photoreceptor cell R7, respectively 18,19. In both cases, it was shown that the receptor is expressed in a broader range of cells than those that it actually affects. Thus, the activity of these receptors seems to be regulated by the localized presentation of the ligand 18,a°. A similar situation appears to apply for the DER protein, which is expressed in a broader tissue spectrum relative to the

tissues affected by f i b mutations9. It is thus likely that the presentation of the active form of the DER ligand(s) provides the temporal and spatial cues for triggering the pathway. Ligands or substrates of DER have not been identified by genetic analyses so far, and there are no other mutations that give rise to a fib-like phenotype. It is possible that the genes for the ligands or substrat.es have a maternal component that masks their zygotic fib-like phenotype. It is also possible that there may be several ligands or substrates, and thus mutations in each one of them would give rise only to a narrow subset of the DER phenotype. In conclusion, the structural conservation of receptor tyrosine kinases between species strongly suggests that the mechanism of signal transduction and some of the elements in the pathway are universal. The challenge is to identify these elements, and to understand how receptor tyrosine kinases are triggered by neighboring cells, and how cells interpret the signals in different developmental contexts.

Acknowledgements We thank all members of the Shilo lab for critical discussions and support. This work was funded by an NIH grant and a Wolfson grant to B.S.

References I Livneh, E. et al. (1985) Cell 40, 599--607 2 Schejter, E.D., Segal, D., Glazer, L. and Shilo, B-Z. (1986) Cell 46, 1091-1101 3 Schejter, E.D. and Shilo, B-Z. (1989) Cell 56, 1093-1104 4 Aroian, R.V. et al. (1990) Nature 348, 693-699 5 Niisslein-Volhard, C., Wieschaus, E. and Kluding, H. (1984) Wilhelm Roux's Arch. Dev. Biol. 193, 267-282 6 Baker, N.E. and Rubin, G.M. (1989) Nature 340, 150-153 7 Schtipbach, T. (1987) Cell 49, 69%707 8 Price, J.V., Clifford, R.J. and Sch(ipbach, T. (1989) Cell 56, 1085-1092 9 Zak, N.B. et al. (1990) Development 109, 865-874 10 Rogge, R.D., Karlovich, C.A. and Banerjee, U. (1991) Cell 64, 39-48 I1 Manseau, L.J. and Schfipbach, T. (1989) GenesDev. 3, 1437-1452 12 Clifford, R.J. and Sch~ipbach, T. (1989) Genetics 123, 771-787 13 Aroian, R.V. and Steinberg, P.W. Genetics (in press) 14 Ullrich, A. and Schlessinger, J. (1990) Cell61, 203-212 15 Cantley, L.C. et al. (1991) Cell 64, 281-302 16 Raz, E., Schejter, E.D. and Shilo, B-Z. Genetics (in press) 17 Steinberg, P.W. and Horvitz, R. (1991) Trends Genet.7, 366--371 18 Sprenger, F., Stevens, LM. and Ntisslein-Volhard, C. (1989) Nature 338, 478-483 19 Hafen, E., Basler, K., Edstrom, J.E. and Rubin, G.M. (1987) Science 236, 55-43 2 0 Kr~imer,H., Cagan, R.L. and Zipursky, S.L. (1990) Nature 352, 207-212

B-Z. SHILO AND E. ~ ARE IN THE DEPARTMENT OF MOLECULAR GENETICS AND VIROLOGY, WEIZMANN INSTITUTE OF SCIENCF, REHOVOT 76100, ISR4EL

TIG NOVEMBER/DECEMBER1991 VOL.7 NO. 11/12