TIG ~ March 1987, Vol. 3, no. 3

During the past ten years, a variety of approaches have led to the isolation of more than 30 different oncogenes implicated in tumorigenesis in vertebrates z. Structural simi- larities among the different oncogenes have allowed their classification into families; three major families are cur- rently recognized. The largest family, src, includes more than ten different oncogenes that share the sequence coding for a tyrosine kinase domain. The ras fanfily contains three members with structural and functional similarities to t.be guanine nucleotide- binding (G) proteins. Finally, the myc/myb family has four genes that code for nuclear proteins. Little is known about the chain of events that led to the generation of oucogene families. When in evolution did each family arise and what might this tell us about the fimction of that oncogene family? Within each of the families, did the gene duplication events occur early in evolution, in an ancestral organism, or did they occur independently later on, in each of the arising superphyla?

The normal cellular counterparts of oncogenes are called proto-oncogenes. Ultimately, the most crucial issue in the study of oncogenes and proto-oncogenes is the elucidation of their function in individual cells and in the organism as a whole. Several oncogene products have been shown to be aberrant versions of essential components in pathways of signal transduction 2 (i.e. growth factors, receptors, or G-like proteins), but the function of many others is not yet known. Are they involved in additional links of signal tran~duction? Current knowledge about other cellular components which interact with oncogene and proto-oncogene products is limited. In addition, we know very little about the developmental contexts in which these pathways are utilized. Are they ~ecessary only for cell division, or do they also partic,pate in the differentia- tion of specific tissue types? Within the context of oncogene families, the functional differences amongst members of a given family are not known.

Proto-oncogenes have been found to be evolution- arily conserved in the genomes of vertebrates. We asked whether proto-oncogenes are also conserved in the genomes of lower eukaryotes such as Drosophila melanogaster 3, reasoning that if there are proto- oncogene homologs in a distantly related species such as Drosophila, their structural and functional characterization could contribute to several major issues in the study of oncogenes. On the one hand, structural analysis of Drosophila proto-oncogenes might allow reconstruction of the chain of events leading to the generation of oncogene families. On the other hand, the genetic tools available in Drosophila would provide a potentially powerful approach to the study of proto-oncogene function during develop- ment, and the identification of other elements in the signal transduction pathways in which they participate.

This review will describe the structural features of proto-oncogenes that have been found in the Drosophila genome. The implications of these findings

~) 1987, EL~vter Publimlzons, ~ 0168 - 9525/87/$02 00

review Proto-oncogenes i n - "

Drosophila melanogaster Ben-Zion Shilo

More than ten different proto-oncogenes have been identified in the genome of Drosophila melanogaster. Their slructural characterizatiou shows a &igh degree of conservation compared wi@ the corresponding genes in vertebrates, and provides interesting clues to the evolution of proto-oncogenes. During Drosophila development proto-oncogenes are expressed in proliferating cells as well as in several differentiated

tissues, suggesting that they seree pleiotropic functions.

foz proto-oncogene evolution and some new insights into proto-oncogene function that have emerged from studies in Drosophila will also be discussed. In addition, another class of Drosophila genes that are potentially involved in regulation of growth will be presented.

Proto-oncogenes in the Drosophila genome Use of the vertebrate oncogenes as hyblidization

probes has enabled the isolation of eleven different proto-oncogene homologs from the Drosophila gen- ome. Although these genes were initially isolated at low stringencies of hybridization, they were demon- strafed to be significantly similar to their vertebrate counterparts at the nucleotide and amino acid levels. Clones considered to be authentic were those in which homology extended throughout more than a single limited domain of the protein. A list of the known proto-oncogenes in Drosophila and their location on polytene chromosomes is given in Table I and presented schematically in Fig. I.

Proto-oncogenes appear to be highly conserved between vertebrates and Drosophila, indicating their essential biological roles. The large number of protoooncogenes found in Drosophila include repre- sentatives correspunding to all three major families of vertebrate oncogenes (src, ras and myc/myb). In this Drosophila is different from yeast, in which only proto- oncogenes of the ras family have been found ~. The functions of the protein products of the src family, represented by at least seven members inDrosophila, may thus be essential to processes which take place

Table 1. Chromosomal location of Drosophila proto-oncogenes

Vertebrate Drosophda Map proto-oncogene homolog position Ref.

s r c

famdy

c-src Dsrc 64B 4,5 Dsrc2 29A 4

c-abl Dash 73B 4, 5 c-orb B (EGF receptor) DER 57F 6

Insulin receptor DIR 93E O. Rosen ~

c-.~s Dfps 85D M. Bsshop ~

c-raf Draf 2F G. Mark ~

Drasl 85D ras famdy Dras2 64B

Dins3 62B c-myb Dmyb 13F

~Personal communications.

¢M o ~ n" Lj..

II I e - - - - I

Fig. 1. Chromosomal locahzation ofDrosophJla proto-oncogenes. The three major Drosophda chromosomes are represented schematically. The locations of the src-famdy proto-oncogenes are shoum above the chromosomes, and the location of the ras and rnyc/myb-related genes below them.

only in multicellular organisms. It is more difficult to argue this in the case of the myc/myb family, since only a single member was found in Drosophila, and the family may simply be iess conserved.

The Drosophila proto-oncogenes show no obvious clustering at particular chromosomal positions: two are on the X chromosome, two on chromosome 2 and seven on chromosome 3. Two chromosomal positions contain more than one proto-oncogene (Dsrc and D~'as2 are at 64B; Dfps and Dram at 85D), hut at the DNA level they could be tens or even hundreds of kilobases apart. None of the identified proto- oncogenes map to known mutations or gene loci in Drosophila. This observation was a sad disappoint- ment to those of us who were not trained as Drosophila geneticists, and naively assumed that after 70 years of Drosophila work, every gene would have been mutated and characterized. Since existing mutations cannot provide us with phenotypes resulting from defective proto-oncogenes, such mt~tations will have to be generated by the approach of 'reverse genetics', which will he discussed below.

S t r u c t u r a l c o n s e r v a t i o n : e v o l u t i o n a r y implications

Our underlying assumption in studying the proto- oncogenes of Drosophila is that they are structurally similar to their vertebrate counterparts and may thus be expected to serve analogous functions. It is essential, therefore, to examine closely the degree of structural conservation. The highest degree of homology was found for the Drasl gene which shares, at the amino acid level, 75% homology with its vertebrate counterpart 7. High degrees of homology were also observed for the other Drosophila genes.

The src-related genes of Drosophila show the highest degree of homology in the tyrosine kinase domain (50% or more). Within this domain, several residues were [ound to be conserved in all tyrosine kinases. They include the ATP binding site and other residues that appear to be essential for the kinase activity. It was interesting to find that conservation of sequences within the kinase domains of the Drosophila

TIG - - March 1987, Vol. 3, no. 3

homologs extends beyond these consensus s e - q u e n c e s . For each one of the kinases there are multiple corresponding residues that are conserved between the Drosophila gene and its vertebrate homolog. This observation focuses our attention on residues within the kinase domain that may dictate the specific function of each member. The most likely possibility is that they participate in the determination of substrate specificity, and that the respective substrates have also been conserved in evolution.

The homology of the src proto-oncogenes extends beyond the kinase domain so that the overall structure of the protein is clearly conserved in each case. Thus at this level, too, each member of the Drosophila src family can be specifically related to a particular member within the vertebrate src gene family. Among the vertebrate src genes, the individual members are distinguished by differences in their overall domain organization and the localization of the tyrosine-kinase region within the protein. Some members (e.g. the EGF receptor, insulin receptor and neu genes) code for receptors with a long extracellular sequence, a single transmembrane domain and a cytoplasmic domain containing the kinase region. In other cases (e.g. c-src or c-)~s), the protein does not traverse the membrane, but is associated with it at its amino- terminus. There are also tyrosine kinases that are not associated with the membrane at all (e.g. some of the forms of c-abl).

Parallel differences in overall domain organization can be recognized among the Drosophila proto- oncogenes. The complete nucleotide sequences of three Drosophila src family members that are homologous to c-erb-B s']°, c.src n and c-abl (M. Hoffmann, pers. commun.) demonstrate remarkable conservation in overall protein stn,cture. The protein of the Drosophila c-erb-B (EGF receptor tune) homolog (termed DER), has the typical structure of a membrane receptor, with a long extracellular sequence, a single transmemhrane domain and a cytoplasmic region containing the tyrosine kinase sequence. Dsrc appears to code for a protein with no transmembrane domains and with a glycine residue, which was shown to be responsible for binding to the membrane, encoded by its second codon. T h e predicted structure of these two Drosophila proteins and their vertebrate counterparts is shown in Fig. 2. The Drosophila gene homologous to c-abl codes for a protein that has the kinase domain at the amino terminus. In the case of the Drosophila insufin receptor homolog, although the seouence of the gene has not yet been fully determined, the ove~aU structure seems to be similar to that of the human receptor. Antibodies directed against the human cytoplasmic domain immunoprecipitate from Droso- phila embryo extracts a 95 kDa protein with tyrosine kinase activity. This activity is stimulated by addition of insulin ~2. Furthermore, a 135 kDa protein (which may represent the c~-chain homolog) was demon- strated to bind insulin ~a.

Each of the Drosophila src-family homologs is thus more similar to its vertebrate counterpart than to the other Drosophila src genes. These findings have profound ramifications upon our view of the evolution of the src family. They indicate that the major gene duplication events which gave rise to the family took place before the divergence of chordates and

TIG - - March 1987, VoL 3, no. 3

arthropods, over 800 million years ago. Once the ancestral gene had been duplicated, each of the members underwent structural changes. After the different members of the family had assumed their unique structures, the coding sequences were conserved in the two superphyla that subsequently arose. This implies that the specific function of each of these kinases is likely to be analogous in Drosophila and vertebrates, and that the study of a gene ,~r protein function in Drosophila will shed fight on the function of its vertebrate homolog as well. It should he pointed out, however, that not all gene duplication events in the src family occurred early on; structural comparisons indicate that some of the src members in vertebrates were generated by later gene dupli- cations. In these cases, a single gene in Drosophila shows a similar degree of homology to two or more vertebrate genes. For instance the Drosophila D E R gene shows a similar degree of homology to both the human EGF receptor and neu receptor genes, and the DIR gene is similar to both human insulin receptor and IGF-I receptor genes.

T i s sue dis t r ibut ion of Drosoph i la proto- oncogene t r ansc r ip t s

An initial assessment of proto-oncogene function can be sought by following the temporal pattern of distribution of proto-oncogene transcripts during Drosophila development. Most proto-uncogenes studied in this way were found to he expressed at several stages of development. (Data obtained from northern blots are summarized in Table 2.) At first glance the results might be taken to imply that these genes serve generalized cellular functions. Alternatively, they could he involved in multiple changing functions specific to different tissue types. Results from in situ hybridization point to the latter possibility.

In situ hybridization offers more precise information about specific spatial localization of transcripts. In particular, discrete compartmentalization of dividing versus non-dividing cells at several stages of Drosophila development makes it possible to ask whether proto-oncogene expression is restricted to proliferating populations. To date, the transcripts of

11 10 19 the Dsrc , DER and three Dras genes have been localized by in situ hybridization. Transcripts of each of these genes were detected during all developmental stages, and in more than one type of tissue. Interest- ingly, at severaJ developmental stages where dividing a__q.r] _non-.d_ivid~tg tissues are clearly distinguishable, proto-oncoge:le transcripts were identified in both tissue types. Among the proliferating tissues expressing proto-oncogenes are the imaginal discs of the larva, which consist of multiplying diploid cells that undergo metamorphosis at the pupal stage to give rise to the adult organs. Dsrc, DER and the Dras genes were all shown to be expressed in @,e larval imaginal discs. Several instances of proto-oncogene expres- sion in terminally differentiated tissues were also observed. Most notably, Dsrc transcripts were detected in the embryonic smooth muscle cells of the gut, and in the eye aniage of the early pupa. The three Dras genes, as well as DER, are expressed in the larval and adult nervous systems.

In order to draw firm conclusions about the spat~fl and temporal distribution of proto-oncogone activity, it remains to he shown that the transcr,pts observed

c-src (chicken)

5 3 3

Dsrc

% C'!

4 1 % I

. . . . . _ . . [

5 6 %

. . . . . . . "¢ 5 5 2

DER

HER = I I °"

37%

° "iiiil |lSS ~'' - - 1

review

J=

EXTRACELLULAR

. 769

CYTOPLASM %

1365

Fig. 2. Structural homology of the putative Dsrc and DER proteins to their vertebrate counterparts. The Drosophila and chicken src proteins are shown on the left, and the percentage of amino acid homclo~ indicated. The Orosine kinase domain is marked by a hatched bar, and the amino-terminal glycine which is essential for membrane binding is marked by a G. The Drosophila EGF receptor homolog (DER) and haman EGF receptor (HER) are shown on the right, and the percentage of amino acid komology indicated. The kinase domain is marked by a katched bar. The dotted boxes in the extracellular reoon represent cysteine-rick domains whose function is not yet known. The open boxes at the amino terminus represent the signal peplides. The numbers indicate amino add positions.

give rise to functional proteins which play an essential role in the tissues that express them. It seems, however, that the function of proto-oncogenes during D r o s o p h i l a development may not be restricted to cell proliferation. The possible involvement of proto- oncogene products in processes of differentiation must also be considered. Since proto-oncogene proteins play a role in pathways of signal transduction, it is possible that a given pathway is common to different cell types and is ,ltilized in a variety of contexts.

Rever se gene t ics Since the proto-oncogenes isolated to date do not

correspond to known genes in Drosophila, an intensive effort has been aimed at isolating mutations in the

Table 2..TranscrOtional expression patterns of Drosophila proto -oncogenes

Tissue a Proto-oncogenes U E L P A Ref. Dsrc +e + + + _+ 11,14 Dash + + - + + 14,15 DER _+ + + + + 16 DIR + + + _+ _+ 12 Drasl + + + + + 17 Dras2 + + + + + 17,18 Dras3 + + + + + 17 Dmyb ND = + ND ND ND 8 aU, unfertilized eggs; E, embryos; L, ia~'ae; P, Pupae; A, adults. e'rranscripfion was analysed by nortbern blots. The indicated levels of transcription represent relative amounts of transcripts detected by a given probe. They do not compare the transcript leve ls .~ detected by the different probes. 7 1 ) ~-D not determined.

e w s endogenous genes. While it is difficult to predict with certainty what spectrum of phenotypes these mutations wdl show, it is assumed that mutations which abolish the synthesis or activity of the protein are likely to have lethal effects. This notion is supported by the high degree of structural conser- vation of the Drosophila proto-ocogenes and their expression throughout development. In the cases where more than one gene coding for a similar protein is found (e.g. the three Dras genes or the two Dsrc genes), inactivation of only one gene may not be sufficient, however, to give rise to a lethal phenotype. Indeed, in the yeast Saccharomyces cerevisiae where two ras genes were identified, a lethal phenotype could be observed only after inactivation of both genes z°.

To isolate mutations in the proto-oncogenes, an 'F2 lethal screen' is used. Briefly, random mutagenesis by X rays or chemical mutagens such as EMS is performed. Recessive lethal mutations corresponding to the relevant chromosomal region are identified by individual crosses to flies bearing a chromosome in which the region around the locus of the proto- oncogene has been deleted. A mutated chromosome is identified by the absence of a defined phenotypic class in the progeny (i.e. those carrying the mutated chromosome and the deleted one). The mutated chromosome is rescued from sibling flies in the same vial. Once the flies are isolated and lines established, it is necessary to prove that a mutation induced in the relevant chromosomal region indeed corresponds to the gene of interest. In the long run, progress in this direction should yield a powerful tool for studying the function of proto-oncogenes in the context of developing multiceliular organisms.

R e c e s s i v e oncogenes The known oncogenes in vertebrates are character-

ized by their dominant phenotype. They induce unregulated growth in spite of the presence of their normal counterparts. A different class of oncogenes, termed 'recessive oncogenes', has also been identified in humans. For these putative 'recessive oncogenes', the loss or inactivation of both normal alleles causes the appearance of tumors. The childhood retino- blastoma tumor is an example of the involvement of this class of genes in the etiology of cancer 21.

A similar phenomenon has been recognized in Drosophila: characterization of a battery of recessive lethal mutations in Drosophila has shown that over 20 mutations follow a similar pattern 22'23. In these cases, lethality is a result of hyperproliferation of a tissue that fails to differentiate. The mutations can be grouped according to the stage at which lethality is induced and according to the affected organs. Most of them exhibit the neoplastic pheuotype at late larval stages, and tumors in the larval brain, imaginal discs and hematopoietic organs have been identified. Neo- plasms during the adult stage are not lethal in most cases; rather, they give rise to a sterile phenotype due to hyperproliferation of the ovaries or testes. The cells of the affected organ in each of the mutants behave like typical cancer cells. They are often invasive, and retain their abnormal growth properties after transplantation into wild-type hosts.

One of the most thoroughly studied mutations of this class is lethal(2) giant larva (l(2)gl), which causes

T I G - March 1987, Vol. 3, no. 3

neuroblastomas and malignant tumors of the larval imaginal discs 23. Because the gene is close to the left telomere of the second chromosome, multiple spontaneously arising alleles have been identified. The 1(2)gi gene has been isolated u and upon injection into the germ line has been shown to rescue the lethal phenotype (B. Mechler, pers, commun.).

The recessive oncogenes inDrosophila define a new and interesting group of genes. Several basic issues regarding their function remain open. At the moment it is not possible to say whether their normal role is to suppress cell proliferation or to induce differentiation, since the observed phenotypes could be accounted for by either model. As several mutations give rise to a similar phenotype, it is tempting to suggest that they represent different stages of the same pathway, but this notion has not yet been tested by looking for interactions among different mutants. Further work at the genetic and molecular level will undoubtedly yield new insights into this special class of genes. Finally, if recessive oncogenes are as well conserved as proto- oncogenes, the Drosophila genes may serve as probes for their vertebrate homologs.

A c k n o w l e d g e m e n t s The author thanks Dr Naomi Zak for critical reading

of [his manuscript. This work was supported by NIH grant GM 35998 and by a grant from the Leukemia Society of America. B.S. is an incumbent of the Haas Career Development chair.

R e f e r e n c e s I Bishop, J. M. (1985) Cell 42, 2.3--38 2 Weinherg, R. A. (1985) Sewnce 230, 770--?76 3 Shilo, B. and Weinberg, R. A. (1981)Proc. NatiAcad. Sci. USA

78, 6789-6792 4 Simon, M. A., Kornberg, T. B. andBishop, J. M. (1983)Nature

302, 837-839 5 Hoffman-Falk, H., Einat, P., Shilo, B. and Hoffmann, F. M.

(1983) Cell 32, 589-598 6 Livneh, E. et al. (1985) Cell 40, 599-607 7 Neuman-Sdberberg, F. S., Schejter, E., Hoffmann, F. M. and

Shilo, B. (1984)Cell 37, 1027-1033 8 Katzen, A. L., Kornberg, T. B. and Bishop, J. M. (1985) Cell

41, 449--456 9 Powers, S. et aL (1984) Call 36, 607-612

10 Schejter, E. D., Segal, D., Glazer, L. and Shilo, B. (1986)Cell 46, 1091-1101

11 Simon, M. A., Drees, B., Kornberg, T. and Bishop, J. M. (1985) Cell 42, 831-640

12 Petruzzelh, L. et al. (1986) Proc. Natl Acad. Sci. USA 83, 4710-4714

13 Petruzzelli, L, Herren, R., Garcia-Arenas, R. and Rosen, O. M. (1985)]. Biol. Chem. 260, 16072-16075

14 Lev, Z., Leibovitz, N., Segev, D. and Shilo, B. (1984)Mol. Cell. Bwl. 4, 982-984

15 Telford, J., Burckhardt, J., Butler, B. and PLrrotta, V. (1985) EMBO]. 4, 2609-2615

16 Lev, Z., Shilo, B. and Kimc.~,~e, Z. (1985) Dev. Biol. 110, 499-502

17 Lev. Z., Kimchie, Z., Hessel, R. and Segev, O. (1985) Moi. Cell. Biol. 5, 1540-1542

18 Mozer, B., Matlor, R., Parkhurst, .f,. and Corces, V. (!085) Mol. Cell. Biol. 5, 885-889

19 Segal, D. and Slnlo, B. (1986) Mol. Cell. B:oi. 6, 2241-2248 20 Kataok~, T. et al. (19t~4) Cell 37, 437--445 21 Knudson, A. G., Jr (1971) Proc. Nati Acad. Scl. USA 68,

820--823 22 Gateff, E. (1978)Science 200, 1448-1459 23 Gateff, E. (1982)Adv. CaneerRes. 27, 33-74 24 Mechler, B. M.. McGinnis, W. and Gehring, W. J. (1985)

EMBO ]. 4, 1551-1557

B-Z. Shilo is at ~.he Department of Virology, The W ~ n Institute of ,Sci~ece, Rekovot, 76100, Israel.