Copyright 0 1992 by the Genetics Society of America

Characterization of Suppressor of fused, a Complete Suppressorof the fused Segment Polarity Geneof melanogaster

Thomas Preat’ Centre de Ginnitique Molkulaire, C.N.R.S., F-91198 G$-sur-Yvette Cedex, France Manuscript received March 10, 1992 Accepted for publication July 25, 1992

ABSTRACT fused vu) is a maternal effect segment polarity of Drosophila melanogaster. In addition, fu females have tumorous ovaries. Two ethyl methanesulfonate mutageneses were carriedout in order to isolate suppressorsof the fu phenotype. A new gene, Suppressor of fused (Sucfu)),was identified. It is located in the 87C8 region of the third . Sucfu) displays a maternal effect and is also expressedlater in development.Although Sucfuyp is a completeloss-of-function mutation, it is homozygous viable and has no phenotype by itself. Sucfu) fully suppresses the embryonic and adult phenotypes of fu mutants. Sucfu) mutations are semidominant anda Sufi)+ duplication hasan opposite effect, enhancing the fused phenotype. It is proposed therefore that the Su(fu)+ product is involved in the same developmental step as the Fu+ kinase. Thus, a new gene interacting with the segment polarity pathway was identified using an indirect approach.

MBRYONIC development is under genetic con- 1988), which define another classof segmentation E trol.In Drosophila melanogaster, extensive ge- . After cellularization of the embryo, the expres- netic and molecular analyses have allowed character- sion of segment polarity genes is under the control of ization of hierarchical interactionsbetween several segment polarity gene products themselves (MARTI- classes of genes (reviewed in INGHAM1988). One of NEZ-ARIAS,BAKER and INGHAM1988; DINARDOet al. thesecorresponds tothe segment polarity genes, 1988; HIDALDOand INGHAM1990; EATONand KORN- which are requiredwithin each embryonic segmentin BERG 1990; HIDALGO199 1; HEEMSKERKet al. 199 1). order to definespecific anteroposterior domains(NUS- -cell interactions are thought to take place during SLEIN-VOLHARDand WIESCHAUS1980). Fourteen such this second phase, but the actual molecular mecha- segment polarity genes have beenidentified so far nisms involved remain largely unknown. Unlike most (reviewed in INGHAMand NAKANO 1990; INGHAM other segmentation mutations, segment polarity mu- 199 1). They form a very particular class of segmen- tations generally affect bothembryonic andadult tation genesas they encode a largevariety of products. development. Furthermore, the corresponding phe- Wingless is a secreted (RIJSEWIJKet al. 1987; notypes and the expression patterns can be related in VAN DEN HEUVELet ai. 1989; GONZiLEZ et al. 1991), some cases (reviewed in WILKINSand GUBBS1991), Armadillo is the homolog of Plakoglobin, a mamma- suggesting that segmentpolarity products have similar lian protein associated to cell junctions (PEIFERand functions at different developmental stages. WIESCHAUS1990), Patched is a transmembrane pro- fused vu)is a maternal effect segment polarity gene tein (HOOPERand SCOTT1989; NAKANOet al. 1989), located onthe X chromosome (1-59.5) (NUSSLEIN- Fused (PREATet al. 1990) and Shaggy (BOUROUISet VOLHARDand WIESCHAUS1980; GERGEN andWIES- al. 1990; SEIGFIREDet al. 1990) are putative serine/ CHAUS 1986; PERRIMONand MAHOWALD1987; Bus- threonine kinases, and Engrailed (FJOSE, MCCINNIS SON et al. 1988) (Figure l).fu embryos derived from and GEHRING1985; DESPLAN,THEIS and O’FARRELL fu females show a deletion of the posterior part of 1985), Gooseberry (BOPPet al. 1986; BAUMGARTNER each thoracic and abdominal segment and, ventrally, et al. 1987), and Cubitus-interruptus Dominant (OR- a mirror duplication of the anterior part which bears ENIC et ai. 1990) are transcriptional regulators which the denticles rows. Abnormal cell death has been possess specific DNA binding domains. The initial shown to occur in the ectoderm and themesoderm of expression of some segment polarity genes is con- developing fu embryos, which may account for the trolled by proteinsencoded by the pair-rulegenes terminal deletion pattern (MARTINEZ-ARIAS1985).fu (HOWARD and INGHAM1986; DINARDOand O’FAR- embryos derived from heterozygousfu/+ females de- RELL 1987; INGHAM, BAKERand MARTINEZ-ARIAS velop normally because of the product (RNA or pro- ’ Present address: Theodor-Boveri-lnstitut fur Biossenschaften, der Univ- tein) accumulated in the oocytes, butfu adults present ersitit Wurzburg, Lehrsthuhl fur Genetik, D-8700 Wiirzburg, Germany. several defects due to its lack during metamorphosis

(knetics 132 725-736 (November, 1992) 726 T. Prkat

productshave opposite roles throughout develop- ment.

MATERIALS AND METHODS

Stocks and culture: fu' is hypomorphic allele recovered as a spontaneous mutation (MORGANand BRIDGES1916). Thefu' stock in thisstudy was marked with thepmutation. fum"63is a strong EMS-induced allele (WURST and HAN- RATTY 1979; BUSSONet al. 1988). DfTI)fuZ4is a 40-kb die- poxybutane-induced deletion (BUSSONet al. 1988) which covers the entire fu (MARIOL, PR~ATand BOUCHON 1987). Some of the genetic properties of these fu alleles are

pupal lethality. described inBUSSON et at. (1988). The Oregon-R and w - d.fu"zH6jY some escapers eelose stocks usedfor themutageneses had been isogenized for the X chromosome in our laboratory and therefore, although - p jjf "'"TFMtj I fUf I not isogenic, the autosomal had been gener- ated outof a small pool. The 1(3)SzC3, 1(3)SzD8, DfT3R)karS'" I fU+ I - P f!J1F,u6 - d FM6,'Y I fu'l (87C7-8;87E5-6), Df(3R)karSz2' (87C7;87C8-9)and - d FMIY I fu+ I DfT3R)kar'Q (87B2-4;87C9-D3) mutations are described in GAUSZet al. (1979) and in GAUSZ,AWAD and GYURKOVICS FIGURE1 .--Genetic properties of fu mutations. (A) Maternal (1980). The third chromosome mutations se, cp, e, cd and effect leading to a segment polarity phenotype. (B) Paternal rescue kar', Tp(3;2)ry" (87C2-3;88C2-3) carrying a Sulfu)+ copy, of the maternal effect. This rescue is very efficient in the case of and the balancer chromosomes MKRS, TM3, M5, FM3and weakfu alleles (recovery of adults), but it is low in the case of strong FM6 are described in LINDSLEYand ZIMM (1985, 1987, alleles (slightly rescued segment polarity phenotype) (PERRIMON fu 1990). Stocks were maintained on a yeast/maize/agar me- and MAHOWALD1987; BUSSONet al. 1988). (C) Zygotic effect of dium (GANS,AUDIT and MASON 1975). Crosses were made weakfu alleles. (D) Zygotic effect of strongfu alleles (pupal lethals). at 23" unless otherwise specified. Ethyl methanesulfonate (EMS) mutageneses: In the first (WURSTand 1979) and adultlife. One of HANRATTY mutagenesis, wild-type Oregon-R males were fed 25 mM these defectsis the fusion of the third and fourth wing EMS as described by LEWISand BACHER(1 968) and mated veins, which gave its name to the mutation (MORGAN topfu'/FM3 virgin females (see Figure 2A) at 20". p and BRIDGES19 16). This phenotypewas reinterpre- fu'/Y*;+*/+;+*/+;+*/+ single maleswith a suppressed fused wing phenotype were screened out of the F1 progeny. tated by FAUSTO-STERLING(1978) as a lack of the The suppressor mutation was further amplified by crossing fourth vein andthickening of the third one. This suppressed F1 individual topfu'/FM3 females. The ex- interpretation matches the oneof the embryonicphe- pressivity of the vein fusionphenotype is rather constant for notype, since in both cases there is failure to form a a particular fu allele raised at a given temperature, so that suppressors only partially rescuing thiswing phenotype posterior structure and duplicationof an anteriorone. could be identified. About 1,700 fu' males were screened. In addition, adultfu females display reduced fecundity This mutagenesis yielded the Sulfu~' allele. The karmoisin due to several ovarian defects including the appear- (kar)gene is located next to Suppressor offused in 87C8, and ance of tumorous egg chambers (KING 1959; SMITH both genes are mutant in the Sulfuy strain (see RESULTS). The SuVuy allele could thus be followed usingthe recessive and KING 1966). We have cloned and sequenced the kar phenotype as a marker. This situation turned out to be fused gene (PR~ATet al. 1990). It encodes a putative useful because homozygous Sulfu) mutations display no vis- serinelthreonine kinase,which indicates thefunda- ible phenotype by themselves.Homozygous stocks p mental role of post-translational modifications in in- fu';Su.y and fu""";Sulfu~were generated. fu" 63/Y males normally die as late pupae (only a few trasegmental pattern formation. ~u"'"~~/Yescapers eclose, which are recognizable as they are I have characterizeda new gene, Suppressor offused, subviable and displa an extreme wing phenotype). On the amorphic forms of which fully suppress all fu pheno- contrary, viable fu"x 63/Y;SuviL)Lp/+flies displaying a par- types.Suppression effects have been previously de- tially suppressed phenotype can be recovered. In the second muta enesis, w males were fed 25 mM EMS and mated to scribed among segmentation mutations(COULTER and fum"p;/FM6 virgin females (see Figure 2B). Forty bottles WIESCHAUS1988; HULSKAMPet ai. 1989; IRISH,LEH- containing 20 EMS-treated males and 20 females were gen- MANN and AKAM 1989;LIMBOURG-BOUCHON, BUSSON erated, and the parents transferred every day for 4 days. and LAMOUR-ISNARD 1991;HIDALGO 1991; E. WIES- Half of the progeny was raised at 23' and half at 25 O. Viable f~""~~/y*;+*/+;+*/+;+*/+ males were screened daily from CHAUS,unpublished result cited in INGHAM,TAYLOR the F1 progeny and crossed to attached-X females. Males andNAKANO 1991), but this is the first reportof displaying an extreme fu phenotype were never found fer- mutageneses specifically designed to isolate suppres- tile, and most likely corresponded to fumH6'/Yescapers. The sors of an abnormal segmentation phenotype. Genetic equivalent of about10,000 individualswere screened. One Sulfu) allele, Sulf~)"~,was isolated. This low mutation properties of Sucfu) mutations have been extensively rate (10-4) could be due in part to the fact that suppression analyzed. The results suggest that the expression of of the pupal lethality by Sulfu) heterozygous mutation is Sucfu) is similar to that offu, but that the two gene only partial. Other mutations were recovered during this Suppressor of Segmentation Mutation 727 screen which do not map to the Sufu) locus (T. PR~AT, unpublished results). As a control, the rate of X chromo- A somes bearing a lethal mutation was estimated. EMS-treated p ? f z6u~'IFM3 x dd +/y w/Y males were crossed to M5/M5 virginfemales. The progeny of individual w*/M5 females crossed to M5/Y males I was scored. Out of 122 fertile females, 32 (26%) gave rise dd f .'6u@ 'Iy fused wing to no w*/Y males, indicating that the mutagenized X chro- mosome carried at least one lethal mutation. df $y:su*/+ normal wng 9 Genetic localization of Szq'fu) mutations: Both Sufuy' and Sufu)Izd are viable mutations, and, except for the kar EMS phenotype displayed by the Sufu)L' allele, these stocks do B not displayany recessive phenotype by themselves. As a 2 consequence, it was not possible to prove that these muta- fumM?FMf, X dd W/y tions actually affect the same gene by adirect SufuY'/ ?? Sufu)Izd complementation analysis. Because the suppressor I pupal lethal phenotype itself issemi-dominant, a complementation analy- cfd ,fu ""6?y sis in a fu/Y;Su(fuy/St@u)'zd combination would not have wable ? been conclusive either. In order to address this problem, d,fu"'".vy:S$/+ both Sufi) mutations were localized after recombination FIGURE2.-Mutagenesis schemes used to isolate fu phenotype with a marked third chromosome in a fu background. M5/ suppressors. (A) EMS-treated males were crossed to heterozygous +; se cp e cd/TM3 virgin females were crossed topfu'/ fi females. Genotypically fu males with a suppressed fused wing Y;Sufu~'/Sufu~males. Virgin M5/4"" fu'; se cp e cd/ phenotype were screened en masse in the F1 progeny (only the Sufuy' females were recovered and crossed to se cp e cd/se relevant genotypes are shown). Presence of the forked v) bristles cp e cd males. The phenotype of 2534"" fu' males was scored marker ensured that males displaying a normal phenotype actually at the Fn generation for the presence of both the semidom- carried afu chromosome. (B) EMS-treated males were crossed to inant Sufu)Lp mutation and the various recessive markers. females heterozygous for the pupal lethal mutation fu""". Viable The Sufuy' mutation was found to be between cp (45.3) fu males were screened en masse in the F, progeny. Suppression of and e (70.7) at the approximate position 53.1 & 2.3. As the their fused wing phenotype was monitored. kar gene is at 51.7, this result is consistent with the hypoth- esis that the viable Su(fuy' mutation is a small molecular were mounted in Euparal following storage in 100% event affecting both the kar gene and the Sue) gene. The ethanol. Ovaries were mounted and stained with fuchsin as SU~U)"~mutation was localized similarly, and was found to described in ZALOKARand ERK(1977). be between cp and e at 20 fi 1.8 units from e. This corre- sponds to the position 50.7 f 1.8, which stron 1 suggests that Sufu)Izd and Sufu)Lp are alleles. A Sufu)" cd stock RESULTS was generated during this experiment. Analysis of 87C complementation groups:To determine Isolation of Sucfu) mutations: Withthe aim of whether any of the previously described 87C complemen- identifying new genes involved thein segment polarity tation groups (GAUSZet al. 1979) interacts with fu, 1(3).!$xC3/ pathway, two EMS mutageneses were carried out in TM3, 1(3)SzD8/TM3,and ka?/karz males were crossed top order to isolate suppressors offu mutant phenotypes fu'/FM3 virgin females, and the fused veins phenotype of the non-TM3 males of the progeny was checked. In addi- (Figure 2 and MATERIALSAND METHODS). The tion, fu"""'lf.""";Sufu~/Sufu~' females were crossed to suppression effect was not identified directly on the 1(3).!$zC/TM3and 1(3).!$rD/TM3males, and the vein pheno- segment polarity phenotype of fu embryos, but onfu type of the fu"H63/U;Su;)"p/l(3)sz males was compared to adult phenotypes. Both mutageneses allowed to isolate the one of their fumH6/U;Sufu)Lp/TM3 siblin s. The eye- en masse partially rescued fu/Y;Su/+ individuals bear- color phenotype observed in Su(fu~'/Sufu~B.flies corre- sponds to arecessive mutation in the kar gene since Sufuyp/ ing a semidominant autosomal suppressor. This would kar' flies display a karmoisin phenot pe. The MKRS balan- nothave been possiblewith anembryonic screen, cer carries the kar' mutation, and?" fu'/Y;MKRS/+ were because a partial rescue of the segment polarity phe- generated to furtherinsure that kar mutations have no effect notype generally does not allow to get a viable adult. on the fused phenotype. To generate Sufu)/Dx3R) individ- In the firstmutagenesis, flieswith suppressed uals, virgin Sufu)/Sufu) females were mated to DN3R)karl fu a TM3 males. wing phenotype were screened (Figure2A). This mu- Embryosand adults preparations: Embryos were tagenesis led to the isolation of the Sucfu)"' allele. In mounted in Hoyer's for cuticle examination as described in the second mutagenesis, suppression of the pupal le- VANDER MEER (1977) and observed under a Zeiss micro- thality of thefumH63 allele was screened (Figure2B). It scope on a dark field. In cases where fu'lf.' females were allowed to get theSucf~)"~ allele. crossed to fully males in various Sufi) backgrounds, both unhatched developed embryos and first instar larvae were Sucfu) is located in the 87C8 band of the third collected. Thisprocedure avoided biasin the statistical chromosome: The Su('fu)Lf and SU~U)"~mutations analysis of the suppression of the segment polarity pheno- were localizedusing recombination with a marked type (embryos which hatch probably display a less severe third chromosome (see MATERIALS AND METHODS). As mutant phenotype). To quantify the efficiency of the deletions covering the 87C region do suppressthe suppression of the embryonic fu phenotype, 200 embryos were collected from each cross. Unhatched developped em- fused phenoytpe (seebelow), precise localization of bryos (UD) and hatched embryos (H) were counted after 3 Sucfu) could be achievedby testing the suppressionof days at 23", and the H/UD + H ratio calculated. Wings the fused vein phenotype using deficiencies in a fu/ 728 T. Prkat TABLE 1 viable and Sucfu)/Sucfu) adults display no visible phe- Complementation patternof 87C mutations notype(Figure 3B, Figure 4B). The Sucfuy/ Df3R)karje,Df?R)ka7""/Df3R)kars"", S~cfu)'~~/ Complementation group with chromosomal Sucfuy and S~cfu)'"~~/Df3R)kar~~combinations are localizationa also viable and the flies display no obvious phenotype 87C8 (except for the kar phenotype itself in the case of the 87C6 87C787C6 87C9 first two combinations). Thusthe amorphic Sucfu) Mutation l(3pls i(3)szC kar l(3pzD Su@) mutation is cryptic. l(jpzc3 + - ++ + Squ)amorphic mutationfully suppresses theseg- kar + + -+ + ment polarity phenotypeof embryos:fu embryos kar' + + - + + fu produced by fu females (Figure 1A) die and display a SUCfU)'2d + + +- + su(jiuy + + - - + segment polarity phenotype (Figure 4C): the posterior l(3)SzDX + + ++ - part of each thoracic and abdominal segment is de- - Dj(?R)kars"' + + - - leted, and the denticle rows are mirror-image dupli- Dj(3R)kar"" + - - - + DJ3R)kdQ - - - - - cated.In orderto analyze the suppression of this phenotype by the Suppressor of fused mutation, fu'/ A "+" sign indicates that the mutation complements alleles of the corresponding group in a trans combination. In the case of fu';Sucfuyp/Sucfuy females weregenerated and Su@) however, it indicates that the heterozygous mutation does crossed to fu'/Y;Sucfuyp/Sucfuyp males. fu';Sucfuy" not suppress the fused phenotype (see MATERIALS AND METHODS). embryos produced by these females display a normal No standard complementation analysis can be performed with Su@) as it is a cryptic mutation (see text). segmentation with no denticleduplication (Figure The respective position of kar and SuY;) is not yet determined. 4D). The penetrance and expressivity of this suppres- sion are total as even a partial "fused" phenotype was Y;Df3R)/+ combination (Figure 3D). Sucfu) was found neverobserved in fu';Su@y embryos. These to be includedin Df?R)kar'Q as well as in Df3R)ka?' fu;Sucfu) embryos hatch and eventually give rise to and Df?R)ka?". Their common deleted region cor- normaladults. Thus,although the mutagenesis responds to the 87C8 band of the third chromosome screens were based on adult phenotype suppression, (GAUSZet al. 1979). GAUSZand collaborators (1979) they identified a suppressorwhich also fullysuppresses saturated the 87C region in a search for viable and the embryonic segment polarity phenotype. lethal mutations. After EMS mutagenesis, they iden- Sucfu) amorphicmutation fully suppressesthe tified mutations corresponding to complemen- 52 5 phenotypes of adults: fu embryos produced by tation groups-l(?)SzA, 1(3)SzB,1(3)SzC, kar and fu heterozygous fu/+ females have a normal embryonic l(3)SzD-with a minimum of 6 alleles per group (Table development due to the maternal expression of fu' 1). I have broughtadditional data to this study as (Figure But adults grown from these embryos regards to Sucfu) (Table 1). No previously known 1C). fu complementation group was found to be associated display several anomalies due the lack of Fuf product with a Su(fu) phenotype, confirming that Sucfu) is a during metamorphosis (WURSTand HANRATTY1979) newly characterized gene. In particular,kar mutations and adult life. When homozygous, Sucfu) fully sup- do not affect the fused phenotype. presses these various defects. Sucfu) mutations correspond to aloss of function: Sucfu)suppresses the wing phenotype: One of these As deficiencies covering the 87C region suppress the adult phenotypes is the abnormal wing vein pattern. fused phenotype (Figure 3D), Sucfu) mutants are due Sucfu) fully suppresses this phenotype, as fumH6j/ to a lossof function.Moreover, the partially sup- Y;Sucfu~/S~cfu~~flies have normalwings (Figure 3F). pressed wing phenotype is identical in f~~~~~/Sucfu) suppresses the tumorous ovarian phenotype: fu/ Y;Sucfuyp/+and in fumH6j/Y;Df(?R)kar'e/+ flies (Fig- fu females demonstrate reduced fecunditydue to sev- ure 3,D and E), defining Sucfuy as an amorphicallele eral defects, including tumorous egg chambers(KING (no activity). The S~cfu)'~~allele behaves as a strong 1959) (Figure 5B). These tumors originate from an hypomorph (strongly reduced activity) because fumH6j/ anarchic division of egg chamber cells. Eight-day-old Y;Sucfu)'2d/Sucfu)'2dflies display a very weak fused fu' females raised at 25 O have about 50% of tumorous veins phenotype(Figure 3G), whereas fumH6j/ egg chambers(SMITH and KING 1966), and onecham- Y;Sucfuy/Sucfu)Lpindividuals are normal. ber may contain as many as 10,000 cells (KING 1970). Sucfu) mutations are viable and displayno visible This phenotype is fully suppressed by Suppressor if phenotype: Sucfu) mutations were screened as heter- fused. fu'/fu';Sucfuy"/Sucfuy females have normal ozygotes in a hemizygous fu background. To analyze ovaries, and no tumorous chamber was found out of the phenotype of Sucfu)/Sucfu)individuals independ- several hundred observed (Figure 5C). In this respect ently of the fu mutation itself-which could affect this Sucfu) mutations can be considered as antioncogenic. phenotype-stocks bearing the Sucfu) mutations in a Sucfu)suppresses the pupal lethalityof strong fu alleles: fu' background were generated. Sucfu) mutations are fu alleles belong to two molecular classes (PR~ATet al. Suppressor of Segmentation Mutation 729

C E

FIGURE3.-Suppression of the fused wing phenotype by Suppressor offused. (A) Wing of a wild-type fly. There arefive longitudinal veins (LV). LVI which runs along the margin is on the left side. The anteroposterior boundary lies between LVS and LV4. (B)Wing of Su@)LpI SU(,~?U)'~fly. This wing cannot be differentiated from wild type. (C) Wing of afu"""/Y escaper displaying an extreme fused phenotype (LV4 is entirely missing). (D)Wing of afumH~'/Y;I)A-)R)Ru~Q/+male, showing a partially suppressed phenotype. (E) Wing of afum""'/Y;Sulfufp/+ nlale. Note that suppression by the heterozygous Suo'.p mutation is as efficient as that due to Dj(?R)kugQ. (F) Complete suppression of the extreme fused phenotype by the homozygous Sulfuf" mutation. Thisfu"ff6'/Y;Sulfu~p/~ulfu~wing is normal. (G)Wing of afu"""/Y;S~lfu)"~/ SU~~U)'~"fly. The arrow indicates the very weak fused phenotype still displayed by these adults. The dark color pigmentation is due to the presence of the ebony marker on this S~fjiir)"~chromosome. 1990). The first class corresponds to largere- pupal lethality of thefu"" allele is also suppressed. arrangements,the smallest one, Df(l)fuz4,being a Sucfu) amorphic mutation does not suppress the 40kb deletion which includes several larval lethality due to Df(I)fuz4: Because of the lack units besidefu (MARIOL,PR~AT and BOUCHON1987). off. deletions of size intermediate between 40 bp and DfTI)fuz4/Ymales produced by heterozygous Df(l)fuZ4/ 40 kb, the lethal stage resulting from a mutationwhich FM6 females die as first instar larvae (BUSSONet al. would remove all fu sequences-but these sequences 1988). The second molecular class corresponds to only-is not known. It was unclear wether the larval events that only affect thefu transcription unit, and lethality of DfTl)fuz4/Ymales was due to the absence which, so far, were found to be smaller than 40 bp. of the fu gene itself, or to another gene included in The strongestfu alleles of this second class lead to a this deficiency. Studying the effect of Sucfu) on this late pupal lethality (WURSTand HANRATTY1979; phenotype allowed examination of this problem. Un- PERRIMON andMAHOWALD 1987) (Figure lD), while like the pupal lethality due to the fumHa3allele, the weakerfu alleles are viable and display only maternal larval lethality due to Dfl)fuz4 is not suppressed by effect lethality. Sucfu) mutations suppress the fu pupal Sucfu): Df(l)fuZ4/FM4;Sucfu~p/Sucfu~pfemales crossed lethality. For example, thefumHa3;Sucfuystock shows to FM6/U;Sufu)"'/Sucfu~males generate Df(l)fuZ4/ no significant pupal lethality. These flies are viable, Y;Sucfu)Lp/Sucfuypindividuals which die as larvae, de- fertile, and display no fu phenotype (Figure 3F). The spite the presence of the homozygous Sucfuy' muta- 730 T. Prht

FIGURE4.--Suppression of the segment polarity phenotype offu embryos by Suppressor of fused. (A) Ventral view of a wild-tvpe embryo. Fhch thoracic and abdominal segment bears denticle belts in the anterior region. Denticles appear white on this dark field image (eight abdominal belts are prominent). (B) Homozygous Su(fiu)'.p/S~(fiu~~embryo displaying normal pattern. (C)fu' embryo. Fach denticle belt is duplicated in a mirror image and the naked cuticle is absent. Note the reduced size of this embryo. (D)fu';Su(fiu)'.pembryo. The segment polarity fu phenotype is fully suppressed. There is no sign of denticle duplication, and the size of the embryo is normal suggesting that no pattern is deleted either. These embryos hatch and develop into normal adults. tion. As all the otherfu phenotypes are suppressed by the fu' allele ("paternal rescue") (Figure 1B). The Sucfu)-including the segment polarity phenotype of following observations indicatethat, likefu, Sucfu)is a Df(l)fu"' itself (T. PR~AT,unpublished result)-this maternal effect gene.But, unlike fu, Sucfu)is not result strongly suggests thatthe larval lethality of "paternally rescued"-although it is very likely also Df(l)fu"' is not due tofu, but to an adjacent gene expressed in the embryo. included in the deficiency. fu'/Y;Sucfuy'/+ andfu 'Ifu';Sucfu)L'/+embryos de- Maternal and zygotic effects of Su($u) on the seg- rived froma fu'Ifu';SucfuY/Sucfuy' X fu'/Y;+/+ ment polarity phenotype of fu embryos: The seg- cross display a normal segmentation pattern (Figure ment polarity gene class includes maternal effect genes 6A) and give rise to adults. On the contrary,embryos and genes expressed fromthe embryonic genome with the same genotype but derived from the recip- (reviewed in Ingham and Nakano 1990). In order to rocal cross fu'lfu';+/+ X fu'/Y;Sucfu)"'/Sucfuydie as study the expression of Sucfu),and also to further embryos with amutant fu phenotype(Figure 6B). analyze the fu-Sucfu)interaction, suppression of the Thus, for Sucfu)as for fu, the phenotype of the em- phenotype of fu embryos in various Sucfu)context was bryos depends on the genotype of their mothers. This assessed. The segment polarity phenotype of fu em- suggests that SuY;)+ is expressed in the ovaries. bryos is determined by the fu/fu genotype of their The complete lack of maternal Su(fu)+ product in mothers. But, in fu/+ embryos derived from a fu/fu fu'/fu';Sucfu)"'/+ and fu'/Y;Sucfuy'/+ embryosde- X +/Y cross, the lack of maternalproduct can be rived froma fu'lfu';Sucfu~'/Sucfu~X fully;+/+ partially compensated by early zygotic expression of cross insures a completesuppression of the fu segment Suppressor of MutationSegmentation 73 1

, tc I

FIGURE5."Suppression of the _.. tumorous ovaries phenotype by Sup- I pressor offused. (A) Wild-type egg chambers at differentdevelopmental stages. The nuclei are stained with fuchsin. The large cells are nurse cells (nc). (B) Tumorous egg-cham- bers from fu"'' females. These chambers are made of hundred of tumorous cells (tc). (C) Eggchamber fromfu';S+f" females. No tumors are observed.

b

polarity phenotype, despite the presence of a Sufu)+ lack of zygotic Su(fu)+ product.fu'/Y;Sufuy/+ males copy in the embryo: the maternal effect of the Sufu) derived from a fu'/FM6;+/+ X +/Y;Sufu)"'/Sufuy mutation is not"paternally rescued" by Sufu)'. It cross display the same partially suppressed phenotype could mean either that the Sufu)+ gene is not ex- as the males of the same genotype derived from the pressed in the embryo, or that the expression of a reciprocal cross fu'/FM6;Sufu~'/Sufu~X +/Y;+/+ single Sufu)' embryonic copy cannot compensate for (not shown). the complete lack of maternal product. I mentioned Sw'fu) is semidominant: The two mutageneses re- that ful/fu';Sufu)LpI+ and fu'/Y;Sufuy'/+ embryos portedhere allowed recovery of semidominant or derived from a fu'lfu';+/+ X fu'/Y;Sufuy/Sufu)"' dominant fu suppressors, as the mutations were cross die with a segment polarity phenotype (Figure screened as heterozygotes in a fu background. The 6B). Actually, their phenotype is on average weaker Sufuy' mutation, which behaves as an amorphic al- than the one displayed by fu'lf.' and fu'/Y embryos lele, is semidominant. For example,f~""~'/Y;Sufu~'/ (Figure 4C). Also, 5% of the embryos derived from a + flies display a partially suppressed wing phenotype fu'lfu';+/+ X fu'/Y;Sufuy'/Sufuy' cross were found while fu"""/Y;Sufu~'/Sufu~' wings are wild type to hatch (Table 2), whereas fu' embryos never do. (Figure 3). A semidominant effect is also observed in The absence of one zygotic Sufu)+ copy allows a the case of thematernal suppression of the fu segment partial suppression of the embryonic fu phenotype, polarity phenotype, since embryos derived from fu'/ a despite the presenceof the maternal Su(fu)+ product. fu';Sufu)L'/+ X fu'/Y cross display a partially sup- Sufu) presents both a maternaland a zygotic suppres- pressed segment polarity phenotype(Figure 6C), sion effect on fu embryos phenotype. Inthe wild type, whereasembryos derived froma fu'lfu';Sufuy/ some functional Su(fu)+ product is made in the em- Sufuy X fully cross are normal (Figure6A). bryo. Therefore, the fact that the maternal effect of The fused phenotype is enhanced by a SW)+ the Sufu) mutation is notaffected by the zygotic duplication: Sufu) loss-of-function mutations sup- expression of a single Sufu)+ allele in embryos derived press the fu phenotype. The effect of overexpression of the Su(fu)+ function was studied by generating fu from a fu'lfu';Sufuy/Sufuy X fu'/Y;+/+ cross-as flies bearing three copies of the Sufu)+ gene. The shown by the complete suppression of the segment fused vein phenotype of fu1/Y;Tp(3,2)ry+,Sufu)+/+;+/ polarity fu phenotype-is probably due to the fact that flies is more severe than that of fu'/Y flies (Figure the amountof zygotic Su(fu)+ product remainsbelow + 7). Thus, increasing the expression of Sufu)+ leads to the required threshold in these embryos. a stronger fu mutant phenotype. The Fu and Su(fu) Zygotic effect of SW)on the phenotype of fu functions are tightly correlated since the strength of adults: Geneticproperties of fu indicate that it is the fu phenotype is directly related to the number of expressed in the female germ line and in the embryo, Sufu)+ copies in the genome. but also later for adult development (FAUSTO-STER- LING 1978; WURSTand HANRATTY 1979).Unlike the DISCUSSION suppression of the embryonic phenotype, the suppres- sion of the phenotype of fu adults does not depend on The search for suppressor mutationshas long been the Sufu) genotype of their mother, but only on the recognized as a fruitful approach for identifying new 732 T. Pr6at

FIGURE6.-Maternal and zygotic effects displayed by Sufi). (A) Em- bryo derived from a fu'/fu';Su(@)'J/ SU~U)'~X fully;+/+ cross. The seg- ments are normal. These embryos hatch and eventually giverise to adults. (B) Embryo derived from the reciprocical cross fu'lfu';+/+ X fu'/ Y;Sufu)""/Sufu~".Most of the seg- ments bear duplicated denticles. However, the segment polarity phe- notype is weaker than that off?' em- bryos (Figure 4C), and 5% of these embryos were found able to hatch. (C) Embryo derived from a fu'/ fu';Sufu)"Pl+ X fu'/Y cross. Maternal suppression by the heterozygous Sufu)"" mutation is not complete since some segments still display an abnormalpattern. About 35% of these embryos hatch but most indi- viduals do not reach the adult stage as they die as larvae or early pupae.

TABLE 2 Genetic propertiesof Suppressor of fused

Hatched Parental genotypes Genotype of offspring Phenotype of offspring embryos -Complete suppression of the segment po- larity fu phenotype -Complete suppression of the adult fu phenotype

-fu/jiu;Su~u)/+ -Complete suppression of the segment po- -fu/Y;Su(fu)/+ larity fu phenotype -Partial suppression of the adult fu pheno- 'Y Pe fulfu;+/+ x fu/Y;sufu)/.sufu) -fu/fu;Su(jiu)/+ Partial suppression of the segment polarity -fu/Y;Sufu)/+ fu phenotype fulfu;Sufu)/+ x fu/Y;+/+ -f~lfu;SUfu)/+ Partial suppression of the segment polarity -fulY;.Su(fu)/+ fu phenotype (the suppression is always -fulfu;+/+fu/Y;+/+ stronger than the one observed in the previous cross) fil/FM6;+/+ x fu/Y;Sufu)/Sufu) -fulfi;sufu)/+" Partial suppression of the adult fu pheno- -fulY;Sufu)/+ 'Y Pe f~l/FM6;.sufu)/.sufu)x fu/Y;+/+ -fulfu;Su(fiu)/+" Partial suppresson of the adult fu pheno- -fulY;Sufu)/+ type (similar to the one observed in the previous cross) f~/FM6;Sufi)/Sufu)X fu/Y;Sufu)/ Complete rescue of the adult fu phenotype Sufu) fu/FM6 X +/Y;Tp(3,2,2)Sufu)+/+ More severe adult fu phenotype

" Only the relevant genotypes are shown. Suppressor of Segmentation Mutation 733

FIGURE7."A Sucfu)+ duplication increases the fused phenotype. (A) Wing offu'/Y;Sucfu)'"'/+ individual. (B) Wing offu'/Y individual.fu' is a hypomorphic allele since wings display a weak fusedphenotype (only the proxinlal portion of LV4 is missing). (C) Wing of fu'/ Y;rp(3,2)ry',Sucfu~+/+;+/+fly. The fused phenotype is severely increased. LV4 is almost completely deleted and LV3 is thickened.

genes involved in a given pathway. At the outset of one,and there is agood correlation between the this study however, there was no description of a strength of these two phenotypes conferred by various mutageneses specifically performed in order to isolate fu alleles (BUSSONet al. 1988). Third, thevarious adult suppressors of segment polarity phenotypes. I report defectsbehaved similarly duringtemperature shift here thecharacterization of anew gene, Suppressor of experiments (WURSTand HANRATTY1979). fused, located in the 87C8 region of the third chro- Maternal and zygotic effects of Sucfu,, are dosage mosome. SuY;) mutationswere isolated after EMS sensitive: The screens described here allowed recov- mutageneses, and screened based on the suppression ery of dominant or semidominant mutations. SuY;) of fu adult phenotype. They correspond to loss-of- alleles belong to the latter group:when heterozygous, function mutations; the amorphic mutation is homo- the Su(fuyp amorphic allele only partially suppresses zygous viable and Sufu)/SuY;) flies do notdisplay any the wing phenotype as well as the segment polarity obvious phenotype. phenotype,but suppression is complete when the When produced by fu/fu mothers, fu embryos die Suvu)Lp mutation is homozygous. Interestingly,a and display a segment polarity phenotype. When pro- Su(@)+ duplication has an effect opposite to that of duced by fu/+ females fu embryos develop normally, Sufu) loss-of-function mutations, as it enhances the fu butfu adults display several anomalies (abnormalwing phenotype. This outlines the close and antagonistic veins, tumorous ovaries. . .). Suyu) fully suppresses relationship betweenfu and Suvu). both embryonic and adultfu phenotypes (Table 2). It The Su(fu)+function, as observed in a fu back- is striking that fu;Sucfu) flies are wild-type and give ground, is very sensitive to any decrease in the amount rise to normal progeny. The fact that all fu defects of Su(fu)+ product. Sufu) presents both a maternal are suppressed by Sufu) suggests that theFu+ product and a zygotic suppression effect on the fu segment is involved in a single molecular process occurring at polarity phenotype. A partial lack of any of the ma- different stages duringdevelopment, despite the ternal or zygotic Su(fu)+ products affects the pheno- pleiotropic effect of the fu mutations. Three inde- type of fu embryos. Also, the total lack of maternal pendent observations supportthis interpretation. First Su(fu)+ product cannot be compensated for by the all fu alleles lead to the segment polarity phenotype, zygotic expression of a single Su@)+ copy. This situ- the wing phenotype, andthe ovarianphenotype ation differs from the Fu function itself, which is less (SMITHand KING 1966; BUSSONet al. 1988). Second, sensitive to a decrease of the amount of Fu+ product. as I mentioned before, both segment polarity pheno- First, fu is a recessive mutation. Second, fu embryos type and wing phenotype can be described as lack of display asegment polarity phenotype onlyif their a posterior structure and duplication of an anterior mothers are fu/fu. The complete lack of fu zygotic 734 T. Prkat

expression alone does not lead to even a partial seg- the molecular mechanisms involved during the fu- ment polarity phenotype. Third,fu presents a paternal Sufu) interaction. rescue since the zygotic expression of thefu+ gene is Sucfu) mutations are viable and do not have any able to partially compensate for the lack of maternal visible phenotype of their own. This raises the ques- product. One explanation could be that, unlike Fu, tion of therole of this gene during development. Su(fu) is notan enzyme but instead interacts with Indirectly the same question arises forfu itself, since anotherprotein in astoichiometric manner. Thus, fu;Sucfu)individuals have normal embryonicand adult any decrease in the amount of Su(fu) product would development in the absence of Fu+product. One affect the activity of this other protein. explanation could be thatSucfu)+ is a segment polarity Interaction between fw' and Sudfu)'; inhibition gene whose function is partially duplicated. Thefu+ and competition models: Sucfu) interacts with allfu gene may have been selected to limit spatially or alleles and, in particular, it suppresses the segment temporally the expression of Sucfu)+. Also, Sufu) and/ polarity phenotype of the 40-kb deficiency Df(l)fuz4 orfu;Sucfu) individuals may have a subtle phenotype (T. PR~AT,unpublished results). Thus, Sucfu) muta- which has not yet been observed.Thus, they may have tions do not act by reactivating the expression of the a decreased fitness compared to fu+;Sucfu)+individ- fu locus itself. In this respect Sufu) differs from most uals, which might explain why both genes were main- tained throughout . other suppressors known in Drosophila, which inter-

act with specificalleles that correspond to the insertion 1 thank JANOS GAUSZ, the Bowling Green stock center, and the of transposable elements (reviewed in KUBLI 1986). Bloomingtonstock center for providing stocks. 1 thank JANINE In these cases, and unlike thefu-Sucfu) interaction, the BLANC for technical assistance during the isolation of theS~(,fiu)'~~ suppression effect is related tothe nature of the allele. 1 am grateful toDENISE BUSSON, JEAN-MAURICE DURA, CLAU- DIE LAMOUR-ISNARD,BERNADETTE LIMBOURG-BOUCHON, PASCAL mutation itself ratherthan to the function of the TH~RONDand HERV~ TRICOIRE forfruitful discussions and for suppressed gene. Two differentmodels-at least-allow their helpful comments on the manuscript. I thank EMMANUELLE interpretation of the suppression of the fu phenotype FABRE and LEONARD RABINOW for their work on the manuscript and for their support. 1 thank the Ministire de I'lndustrie et de la by Sufu). In the first one (inhibition model),the main Recherche and the Association pourla Recherche contrele Cancer function of the Fu+ ser/thr kinase in wild-type individ- for financial support. uals would be to inactivate the Su(fu)+ protein. In the absence of Fu+ product the Su(fu)+ functionwould be LITERATURECITED overexpressed,leading to a fu phenotype. If no Su(fu)+ product is synthesized, the presence of the BAUMGARTNER,S., D. BOPP, M. BURRIand M. NOLL, 1987 Structure of two genes at the gooseberry locus related to Fu+ product would no longer be required and there- the paired gene and their spatial expression duringDrosophila fore fu;Sucfu) individuals would benormal. In an embryogenesis. Genes Dev. 1: 1247-1267. alternative model (competition model), the Fu+ and BOPP, D., M. BURRI,S. BAUMGARTNER,G. FRIGERIOand M. NOLL, Su(fu)+ products have antagonistic effects and act 1986 Conservation of a large in the segmen- competitively. The Fu+ ser/thr protein kinase and the tation gene paired and in functionally related genes of Dro- sophila. Cell 47: 1033-1040. Su(fu)+ protein would have a common substrate, but BOUROUIS,M., P. MOORE,L. PUEL, Y. GRAU,P. HEITZLER andP. their actions would have opposite effects on the activ- SIMPSON, 1990 Anearly embryonic product of the gene ity of this third protein. The absence of both Fu+ and shaggy encodes a serine/threonine protein kinase related to Su(fu)+ would result in a close to normal balance CDC28/cdc2+ subfamily. EMBO J. 9: 2877-2884. status. BUSSON,D., B. LIMBOURG-BOUCHON,M.-C. MARIOL,T. PR~ATand C. LAMOUR-ISNARD, 1988 Geneticanalysis of viable and lethal Of course, the interaction between the Fu+ kinase fused mutants ofDrosophila melanogaster.Wilhelm Roux's Arch. and Su(fu)+ could involve some intermediate.For Dev. Biol. 197: 221-230. example in the inhibition model, another protein ki- COULTER,D. E., and E. WIESCHAUS,1988 Gene activities and nase could be activated after phosphorylation by the segmental patterning in Drosophila: analysis of odd-skipped and Fu+ kinase and inactivate the Su(fu)+ protein. Also, pair-rule double mutants. Genes Dev. 2: 1812-1823. the Fu+ and Su(fu)+ functions could be expressed in DESPLAN,C., J. THEISand P. O'FARRELL, 1985 The Drosophila developmentalgene, engrailed, encodesa sequence specific different parts of the segments and interact thanks to DNA binding activity. Nature318: 630-635. some other segment polarity products involved in cell- DINARDO,S., andP. H. O'FARRELL, 1987 Establishement and cell communication. However, the observation that refinement of segmentalpattern in the Drosophila embryo: the suppression of thefu syndrome by Sucfu) is total spatial control of engrailed expressionby pair-rule genes. Genes suggests that Fu+ and Su(fu)+ are directly involved in Dev. 1: 1212-1225. DINARW,S., E. SHER, J. HEEMSKERK-JONGENS,J. KASSIS and P. the same molecular process. Cloning of the Sucfu)+ O'FARRELL,1988 Two-tiered regulation of spatially pat- gene is underway,together with the study of the terned engrailed duringDrosophila embryogen- expression of some other segment polarity genes in esis. Nature 332: 604-609. fu;Sucfu) embryos. These results should shed light on EATON,S., and T. B. KORNBERG,1990 Repression of ci-D in Suppressor of Segmentation Mutation 735

posterior compartments ofDrosophila by engrailed. GenesDev. KING, R. C., 1959 Oogenesis infulfu females. Drosophila Inform. 4: 1068-1077. Serv. 33: 142. FAUSTO-STERLING, A.,1978 Pattern formation in the wing veins KING,R. C., 1970 Ovarian Development in Drosophila melanogaster. of the fused mutant (Drosophila melanogaster). Dev. Biol. 63: Academic Press, New York. 358-369. KUBLI,E., 1986 Mechanisms of suppression inDrosophila. Trends FJOSE,A. W., W. J. MCGINNISand W. GEHRING, 1985 Isolation Genet. 2: 204-209. of a homeobox containing gene from the engrailed region of LEWIS, E. B., and F. BACHER,1968 Methodof feeding ethyl Drosophila and the spatial distribution of its trancripts. Nature methane sulfonate (EMS) to Drosophila males. Drosophila In- 313: 284-289. form. Serv. 43: 193. GANS,M., C. AUDITand M. MASSON,1975 Isolation and charac- LIMBOURG-BOUCHON,B., D. BUSSON andC. LAMOUR-ISNARD, terizationof sex-linked female-sterile mutants in Drosophila 1990 Interactions betweenfused, a segment polarity gene in melanogaster. Genetics 81: 683-704. Drosophila, and other segmentation genes. Development 112: GAUSZ,J., A.A. M. AWAD and H.GYURKOVICS, 1980 New defi- 41 7-429. ciencies for thekar locus ofD. melanogaster. Drosophila Inform. LINDSLEY, D.L., and G. ZIMM, 1985 The genome of Drosophila Serv. 45: 45-46. melanogaster, Part 1: Genes A-K. Drosophila Inform. Serv. 62. GAUSL,J., G. BENCZE,H. GYURKOVICS,M. ASHBURNER,D. ISH- LINDSLEY, D.L., and G. ZIMM, 1987 The genome of Drosophila HOROWICZand J. J. HOLDEN,1979 Genetic characterization melrnogaster, Part 3: Rearrangements.Drosophila Inform. of the 87C regionof the third chromosome of Drosophila Serv. 65. melanogaster. Genetics 93: 917-934. LINDSLEY, D. L., and G. ZIMM, 1990 The genome of Drosophila GERGEN,J. P., and E. WIESCHAUS,1986 Localized requirements melanogaster, Part 4: Genes L-Z, balancers, transposable ele- for geneactivity in segmentation ofDrosophila embryos: analy- ments. Drosophila Inform. Serv. 68. sis of armadillo,fused, giant and unpaired mutations in mosaic MARIOL, M.-C., T. PR~Tand B. LIMBOURG-BOUCHON, embryos. Wilhelm Roux’s Arch. Dev. Biol. 195: 49-62. 1987 Molecular cloning offused, a gene required for normal GONZALEZ,F., L. SWALES,A. BEJSOVEC,H. SKAER and A. MARTINEZ segmentation in the Drosophila melanogaster embryo. Mol. Cell. ARIAS,1991 Secretion and movement of wingless protein in Biol. 7: 3244-3251. the epidermis of the Drosophila embryo. Mech. Dev. 35: 43- MARTINEZ-ARIAS,A., 1985 The development offused embryos of 54. Drosophila melanogaster. J. Embryol.Exp. Morphol. 87: 99- HEEMSKERK,J., S. DINARDO,R. KOSTRIKENand P. H. O’FARRELL, 114. 199 1 Multiple modes of engrailed regulation in the progres- MARTINEZ-ARIAS, A.,N. E. BAKERand P. W. INGHAM,1988 Role sion towards cell fate determination. Nature352: 404-410. of segment polarity genes in the definition and maintenance of HIDALGO, A,,1991 Interactions between segment polarity genes cell states in the Drosophila embryo. Development 103: 157- and the generation of segmental pattern in Drosophila. Mech. 170. Dev. 35: 77-87. MORGAN,T. H., and C. B. BRIDGES,1916 Sex-linked inheritance HIDALGO, A., and P. INGHAM, 1990 Cell patterning in the Dro- in Drosophila. Carnegie Inst. Wash. Publ. 237: 55-58. sophila segment: spacial regulation of the segment polarity gene NAKANO,Y., I. GUERRERO,A. HIDALGO,A. M. TAYLOR,J. R. S. patched. Development 110 291-301. WHITTLEand P. W. INGHAM,1989 The Drosophila segment HOOPER,J., and M. P. SCOTT, 1989 The Drosophila patched gene polarity genepatched encodes a proteinwith multiple potential encodes a putative membrane protein required for segmental membrane spanning domains. Nature 341: 508-513. patterning. Cell 59 751-765. NUSSLEIN-VOLHARD,C., and E. WIESCHAUS,1980 Mutationsaf- HOWARD,K., andP. W. INGHAM, 1986 Regulatoryinteractions fecting segment number and polarity in Drosophila. Nature between the segmentation genefushi tararu,hairy and engrailed 287: 795-801. in the Drosophila blastoderm. Cell 44: 949-957. ORENIC,T. V., D. C. SLUSARSKI,K. L. KROLL~~~R. A. HOLMGREN, HULSKAMP,M., C.SCHRODER, C. PFEIFLE,H.JbcKLEand D. TAUTZ, 1990 Cloningand characterization of the segment polarity 1989 Posterior segmentation of theDrosophila embryo in the gene Cubitus interruptus Dominant of Drosophila. Genes Dev. 4: absence of a maternal posterior organizer gene. Nature 338: 1053-1067. 629-632. PEIFER, M., and E. WIESCHAUS,1990 The segment polarity gene INGHAM,P. W., 1988 The molecular genetics of embryonic pat- armadillo encodes a functionally modular protein that is the tern formation in Drosophila. Nature 335 25-34. Drosophila homologof plakoglobin. Cell 63: 1167- INGHAM,P. W., 1991 Segment polarity genes and cell patterning 1178. within the Drosophila body segment. Curr. Opin. Genet. Dev. PERRIMON,N., and P.MAHOWALD, 1987 Multiplefunctions of 1: 261-267. segment polarity genesin Drosophila. Dev. Biol. 119: 587-600. INCHAM, P. W.,N. E. BAKERand A. MARTINEZ-ARIAS, PR~AT, T.,P. TH~ROND,C. LAMOUR-ISNARD, B. LIMBOURG-BOU- 1988 Positive andnegative regulation of segment polarity CHON, H. TRICOIRE,I. ERK, M.-C. MARIOLand D.BUSSON, genes in the Drosophila blastoderm by the pair rule genesfushi- 1990 A putative serine/threonine protein kinase encoded by tarazu and even-skipped. Nature 331: 73-75. the segmentpolarityfused gene ofDrosophila. Nature 347: 87- INGHAM,P. W., and Y. NAKANO,1990 Cell patterning and seg- 89. mentpolarity genes in Drosophila. Dev.Growth Differ. 32: RIJSEWIJK,F., M. SCUERMANN, E. WAGENAAR, P. PARREN,D. WEI- 563-574. GEL and R. NUSSE, 1987 The Drosophila homologue of the INGHAM, P. W., A. M. TAYLORand Y.NAKANO, 1991 Role of the mouse mammary oncogene int-l is identical to the segment Drosophila patched gene in positional signalling. Nature 353: polarity gene wingless. Cell 50: 649-657. 184-187. SEIGFIRED,E., L.A. PERKINS, T. M. CAPACIand N. PERRIMON, IRISH,V., R. LEHMANNand M. AKAM, 1989 The Drosophila 1990 Putative kinase product of the Drosophila segment-po- posterior-group gene nanos functions by repressing hunchback larity gene zeste-white3. Nature 345: 825-829. activity. Nature 338: 646-648. SMITH, P. A., and R. C. KING, 1966 Studies onfused, a mutant 736 T. P&at

gene producing ovarian tumors in Drosophilamelanogaster. embryo and imaginal discs of Drosophila: what are the links? JNCI 36: 445-463. Dev. Biol. 145: 1-12. VANDEN HEUVEL,M., R. NLJSSE,P. JOHNSTON and P. A. LAWRENCE, WURST,G. G., and W. P. HANRATTY,1979 Studies of the devel- 1989 Distribution of the wingless geneproduct: a protein opmental characteristics of fused mutants of Drosophila mela- involved in cell-cell communication. Cell 59: 739-749. nogaster. Can. J. Genet. Cytol. 21: 335-346. VANDER MEER,J. M., 1977 Optically clean and permanent whole ZALOKAR, M., and I. ERK,1977 Phase-partition fixation and stain- mount preparation for phase contrast microcospy of cuticular ing of Drosophila eggs. Stain Technol. 52: 89-95. structures of insect larvae. Drosophila Inform. Serv. 52: 160. WILKINS,A. S., and D. GLJBB,1991 Pattern formation in the Communicating editor: T. SCHUPBACH