The Apterous Locus in Drosophila Melanogaster

APPARENT GENETIC COMPLEXITY GENERATED BY DEVELOPMENTAL THRESHOLDS: THE APTEROUS LOCUS IN DROSOPHILA MELANOGASTER

MARY E. STEVENS' AND PETER J. BRYANT Developmental Biology Center and Department of Deuelopmental and Cell Biology, University of Cal$ornia, Imine, Calij-ornia 9271 7 Manuscript received June 15, 1984 Revised copy accepted February 14, 1985

ABSTRACT Mutations at the apterous (up) locus in Drosophila melanogaster give rise to three distinct phenotypes: aberrant wings, female sterility and precocious adult death. The wing phenotype includes five types of abnormality: blistering, deficiencies, duplications, high-order repetitions and transformation of structures. The mildest phenotype is seen with homozygous up"" animals which have either normal or slightly blistered wings. Most alleles produce, in the homozygote, a deficient wing in which part or all of the wing margin and wing blade is missing, but wing hinge and notum regions are normal. Animals hemizygous for each of 20 ap alleles, as well as ap*D/a$xnheterozygotes, show duplication of parts of the notum associated with complete wing deficiency. Animals heterozygous for up' and the other tested ap alleles show repetitions of parts of the anterior wing margin, an engrailed-like transformation of posterior wing margin into anterior margin or both. Both up'" and upc show similar phenotypes in homozygotes and hemizygotes, yet both produce a less extreme phenotype than that of the other hemizygotes, suggesting that neither mutation causes loss of the entire up' function. The 15 alleles that cause precocious death and female sterility occur in six complementation groups based on complementation for these phenotypes. This supports the previous conclusion that the effects of apterous mutations on the wing do not correlate with their effects on viability and fertility. We propose an explanation for the effects of apterous mutations on the wing in which quantitative reductions in the activity of gene product give rise to qualitatively different phenotypes because of different threshold requirements of the up+ function for critical events in wing disc development.

HE complexity of complex genetic loci in higher organisms takes several T forms. Different mutations of such loci often give different phenotypes, and the mutant phenotypes are often pleiotropic. Furthermore, pairs of alleles often show partial or complete interallelic complementation, making definition of functional genetic units difficult. Consequently, there have been several interpretations of the structure and function of complex loci (for reviews, see CARLSON1959; JUDD 1976). One view is that such a locus may code for a single multifunctional protein with several domains, each of which is respon-

Present address: Laboratory of Radiobiology, University of California, San Francisco, California 94 143.

Genetics 110: 281-297 June, 1985. 282 M. E. STEVENS AND P. J. BRYANT sible for a particular function; for example, in Neurospora, the mom locus encodes a polypeptide with five enzyme activities (GAERTNERand COLE1977). Similarly, in Drosophila three enzyme activities controlled by the rudimentary locus appear to be associated with a single polypeptide (BROTHERSet al. 1978), and GRACE(1980) has proposed that the dumpy locus may also encode a multifunctional polypeptide. Complex loci have also been interpreted as con- sisting of clusters of closely linked genes which show genetic interaction; for example, bithorax (LEWIS 1978), Antennapedia (LEWISet al. 1980), decapentaplegic (SPENCER, HOFFMANand GELBART1982) and Notch (PORTIN1975) of Drosophila have all been considered as multigene complexes. The apterous (up;2-55.2) function in Drosophila behaves as a complex locus in that different mutations give different phenotypes which are often pleiotropic, and that complementation occurs between some pairs of alleles. The phenotypes caused by apterous mutations include several types of abnormal wings, female sterility, precocious adult death, abnormal gut morphology, per- sistence of larval fat body cells in the adult (BUTTERWORTHand KING 1965; BUTTERWORTH1972; WILSON 1980), leg deficiencies and duplications and oc- casional antenna-to-leg transformations (M. STEVENS,unpublished observations). The female sterility is associated with nonvitellogenesis which is thought to be caused by juvenile hormone deficiency, since application of a juvenile hormone analog can stimulate vitellogenesis in up4 females (POSTLETHWAIT and WEISER1973). However, application of this compound before or during the temperature-sensitive period for adult death is not sufficient to prolong adult survival (WILSON 198 la). The relationship between juvenile hormone and the other up phenotypes has not been studied. By using a temperature-sensitive allele, WILSON (1981a) showed that the apterous locus has two distinct temperature-sensitive periods, one during the larval period for the wing phenotype and a second one during the pupal period for the precocious death and fertility phenotypes. The latter phenotypes are highly correlated with each other among genotypes (WILSON1980), but neither adult death nor female sterility shows correlation with the wing phenotype. These results indicate that the up gene product functions in different body parts at different developmental times and that mutations may preferentially interfere with one or another of its functions. The locus may, therefore, provide an instructive example of gene activity under both temporal and spatial control. In this paper, we describe the phenotypes of several up mutants in detail, and by studying their behavior in homozygous, hemizygous and heterozygous conditions, we arrive at a new interpretation of the apparent genetic complexity of this locus.

MATERIALS AND METHODS Stocks: Sixteen of the 24 ap alleles were kindly provided by T. WILSON;the other eight were obtained from Drosophila stock centers. We have classified the 24 alleles into six classes: in classes containing more than one allele, we refer to any allele within a particular class by using a superscript for that class. For example, ap” designates all of the class I1 alleles. WILSON(1980) noted that b pr ap*”’ homozygotes showed a much stronger phenotype than that described by LINDSLEY and GRELL(1968) for ap*’’;this observation and our genetic analysis indicate that ap”“ and ap*” COMPLEXITY OF THE APTEROUS LOCUS 283 are different alleles. Due to the low viability of most homozygotes, only the ap”’, up”’, apbUand a#‘ alleles were kept as homozygous stocks; 18 others were balanced over the SM5 chromosome (LINDSLEYand GRELL 1968). The two dominant alleles, ap’D and up-, were kept heterozygous with a chromosome carrying the homozygous lethal bw” mutation. A chromosome carrying M(2)S4, which is deficient for apterous, was used to produce animals hemizygous for various ap alleles. Except where noted otherwise, all stocks were maintained at 25” using a standard cornmeal, yeast, corn syrup and agar medium. Morphological phenotypes: Adult structures were prepared by removing the head, legs and abdomen, heating the thorax and wings in 20% KOH for 5 min and mounting the parts between coverslips in either Euparal or Faure’s mounting medium. The wing disc derivatives were examined under a Zeiss microscope at X125 and X300 magnification and were scored using BRYANT’S(1975) fate map of the wing disc. Characterization of sterility and precocious death pheno*e: Ten pairs of animals from each genotype were allowed to lay eggs for 24 hr at 25”. FI homozygotes or heterozygotes were scored for time of eclosion, and females were then tested for fertility and length of survival after eclosion by placing the animals into vials with several wild-type males and checking daily for dead animals and for the presence of Fa larvae. Heteroallelic combinations: All possible heteroallelic combinations of the 24 apterous alleles were made, using ten males and ten females in each cross. Reciprocal crosses were carried out, but since no significant differences were found between the results of the two crosses, the data for each combination were pooled. The parents were allowed to lay eggs for 6 days, transferred to fresh bottles for an additional 6 days and then discarded. Progeny from both sets of bottles were counted daily, with a minimum of 100 FI counted for each cross. The presence of the Cy marker on the SM5 chromosome (which carries a$+) enabled us to identify and use animals heterozygous for up+ as controls in the same culture vials. Some of the mutants that showed a clearly abnormal wing phenotype were nevertheless fertile and long-lived in both homozygous and hemizygous conditions. The normal phenotype seen in these hemizygotes indicates that one copy of such an allele is sufficient for a nonmutant phenotype; therefore, heterozygotes with these alleles were not used in the complementation analysis for sterility and precocious death. Heterozygotes between alleles that produce these mutant phenotypes when homozygous were checked daily for fertility and early death. The appearance of larvae in the vial indicated complementation for the sterility phenotype. In genotypes producing precocious death phenotypes, escapers occurred but with a frequency of 10% or less. We used the value of 20% of adults living for more than 5 days as our criterion for complementation of the precocious death phenotype.

RESULTS The wing phenotypes of apterous: Several wing phenotypes other than the structural deficiencies described by BUTTERWORTH and KING (1965) and WIL- SON (198 1a) were observed with various apterous genotypes during the course of this study. The wing abnormalities observed were blistering of the wing blade, various degrees of deficiency of wing structures (called “mitten,” “strap” and “nubbin,” in order of increasing severity), deficiency and duplication of notal structures, repetition of anterior wing margin elements and transformation of the posterior wing margin into structures characteristic of the anterior wing margin. These phenotypes are illustrated in Figures 1 and 2 and described in more detail below. Wing phenotypes in homozygotes, hemizygotes and heterozygotes: The apterous alleles can be divided into six classes according to their phenotypes in homozygotes, hemizygotes and heterozygotes with up’. The first four classes are recessive; classes V and VI are dominant. The characteristics of the alleles are listed in Tables 1 and 2. 284 M. E. STEVENS AND P. J. BRYANT

FIGUREI .-Wing phenotypes produced by representative class I-VI apterous genotypes. A. ap''/apb" blistered wing. Arrow indicates blister. B, ap77/lap'7"-ap wing (25").C, aph/aph nubbin wing. Arrow indicates wing material. D. apr/apr nubbin wing. Arrow, indicates wing material. E, apx"/ap' mitten wing. Arrow indicates notched margin. F. ap"/ap' strap wing (18'). Bar = 0.1 nim.

Class 1 (upb") homozygotes showed a relatively normal wing with blistering (i.e., failure of dorsoventral fusion) of the interior wing blade (Figure 1A). In hemizygous animals. the wings were slightly more blistered than in homozygous upb". Slight scalloping of the posterior wing margin and loss of marginal pattern elements were seen in 30% of upb'' homozygotes. WHITTLE (1979) has reported an engrailed-like phenotype in homozygous upb'' animals, but this has not been observed in our stocks. Class I1 alleles (up4', up54,up"', uph7*') are temperature sensitive. At 22", homozygotes showed a strap wing (Figure 1B) in which all of the wing margin and much of the wing blade were missing, although the wing hinge and notum COMPLEXITY OF THE APTEROUS LOCUS 285

I 1

A B

FIGURE2.--Wing phenotypes produced by hemizygous (A) and heteroallelic (B-D) apterous genotypes. A, M(2)"2'/ap'' reduced thorax. Arrow indicates loss of scutellum. B, apr/aph triplicated anterior margin. Arrow indicates triplicated triple row elements. C, ap'/aph transformed posterior margin. Arrow indicates triple row (anterior) bristles in posterior margin. D, ap"/ap" duplicated thorax (27"). Dashed line shows line of mirror symmetry. Bar = 0.1 mm. were normal. At 29", a more reduced nubbin wing was produced (see below). Hemizygotes grown at 18" showed a nubbin phenotype, and at 25" they showed the reduced thorax phenotype (see below). Class 111 (upaa,upsa, up4, up4a,upt3, up'*, up2', ups2, up4', up4', up4", up""/, up5'. up'*, up7*', up*"*)homozygotes produced the typical apterous phenotype, a very reduced nubbin wing in which nearly all of the wing blade material was absent (Figure 1C). Wing hinge elements were occasionally missing, but more often all hinge and notal elements were present. When these alleles, or the class I1 alleles grown at temperatures greater than 25", were made hemizygous using the deficiency A~f(2)S2~,a new phenotype was produced in which the entire scutellar region and most of the notal macrochaetes were missing, and the remains of wing hinge elements existed as a somewhat disorganized mass (Figure 2A). The anterior part of the notum was present, and separated from it by featureless cuticle there was occasionally a small piece of additional notum. Although the lack of macrochaetes made these extra structures difficult to score, the polarity of their microchaetes was reversed compared to those of the central, more complete notum, and we, therefore, scored them as dupli- 286 M. E. STEVENS AND P. J. BRYANT

zz>zoogo z>zzzz0~0000 z>zz>ogoog

A 999'4 9199" 999 Gg I-moo0 zoo +- 043gm CC Q, 10

OW" 0000000 "OOm moo T CO m n 3

Ion-I-m CO30 cci d+- -*m

E D Y c 2 z

0 0

-3m- mwomn

h mmmm ci Y C 3333 X c- COMPLEXITY OF THE APTEROUS LOCUS 287

TABLE 2 Characteristics of apterous hemizygotes

Hemizygous for ap Alive at

M(Zp2‘/SM I Total 4 days 5+ days 5+ days X NO. FI n % (a) (a) (%) Fertile Class I ap bl‘ 105 63” 60.0 2 61 96.8 Yes Class 11 ap4’/SM5 66 9 13.6 4 0 No ap54/SM5 110 46 41.8 34 0 No ap7’f/SM5 101 12 11.9 5 16.7 No ap””3/SM5 102 40 39.2 40 0 No Class 111 apso/SM5 146 1 0.6 1 0 0 No ap5”/SM5 174 27 15.5 22 9 29.0 Yes ap4/SM5 137 14 10.2 12 0 0 No ap4a/SM5 112 16 14.3 16 0 0 No ap”/SM5 67 10 8.9 10 0 0 No ap”/SM5 76 17 22.3 18 0 0 No app5/SM5 171 35 20.5 13 12 48.0 Yes apsp/SM5 51 14 27.5 15 0 0 No ap‘OlSM5 105 30 28.6 14 12 46.2 Yes ap46/SM5 153 33 21.6 37 0 0 No ap4’j/SM5 252 78 30.9 65 0 0 No ap56f cn 32 10 31.3 1 9 90 Yes ap57/SM5 160 6 3.8 2 0 0 No ap 58/SM5 94 10 10.6 2 0 0 No ap7%/SM5 114 33 28.9 24 0 0 No b pr apb“’/SM5 133 17 12.8 17 0 0 No Class IV apt 92 17 18.5 0 17 100.0 Yes Class V apxa/bw” 146 0 0.0 Class VI ap’’/bwv 44 0 0.0

a Blistered. cations. The change in phenotype is due to hemizygosity for apterous rather than to interaction between the apterous and Minute phenotypes, since homozygotes for ap3”,apbtr and apc in genetic backgrounds containing either of two other equally strong Minutes, M(2)c” and M( l)o‘J’, showed phenotypes typical of simple homozygotes. Class IV (up‘) homozygotes and hemizygotes produced the same phenotype; namely, the typical nubbin phenotype described above (Figure 1D). The class V dominant allele (a$’.; BUTTERWORTHand KING 1965) when heterozygous with ap+, produced a mitten wing which had part of the anterior wing margin and wing blade missing (Figure 1E). The wing hinge and notum were normal. This allele is homozygous and hemizygous lethal. The class VI temperature-sensitive dominant allele (up”), when heterozygous with ap+, produced the strap phenotype at 18 O (Figure 1F); at higher temper- 288 M. E. STEVENS AND P. J. BRYANT atures the wing became progressively more reduced. It is homozygous and hemizygous lethal. Wing phenotypes produced by heteroallelic combinations: The wing phenotypes produced by heteroallelic combinations between the different classes are listed in Table 3. Heterozygotes of upb1' (class I) with class I1 and class 111 alleles produced a phenotype similar to that of apb''/apb1'homozygotes. The wing phenotype of ap bl'/apc(class I/class IV) heterozygotes differed from that produced by either of these alleles in other heteroallelic combinations. This genotype did not produce any triplications, and one wing of 38 (3%) was completely normal. Posterior margin elements were missing in 84% of the wings examined, and the margin showed transformation in 68% of cases. Of these transformatiom, the posterior margin was transformed into double row (BRYANT1975) in 58% of cases, to triple row in 16% of cases and into both double and triple row elements in 13% of cases. The transformations were position specific, with the proximal part of the posterior margin replaced by triple row bristles and the distal part transformed to double row. One wing showed triplicated medial and distal costa on the anterior margin, although the triple row was unaffected. The vein pattern of 97% of apblt/apcheterozy- gotes was abnormal, ranging from no apparent veins to extra venation similar to the vein pattern of fused;engrailed double mutants (FAUSTO-STERLINGand SMITH-SCHIESS1982). Extra sensilla were often observed in the wing blade; two cases (6%) had an extra copy of vein I11 as shown by the distribution of sensilla identical with that of the normal vein 111. The wing phenotypes observed in apblt/apcwere not seen in our apblt/apbl'stocks or in any other allelic combination we have studied. Heterozygotes between upb1'(class I) and either of the two dominant alleles in classes V and VI showed the dominant phenotypes. In apb"/apxaheterozy- gotes, the wing, although blistered, was mitten shaped as in apxa/ap"; these were the only two genotypes in which the mitten phenotype was observed. There was insufficient wing blade in ap bl'/ap'D heterozygotes to determine whether blistering occurred in this genotype. Heteroallelic combinations within class I1 (25") and within class I11 gave the same phenotype (strap and nubbin, respectively) as the homozygotes in these classes. No complementation or new phenotypes were observed. Repetition of wing structures was seen in animals heterozygous for upc (class IV) and all class I1 and 111 alleles tested. The upc homozygote showed the nubbin phenotype as described above, but when made heterozygous with other alleles, upc produced a large abnormally shaped wing blade (Figure 2B). We have analyzed 256 wings of' one apc/apmgenotype, apC/ap3",in detail, although similar phenotypes were produced by the other upclap" and apc/apm heterozygotes. The apC/ap3"genotype produced 12.2% normal wings, 55.5% wings with repeated triple row (the three rows of bristles lying along the anterior wing margin; BRYANT1975), 54.2% with the posterior margin transformed into anterior margin structures (Figure 2C) and 26.1 % wings with both repetition and transformation. The repetition varied from three complete parallel triple rows (nine rows total) to partial repetitions in which a longitudinal line of symmetry lay within the triple row itself or along a shared row of dorsal or COMPLEXITY OF THE APTEROUS LOCUS 289

- sa 3 2 90 M. E. STEVENS AND P. J. BRYANT ventral triple row bristles. Although three copies of the triple row was the most common number produced, in 2.8% of apc/up3"animals five or more copies of triple row were observed lying along the anterior margin with additional disorganized patches of triple row bristles in the central area of the wing blade. The costal region of the wing was also sometimes present in multiple copies as was the tegula, although the tegula was usually apparently normal or duplicated rather than triplicated. Engrailed-like transformations, recognized as the presence of triple-row elements along the posterior wing margin (Figure ZC), were observed in animals heterozygous for up" and other class I, class I1 and class 111 alleles. Although the wing phenotype occasionally resembled the actual engrailed phenotype (LAWRENCEand MORATA 1976), the anterior compartment was also affected; 87% of upc/ap3awings were misshapen, blistered and/or enlarged, and the anterior margin showed the triplications described above. In heteroallelic combinations between class 11, I11 and IV alleles and the dominant class V and VI alleles, the phenotype was more severe than in either ap"/ap' or ap"/ap'. Heterozygotes of ap"/ap", ap"/aP" and apc/apxa showed a strap phenotype, whereas heterozygotes of ap"/ap", ap"'/ap" and apc/apfDproduced strap wings, nubbin wings and duplicated thoraces. The heteroallelic combination apc/upxaproduced additional margin elements.

Eighty-one percent of apx"/apiDheterozygotes grown at 25 O had strap wings, whereas the other 19% showed a deficiency/duplication phenotype. The wing hinge, wing blade and variable amounts of the notum were missing on the affected side, and remaining parts were clearly duplicated with mirror-image symmetry (Figure 2D). Only 5% of the animals had both sides duplicated, indicating that the occurrence of a duplication on one side does not influence the formation of a duplication on the other side. The duplications resembled the notal duplications reported in wingless animals (SHARMAand CHOPRA 1976); in fact, the apID (up"') allele interacts with the wingless locus to produce duplications in 8 1% of wg' ap"/wg up' double heterozygotes even though none are seen when only one locus is heterozygous and the other is wild type (M. STEVENS,unpublished observations). The extent of the duplication ranged from the entire notum duplicated to just a few bristles, usually including the scutellar bristles, present in two copies. Viability, fertility and longevity: Ten of the alleles appeared to give lowered preeclosion viability since the expected Mendelian ratio of homozygotes to heterozygotes was not observed (Table 1). We have not determined when during development this reduction in viability occurred. The female sterility and precocious adult death phenotypes were completely correlated in homozygotes as reported by WILSON (1980). Fourteen alleles showed the sterility phenotype as well as precocious death in the homozygote. Four of these (ups2,ups4, ups8 and up7'') produced some escapers from lethal- ity. Six alleles, up3", up4', ups6/,up"/, apblt and up", proved to be homozygous fertile and long-lived, and it was possible to maintain homozygous stocks with all of these alleles except ap". Alleles that showed long-term survival and fertility in the homozygote also COMPLEXITY OF THE APTEROUS LOCUS 291 2u 4 4u 13 18 49 49j 57 78e bit*

58 46

32 54 78j

FIGURE3.-Complementation map for the sterility/precocious death phenotypes. See text for explanation. showed these features in the hemizygote (Table 2), although in four of the six cases the frequency of survival was much lower in hemizygotes. One allele, up”f, was long-lived and fertile as a homozygote, but its low survival rate as a hemizygote precluded a statistically valid test of fertility. Another allele, up25, showed fertility and longevity as a hemizygote, although no homozygous up25 animals have been obtained. We also analyzed heteroallelic combinations for the precocious death and sterility phenotypes. WILSON (1980) found these two phenotypes to be highly correlated with each other in homozygotes, and our data support this conclusion with heterozygotes. Ninety-three percent of heterozygous genotypes that allowed fertility had increased longevity; conversely, 80% of the long-lived heterozygotes were fertile. Eight-two of 90 heteroallelic combinations did not show complementation for either phenotype. We have constructed a complementation map for the combined phenotypes of sterility and precocious death. The map shows 15 alleles occurring in six complementation groups (Figure 3).

DISCUSSION Comparisons between the mutant phenotypes of hemizygotes and homozygotes can often provide an indication as to whether a given mutation results in complete or only partial loss of gene product function. Amorphic mutations, with complete loss of function, show the same phenotype in hemizygous and homozygous genotypes, and the phenotype is the same as that of the homozygous deficiency when the latter can be determined. Hypomorphic mutations, with partial loss of function, show a more extreme phenotype in the hemizygote than in the homozygote, and these phenotypes are less extreme than that of an amorph or of the homozygous deficiency. Neomorphic mutations produce an effect not produced to any appreciable extent by the original normal gene and are not affected by additional doses of the normal gene (MULLER 1932). The apterous mutations do not behave entirely as expected from the above definitions. Twenty of the recessive apterous alleles, including the most extreme ones, show a more severe phenotype in the hemizygote than in the 292 M. E. STEVENS AND P. J. BRYANT homozygote and, therefore, represent hypomorphic mutations. The phenotype of a homozygous deficiency for apterous is not known, but the phenotypes produced by two alleles, upb'' (class I) and up" (class IV), are extremely similar in both homozygotes and hemizygotes. However, these alleles cannot be amorphic mutations, since their phenotypes differ from one another and both are less extreme than that of hemizygotes of the hypomorphic alleles. The two dominant alleles, upXa(class V) and upID (class VI), are both homozygous and hemizygous lethal. The addition of an additional dose of up' to produce upxu/up+/up' or up'D/up'/up' doses not give a different phenotype from the apXu/up+or up'D/up' heterozygotes (M. STEVENS,unpublished observations), suggesting that these are neomorphic mutations. The apterous locus clearly encodes a product whose function is necessary for normal wing disc development. Most of the alleles, including upb1' but excluding the dominant alleles and up", behave as though they reduce the activity of the gene product to varying extents. We propose that quantitative reductions in the activity of the up gene product lead to qualitatively different wing phenotypes because of threshold requirements of gene product for critical events in wing disc development. Such a developmental threshold model can account for many of our results, as shown in Figure 4. Each phenotype is assumed to be produced over a given range of up gene product activity, and total gene product activity for each genotype is simply the sum of the two allele-specific activities. If the up'/up+ genotype is assumed to produce full activity (1OO%), then up' allele-specific activity must be 50%. All of the mutant genotypes must decrease the total up activity to less than 50% since an up' hemizygote (assumed to have 50% of up+/up+ activity) is completely normal. Of all the mutant alleles upb1'appears to have the greatest amount of up' function since the homozygote shows only slight blistering and/or normal wings. The fact that the hemizygous phenotype is so similar to the homozygous phenotype indicates at least a two-fold range of gene activity for the blister phenotype. Allele-specific activity for upb'' is set at 16-25%, and both upb''l/Def and upbL*/upbl'are within the blister range. Lower levels of activity would produce more severe phenotypes in the following order: blistering, transformation, repetition, strap wing, nubbin wing, reduced thorax. Since some alleles produce strap wings in the homozygote but reduced thorax in the hemizygote, the intermediate nubbin phenotype must correspond to less than a two-fold range of gene product activity. The ranges of gene product activity for other phenotypes cannot be determined quantitatively from our data and are, therefore, expressed arbitrarily in Figure 4. The up" (class IV) allele and the dominant alleles do not fit into a simple additive model. If they behaved additively, two alleles that cause nubbin wings in the homozygotes would produce nubbin wings in the heterozygotes. How- ever, heterozygotes with up" show complementation, producing both normal wings and wings with margin triplications and/or transformations. It, therefore, appears that the heteroallelic combination of up"/up" and up"/up"' has more activity than expected from the homozygotes. This would imply that an allelic COMPLEXITY OF THE APTEROUS LOCUS 293 Phenotype Genotype hemizygote homozygote heterozygote 100 100 90 90 80 80 70 70 60 60 50 50

40 40

30 30 ,m

20 20

10 10 9 9 8 8 7 7 6 6 5 5 thorax 4 4

3 FIGURE4.-Threshold model for apterous phenotypes. Qualitatively different phenotypes are assumed to be produced by quantitative changes in the amount of up+ activity (see text). All alleles within any given class are symbolized by a superscript for that class, e.g., up" = up4', ap54,ap''f, ap ''j. 294 M. E. STEVENS AND P. J. BRYANT interaction occurs between up" and the other alleles, with heteroallelic combinations producing more gene product activity than expected from a simple additive mechanism. The behavior of the two dominant alleles, up'" and upxa, can also be explained by allelic interaction, assuming that these alleles interact with the up' gene product to decrease its activity. Heterozygotes of uprD and upxa with the hypomorphic class I1 and 111 alleles should then show a more severe phenotype than heterozygotes of up" and upxa with up'; this is observed. Although allelic interaction provides a convenient explanation for the aberrant behavior of upc and the dominant alleles, other explanations cannot be ruled out. For example, the effects of these mutations could involve changes in the timing or position specificity of mutant gene expression in the wing disc or defective interaction with other gene products during wing disc development. A series of vestigial alleles shows a quantitative order of wing deficiency phenotypes similar to that shown by apterous, with wing emargination ranging from 0 (normal wing) to 100% (nubbin) (GOLDSCHMIDT1938). Duplications, triplications and transformations were not reported in this series, although WADDINCTON(1 953) and JAMES and BRYANT(1 98 1) have described duplications of the thorax, with complete wing deficiency, in vestigial homozygotes. GOLDSCHMIDT(1938) suggested that the quantitatively different grades of deficiency are due to differences in the timing of a degeneration process. He proposed that the degeneration is due to a deficiency of some crucial substance necessary for normal differentiation, and that the deficiency of this substance must begin earlier with the more severe alleles. Similar differences in timing could be important in the case of apterous. The wing phenotypes of several up mutants suggest that the intact locus is necessary for cell survival in the imaginal wing disc of the larva. FRISTROM (1969) reported cell death in upxa third instar wing discs, and we have observed trypan blue staining, indicating cell death, in the wing discs of apblt/upb't(SED- LAK, MANZO and STEVENS1984), ap*"/upXa,upxa/ap+, ap"/apxa and apc/ap3 larvae (M. STEVENS, unpublished observations). Although WILSON(198 la) did not observe trypan blue staining in ~p"~~j/up"~Jmature wing discs, this does not rule out cell death in this genotype. We have observed cellular degeneration and debris in methylene blue-stained sections of up"/upcwing discs, which are also trypan blue negative (M. STEVENSand P. BRYANT,unpublished observations). The various wing deficiencies are easily explained by removal of the presumptive wing structures by cell death; the lack of regulation (regeneration or duplication) in response to this loss of tissue may be due to continual death of cells in this position or to insufficient time for regulation to occur. WILSON (1 980) suggests that the inability of up"78j wing discs to regenerate the degen- erated parts may be related to the juvenile hormone deficiency, but this seems unlikely since animals of fertile genotypes which presumably have juvenile hormone function often still have grossly deficient wings. Very late cell death may also explain the blistering seen in the wings of animals; SEDLAK, COMPLEXITY OF THE APTEROUS LOCUS 295 MANZOand STEVENS(1 984) have shown autophagy and basal extrusion of cell fragments in late third instar wing discs, indicating that late cell death does occur. The production of pattern duplications provides an example of the way in which quantitative differences in a primary abnormality might produce qualitatively different phenotypes by a threshold mechanism. Removal of less than about half of the wing disc leaves a fragment that can regenerate during subsequent culture, whereas removal of more than half leaves a smaller piece that produces a mirror-image duplicate instead (BRYANT1975). This has been explained by the shortest intercalation rule (FRENCH,BRYANT and BRYANT 1976), according to which circumferential intercalation following wound healing proceeds via the shorter of the two available circumferential routes until pattern continuity is restored. If less than half of the wing disc degenerates in apterous, regeneration might be expected but might fail to repair the defect because of degeneration of the regenerated cells. Degeneration of more than half of the disc, on the other hand, would lead to the qualitatively different phenotype of pattern duplication, as seen in uprD/upxaheterozygotes as well as class I1 and I11 hemizygotes. The wing margin transformations and triplications seen in upb"/upC,up"/up" and upc/upfrlheterozygotes are more difficult to explain. SZABAD,SIMPSON and NOTHICER(1 979) have shown that cell death in situ caused by damaging wing discs with a tungsten needle or by using a cell-lethal mutation can result in posterior margin transformations; BRYANT(1 97 1) produced wing and leg triplications in a similar manner. Irradiation of 12- to 24-hr-old embryos also produces wing triplications (POSTLETHWAIT1975). Therefore, it seems clear that treatments causing cell death can result in these phenotypes even though the exact mechanism is not known. GIRTON(1981) and BRYANT,FRENCH and BRYANT(1 98 1) have proposed that the leg triplications observed in I( 1)s-726 flies result from cell death followed by a particular mode of wound healing and regulative growth. A small cell death patch removing a minority of circumferential positional values followed by intercalation and distalization could result in a pattern triplication. The wing margin triplications seen in upc/uprf and upc/upffranimals may result from a similar mechanism. Many mutations, including several up and vg alleles, appear to preferentially affect the wing margin in Drosophila. With the temperature-sensitive allele up"'*j, WILSON(1 98 la) showed that the least extreme wing deficiencies involve the wing margin and that progressively more of the wing blade is missing in the more extreme phenotypes produced at higher temperatures. We have found that the wing margin is often the only structure affected in apc/uprrand upc/upm heterozygotes. The ap' gene product may, therefore, function in a pattern-forming process beginning at the wing margin. Alternatively, the margin region of the wing may simply be more susceptible to any perturbation, so that phenotypic abnormalities would be preferentially localized at this position, O'BROCHTAand BRYANT(1985) have observed a zone along the presumptive wing margin in late third instar wing discs that shows no DNA 296 M. E. STEVENS AND P. J. BRYANT synthesis; this growth repression may be responsible for the lack of regeneration of this area following cell death and for the preferential loss of these pattern elements in vg and up mutant adults. The various wing phenotypes associated with apterous may be explained by localized cell death at the appropriate time and location in the wing discs during larval development, but it is unclear whether a cell death mechanism can also explain the sterility and precocious death phenotypes. If a juvenile hormone deficiency causes sterility in apterous flies, this could be a result of cell death somewhere in the endocrine system. Cellular degeneration in many tissues may also contribute to early death of adult animals. WILSON (1981b) has shown using genetic mosaics that both sterility and precocious death appear to result from effects of the mutation on tissues localized in the posterior region of the abdomen. He proposes that both phenotypes result from cyto- toxicity in several tissues due to abnormal hemolymph produced by the mal- function of malphigian tubules. Thus, the primary effect of apterous mutations may be simply to decrease the viability of certain types of cells; the various phenotypes associated with the locus may all be secondary effects following this cell death.

This work was supported by grant HD06082 from the National Institutes of Health. We thank TIMSLITER for helpful discussions.

LITERATURE CITED BROTHERS,V. M., S. I. TUBOTA,S. E. GERMERAADand J. W. FRISTROM,1978 Rudimentary locus of Drosophila melanogaster: partial purification of a carbamyl phosphate synthase-aspartate transcarbamylase-dihydroorotase complex. Biochem. Gen. 16: 32 1-322. BRYANT,P. J. 1971 Regeneration and duplication following operations in situ on the imaginal discs of Drosophila melanogaster. Dev. Biol. 26: 637-651. BRYANT,P. J., 1975 Pattern formation in the imaginal wing disc of Drosophila melanogaster: fate map, regeneration and duplication. J. Exp. Zool. 193: 49-78. BRYANT,S. V., V. FRENCHand P. J. BRYANT,1981 Distal regeneration and symmetry. Science 212: 993-1002. BUTTERWORTH,F. M., 1972 Adipose tissue of Drosophila melanogaster. V. Genetic and experi- mental studies of an extrinsic influence on the rate of cell death in the larval fat body. Dev. Biol. 28 31 1-325. BUTTERWORTH,F. M. and R. C. KING, 1965 The developmental genetics of apterous mutants of Drosophila melanogaster. Genetics 52: 1 153-1 174. CARLSON,E. A., 1959 Comparative genetics of complex loci. Q. Rev. Biol. 34: 33-67. FAUSTO-STERLING,A. and H. SMITH-SCHIESS,1982 Interaction between fused and engrailed, two mutations affecting pattern formation in Drosophila melanogaster. Genetics 101: 71-80. FRENCH,V., P. J. BRYANTand S. V. BRYANT,1976 Pattern regulation in epimorphic fields. Science 193: 969-98 1. FRISTROM,D., 1969 Cellular degeneration in the production of some mutant phenotypes in Drosophila melanogaster. Mol. Gen. Genet. 103: 363-379. GAERTNER,F. H. and K. W. COLE, 1977 A cluster gene: evidence for one gene, one polypeptide, five enzymes. Biochem. Biophys. Res. Commun. 75: 259-262. GIRTON,J. R., 1981 Pattern triplications produced by a cell-lethal mutation in Drosophila. Dev. Biol. 84: 164- 172. COMPLEXITY OF THE APTEROUS LOCUS 297

GOLDSCHMIDT,R., 1938 Physiological Genetics. McGraw-Hill, New York. GRACE,D., 1980 Genetic analysis of the dumpy complex locus in Drosophila melanogaster: complementation, fine structure and function. Genetics 94 647-662.

JAMES, A. A. and P. J. BRYANT,1981 Mutations causing pattern deficiencies and duplications in the imaginal wing disc of Drosophila melanogaster. Dev. Biol. 85 39-54. JUDD,B. H., 1976 Genetic units of Drosophila complex loci. pp. 767-799. In: The Genetics and Biology of Drosophila, Vol. lb, Edited by M. ASHBURNERand E. NOVITSKI.Academic Press, New York LAWRENCEP. A. and G. MORATA,1976 Compartments in the wing of Drosophila: a study of the engrailed gene. Dev. Biol. 48 132-147. LEWIS,E. B., 1978 A gene complex controlling segmentation in Drosophila. Nature 276: 565- 570. LEWIS,R. A., B. T. WAKIMOTO,R. E. DENELLand T. C. KAUFMAN,1980 Genetic analysis of the antennapedia gene complex and adjacent chromosomal regions of Drosophila melanogaster. 11. Polytene chromosome segments 84A-84B1,2. Genetics 95: 383-397. LINDSLEY,D. L. and E. H. GRELL,1968 Genetic variations of Drosophila melanogaster. Carnegie Inst. Wash. Publ. 627. MULLER,H. J., 1932 Further studies on the nature and causes of gene mutations. Proc. 6th lnt. Congr. Genet. 1: 213-255. O'BROCHTA,D. A. and P. J. BRYANT,1985 A zone of non-proliferating cells at a lineage restric- tion boundary in Drosophila. Nature 313: 138-141. PORTIN,P., 1975 Allelic negative complementation at the abruptex locus of Drosophila melanogaster. Genetics 81: 121-133. POSTLETHWAIT,J. H., 1975 Pattern formation in the wing and haltere after irradiation of Dro- sophila melanogaster first instar larvae. Wilhelm Roux Arch. 178 29-50. POSTLETHWAIT,J. H. and K. WEISER,1973 Vitellogenesis induced by juvenile hormone in the female sterile mutant upterous-four in Drosophila melanogaster. Nature (New Biol.) 244: 284- 285. SEDLAK,B. J., R. MANZOand M. E. STEVENS,1984 Localized cell death in Drosophila imaginal wing disc epithelium caused by the mutation apterous-blot. Dev. Biol. 104 489-496. SHARMA,R. P. and V. C. CHOPRA,1976 Effect of the wingless (wg) mutant on wing and haltere development in Drosophila melanogaster. Dev. Biol. 48: 46 1-465. SPENCER,F. A., F. M. HOFFMANand W. M. GELBART,1982 Decapentaplegic: a gene complex affecting morphogenesis in Drosophila melanogaster. Cell 28 45 1-46 1. SZABAD,J., P. SIMPSONand R. NOTHIGER,1979 Regeneration and compartments in Drosophila. J. Embryol. Exp. Morphol. 49 229-241. WADDINGTON,C. H., 1953 The interactions of some morphogenetic genes in Drosophila melanogaster. J. Genet. 52: 243-258. WHITTLE,J. R. S., 1979 Replacement of posterior by anterior structures in the Drosophila wing caused by the mutation apterous-blot. J. Embrol. Exp. Morphol. 53: 291-303. WILSON,T. G., 1980 Correlation of phenotypes of the apterous mutant in Drosofihila melanogaster. Dev. Genet. 1: 195-204. WILSON,T. G., 1981a Expression of phenotypes in a temperature-sensitive allele of the apterous mutation in Drosophila melanogaster. Dev. Biol. 85: 424-433. WILSON,T. G., 1981b A mosaic analysis of the apterous mutation in Drosophila melanogaster. Dev. Biol. 85: 434-445. Communicating editor: T. C. KAUFMAN