Genetic control of sex determination in the germ line and soma of the housefly, Musca domestica

A number of different mechanisms for sex determination exist among living organisms. In the class of insects alone, sex can be determined by Y-chromosomal or by autosomal factors, by the ratio of X-chromosomes to autosomes, by haploidy versus diploidy, by maternal effects or by environmental factors (for a review see Nöthiger and Steinmann-Zwicky, 1985). Interestingly, different sex-determining mechanisms have evolved even within a single species, the common housefly Musca domestica (for a review see Dübendorfer et al., 1992).

In standard strains of Musca domestica, females are XX and males are XY whereby the Y chromosome carries a male-determining factor called M Y (Perje, 1948). Other strains exist in which the factor M is located on other chromosomes including the X (M X) and four of the five autosomes M I, M I I, M I I I, M V (Wagoner, 1969; Inoue and Hiroyoshi, 1982, 1984). In all these strains, the males are the heterogametic sex, generally designated as M/+, and females are +/+. It has been proposed that M has characteristics of a mobile element (Green, 1980; Nöthiger and Steinmann-Zwicky, 1985), similar to the male-determining factor in another dipteran insect, Megaselia scalaris (Mainx, 1966; Willhoeft and Traut, 1990).

There are, however, also strains in which males and females are homozygous for M, and females, in addition, are het-erozygous for a dominant female-determining factor on autosome IV, FD, that is epistatic to M (Rubini, 1967; McDonald et al., 1978). This factor FD is visualized as a constitutive mutation of a postulated key gene F that is differentially regulated in males and females of standard strains. We have proposed that an active wild-type F+ gene leads to female development, and that in M/+ strains, F+ is repressed by M whereas FD is insensitive to such repression (Nöthiger and Steinmann-Zwicky, 1985; Inoue and Hiroyoshi, 1986; Hilfiker-Kleiner et al., 1993).

Furthermore, mutations are known that lead to sexual transformation by a maternal effect. Two laboratory strains, Arrhenogenic (Ag) (Vanossi Este and Rovati, 1982) and transformer (tra) (Inoue and Hiroyoshi, 1986), have been isolated in which genotypically female offspring (no M and no FD) from mutant mothers are transformed into fertile males and intersexes. Using Ag, a stock could be established in which sex is solely determined by a maternal effect (Dübendorfer et al., 1992).

The sex-determining system of Musca thus appears flexible and open for modification and rapid evolution. A similar plasticity also exists in the nematode Caenorhabditis elegans in which sex is normally determined by a chromosomal signal, the X:A ratio. By introducing mutations in single control genes, J. Hodgkin could synthesize new sex-determining systems that now operate on the basis of allelic differences at single genes of the hierarchy (for reviews see Hodgkin, 1987, 1992). Caenorhabditis and Musca demonstrate the ease with which sex-determining mechanisms can change in certain species. Musca, in addition, offers the unique opportunity to study different mechanisms within a single species and to compare them to those found in another related and well-analyzed species, Drosophila melanogaster. Such a comparison should provide insights into the evolution of sex-deter-mining mechanisms in insects.

This paper is mainly devoted to the questions of how sex is determined in the germ line and what role the soma plays in this process. By transplantation of pole cells, which are the pre-cursors of the germ line, we show that both M and FD, when present in the female germ line, are active during oogenesis as indicated by maternal sex-determining effects on the progeny. The sexual differentiation of the germ line itself, however, is not affected by these factors and is solely dictated by the sex of the host.

Stocks and flies for mass crosses were kept in acrylic glass cages (14×14×19 cm) and fed on water, sucrose and milk-powder. Since the flies prefer to lay eggs in dark places, black cylindrical plastic boxes (ordinary film boxes), half way filled with larval rearing medium and topped with punched lids, were used to collect eggs. Larvae were reared at 25°C and 70% r.h. on a diet prepared according to the following prescription: 900 ml of water, 20 ml ethanol containing 0.67 g Nipagin and 1.33 g Nipasol as fungicides, 480 g wheat bran, 60 g wheat flour, 40 g milk powder, and 20 g brewer’s yeast.

Virgin females were collected within 6 hours of emergence. For single pair crosses, flies were kept in transparent plastic beakers (180 ml volume) and fed on water, sucrose and milk-powder until females were ready for egg laying. Then, the milk powder was replaced by larval rearing medium. Offspring from single crosses were reared at 29°C and 70% relative humidity.

We collected embryos of the syncytial and early cellular blastoderm stage, dechorionized them with 3% sodium hypochlorite at 19–20°C, and transplanted pole cells according to the method of Van Deusen (1976) with the following modifications. The dechorionated embryos were placed on a microscope cover slip without any glue and dried for 2–4 minutes before they were covered with Voltalef 15S fluoro-carbon oil. The whole syringe system and the micropipette were filled with Voltalef 3S (Steinmann-Zwicky et al., 1989). After transplantation of the embryos and removal of the oil, the coverslips with the embryos were transferred to glass dishes containing about 30 ml of stained solid agar (6 g agar, 2 g milk-powder, 500 ml water, some nipagin and food colour). These dishes were kept under O2 for the next 20 hours at room temperature (22–24°C). Hatched larvae were collected and placed on larval medium at 29°C where they could develop to adulthood. Since larvae only develop satisfactorily in large populations, genetically labelled ‘helper’ larvae, preferentially of the genotype used for later outcrossing of the host animals, were added to the cultures of host larvae.

Genes, mutations, symbols and linkage maps are described in Milani (1967) and Hiroyoshi (1977). The autosomal markers used in this study were ar (aristapedia) and cm (carmine) on linkage group II, bwb (brown body), ge (green eyes) and pw (pointed wings) on linkage group III, and ocra (ochre eyes) on linkage group V. In this study, we are using the symbol M for the male determiner and a superscript (X, Y or roman number) to specify the linkage group, e.g. M I I I for the male determiner on linkage group III.

A wild-type strain with standard XX-XY sex determination and four strains with autosomal sex determination were used in this study: (1) cm M I I / +, (2) bwb ge + M I I I / bwb ge pw +, (3) + M I I I / bwb + ; ocra / ocra, (4) bwb ge M I I I / bwb ge M I I I ; FD/+. The strains (1), (2) and (4) were isolated from FD strains obtained from the Osaka University Medical School in Japan, and strain (3) was provided by the Genetics Department of the University of Pavia in Italy.

Strains and crosses designed to produce donor and host embryos, and the genotypes of the test partners, are given in Table 1. Some of these experiments, however, require further explanation:

In all experiments, the transplanted pole cells carried a marker mutation (ocra in series 1 and 3, bwb in series 2 and 4, and ar in series 5) that allowed us to distinguish unambiguously between donor-derived and host-derived offspring. The sexual genotype (M/+, or +/+, or M;FD/+) of the transplanted germ cells could later be identified by the phenotype of the resulting donor-derived offspring.

In series 1 and 3, pole cells from female donors were homozygous for the body color marker bwb, while pole cells from male donors were trans-heterozygous for bwb and M. After crossing adult hosts to bwb test partners (bwb females and bwb M I I I /bwb males, respectively), all donor-derived offspring originating from genetically female germ cells exhibited a bwb phenotype, whereas offspring derived from genetically male germ cells were either bwb+ or bwb.

In series 2, donor embryos were obtained from an FD strain in which females and males were homozygous for M I I I, and females carried, in addition, one copy of FD. After crossing host females to males homozygous for M I I I, and host males to standard females, respectively, donor offspring of both sexes were only expected if female pole cells containing FD had been integrated, whereas inte-grated male pole cells were expected to yield exclusively male offspring.

For unknown reasons, host embryos from the bwb ge M I I I ; FD/+ strain did not survive the transplantation procedure. Recipients carrying M and FD were therefore created by crossing bwb ge M I I I ; FD/+ females to cm M I I /cm+ males (series 3).

To determine the genotypes of the male offspring that developed from M I I I /+ pole cells (series 1c and 3c, Table 2), we crossed such males individually to virgin bwb; ocra females and analyzed the sex and phenotype of the resulting offspring as shown in Fig. 1A.

In series 4 and 5, a cross of standard females homozygous for pw to males homozygous for M I I I was used to produce a unisexual population of male donor embryos that were all trans-heterozygous for M I I I and pw.

The roles of soma and germ line in sexual development of Musca domestica were studied by transplanting pole cells of male and female genotypes into different hosts. The genotypes of donors, hosts and test partners were chosen such that donorderived and host-derived offspring could be distinguished, and the sexual genotype of the transplanted pole cells could later be inferred from the phenotype of the offspring. The complete genotypes and the numbers of injected, surviving, and mosaic hosts are given in Table 1. In the text and in Tables 2 and 3, we will omit the markers and only give the sexual genotypes.

The data, summarized in Table 2, demonstrate that all types of genetically female pole cells, when transplanted into female hosts, form functional eggs (series 1a, 2a, 3a), and all types of genetically male pole cells, when transplanted into male hosts, form functional sperm (series 1b, 2b, 3b). These eggs and sperm give rise to normal offspring.

The results shown in Table 2 also demonstrate that genetically male pole cells heterozygous (series 1c, 3c) or homozygous (series 2c) for M, when transplanted into female embryos, are able to form normal eggs and offspring. Female pole cells with FD (series 2d) or without FD (series 1d, 3d), when transplanted into male embryos, develop into functional sperm that give rise to normal offspring. These results indicate that sexual differentiation of the germ line is a non-autonomous inductive process controlled by the surrounding host tissues.

In series 2c (Table 2), the male pole cells, transplanted into female hosts, were homozygous for M. These pole cells formed eggs which all carried M. When fertilized by sperm from M/+ tester males, all offspring had at least one M, and, as expected, they all developed as males.

The male pole cells transplanted in series 1c and 3c (Table 2), however, were heterozygous for M (see footnotes in Table 2), and upon transplantation into female hosts they should have produced eggs with M and eggs without M. When these host females were mated to M/+ tester males, donor-derived progeny consisted exclusively of males (series 1c and 3c), although females without M were also expected (boxed genotype in Fig. 1A). Since their absence could either be due to sex-specific lethality or to transformation of genetically female offspring (without M) into males, we analyzed the genotypes of 43 randomly chosen donor-derived males by crossing them individually to standard females. Nine of these males (four out of nine in series 1c, and five out of 34 in series 3c) produced only daughters (average of 26 offspring per male), showing that they did not carry M and thus had a female genotype. Since these males have no M, we will call such animals ‘NOM-males’. Of the remaining males, 32 proved to be heterozygous and two homozygous for M.

In series 1c and 3c, the evidence for absence of M in the sons deriving from transplanted pole cells had to be verified by progeny tests. We also designed an experiment in which we could show this more directly, using the mutation pointed wings (pw) to mark the chromosome without M. The gene pw is closely linked to M I I I (<1 map unit; Inoue and Hiroyoshi, 1984), i.e., homozygosity for pw is indicative of the absence of M. The donors of pole cells were trans-heterozygous for M I I I and pw, and the hosts were either standard females (Table 3, series 4a) or M I I I /+; FD/+ females (Table 3, series 5). In these two series, all donor-derived offspring of female hosts were males (except for a few intersexes in series 5). One fourth of these males (144 out of 591) had pointed wings, i.e., a female genotype. Such pw males were mass-mated to standard females and produced exclusively female progeny, demon-strating that they were all NOM-males (Fig. 1A).

In a control experiment (Table 3, series 4b), we transplanted M I I I /pw pole cells into male hosts that were then crossed with pw females. In this case, pw daughters did occur, which shows that the predetermining effect of M is restricted to the female germ line.

In somatic cells, the female-determining activity of FD is epistatic to the male-determining effect of M. The interaction between M and FD in the female germ line was tested by crossing M I I I /pw; FD/+ females to M I I I /pw males (Fig. 1B). In the progeny of 1562 flies, we counted 607 females (carrying FD and M) and 630 males (carrying M as indicated by their pw⁺ phenotype). The remaining 325 flies were females of the pw phenotype (homozygous for pw, and thus devoid of M). No pw/pw males were observed. Presence or absence of FD in the pw females was determined by crossing them individually to homozygous M/M males. Of 11 successful single matings, three produced daughters and sons, indicating that the tested pw females were carriers of FD; and eight females produced only sons (nine or more offspring from each female) and therefore did not carry FD. This demonstrates that presence of FD in the female germ line overrules or prevents the mas-culinizing maternal effect of M so that zygotes with a female genotype now develop as females. Note that the framed genotypes in Fig. 1 of which one (A) develops as male, the other (B) as female, are identical, both lacking M and FD.

When M/+ pole cells were transplanted into M/+;FD/+ female embryos (Table 2, series 3c, and Table 3, series 5), seven out of 31 recipients also produced some intersexual donor-derived offspring. In series 5, where M was marked with pw+, all inter-sexes were pw and thus devoid of M. In both series, however, we also found some intersexes developing from host-derived pole cells. Standard-type female hosts never gave rise to inter-sexual offspring, either host-derived or donor-derived, nor did we ever observe intersexes in our FD-strains.

Microscopical analysis of all intersexual animals revealed a mosaic type with areas of clearly female and clearly male differentiation. We never found structures of intermediate sexual morphology. The intersexes were sterile, but the pw phenotype shows that at least those of series 5 were clearly of a female genotype. It thus appears that the maternal effect of M in donor cells can be partially reversed by the presence of FD in the host cells. Whether this is an effect of FD in somatic cells or in germ cells of the host cannot be decided. At present, we are unable to offer a plausible interpretation for the occurrence of the described intersexes.

Recombination in the male germ line of Musca domestica is virtually absent (Inoue and Hiroyoshi, 1980; Hiroyoshi et al.,1982). The frequency of recombination in males between bwb and M I I I is only 0.156% (Hiroyoshi et al., 1982), whereas in FD females, the distance is some 55 map units (Inoue and Hiroyoshi, 1984). In our series 1 and 3 (Table 1), the transplanted pole cells were trans-heterozygous for M and bwb. We analyzed 27 animals of the all-male offspring originating from M/bwb pole cells transplanted into female hosts (Table 2, series 1c and 3c) and found that 10 of them were carriers of recom-binant chromosomes. The recombination frequency of 37% corresponds to a genetic distance of some 50 to 60 map units. This result demonstrates that germ cells with a male genotype, when forced to undergo oogenesis, exhibit the same frequency of recombination as germ cells with a female genotype.

In multicellular organisms, the sex of the germ cells and the sex of the soma must be matched so that eggs develop in ovaries and sperm in testes. This is most easily achieved if the germ cells non-autonomously adopt the sexual pathway taken by the surrounding somatic cells.

Our results obtained with genetic mosaics show that, in Musca domestica, sexual differentiation of germ cells is dictated by the phenotypic sex of the surrounding soma and is completely independent of the genotypic sex of the germ cells. Pole cells without the male-determiner M, and even when carrying the dominant female-determining factor FD, form functional sperm in a male host. Conversely, pole cells with one or even two M factors form functional eggs in a female host. Somatic induction of germ cell development also operates in other Diptera, such as the blowfly, Chrysomya rufifacies (Ullerich, 1984).

In Drosophila, another dipteran insect, autonomous and inductive signals cooperate to control the sexual differentiation of germ cells. When autonomous sex-determining signals act in the germ cells, the genetic hierarchies controlling the sex of the soma and the sex of the germ line must be coupled to coor-dinate the two regulatory systems. In Drosophila, this is accomplished by the X:A ratio (Van Deusen, 1976; Schüpbach, 1985; Steinmann-Zwicky et al., 1989; Steinmann-Zwicky, 1993). XY and X0 germ cells, transplanted into female hosts, still enter spermatogenesis, thus displaying autonomous sexual differentiation. XX and XXY germ cells, on the other hand, non-autonomously enter spermatogenesis in male hosts. The fact that XX cells enter, but do not complete, spermatogenesis in the perfect environment of a male host reveals a second autonomous component that acts later during the process of sperm development, but is not involved in sex determination.

The failure of XX cells to undergo complete spermatogenesis may be due to strongly dimorphic sex chromosomes and dosage compensation, or to X-inactivation: the single X chromosome of males must be inactivated during a critical phase of spermatogenesis (Lifschytz and Lindsley, 1972) and inactivation of both X chromosomes may not occur. Germ cells of Musca, however, do not encounter any problems of unequal gene dosage between the sexes, since X and Y chromosomes, although cytologically distinguishable, are functionally equiv-alent and interchangeable (Rubini and Franco, 1965; Milani, 1967).

In summary, it appears that somatic induction is the ancestral mechanism for sex determination of the germ line in Diptera. It is operative in more primitive species, such as Musca and Chrysomya, whereas in higher Diptera, such as Drosophila, an autonomous component has evolved in addition to somatic induction.

In Drosophila, the sex-determining effect on the germ line is clearly somatic, as shown by transplantation of pole cells into agametic hosts (Steinmann-Zwicky et al., 1989; Steinmann-Zwicky, 1994a). In Musca, on the other hand, the donor pole cells had to be deposited among those of the host and remained in close contact with them. It is thus possible that the inductive signal is given by the host’s germ cells rather than by its somatic cells. Two observations, however, render this hypoth-esis untenable: (i) in NOM-males (framed in Fig. 1A), all germ line cells are genetically female, embedded in phenotypically male somatic tissue, and these germ cells differentiate in the male mode. (ii) The intersexes produced in our Ag-stock are mosaic for male and female tissues, and the mosaicism occasionally affects the gonads. In these cases, gametes of both sexes are found in the same gonad whereby spermatogenic cells are surrounded by testicular tissue and oogenic cells by ovarian tissue (A. Dübendorfer, D. Hilfiker-Kleiner, R. Schmidt and R. Nöthiger, unpublished data). The observations that completely sex-transformed flies are fertile and that inter-sexes may contain differentiated germ cells of both sexes, matched to the surrounding somatic sex of the gonad, suggest that somatic cells have an inductive, sex-determining effect on the developing cells of the germ line. The intersexes further-more show that the inductive influence results from short-range effects, most probably emanating from the somatic cells of the gonad.

One wonders why sexual differentiation is a cell-autonomous process in the soma of Musca and Drosophila, but not in the germ line. A plausible hypothesis is that germ cells possess receptors which are capable of receiving and trans-ducing inductive signals from the somatic parts of the gonad. Such receptors would have to be absent from somatic cells.

Neither M nor FD, when present only in the germ line, has a direct effect on sexual differentiation of germ cells. Neverthe-less, the two factors are apparently expressed during oogenesis. This follows from the maternal effects which M and FD have on the sex of the resulting zygotes (Fig. 1). The masculinizing effect of a maternal M on genetically female offspring is con-sistent with our previous observation that a zygotic M acts early and transiently to repress the F gene permanently, thus dictating male development (Hilfiker-Kleiner et al., 1993). The finding, however, that the presence, i.e. activity, of FD in the germ line of M/+; FD/+ females can offset the masculinizing maternal effect of M so that +; + zygotes now can develop as females, leads to new insights into the regulation and expression of F.

We propose that F is normally expressed during oogenesis and that this maternal product is a prerequisite for the zygotic F gene to become active. Unless a zygotic M prevents the activity of F in offspring of standard or M/+; FD/+ mothers, the zygotes will develop as females. However, when M/+ pole cells, transplanted into female hosts, undergo oogenesis, M prevents the synthesis or activity of maternal F product. Offspring from such germ cells will thus develop as NOM males, despite their female genotype and the presence of two F+ alleles which, however, cannot become active.

The model makes two predictions for a null mutation of F: (i) Since F activity is zygotically required for female development, F—/F— embryos should develop as males, and mutant cell clones should form male structures. (ii) Unless F activity is required for oogenesis, F—/F— pole cells, when transplanted into female hosts, should produce NOM-male offspring, but female development should result when FD is introduced via sperm into the zygote. If our predictions are fulfilled, we would conclude that it is not the presence of maternal M product that transforms genetically female zygotes into NOM males, but rather the absence of maternal F product.

Our view of the roles of M and F during oogenesis also suggests a new interpretation of the maternal effect mutations, Ag and tra, which both transform genetically female zygotes into NOM males and intersexes. Rather than identifying two more sex-determining genes, Ag and tra could be hypomorphic alleles of M and F, respectively. This hypothesis is supported by the fact that in one of the Ag strains, an M factor sponta-neously appeared which genetically mapped to the same position as Ag (Vanossi Este et al., 1974; Rovati et al., 1983). Furthermore, the experimentally produced maternal effect of M mimics the effect of Ag which is, like the maternal effect of M, also overridden by FD (Rovati and Vanossi Este, 1978). The second mutation, tra, characterized by a strong maternal effect that is enhanced by zygotic homozygosity, could not be separated from FD on chromosome IV (Inoue and Hiroyoshi, 1986). If, as we propose, expression of F is required mater-nally and zygotically to implement female development, tra could be a special allele of F that has lost most of its activity in the germ line, but has retained sufficient zygotic activity. Similarly, Ag could represent a mutant M that has no or insufficient zygotic activity, but still acts in the germ line to prevent the function of F.

Using the mutation Ag, we succeeded in creating a stock of Musca domestica in which sex is solely determined by a maternal effect. In this new synthetic system, females are either Ag/+ and produce only sons, or +/+ and produce only daughters. Males are either Ag/+ or +/+, and have no influence on the sex of their offspring (see Fig. 3 in Dübendorfer et al., 1992). The stock is self-maintaining and runs without selection, producing some 70% females and 30% males, as well as a few intersexes of the mosaic type.

Sex determination by maternal effect is the natural way in the dipteran insects Sciara coprophila (Crouse, 1960, 1965) and Chrysomya rufifacies. In Chrysomya, two types of females exist, F ′/f and f/f, and males are f/f . F ′/f females produce only daughters (which are either F ′/f or f/ f ); f/f females produce only sons (which are f/ f ). F ′ thus functions as a dominant, maternally acting female determiner (Ullerich, 1984).

In Drosophila, all aspects of sexual differentiation are controlled by the gene Sex-lethal (Sxl) which is active in females and inactive in males (for reviews see Belote, 1992; Cline, 1993; Parkhurst and Meneely, 1994). Sxl stands at the top of three short regulatory cascades which control sexual differentiation in the soma, dosage compensation, and sexual differentiation in the germ line. Downstream of Sxl, a small number of subordinate control genes act in somatic cells to control sexual differentiation (tra, tra2, dsx, ix) and dosage compensation (msl-1, msl-2, msl-3, mle). Subordinate control genes probably also exist to govern sexual differentiation in the germ line, but they are not yet well defined (Pauli and Mahowald, 1990; Steinmann-Zwicky, 1992, 1993, 1994b).

It was previously proposed (Nöthiger and Steinmann-Zwicky, 1985; Inoue and Hiroyoshi, 1986; Dübendorfer et al., 1992; Hilfiker-Kleiner et al., 1993) that F of Musca might functionally correspond to Sxl or tra of Drosophila, and FD to a constitutive mutation of Sxl or tra. There are, however, dis-crepancies that speak against a simple correspondence. Although Sxl is expressed in the female germ line of Drosophila, which is a conspicuous parallel to the activity of F in Musca, no SXL protein is detectable in freshly deposited Drosophila eggs or embryos before the blastoderm stage (Bopp et al., 1993). A sex-determining maternal effect of Sxl is thus very unlikely, and for tra it has effectively been ruled out (Marsh and Wieschaus, 1978). Furthermore, cell clones produced by mitotic recombination show that an active F gene, such as FD, is apparently not able to initiate an autoregulatory loop in a wild-type allele, at least not as long as M is present in the same cell (Hilfiker-Kleiner et al., 1993).

Whereas it is useful and stimulating to look for parallels and differences, we will have to await more genetic, developmental and eventually molecular data to see how related the sex-determining mechanisms of Drosophila and Musca are. It is formally conceivable to modify the Drosophila sex-determin-ing system by introducing mutations at different levels of the genetic hierarchy. For example, a partial loss-of-function mutation exists in the gene daughterless (da): offspring from da/da mothers fail to activate their Sxl-gene, and so only sons are produced (Cline, 1978). One could now imagine a system in which all animals are XX, with two types of females occurring, namely da/da+ and da/da. When crossed to X/X; da/da males, the former females would produce only daughters, the latter only sons.This is exactly what is found in Chrysomya rufifacies, and in fact, a probe of the Drosophila gene da hybridizes in the region of the chromosome where F’ is located (Clausen and Ullerich, 1990). In Drosophila, however, the system does not work, (i) because Sxl also controls dosage compensation so that XX zygotes that do not activate Sxl die, and (ii) because XX germ cells fail to complete spermatogenesis so that transformed XX males are sterile.

Similarly, one could devise in Drosophila a system in which both sexes are XX, with females being tra+/tra—, and males being tra—/tra—. The mutation tra— was discovered by Sturte-vant (1945); it transforms XX animals into males, which, however, are sterile. Whereas this system has no problems with dosage compensation, it fails because the autonomous component of the germ line prevents XX germ cells from becoming functional sperm, even when a Y chromosome is present (Sturtevant, 1945; Nöthiger et al., 1989).

The two examples show that dosage compensation and autonomous genetic control in the germ line have led Drosophila into an evolutionary dead-lock. In this species, modifications of the sex-determining mechanism by mutations in single genes of the genetic control system are doomed to failure and result either in lethality or sterility.

This is in gross contrast to Musca domestica where we find a variety of sex-determining mechanisms coexisting within one species. The variability is truly impressive and comprises an XY mechanism with a male-determining Y, mechanisms with a single factor M located on different chromosomes, or with FD and female heterogamety, or with a maternal effect (Ag). It should, however, be emphasized that these superficially very different mechanisms are simple modifications of a basic strategy in which a factor, M, regulates a key gene, F. Mutations in either of the two elements can account for the diversity that we encounter today in Musca domestica. Musca, in which there are no problems with dosage compensation and in which the sexual differentiation of the germ line is non-autonomous, has an open future with respect to its sex-deter-mining mechanisms.

Our work was supported by the Swiss National Science Foundation grant Nr. 31-27953.89, the Sandoz-Stiftung zur Förderung der medizinisch-biologischen Wissenschaften, and the Julius Klaus-Stiftung. Maria-Gabriella Franco and Pier Giorgio Rubini of the University of Pavia (Italy) and Toshiki Hiroyoshi of the University of Osaka (Japan) kindly provided fly stocks. We are grateful to Jennie Schöpfer-Bons, Pia Meier and Raymond Grunder for technical assistance, and to Mariana Wolfner, Monica Steinmann-Zwicky, Daniel Bopp, Pier Giorgio Rubini and Regula Schmidt for fruitful discussions and critical comments on the manuscript.