Sex determination in Drosophila: sis-b, a major numerator element of the X:A ratio in the soma, does not contribute to the X:A ratio in the germ line

The sex of Drosophila germ cells is determined by a mech-anism that is different from that acting in somatic cells (reviewed in Pauli and Mahowald, 1990; Steinmann-Zwicky, 1992a,b). XX cells enter the male pathway when developing in a male host animal. This shows that their sex is determined by induction. XY and XO cells, in contrast, form spermatocytes even when developing in a host ovary. They have an autonomous information for maleness and they do not respond to inductive signals (Steinmann-Zwicky et al., 1989). The sex of germ cells is thus deter-mined by cell-autonomous and inductive signals. The sex of somatic cells, however, is determined solely by a cell-autonomous signal called the X:A ratio, which arises from relating the number of X chromosomes to the number of sets of autosomes (reviewed in Baker, 1989; Steinmann-Zwicky et al., 1990; Belote, 1992).

Both somatic tissue and germ cells require Sxl activity to enter the female pathway (Cline, 1978; Sánchez and Nöthiger, 1982; Schüpbach, 1985; Steinmann-Zwicky et al., 1989). In the soma, Sxl is regulated at the level of tran-scription (Torres and Sánchez, 1991; Keyes et al., 1992) and alternative splicing (Bell et al., 1988). Early female-specific transcripts are found in embryos with an X:A ratio of 1. Two X-chromosomal elements of the X:A ratio, sis -terless-a (sis-a) and sisterless-b (sis-b) induce these early Sxl products together with the maternally provided tran-scription factor daughterless (da) and maybe other gene products. Later, the products of fl(2)d, liz (also called fs(1)1621 and snf) and Sxl itself are required for maintain-ing Sxl active, probably for female-specific splicing of the Sxl pre-mRNA. XX animals that lack Sxl activity, or XX animals that lack sis-a or sis-b, die because both X chro-mosomes are transcribed at a high level, which is typical of the single X chromosome of males (Lucchesi and Skrip-sky, 1981; Cline, 1988; Steinmann-Zwicky, 1988; Granadino et al., 1990; Bell et al., 1991; Torres and Sánchez, 1991; Keyes et al., 1992; reviewed in Belote, 1992).

In germ cells, the products of fl(2)d and liz are also required for Sxl activity (Steinmann-Zwicky, 1988; Granadino et al., 1992; Salz, 1992). Little, however, is known about other genes regulating Sxl in the germ line. XX germ cells carrying the mutation Sxl^M1, which consti-tutively expresses functions of the gene Sex-lethal, can become oogenic even when developing in a host testis (Steinmann-Zwicky et al., 1989). Sxl^M1 therefore provides XX germ cells with an autonomous information for female-ness and renders them insensitive to induction. This shows that the somatic inductive signal that determines the sex of XX germ cells exerts its action by regulating the gene Sxl.

Due to analogies to the situation in the soma, the cell-autonomous signal that renders XX germ cells sensitive to induction, while leaving XY and XO cells insensitive has been called ‘X:A ratio’ (reviewed in Steinmann-Zwicky, 1992a,b). Here I tested whether one of the elements form-ing the X:A ratio in somatic cells also participates in build-ing the X:A ratio in germ cells. For this, I transplanted XX germ cells lacking sis-b function into host females. Such germ cells formed functional eggs, which shows that sis-b is not required for oogenesis. To test whether Sxl expression is sufficient to drive XY cells into the female pathway, I transplanted XY cells carrying the constitutive mutation Sxl^M4 into host animals of either sex. XY germ cells car-rying this mutation did not become oogenic even when developing in ovaries.

Pole cells were transplanted as described in Van Deusen (1976) and Steinmann-Zwicky et al. (1989). Agametic host embryos with-out germ cells were derived from mothers homozygous for osk³⁰¹ kept at 18°C. Adult host flies were crossed to test partners. Ster-ile flies were dissected and their gonads were analysed with a microsope. Criteria used to identify the sex of germ cells are listed in Steinmann-Zwicky et al. (1989).

The stock used to obtain XX embryos lacking sis-b function was: sc^10-1f^36a/FM6/y²Y 67 g. To test the genotype of transplanted germ cells, adult host flies were individually crossed to y w f part-ners. Between 50 and 100 progeny from each fertile fly were scored.

To obtain donor XY embryos carrying Sxl^M4, females of geno-type cm Sxl^M4/FM7 were crossed to T(X;Y)22-3, y v f . Y^LRsp^sB^S/Y; E(SD)Rspⁱbw/SD-ARM VO17, lt males. These males carry mutations causing a segregation distortion so that they only trans-mit their Y chromosome (Walker et al., 1989).

To test the genotype of transplanted germ cells, I crossed each host male to three different types of females: (a) cm Sxl^M4/FM7, (b) y cm Sxl^M1/FM6, (c) cn bw. Host females were crossed to males of genotype cn bw. Mutations and balancer chromosomes are described in Lindsley and Zimm (1992).

The X-chromosomal element of the X:A ratio that has best been analysed is sis-b (Cline, 1988; Torres and Sánchez, 1989; Erickson and Cline, 1991). The sis-b function is pro-vided by one of the transcripts of the achaete-scute com-plex (AS-C), T4. The allele sc^10-1 lacks all sis-b activity (Torres and Sánchez, 1989) since it contains a point muta-tion that places a stop codon within T4 (Villares and Cabr-era, 1987). To test whether sis-b activity is required in the germ line, I investigated the developmental capacities of XX germ cells homozygous for sc^10-1. I transplanted pole cells from progeny of sc^10-1/FM6 females crossed to sc^10-1/Y males carrying a sc⁺ duplication on their Y chro-mosome. From here on, the chromosome carrying sc^10-1 will be called sis-b.

Table 1 shows the results of this experiment. 22 host females formed eggs and had therefore integrated XX germ cells. 4 of them did not lay their eggs, such that the geno-type of these could not be identified. 7 females had prog-eny some of which carried the balancer chromosome FM6, showing that they had integrated germ cells of genotype sis-b/FM6. 11 fertile females, however, transmitted only the chromosome carrying sis-b to their progeny. These females had integrated germ cells that were homozygous for sis-b. The results show that germ cells do not require the sis-b function to enter or to complete oogenesis. Of the remaining females that had no progeny, 23 contained spermatocytes and had therefore integrated XY germ cells, 27 had empty ovaries, and 2 died during the test crosses.

24 host males produced sperm showing that they had integrated XY germ cells. 4 had no progeny, 4 transmitted the chromosome FM6 and 16 transmitted the chromosome carrying sis-b. 9 sterile males had spermatocytes in their testes and had therefore integrated XX cells. In one case, these displayed the crystals that are specifically formed by spermatogenic germ cells lacking a Y chromosome (Hardy et al., 1984; Livak, 1984; Steinmann-Zwicky et al., 1989). 26 males had empty gonads.

The fertile female and male hosts had integrated germ cells homozygous or hemizygous for sis-b more often than germ cells heterozygous or hemizygous for FM6. In the case of male hosts, this is especially striking. Either het-erozygous females must transmit their sis-b chromosome more often than their FM6 chromosome, or XY embryos carrying FM6 might be selected against in the transplantation experiments, either because they develop at a different speed than their sis-b carrying brothers, or because many of them die before blastoderm. No biased transmission of chromosomes was observed in other transplantation experiments involving FM6 (Steinmann-Zwicky et al., 1989). When counting progeny from the experimental cross that were allowed to survive to adulthood, a small excess of males carrying the sis-b chromosome was observed. In one experiment, I counted 84 sis-b/Y males, 54 FM6/Y males and 70 sis-b/FM6 females.

Sxl^M1 and Sxl^M4 are mutations that express female-specific Sxl functions even in the absence of factors normally required for Sxl expression. XX animals lacking da or liz product are rescued by both alleles (Cline, 1978; Maine et al., 1985; Steinmann-Zwicky, 1988; Salz, 1992). Since XX germ cells that lack liz function are also rescued (Stein-mann-Zwicky 1988; Salz, 1992), we know that XX germ cells carrying Sxl^M1 or Sxl^M4 express Sxl functions without the requirement of liz.

XY animals carrying Sxl^M1 show no Sxl expression in early embryogenesis (Gergen, 1987), but they die as larvae and their X chromosome is only half as wide as that of con-trol larvae, which probably reflects its hypoactivity (Luc-chesi and Skripsky, 1981). In some cases, adult tissue shows female-specific traits (Cline, 1979). XY germ cells carry-ing Sxl^M1 are spermatogenic (Cline, 1983; Steinmann-Zwicky et al., 1989). This either means that Sxl is not expressed in these germ cells or that expressing Sxl is not sufficient for XY germ cells to become oogenic.

Sxl^M1 is still at least partially regulated by elements of the X:A ratio, as it is possible to make a stock in which females are liz Sxl^M1/liz Sxl^M1 and males are liz Sxl^M1/Y (Steinmann-Zwicky, 1988). The observation that one Sxl^M1 allele cannot fully rescue females mutant for liz or sc^3-1 in the absence of Sxl⁺ or a second Sxl^M1 allele, also suggests that Sxl^M1 does not express Sxl functions quite constitu-tively (Steinmann-Zwicky, 1988; Torres and Sánchez, 1989).

Sxl^M4 seems to depend on factors less than Sxl^M1. XY animals carrying Sxl^M4 die before hatching. Males of geno-type liz Sxl^M4/Y also die (Salz, 1992). Since the chromo-some carrying Sxl^M4 carries no lethal mutation (see below), it can be concluded that these animals are not rescued by introducing a liz mutation. They therefore seem to express Sxl functions independently of the X:A ratio and liz. There-fore, I chose to test whether XY germ cells carrying Sxl^M4 can become oogenic.

Pole cells were taken from embryos of genotype Sxl^M4/Y or FM7/Y (see Materials and Methods) and transplanted into XX or XY embryos that had no germ cells of their own. Table 2 shows the results of this experiment. Of 20 sur-viving males, 15 were fertile. Among these, 7 had inte-grated FM7/Y cells and 8 had germ cells of genotype Sxl^M4/Y. One male differentiated only spermatocytes and could therefore have integrated XX cells, 3 males had empty testes and one animal died before it could be tested. 9 host females had ovaries filled with spermatocytes, which shows that they had integrated XY germ cells, 6 had empty gonads and one died during the tests. One female produced eggs and was fertile, but her progeny displayed that she had integrated germ cells carrying both chromosomes Sxl^M4 and FM7. The donor embryo must therefore have arisen by maternal non-disjunction and it must have had two X chro-mosomes and one Y chromosome.

The results show that XY germ cells carrying Sxl^M4 do not become oogenic even when developing in host ovaries. The nine females containing spermatocytes had probably integrated germ cells of genotype either Sxl^M4/Y or FM7/Y. The results obtained with males show that germ cells from Sxl^M4/Y embryos were transplanted equally often as germ cells from FM7/Y embryos. As I did not observe two dif-ferent classes of host females with spermatocytes, there seems to be no difference in developmental performance between Sxl^M4/Y and FM7/Y germ cells.

To test the genotype of germ cells of fertile males, I crossed them to several females carrying various markers. One type of female was of genotype Sxl^M4/FM7. When males carrying Sxl^M4/Y germ cells were crossed to these females, female progeny arose that were homozygous for the Sxl^M4 chromosome. Although these were less viable than their Sxl^M4/FM7 sisters (on average I counted about 20 % of the expected Sxl^M4/Sxl^M4 females), they showed that the chromosome carrying Sxl^M4 has acquired a muta-tion affecting the eye similar to lz, but no lethal mutation, which means that males carrying this chromosome die because of Sxl^M4. This result is important for any analysis that tests the viability of animals carrying the Sxl^M4 chro-mosome.

Three X-chromosomal elements, sis-a, sis-b and runt, are known to regulate the expression of Sxl in somatic cells in a dose-dependent manner (Cline, 1988; Torres and Sánchez, 1989, 1992; Duffy and Gergen, 1991). They are therefore called numerator elements of the X:A ratio. The products of sis-a and sis-b are probably transcriptional activators that control the expression of Sxl (Torres and Sánchez, 1991; Keyes et al., 1992). For the third element, the segmenta-tion gene, runt, the situation is different. XX embryos that lack runt form no Sxl product in the middle region where runt is normally expressed. They, however, express Sxl at both terminal regions, anterior and posterior (Duffy and Gergen, 1991; Torres and Sánchez, 1992). This shows that runt product is required in some regions of the embryo but not in others to activate Sxl. The runt function might there-fore participate indirectly in the regulation of Sxl, maybe by repressing a segmentation gene whose product, when abnormally expressed, could interfere with proper activa-tion of Sxl.

Because of their direct involvement in the activation of Sxl, the elements sis-a and sis-b seem better suited than runt to test whether genes that regulate Sxl in the soma, also regulate Sxl in the germ line. The only sis-a mutation available is known to be a hypomorphic allele, and homozy-gous sis-a females can occasionally survive. Germ cells that became homozygous for sis-a as a consequence of mitotic recombination induced after 48 hours of development were oogenic (Cline, 1986). This could mean that sis-a function is required early for oogenesis, but not after 48 h, for exam-ple because the state of activity of Sxl is already irreversibly fixed at that time, which is the case in somatic cells (Sánchez and Nöthiger, 1983). Alternatively, this could mean that this hypomorphic allele of sis-a provides enough gene function for oogenesis. The third possibility is that sis-a function is not required at all in the female germ line. Transplanting germ cells would not enable us to distinguish between the latter two alternatives.

I therefore decided to test whether sis-b, for which a null allele is available, is required for oogenesis. My results show that XX germ cells lacking sis-b produce functional eggs when allowed to develop in a host female. Thus, germ cells, unlike somatic cells, do not require sis-b product to enter the female pathway. XX flies lacking sis-b die as embryos because they cannot activate their Sxl gene (Cline, 1988; Torres and Sánchez, 1989, 1991). From previous work, we know that oogenic germ cells require Sxl (Schüp-bach, 1985; Steinmann-Zwicky et al., 1989). Thus, since XX germ cells lacking sis-b become oogenic, they must express Sxl without requiring sis-b. The transcription factor da had previously been shown not to be required for ooge-nesis (Cronmiller and Cline, 1987). This already suggested that, in germ cells, Sxl is activated by a mechanism that is different from that acting in the soma. In somatic cells, the products of da and sis-b, which are both helix-loop-helix (HLH) proteins, associate and the heterodimer probably activates Sxl (Dambly-Chaudière et al., 1988; Murre et al., 1989; Van Daren et al., 1991). The autosomal gene da, however, is not a numerator element of the X:A ratio. The da product is provided maternally to all eggs and plays no discriminative role in the process of activating Sxl in females, but not in males. The finding that da is not required in the germ line does not exclude that sis-b activates Sxl in the germ line without the help of da product. My results show that this is not so.

Although the X:A ratio provides a sex-determining signal in the germ cells, the elements forming this signal are dif-ferent from those forming the X:A ratio in somatic cells. The target of the somatic X:A signal is Sxl. We can now ask whether the target of the germ line X:A signal is also Sxl. If both the inductive signal that determines the sex of XX germ cells and the autonomous germ line X:A signal that makes germ cells responsive to induction regulate the gene Sxl, expressing Sxl should be sufficient to direct XY germ cells into oogenesis. We already knew that Sxl^M1 does not feminize XY germ cells (Cline, 1983; Steinmann-Zwicky et al., 1989). In this paper, I show that even Sxl^M4 which is known to constitutively express Sxl functions in somatic XX and XY cells and in germ cells carrying two X chromosomes, does not drive XY germ cells into ooge-nesis, not even when developing in an ovary. This leaves us with two alternatives. Either Sxl^M4 does not express Sxl functions in XY germ cells, or Sxl expression is not suffi-cient for germ cells to enter the female pathway.

The sex of XX germ cells is determined by an inductive signal that emanates from somatic cells. This signal regu-lates Sxl by either of two mechanisms. (1) XX germ cells might enter the male pathway unless they are feminized by an inductive signal that leads to the activation of Sxl. (2) XX germ cells might enter the female pathway unless they are masculinized by an inductive signal that leads to the repression of Sxl.

If XX germ cells are female in the absence of an induc-tive signal, it follows that the X:A ratio must confer female-ness on them by activating Sxl (Fig. 1A). If XX germ cells are male in the absence of an inductive signal, a female X:A ratio is necessary but not sufficient to drive germ cells into oogenesis, either because elements of the X:A ratio provide factors that help to activate Sxl together with the inductive signal (Fig. 1B), or because the X:A signal con-trols other genes that are required for oogenesis in parallel to Sxl activity (Fig. 1C). This last possibility was largely ignored in previous publications. If Sxl expression is not sufficient to drive XY germ cells into oogenesis, as suggested by the finding that Sxl^M4 does not feminize XY germ cells, this possibility becomes the one that is most likely. However, unless it can be shown that a functional Sxl product is present in Sxl^M4/Y germ cells whenever this product is required for germ cells to enter and maintain the female pathway, we cannot be sure that this is the right interpretation.

I thank Rolf Nöthiger for critically reading the manuscript, Eva Niederer for excellent technical assistance and Tatjana Kabat for patient typing. The project was supported by a grant of the Swiss National Science Foundation and by the Kanton Zurich.