The sisterless-b function of the Drosophila gene scute is restricted to the stage when the X:A ratio determines the activity of Sex-lethal

In Drosophila melanogaster 2X;2A individuals (X, X chromosome; A, haploid autosomal set) are female and XY;2A individuals are male. The Y chromosome does not play any role in sex determination. This process occurs by the sex-specific expression of a group of genes hierarchically organized (reviewed in N öthiger and Steinmann-Zwicky, 1985; Baker, 1989; Hodgkin, 1989; Steinmann-Zwicky et al. 1990). The regulation of these genes takes place during all development and in adult life by alternative splicing of their transcripts (Bell et al. 1988; Boggs et al. 1987, Burtis and Baker, 1989). Dosage compensation (hypertranscription of the male X chromosome) is a process linked to sex determination (reviewed in Lucchesi and Manning, 1987). By means of this process, the products of the X-linked compensated genes are present at the same levels in females and males. Both sex determination and dosage compensation are triggered by a common initial signal: the ratio between the number of X chromosomes and the number of autosomal sets in each cell (X:A) (Bridges, 1925; Maroni and Plaut, 1973). The X:A ratio determines the state of activity of the gene Sex-lethal (Sxl): in females Sxl will be ON, while in males Sxl will be OFF (Cline, 1978). Activation of Sxl also requires the maternal daughterless (da) product (Cline, 1978). Once the state of activity of Sxl is defined, which occurs around the blastoderm stage, the X:A ratio is no longer relevant, and both sex determination and dosage compensation come under the control of Sxl (Sánchez and N öthiger, 1983; Bachiller and Sánchez, 1991). The capacity of this gene to function as a stable genetic ‘switch’ is due to a positive autoregulatory function of the Sxl product (Cline, 1984; Bell et al. 1991). The regulation of the gene Sxl throughout most of development and in adult life occurs by alternative splicing of its primary transcripts: the male transcripts give rise to inactive truncated proteins due to the presence of a translation stop codon in an additional exon (Bell et al. 1988); in contrast, this exon is spliced out from the female transcripts and, consequently, functional Sxl proteins are produced. The gene fl(2)d is required for female-specific splicing of Sxl transcripts (Granadino et al. 1990). Sxl controls both sex determination and dosage compensation by regulating two independent sets of genes (Lucchesi and Skripsky, 1981). Failures in dosage compensation, producing hypertranscription in females or hypotranscription in males, are lethal (Cline, 1978; Lucchesi and Skripsky, 1981). Sex determination is not a vital process, and so, failures in sex determination lead to viable sex transformed phenotypes (reviewed in Baker and Belote, 1983 and in N öthiger and Steinmann-Zwicky, 1985). For these reasons, misexpression of Sxl can produce sex-specific lethality and/or sexual transformation to either males or females.

The genetic basis of the X:A signal is unknown. It is thought that it results from the interaction between X-linked elements, ‘numerator elements’, and autosomal elements, ‘denominator elements’. So far, two numerator elements of this signal have been identified, sisterless-a (sis-a) (Cline, 1986) and a region of the achaete –scute complex (AS –C) that has been named sisterless-b (sis-b) (Cline, 1988) and which corresponds to the gene scute (sc) (Torres and Sánchez, 1989). Recently, two sets of experiments have confirmed that the gene sc is responsible for the sis-b function. On the one hand, it has been shown that misexpression of sc induces ectopic Sxl expression and male-specific lethality (Parkhurst et al. 1990). On the other hand, transformation experiments with wild-type and modified sc transgenes have shown that both the rescue of females lacking sis-b function and the lethality of males due to inappropriate Sxl expression, results from the presence of wild-type sc transgenes (Erickson and Cline, 1991). Thus, the gene sc has a dual function: the sc function, which is involved in neurogenesis (reviewed in Ghysen and Dambly-Chaudière, 1988), and the sis-b function which is involved in generating the X:A signal (Torres and Sánchez, 1989; Parkhurst et al. 1990; Erickson and Cline, 1991; this report).

The sc gene shows two patterns of embryonic expression: an expression during syncytial blastoderm leading to an homogeneous distribution of the sc transcript, and a postblastodermal expression restricted to the regions from which the precursors of the nervous system will differentiate (Romani et al. 1987; Cabrera et al. 1987). The temporal aspects of the regulation of Sxl by the X:A signal were not explored in the experiments reported by Parkhurst et al. (1990) and Erickson and Cline (1991). Here, by means of a hsp70 –sc chimeric gene (HSSC-3) (Rodríguez et al. 1990), we have administered pulses of sc expression to determine the temporal specificity of its sis-b function. Our results show that the ectopic sc product is able to activate Sxl only at the syncytial blastoderm stage. We also present evidence of the transcriptional control of Sxl by the elements of the X:A signal.

Flies were raised on standard Drosophila medium under noncrowded conditions. The temperature of cultures was 25°C unless otherwise stated. For full description of markers and chromosomes used, see Lindsley and Zimm (1985, 1987, 1990). For description of the hsp70 –sc chimeric gene construction and transformation, see Rodriguez et al. (1990).

Flies were macerated in KOH 10 % at 50°C and the cuticle was mounted in Faure’s solution.

Embryos were hybridized with the h1 genomic fragment (Bell et al. 1988) that detects all the Sxl transcripts (Saiz et al. 1989). The non-radioactive labeling method described by Tautz and Pfeifle (1989) was used, with an additional fixation of the embryos in glutaraldehyde 0.5 % for two minutes at 4°C. To quantify the staining, light absorption in the 560 –620 nm range was monitored in the cellular region at the margin of the embryos, with a Vickers Microdensitometer M60A. For each preparation, values were corrected with the mean obtained for embryos in stages 7 –11 (Campos-Ortega and Hartenstein, 1985) in which an unimodal distribution was found.

Hw^49cR5/FM7c, y^31dw^asn^x2v^Ofg⁴B × Hw^49cK5/v⁺Yy⁺

sc^3-1w f^36a/FM7c, y^31dw^asn^x2 v^ofg⁴B X sc^10-1f^36a/y²Y67g

sc^10-1f^36a/FM6, y^31dsc⁸dm B × sc^10-1f^36a/y²Y67g

sc^10-1f^36a/FM6, y^31dsc⁸dm B × sc^3-1w cm Sxl^f1ct⁶f^36a/ Y

sc^10-1Sxl^M1/FM6, y^31dsc⁸dm B × sc^10-1f^36a/y²Y67g

sc^10-1f^36a/FM6, y^31dsc⁸dm B × Df(1)N71,sis-a^-/v⁺Yy⁺

Df(1)N71, sis-a^-/FM7c, y^31dw^asn^x2v^ofg⁴B × sc^3-1w cm

Sxl^f1ct⁶f^36a/Y

Df(1)N71, sis-a^-/FM7c, y^31dw^asn^x2v^ofg⁴B X y sis-a/Y

Females defective for Sxl show an embryonic lethal phase (our unpublished data). Thus, the fact that sc is responsible for the sis-b function, involved in Sxl activation, predicts an embryonic lethal phase for females deficient for sc. This prediction is fulfilled. Females carrying different defective sc alleles show different degrees of embryonic lethality (Fig. 1). This lethality is more severe in females containing only a single dose of Sxl⁺. Viability is recovered when one of the Sxl copies is replaced by Sxl^M1, a mutation that expresses the female-specific functions of Sxl largely independently of the X:A ratio (Cline, 1978). Females homozygous for sc^10-1, carrying Sxl^M1, survive to the adult stage. Significantly, these females show the same extreme sc phenotype as the sc^10-1 males. This clearly demonstrates the dual function of sc and the independence of its two functions: Sxl^M1 is able to rescue the sis-b function but not the proneural function of sc. Finally, as a consequence of the involvement of sis-a in the X:A signal, females mutant for sis-a, or doubly heterozygous for sis-a and sc mutations, also show embryonic lethality (Fig. 1).

There are female-lethal transheterozygous synergistic interactions between Sxl mutations and sis-a and sc mutations (Cline, 1986, 1988; Torres and Sánchez, 1989). These interactions are thermosensitive, independently of the mutations employed, which include deficiencies of the respective genes. The permissive temperature is 18°C and the restrictive one 29°C. Similar thermosensitivity is found for the lethal phenotype of females homozygous for sis hypomorphic mutations (Cline, 1986; Torres and Sánchez, unpublished data). In contrast, males die if they contain three doses of either sis-a⁺ or sc⁺, or a duplication of both genes, or a duplication of either of them and a duplication of Sxl (Cline, 1988; Torres and Sánchez, 1989). This lethality is also thermosensitive, but 18°C is the restrictive and 29°C is the permissive temperature. These results indicate that the thermosensitivity is not due to the nature of the mutations employed; rather, it seems characteristic of the Sxl activation process. At 18°C activation of Sxl would be more efficient, making this temperature permissive for females and restrictive for males. Temperature does not affect Sxl regulation in wild-type flies; it only becomes a relevant factor in mutant flies in which the signal for Sxl activation is ambiguous.

We have determined the thermosensitive phase (TSP) for the interaction between sc^3-1 and Sxl^f1 in females. The TSP is extremely short and occurs very early in development (Fig. 2A). It begins before the first hour of development is completed and ends around the second hour of development at 29 °C. In males carrying duplications of sis-a and sc, the TSP ends at about the same time as in females (Fig. 2B). Although the interaction is strictly zygotic (Cline, 1988), we have not been able to define the beginning of the TSP. Males that escape the lethal effect of the sis-a and sc duplications show morphological alterations. The most frequent alteration is the abnormal development of the external terminalia which are absent or reduced to various degrees and show intersexual traits (Fig. 3A). Exceptionally, some of these males also show mosaic sexcombs composed of male and female bristles (Fig. 3B). These phenotypes are most likely due to the functioning of Sxl in the female-specific mode in some of the cells. Thus, the TSP of the lethal interaction between sc and Sxl coincides in males and females with the syncytial blastoderm stage, when generalized sc expression takes place (Romani et al. 1987; Cabrera et al. 1987).

We have used pulses of sc expression administered at different developmental times to define precisely its temporal requirement during development to activate Sxl. To this end we used a hsp70 –sc chimeric gene (HSSC-S) (Rodr íguez et al. 1990). sc^3-1/sc^10-1 females show a strong sis-b phenotype and consequently a very low viability. We have given heat pulses to embryos from a cross in which sc^3-1/sc^10-1 females carrying the HSSC-3 are produced. Expression of sc rescues sc^3-1/ sc^10-1 females during the first two hours of development (Fig. 4A). This coincides with the time of the transient generalized expression of sc in wild-type embryos (Romani et al. 1987; Cabrera et al. 1987). The fact that there is only partial, rescue is probably due to asynchronization of the egg sample at the time of the heat shock. Egg retention by females would cause contamination of samples with embryos in a more advanced stage than that expected from the oviposition time. This explains why the 0 –1 h class shows recovery, despite the fact that embryos should not be competent for heat-shock response until the syncytial blastoderm stage (Dura, 1981).

We have followed a different procedure to overcome the problem of the asynchronization of the samples. We carried out the same cross as in experiment-2 of Fig. 4A, in which the HSSC-3 comes from the father, but the heat shock was not given to the embryos but to their mothers. Maternal heat shock could induce transcriptional activation from the HSSC-3 in the egg only when the zygotic genome is already competent for transcription; i.e. at the ninth nuclear division (Anderson and Lengyel, 1980). As the HSSC-3 comes from the paternal genome, its transcription would be caused by accumulation in the egg of activated heat-shock transcription factor (HTF) (Wu et al. 1987) of maternal origin. With this procedure the heat shock mediated expression of sc will occur at the same developmental time in all the embryos. Even 3 –6 h hours after the maternal heat shock, the embryos at the syncytial blastoderm stage contain sc mRNA in larger amounts than the non-heat-shocked wild-type embryos at the same stage; moreover, in heat-shocked embryos, the sc mRNA lasts through the cellularization stage (data not shown). This indicates that sc expression is achieved using this procedure and, as expected, full recovery of experimental females is observed in the 3 –6 h class (Fig. 4B). Thus, sc expression is fully able to suppress the sis-b phenotype. In the egg collections after 6h or more from the heat shock, the degree of recovery gradually decreases, presumably due to the disappearance of the consequences of the heat shock in the mother. The time elapsed until the eggs laid are free of heat-shock effects is surprisingly long. This suggests a high perdurance of the activated HTF in the egg. In the first class (0 –3 h), there is only partial recovery. This is probably due to the presence of eggs that at the moment of the heat shock were already independent of the maternal influence, but not yet competent for heatshock response. The control cross (HSSC-3^-) shows some degree of recovery in the first two classes. This may be due to the maternal heat shock but not to the expression of sc. This result suggests that maternal heat shock affects some maternal process involved in Sxl early activation.

The above results show that the sis-b function is provided by the sc gene, confirming that sc is responsible for the sex-determining function of the AS –C. Moreover, our results eliminate the possibility that the sis-b function could be attributed to the existence of sequences in the gene sc with capacity to absorb autosomal trans-acting factors, since we find that the sis-b activity is not related to the mere presence of the HSSC-3 but to its expression.

If the amount of sc transcripts produced during the syncytial blastoderm is making up the X:A signal that determines Sxl activity, sc expression should induce male lethality specifically at this stage due to Sxl activation. We have heat shocked embryos carrying the HSSC-3 at different stages of early development. Two types of males arise from the experimental cross (see legend to Fig. 5) from which eggs were heat shocked: those with Sxl⁺ and those with Sxl^7BO, a deficiency for the entire gene (Saiz et al. 1987). Both types carry the HSSC-3. Sxl^7BO males are used as viability controls for their Sxl⁺ brothers. In this way we should exclusively detect the lethality associated with ectopic Sxl expression. Overexpression of sc indeed promotes Sxl-dependent male lethality (Fig. 5) only during the first 3h of development at 18°C. Although only partial, lethality is stronger in the 1 –2 h class than in the 2 –3 h class. Viability of the Sxl^7BO males is unaffected. We see that the developmental time at which male viability is sensitive to excess of sc expression coincides with the effective period in which this same expression rescues the sis-b phenotype in females and it corresponds to the syncytial blastoderm stage.

The initial regulation of Sxl depends on the X:A signal, and seems to occur at the transcriptional level (Saiz et al. 1989). Therefore, alteration in the X:A signal due to sc and sis-a mutations should affect this initiation process. To test this hypothesis, we have looked at the early Sxl transcription in a sample of embryos in which all of the females were simultaneously heterozygous for sc and loss-of-function sis-a mutations and a deficiency for Sxl. These females die during the embryonic stage due to their inability to activate Sxl (data not shown). The Sxl transcription has been analyzed by in situ hybridization of embryos with the digoxigenin-labeled hi probe (Bell et al. 1988) that detects all of the known Sxl transcripts. The degree of hybridization has been quantified by measuring the intensity of the staining in the individual embryos (legend to Fig. 6). As controls, we have used Oregon-R wild-type embryos. In this case, around cellular blastoderm stage, we find two types of embryos that differ in the intensity of the staining (Fig. 6) and fall into a bimodal distribution with a 1:1 ratio of the two types. This bimodal distribution most likely reflects differences between females and males, with respect to the early expression of Sxl. An X:A ratio of 1 in females would produce a regulatory signal that allows the initiation of Sxl expression, while in males an X:A ratio of 0.5 would prevent this expression. In the cross where all of the female embryos are heterozygous for sc, sis-a and Sxl, the embryos fall into a unimodal distribution (Fig. 6), which corresponds to the weakly stained class of wildtype embryos. This variation in the pattern of Sxl expression cannot be attributed to the presence of a single dose of Sxl in the experimental females, since Df(Sxl)/+ females containing two doses of both sc and sis-a are fully viable (data not shown). Thus, the mutant condition of the females for sc and sis-a coincides with the disappearance of the strongly stained class. This indicates first, that the strongly stained class observed in the wild-type sample indeed corresponds to females and second, that the strong Sxl expression depends on the doses of sc and sis-a. This result supports the transcriptional control of Sxl by the X:A signal.

The X:A ratio determines the state of activity of Sxl around the blastoderm stage; thereafter Sxl maintains its determined state (either ON in females or OFF in males) independently of the X:A ratio (Sánchez and N öthiger, 1983; Bachiller and Sánchez, 1991). Thus, the function of this ratio as a genetic signal is temporally restricted to an early stage in development. Lack-of-function Sxl mutations produce embryonic lethality (our unpublished results). Since the X:A signal determines Sxl activity and the gene sc behaves as a numerator element of this signal, it is expected that females lacking sc function would show embryonic lethality. The experiments reported here confirm the prediction. The lethal phase of sc^- females is embryonic. The cause of the lethality is the lack of Sxl function, since this lethality is suppressed by Sxl^M1, a mutation that constitutively expresses the Sxl function (Cline, 1984). The same holds for mutations at the gene sis-a, thus confirming the involvement of this gene in the X:A signal.

Loss-of-function mutations at sc and Sxl interact synergistically causing female lethality, while duplications of sc and Sxl cause male lethality due to expression of Sxl (Cline, 1988; Torres and Sánchez, 1989). This interaction between sc and Sxl is thermosensitive. We have shown here that the TSP of this interaction is very short and spans approximately the syncytial blastoderm stage, being the same for both females and males. By means of a hsp70 –sc chimeric gene (HSSC-3) (Rodríguez et al. 1990), we have expressed sc at different developmental times and checked when this expression suppresses the sis-b mutant phenotype in females,, and causes lethality in males due to expression of Sxl. We have found that expression of sc activates Sxl only at a very specific stage in development, coinciding with the TSP delimited by the temperature-shift experiments and corresponding to the syncytial blastoderm stage. At this stage, the X:A signal determines Sxl activity (Sánchez and N öthiger, 1983; Bachiller and Sánchez, 1991) and the gene sc undergoes an homogeneous expression (Romani et al. 1987; Cabrera et al. 1987). We conclude that the sc expression al the syncytial blastoderm is responsible for its sis-b function. Moreover, since the expression of sc from the HSSC-3, but not the mere presence of the gene, fully suppresses the sis-b phenotype, the sis-b function must be provided by the sc protein.

Ectopic expression of sc in the heat-shock experiments only activates Sxl in males or in females when expressed at the time when generalized sc expression takes place in wild-type embryos. Moreover, sc expression in neurogenic regions does not induce the activation of Sxl in males. Consequently, the stage-specificity of sc expression cannot account for the stage-specificity of the Sxl activation process. Two hypotheses can be put forward to explain such time-specificity. First, since the X:A signal is made up of several discrete genes, such as sis-a and sc and it also needs the activity of transducer genes, such as da, it is possible that one or more genes are indispensable for Sxl activation and expressed only at the syncytial blastoderm stage, da cannot confer temporal specificity to the process, as it is also expressed zygotically and required, together with sc, in neurogenesis (Caudy et al. 1988a). However, the molecular nature and pattern of expression of sis-a are unknown. This gene thus remains as a candidate to be responsible for the time-specificity observed. The second hypothesis is that Sxl could be susceptible to activation by the X:A signal, only at the syncytial blastoderm stage; later its active or inactive states would be independent of this signal. The gene Sxl seems to contain two promoters: an early promoter responsible for the early Sxl expression, which would specifically respond to the X:A signal, and a constitutive promoter,. which functions later both in males and females (Saiz et al. 1989). The accessibility of the early promoter exclusively at the beginning of development might account for the time-specificity in receiving the X:A signal. One possibility is that the activation of the late promoter could hinder the accessibility of the more internal early promoter.

The X:A signal results from the interaction between X-linked (sisterless) and autosomal elements, sis elements would be .genes expressed around blastoderm stage in a non-compensated way, so that female embryos would have twice the amount of sisterless products than male embryos. Only the molecular nature of one of the members of the X:A signal, sc, is known. This gene encodes a helix-loop-helix (HLH) protein (Villares and Cabrera, 1987). HLH proteins are transcriptional regulators whose activity depends on homo- or heterodimerization with other HLH proteins (Murre et al. 1989a,b). Association of a particular HLH protein with different members of the family produces dimers that differ in the affinity for their DNA-binding sites (Benezra et al. 1990; Garrell and Modolell, 1990; Ellis et al. 1990; Sun and Baltimore, 1991). Parkhurst et al. (1990) have proposed that the X:A signal is formed by X-linked sis products (numerator elements) which could be titrated by autosomal HLH products (denominator elements) so that an effective concentration of sis products would only be attained in females. The sis products would form heterodimers with the da maternal product, which is also an HLH protein (Caudy et al. 1988b), so that heterodimers between da and sis products would promote early transcription of Sxl. Evidence supporting the transcriptional regulation of Sxl by the X: A signal is presented here. The amount of Sxl transcripts in wild-type young embryos fall into a bimodal distribution, with an equal number of embryos in each class. This distribution would reflect the two types of embryos, female and male, with different levels of Sxl transcription. However, while females with a single dose of Sxl are wild-type, female embryos containing a single dose of Sxl, sc and sis-a, are indistinguishable from their sibling male embryos by their amount of Sxl transcripts. This indicates a transcriptional control of Sxl by the doses of sis-a and sc around the blastoderm stage. The proteins that directly bind to Sxl for its initial activation are unknown. The fact, however, that sc codes for an HLH protein and the very short time between sc and Sxl early expression suggests that the sc protein may be one of the factors binding directly to the Sxl early promoter.

We are grateful to L. Mateos and R. de Andrés for technical assistance. We thank I. Rodríguez and M. Ruiz-Gómez who kindly provided the HSSC-3 stock and S. Campuzano for her help with the in situ hybridizations. We are grateful to B. Granadino, M. de la Escalera, J. Modolell and S. Campuzano for helpful comments on the manuscript. This work was supported by a grant PB87-0239 from the Comisión Interministerial de Ciencia y Tecnología (CICYT). M.T. is recipient of a predoctoral fellowship from the Ministerio de Educación y Ciencia.