Sex-Specific Transcript Splicing

Sex-lethal ( Sxl) gene products induce female development in Drosophila melanogaster and suppress the transcriptional hyperactivation of X-linked genes responsible for male X-chromosome dosage compensation. Control of Sxl functioning by the dose of X-chromosomes normally ensures that the female-specific functions of this developmental switch gene are only expressed in diplex individuals. Although the immediate effect of X-chromosome dose is on Sxl transcription, during most of the life cycle “on” us. “off” reflects alternative Sxl RNA splicing, with the female (productive) splicing mode maintained by a positive feedback activity of SXL protein on Sxl pre-mRNA splicing. “Male-lethal” ( S x l M ) gain-offunction alleles subvert Sxl control by Xchromosome dose, allowing female Sxlfunctions to be expressed independent of the positive regulators upstream of Sxl. As a consequence, SxlM haplexanimals (chrome soma1 males) die because of improper dosage compensation, and SxlM chromosomal females survive the otherwise lethal effects of mutations in upstream positive regulators. Five independent spontaneous SxlM alleles were shown previously to be transposon insertions into what was subsequently found to be the region of regulated sex-specific Sxl RNA splicing. We show that these five alleles represent three different mutant types: Sxl”, SxlM3, and SxlM4. Sxl” is an insertion of a roo element 674 bp downstream of the translation-terminating male-specific exon, SxlM3 is an insertion of a hobo transposon (not 297 as previously reported) into the 3’ splice site of the male exon, and SxlM4 is an insertion of a novel transposon into the male-specific exon itself. We show that these three gain-of-function mutants differ considerably in their ability to bypass the sex determination signal, with SxlM4 being the strongest and Sxl” the weakest. This difference is also reflected in effects of these mutations on sex-specific RNA splicing and on the rate of appearance of SXL protein in male embryos. Transcript analysis of doublemutant male-viable SxlM derivatives in which the SxlM insertion is cis to loss-of-function mutations, combined with other results reported here, indicates that the constitutive character of these SxlM alleles is a consequence of an alteration of the structure of the pre-mRNA that allows some level of female splicing to occur even in the absence of functional SXL protein. Surprisingly, however, most of the constitutive character of SxlM alleles appears to depend on the mutant alleles’ responsiveness, perhaps greater than wild-type, to the autoregulatory splicing activity of the wild-type SXL proteins they produce. G AIN-OF-FUNCTION (g-o-f) alleles of the X-linked regulatory gene, Sex-lethal, partially bypass the primary sex-determination signal of Drosophila melanogaster. These alleles have played an important part in the elucidation of the mechanism by which the sex of D. melunoguster is determined (CLINE 1978, 1979, 1983, 1984), and they continue to be useful in such studies ( STEINMANN-ZWICKY et ul. 1989; OLIVER et al. 1990,1993; SALZ 1992; PAULI et al. 1993; STEINMANN-ZWICKY 1993). However, little is known about the specific molecular nature of their mutant lesions, the mechanistic basis for their constitutive expression, or the functional differences that might exist among alleles of this class. Cmsponding authw: Thomas W. Cline, Division of Genetics, Department of Molecular and Cell Biology, University of California, 401 Barker Hall, Berkeley, CA 94720-3204. E-mail: [email protected] ’Present address: Yale University, New Haven, CT 06520-8103. Genetics 139: 6.71648 (February, 1995) Genetic and molecular characterization presented here addresses these questions. It should enhance the usefulness of these alleles, as well as provide insights into mechanisms leading to ectopic gene expression in higher eukaryotes. Sex determination in Drosophila is controlled by the activity of Sxl, which, in response to the primary sexdeterminant, X-chromosome dose, is active in females ( X X ) and inactive in males ( XY) (recently reviewed by BELOTE 1992; CLINE 1993). Sxl in turn determines the functional states of more developmentally specialized regulatory genes downstream. Loss-of-function mutations in Sxl are specifically lethal to chromosomal females because of dosage compensation upsets (LUG CHESI and SKRIPSKY 1981; CLINE 1983; GERGEN 1987; BERNSTEIN and CLINE 1994). This lethality generally obscures the masculinizing effect of such mutations; however, effects on female sexual differentiation can 632 M. Bernstein et al. be observed readily in genetic mosaics and triploid intersexes (CLINE 1979; SANCHEZ and NOTHICER 1982; CLINE 1983) . G “ f alleles in Sxl were recovered as mutations that rescued chromosomal females that otherwise would have died because of the lethal maternal effect of mutations in daughterless (CLINE 1978). Although the maternal activity of d a normally is required for the activation of Sxl+ in the zygote, Sxl g-o-f mutations partially bypass this requirement (CLINE 1980, 1983). In bypassing the normal mechanism that activates Sxl in response to Xchromosome dose, these mutations also behave as dominant male-specific lethals due to their inappropriate expression of female-specific Sxl activities in haplo-X individuals (CLINE 1984). For this reason, they are known as SxlM alleles, in contrast to the female-lethal Sxl’ loss-of-function alleles. The feminizing effect of these male-specific lethals on adult differentiation is apparent in genetic mosaics and triploid intersexes (CLINE 1979, 1983). Apart from this regulatory defect, SxlM alleles are functionally wild-type, as evidenced by their ability to support wild-type female development even when hemizygous. The initial control of Sxl’ by X-chromosome dose is at the level of transcription and reflects effects on the activity of a “sexual pathway establishment” promoter, Sxlp, (KEYES et al. 1992). A short burst of SXL protein is generated very early in development in females as a result of the transitory expression of Sxl,+ in response to their double dose of X-linked “XA numerator element” genes. The single dose of these same genes in males is not sufficient to activate Sxl,+. Throughout most of development, however, Sxl,+ is silent in both sexes, and the activity state of Sxl reflects alternative RNA splicing and the operation of a positive feedback loop for transcripts derived from a “sexual pathway maintenance” promoter, Sxl,+n, located 5 kb upstream of SxZ,+ (BELL et al . 1988, 1991; SAMUEIS et al. 1991). Only females splice SxlIJVL transcripts into mRNAs encoding full-length active SXL proteins. This is because SXL proteins exhibit an RNA binding activity ( SOSNOWSKI et al. 1989; INOUE et al. 1990; SAKAMOTO et al. 1992; VALCARCEL et al. 1993; SAMUELS et nl. 1994) that inhibits inclusion of a male-specific translation-terminating exon that would otherwise be present in Sxl,>,-derived mRNAs and prevent the production of functional Sxl products (BELL et al. 1988, 1991 ) . This positive feedback on RNA splicing is responsible for maintaining cells’ commitment to the female developmental pathway after X-chromosome dose has been assessed. Transcripts from Sxlpe, unlike those from Sxl,,m, are productively spliced even in the absence of SXL protein. Thus, it is the early burst of SXL protein from transcripts originating at SxZpe that initiates the female commitment by setting the autoregulatory feedback loop in operation for Sxlh transcripts. Although Sxl is nonfunctional in males, Sxlpm is active in this sex, beginning at the blastoderm stage and continuing through adulthood (SALZ et al . 1989; SAMUELS et al . 1991) ; however, because males lack the initial burst of SXL protein derived from Sxli+ expression, they never generate functional protein products because the translation-terminating male-specific exon is always present in their Sxl mRNA. In this way, the nonfunctional state of Sxl , the committed male developmental pathway, is maintained by default. When Sxl was first cloned, it was shown that the DNA alterations associated with five spontaneously generated SxlM alleles are transposon insertions within a 1-kb region of the gene ( MAINE et al. 1985) that was subsequently shown to contain the male-specific exon (BELL et aZ. 1988; SALZ et al. 1989). Given our current understanding of SxZ regulation, there are at least three ways that the transposon insertions present in the SxZM alleles might bypass the normal sex-determination signal: ( 1 ) the transposon might activate Sxl,, in the absence of the positive regulators normally required for its expression, ( 2 ) promoters in the transposon itself might generate novel transcripts that could encode functional SXL proteins, or ( 3 ) the transposons might disrupt the structure of the Sxlt+,t-derived RNA so that SXL protein is no longer required to promote the productive transcript splicing mode. The finding that all five transposon insertions in the Sxl” alleles inserted in the vicinity of the male-specific exon, rather than in the vicinity of Sxlpp, would seem to favor a direct effect on Sxl splicing, but by itself this finding does not eliminate other alternatives. Because transcriptional enhancers can work at considerable distances, enhancers carried by SxZM transposons might be able to activate Sxlpp. Also, the discovery that translation of Sxl mRNA seems to reinitiate in adult males, albeit at very low efficiency downstream of the male-specific exon ( BOPP et al. 1991 ) , raises the possibility that transcripts initiating from transposons in the vicinity of the male-specific exon might be able to generate significant levels of functional SXL protein (it should be noted, however, that there was no indication of activity for the truncated adult male proteins mentioned). Independent of the mechanism leading to the initial burst of ectopic SxZ expression, the autoregulatory activity of SXL proteins generated could be expected to influence the splicing and hence functioning of subsequent transcripts; therefore, it is of interest to know the extent to which the constitutive functioning of SxZM alleles depends on SxZ autoregulation. Here we address the question of what is responsible for the constitutive behavior of SxlM alleles by defining more precisely the specific molecular nature of the SxlM lesions and by examining their effects on the kinds of SxZ mRNAs that are generated in the presence and absence of SXL autoregulatory activity. Our results indicate that at least one SxZM allele directly affects Sxl splicing [alternative ( 3 ) above] but also indicate that the level of female transcripts generated in males-the Male-Lethal Insertions in Sxl 633 level of constitutive expression-is remarkably dependent on the autoregulatory activity of the SXL proteins produced. Moreover, those levels may reflect an increased sensitivity of the mutant alleles to such autoreg-