Maternal Repression of the P Element Promoter in the Germline of Drosophila melanogastec A Model for the P Cytotype Bruno Lemaitre , ’ Stephane Ronsseray and Dario Coen

The transposition of P elements in Drosophila melanogaster is regulated by products encoded by the P elements themselves. The P cytotype, which represses transposition and associated phenomena, exhibits both a maternal effect and maternal inheritance. The genetic and molecular mechanisms of this regulation are complex and not yet fully understood. In a previous study, using P-lacZ fusion genes, we have shown that P element regulatory products were able to inhibit the activity of the P promoter in somatic tissues. However, the repression observed did not exhibit the maternal effect characteristic of the P cytotype. With a similar approach, we have assayed in vivo the effect of P element regulatory products in the germline. We show that the P cytotype is able to repress the P promoter in the germline as well as in the soma. Furthermore, this repression exhibits a maternal effect restricted to the germline. On the basis of these new observations, we propose a model for the mechanism of P cytotype repression and its maternal inheritance. T HE transposition of the Drosophila melanogaster P element is genetically regulated. It occurs at high frequencies in the progeny of a cross in which males from a P element-containing strain (P strain) are mated to females devoid of P elements (M strain). The progeny of both sexes resulting from this type of cross display the P-M hybrid dysgenesis syndrome as a consequence of P element mobiIization (reviewed in ENGELS 1989; RIO 1990). This syndrome includes high rates of mutations, chromosome rearrangements, male recombination and a thermosensitive agametic sterility (GD sterility). The virtual absence of these dysgenic traits in the progeny of the reciprocal cross (P females crossed to M males) demonstrates maternal repression of P . This maternal effect is attributed to a cytoplasmic condition not permissive for P element transposition called the P cytotype. Moreover, this P cytotype state can be maternally inherited for several generations. Both the maternal effect and the maternal inheritance are specified by the genomic P elements themselves (reviewed in ENGELS 1989; RIO 1990). P element transposition is normally restricted to the germline. This germline specificity is regulated at the level of pre-mRNA splicing of the third P element intron between ORFs (open reading frames) 2 and 3 (LASKI, RIO and RUBIN 1986). In germline tissues, the full length 2.9-kb P element expresses a trans-acting protein of 87 kDa, demonstrated to be the transposase (RIo, LASKI and RUBIN 1986). In somatic tissues, the ’ Present address: UniversitC Louis Pasteur, Laboratoire de Biologie Ginirale, 12, rue de I’Universiti, F-67000 Strasbourg, France. Genetics 135 149-160 (September, 1993) retention of the third intron results in the production of a smaller protein of 66 kDa specified by the first three ORFs. This truncated protein was postulated to be the repressor of transposition involved in the P cytotype (RIo, LASKI and RUBIN 1986). T o test this hypothesis, in vitro modified P elements have been studied regarding their ability to specify the P cytotype (ROBERTSON and ENCEJS 1989; MISRA and RIO 1990). Elements having lesions affecting the fourth transposase exon but leaving intact the first three exons have the ability to engender many aspects of the P cytotype. However, none of the lines carrying these in vitro modified P elements exhibits the maternal effect characteristic of the P cytotype, and they are unable to suppress totally the GD sterility induced by a P strain. The authors have interpreted this observation as resulting from an insufficient amount of repressor in the appropriate tissue ( i e . , the germline)(ROBERTSON and ENGELS 1989; MISRA and RIO 1990). This could be due either to a position effect or to the nature of the repressor produced by the element tested or to both. In addition, although the 66kDa protein clearly has regulatory properties, the molecular mechanism of its repression is still unknown. Competition at the DNA binding site between the transposase and the repressor, transcriptional repression, action on RNA splicing or RNA stability and protein-protein interactions are all considered possible mechanisms (reviewed by RIO 1990). In a previous study, we have shown that P-encoded regulatory products, including the 66-kDa protein synthesized by a modified P element, inhibit in vivo the expression of the P promoter (LEMAITRE and 150 B. Lemaitre, S. Ronsseray and D. Coen COEN 1991). These results were obtained by using “enhancer-trap” insertions (P-lac2 fusion genes). In these constructs, the Escherichia coli lac2 gene is fused in frame with the P element transposase gene (for a review on “enhancer trap” see WILSON, BELLEN and GERHINC 1990). Therefore, they function as reporter genes of P promoter activity. LEMAITRE and COEN (1 991) assayed the repression by P regulatory products on P-lac2 expression only in the somatic tissues. It was noted that in these tissues the repression did not show the maternal effect characteristic of P cytotY Pee Here we have extended our investigation to the germline by using P-lac2 insertions that are expressed in this tissue. We show that P cytotype repression of P-lacZ expression is also observed in the germline. The intensity of this repression is consistently stronger in this tissue than in the somatic tissues. Furthermore, a maternal effect of the repression was clearly detected and appeared to be restricted to the germline. These observations allow us to propose a model for the mechanism of P cytotypic repression and its maternal inheritance. MATERIALS AND METHODS P-laeZ fusion genes: P[lac, ry+]A contains an in-frame translational fusion of the E. coli &galactosidase gene ( lac2) to the second exon of the P transposase gene and also contains the rosy+ gene as a marker for transformation (O’KANE and GEHRINC 1987). P[lwB] is a similar P-lac2 construction except that the marker for transformation is the mini-white gene (KLEMENZ, WEBER and GEHRINC 1987) and that it contains bacterial plasmid sequences allowing rapid cloning by plasmid rescue (BELLEN et al. 1989). Drosophila stocks: Canton (KIDWELL, KIDWELL and SVED 1977), Gruta ( A N X O L A B ~ ~ R E et al. 1987): typical M strains containing no P element sequences. Harwich (KIDWELL, KIDWELL and SVED 1977): a reference P strain. HS2-25: this P strain was derived from the Gruta M stock by germline transformation with complete P element DNA (ANXOLAB~H~RE et al. 1987) and this stock is therefore essentially isogenic to the Gruta strain (LEMAITRE and COEN y506 P[ry+A2-3](99B), hereafter designated A2-3(99B), contains an essentially immobile P element insertion that lacks the last intron and produces a high level of transposition in the soma as well as in the germline (ROBERTSON et al. 1988). R20a is a line harboring an insertion of a P-lac2 element on the third chromosome at cytological site 100CD. It shows ubiquitous lac2 expression (LEMAITRE and COEN 199 1). BG07, BQ16, BC69, BA37, ABOO, LJ25, BP76, BA34, BA32, BP73, BL54 are 11 lines harboring an autosomal insertion of a P-lac2 element (P[lac, ry+]A) balanced by T(2;3),CyO; TM6. They were isolated in a screen for female sterile mutations and express the P-lac2 transgene in ovaries and testes (J. L. COUDERC and F. LASKI, unpublished data). These strains are described in Table 1 and in the text. KP-D, KP-F2, KP-U are three lines bearing one autosomal KP insertion and displaying some regulatory properties 1991). (W. R. ENCELS, G . GLOOR and C. PRESTON, unpublished data; see DISCUSSION for description). The U line contains a nested element composed of two KPs, one inserted in the 3’ end of the other (W. R. ENGELS, G. GLOOR and C. PRESTON, unpublished data). The Lk-P(lA), P[ry+, SalI](89D), v6 and WY 11 3 strains are described in the text. For complete description of marker genes and balancer chromosomes, see LINDSLEY and GRELL (1 968). &Galactosidase localization: Ovaries and testes were fixed in 0.5% glutaraldehyde, 1 mM MgCI, in PBS pH 7.5 for 4 min, washed in PBS buffer, then submerged in 0.2% 5-bromo-4-chloro-3-indolyl-~-~-galactopyranoside (X-gal), 3.5 mM K4Fe(CN)6, 3.5 mM KsFe(CN)s, 1 mM MgC12, 150 mM NaCI, 10 mM Na2HP04, 10 mM NaH2P04 and incubated overnight at 37” (HIROMI, KUROIWA and GEHRING 1985). They were then mounted in glycerol. Quantitative measurement of &galactosidase activity: Three ovaries were homogenized in Z buffer (MILLER 1972) and centrifuged for 10 min at 10,000 rpm (4”) to remove debris. P-Galactosidase activity was measured as described in MILLER (1972). The protein concentration was determined in each sample by the Bradford assay (BioRad), using BSA as a standard. Results are given in nM/min/mg of proteins. Statistical analysis: Comparisons of means were performed with the Fisher-Student t-test when the data analyzed fitted the conditions of validity. The use of the t-test implies two conditions. (1) The data must be normally distributed. We have checked this point by testing (with a x* test) the distribution among intervals of equiprobability of a large number of preliminary measurements of the BC69 P-lac2 activity in various backgrounds. The distribution of P-lac2 activity measurements was clearly consistent with the hypothesis of normality. In fact, this result is not surprising since biological characters are often normally distributed. (2) The variances must be identical throughout the groups compared. The t-test needs then the calculation of a common weighted SD value calculated with the whole set of data. For the results in Tables 2 and 4, the SD values obtained for the different sets of measurements are consistent with the hypothesis of homoscedasticity when tested with the Hartley-test. Consequently, within each of these two tables, the comparisons of means were performed with the t-test using a common weighted SD calculated from all the data in the table. This common SD is given with its degree of freedom (df) in the legend of each table. In Table 3, some sets of measurements produced unexpectedly high SD values. So in this table, comparisons of means were performed with the nonparametric Mann-Whitney test, which does not imply homoscedasticity.