A Novel Repeat-Associated Small Interfering RNA-Mediated Silencing Pathway Downregulates Complementary Sense gypsy Transcripts in Somatic Cells of the Drosophila Ovary

Fly care.

Unless otherwise indicated, genetic materials and fly stocks are described in Flybase (http://flybase.bio.indiana.edu). Flies were grown at 25°C on standard Drosophila medium (17).

The permissive wOR(P) and restrictive wRev(R) strains were described previously (35). They are called empty strains because they are devoid of functional gypsy proviruses and only contain the defective heterochromatic proviruses found in all present-day flies (27, 63). In contrast, the A237 (unpublished) and MG1 (44) restrictive strains both contain, in addition, many functional gypsy proviruses, located on the X chromosome and the whole genome, respectively. The MG1(rec) stock was derived from MG1 as follows. After a series of eight successive backcrosses with attached-X females of an empty strain, the MG1 autosomes were replaced by autosomes devoid of functional gypsy proviruses. We checked that only the X chromosome of the backcrossed stock contained sequences able to be lighted up by fluorescence in situ hybridization (FISH) of a gypsy probe with salivary gland chromosomes. These X-linked proviruses were eliminated together with the associated y v f mal markers, by recombination of the flam^MG1(R) restrictive allele into the wOR(P) background.

The following mutant fly stocks were also used: y w eyFlp; FRT42D dcr-2^L811fsX (31), w¹¹¹⁸;cn piwi¹/CyO and cn piwi³/CyO;ry⁵⁰⁶ (12), aub^HN2 cn bw/CyO and aub^QC42 cn bw/CyO (54), w UASt-IRlacZ (12A) (23), and GMR-wIR (30, 31).

Detection of small RNA species.

The detection of small RNAs was performed essentially as described previously (52), with the following modifications. Total RNA was isolated from 100 pairs of ovaries by using TRIzol. The RNA was enriched for low-molecular-weight species by spinning for 20 min at 1,000 × g through Microcon YM-100 concentrators (Millipore). Alternatively, enrichment was performed by extracting the RNA with a mirVana kit (Ambion). About one-fourth of the ovarian extracts were loaded onto a 15% denaturing acrylamide-bisacrylamide (19:1) gel, transferred onto a Hybond NX membrane (Amersham), and fixed by UV cross-linking (0.5 J/cm²). Hybridizations with hydrolyzed RNA probes (50 nt, on average) were performed overnight at 40°C in 45% formamide-0.3 M NaCl-50 mM NaPO₄ (pH 6.8)-1× Denhardt solution-7% sodium dodecyl sulfate. Membranes were washed twice for 30 min at 60°C in 2× SSPE (1× SSPE is 0.18 M NaCl, 10 mM NaH₂PO₄, and 1 mM EDTA [pH 7.7])-0.5% sodium dodecyl sulfate and exposed for 2 days on a PhosphorImager screen (Molecular Dynamics). The sizes and specific radioactivities of both cRNA probes were compared by comigration on 8% denaturing acrylamide gels. The hybridization efficiencies were also calibrated by cohybridizing the Northern membranes with duplicate dots of dsRNA which was expected to be similarly lighted up by both RNA probes. To ensure that the RNA transcripts used as probes were either only sense or antisense, oligonucleotides corresponding to sense and antisense polarities were also spotted on these control membranes. After narrowing down the linear sensitivity range by using the “Gray/Color Adjust” function of ImageQuant, the exported tiff images were mounted with Adobe Photoshop without any further editing. The amount and integrity of the RNA were checked both by staining the gels with 2 mg of ethidium bromide/ml and by reprobing the unstripped membranes with an end-labeled oligonucleotide complementary to the mir-13b miRNA (5′-ACTCGTCAAAATGGCTGTGATA-3′).

NaIO₄ reaction and β-elimination.

This chemical reaction was performed essentially as described previously (1). Up to 20 μg of low-molecular-weight RNA was resuspended in 17.5 μl (30 mM borax, 30 mM boric acid [pH 9.0]). After the addition of 2.5 μl of freshly dissolved 200 mM NaIO₄, a 10-min incubation was performed in the dark at room temperature. To quench the unreacted NaIO₄, 2 μl of glycerol was added, followed by incubation for an additional 10 min in the dark at room temperature. Samples were almost completely dried by centrifugation under vacuum for 1 h at room temperature, resuspended in 50 μl of borax buffer (33.75 mM borax and 50 mM boric acid [pH 9.5] with NaOH), and incubated for 90 min at 50°C in dark. RNA was precipitated overnight at −20°C with 6 μl of 3 M sodium acetate and 168 μl of ethanol, collected by centrifugation, and then dissolved in the denaturing gel loading buffer. As a control, the whole procedure was performed by replacing the NaIO₄ with water.

Plasmid construction and P element-mediated germ line transformation.

As described previously (52), the pES2 and pES5 constructs (Fig. 1D) are carried by a transformation vector that includes the white gene as a selection marker. The parental pES2 plasmid consists of the yp3 ovarian promoter, driving the expression of the lacZ reporter in the follicle cells (47). The 4342-4523 gypsy fragment was PCR amplified with primers containing SalI restriction sites at their non-gypsy 5′ extensions. The pESpol− and pESpol+ plasmids were constructed by cloning this SalI-restricted 187-bp fragment into the 5′UTR of pES2 (XhoI restriction site) in the reverse and forward orientations, respectively.

P element-mediated transformation was performed as described previously (49). The constructs were coinjected with the pUChsΔ2-3 helper (a gift from D. C. Rio) into embryos of the permissive wOR(P) stock. Flies carrying the insertion were identified by rescue of the white phenotype. Inserts in transgenic flies were made homozygous in this w flam^OR(P) permissive background by endogamous crosses and selection for the most highly pigmented flies. The transformants were also backcrossed twice to wRev(R) females and then made homozygous in this restrictive background as well. Each pair of transgenes was analyzed (XbaI-restricted genomic DNA probed with the simian virus 40 trailer) to confirm that the same transgene was indeed present in both backgrounds.

Histochemical analysis of β-galactosidase.

Ovaries were fixed in 2% formaldehyde-0.2% (vol/vol) glutaraldehyde-phosphate-buffered saline (PBS) for 5 min at room temperature, washed twice with PBS, rinsed once in staining buffer [1 mM MgCl₂, 4 mM K₃Fe(CN)₆, 4 mM K₄Fe(CN)₆, 1% Triton X-100], and stained for 0.5 to 4 h at 37°C in staining buffer containing 0.27% X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside). For the sake of comparison, each pair of permissive and restrictive genotypes containing the same homozygous transgene was treated simultaneously. The reaction was stopped in PBS, and the ovaries were mounted in 90% (vol/vol) glycerol in PBS and viewed in a Leica DMRB microscope. Digital images were captured using the same settings and processed together after having been mounted in a single Adobe Photoshop file.

In situ hybridization.

Detection of RNA in whole-mount ovaries and female larval gonads was essentially performed as previously described (41) with the following modifications. Glutaraldehyde, dimethyl sulfoxide, heparin, levamisole, and RNase treatments were omitted. Hybridization was performed on wandering female larvae that had been inverted after cutting off the head (the proteinase K treatment was omitted, too). The prehybridization and hybridization temperatures were 55 and 65°C, respectively. Digoxigenin-labeled RNA probes, a 1.2-kb gypsy antisense fragment (coordinates 1425 to 199), and a 2.5-kb lacZ antisense fragment (accession number J01636, coordinates 2100 to 4600) were not hydrolyzed. Washings were performed at 55°C in hybridization buffer and then in PBS containing 0.1% Tween 20. The background was destained by soaking overnight in 95% ethanol. Ovaries and larvae were stored in PBS-75% glycerol, dissected, mounted in PBS-90% glycerol, and viewed under a Leica DMRB microscope. Digital images were captured using the same settings and processed together after having been mounted in a single Adobe Photoshop file.

RNA secondary structure prediction.

The sequence analysis was done online at http://www.genebee.msu.su/genebee.html. Each selected sequence, flanked by at least 100 nt at both ends, was checked for the absence of stems longer than 20 nt.

Unlike miRNAs, both gypsy rasiRNAs and hairpin-induced siRNAs are modified at their 3′ termini.

As reported recently (62), the 3′ terminus of rasiRNAs respectively produced by the Su(Ste) and roo repeated genetic elements in the testis and ovary differ from those of the miR-8 and miR-311 microRNAs. The absence of either a 2′ or a 3′ hydroxyl group makes them resistant to the β-elimination reaction which, on the contrary, causes the loss of the terminal nucleoside of NaIO₄-pretreated microRNAs. We repeatedly observed that the 24- to 29-nt rasiRNAs detected by riboprobes complementary to antisense gypsy sequences from within the long terminal repeat and env regions (Fig. 1C) were not affected by the NaIO₄ and β-elimination treatments (Fig. 2A, top, lanes 2 and 3). In contrast, in the very same samples, the miR-13b-positive control RNA reacted to completion with periodate, as shown by the apparent 2-nt shift downward in mobility resulting from the loss of one nucleotide and the gain of one negative charge (Fig. 2A, bottom, compare lanes 2 and 3).

Unlike plant mi- and siRNAs (15, 32, 66), animal siRNAs are thought to have terminal hydroxyl groups at both the 2′ and the 3′ positions and to be therefore sensitive to β-elimination. However, we observed (Fig. 2B) that the 21-nt RNAs resulting from the expression in the somatic ovarian tissues of a lacZ hairpin construct known to trigger RNAi (23) did not react with NaIO₄ altogether. Similarly, we inferred from the result of this β-elimination assay that the 21-nt RNAs derived from the expression in the head of the GMR-wIR hairpin RNAi trigger (30, 31) also lack either or both of the 2′ and 3′ terminal hydroxyl groups (Fig. 2C). Modification of their 3′ ends do not, however, prevent such hairpin-induced siRNAs from performing effective in vivo silencing as shown below (see Fig. 6 and 7). In this respect, they therefore look like plant mi-and siRNAs (15, 32, 66), as well as fly rasiRNAs.

Unlike siRNAs, gypsy rasiRNAs are strongly unbalanced in favor of the antisense strands.

The presence of a small fraction of gypsy dsRNA in tissue cell culture total RNA was reported long ago (21). Like siRNAs and miRNAs, Drosophila rasiRNAs were assumed to be the products of RNase III processing of the dsRNA precursors (4). As a control, we first checked that both strands of the GMR-wIR hairpin RNA precursor were equally represented in the population of their siRNA products (Fig. 2C). We then wondered whether gypsy rasiRNAs would also contain both strands of the gypsy sequence. To address this question, we first used both strands of the 120-nt gypsy 5′UTR fragment as probes (Fig. 1C). Although sense-oriented RNAs with the same size as the antisense strands were indeed detected, they were notably less abundant than their antisense counterparts (data not shown). Three additional small fragments were then randomly chosen from the complete gypsy proviral sequence (accession number M12927, coordinates 2188 to 2368, 4342 to 4526, and 6541 to 6718) to produce ∼180-nt strand-specific RNA probes (Fig. 1C).

The results obtained with these three gypsy fragments (Fig. 3) were in agreement with the observations made with the 120-nt fragment from within the gypsy 5′UTR. (i) We were able to detect small RNAs corresponding to each fragment, suggesting that most (if not all) gypsy sequences are likely to be represented in the population of gypsy rasiRNAs. (ii) Their sizes ranged from 23 to 29 nt. (iii) Both strands of the gypsy sequence could be detected, as if they were born double-stranded from long dsRNA precursor(s). (iv) However, regardless of the particular fragment used as a probe, the antisense gypsy rasiRNAs (Fig. 3A, B, E, F, I, and J) were always clearly more abundant than the corresponding sense strands (Fig. 3C, D, G, H, K, and L, respectively). An even greater, although also unexplained, imbalance has just been reported for the Su(ste) and roo rasiRNAs in Drosophila (62). In the other way, plants infected by positive RNA viruses have been reported to preferentially accumulate sense-stranded, hairpin-derived viral ds-siRNAs (37). A similar explanation for our results, which would assume that these mostly antisense gypsy rasiRNAs are processed from antisense gypsy hairpins, was ruled out by the following two observations: first, we could not predict any 20-nt hairpins for any of the three tested sequences (see Materials and Methods), and second, even if hairpins did exist, they would probably be mostly sense oriented, in view of the excess of sense gypsy transcripts (data not shown), which would therefore mostly produce sense rasiRNAs. (v) The three gypsy probes disclosed the interstrain variation in the amount of rasiRNAs previously observed with the 5′UTR RNA probe (52). Indeed, for each of the six strand-specific probes, the restrictive MG1 strain (Fig. 3B, C, F, G, J, and K) showed a significantly stronger signal than the restrictive wRev(R) strain (Fig. 3A, D, E, H, I, and L, respectively). In the latter strain, the amount of antisense rasiRNAs complementary to the 4342-4526 sense probe was so low that it was indistinguishable from that of the permissive wOR(P) strain (data not shown). If, as hypothesized earlier, these rasiRNAs are actually the effectors of gypsy regulation, then the strongest repression should occur against sense-oriented gypsy targets in the MG1 ovaries, whereas the 4342-4526 sense sequence should not be targeted by the repression in the wRev(R) strain.

The amounts of gypsy rasiRNAs are correlated with the sensitivity of the complementary targets to repression.

As shown in Fig. 1D, the targets that we decided to test correspond to both orientations of the small “pol” fragment that has been previously used to detect rasiRNAs. Four pESpol+ (sense orientation) and three pESpol− (antisense orientation) independent transgenic inserts were tested in three different genetic backgrounds, including the wOR(P) permissive strain and the F₁ progeny of crosses between the wRev(R) and MG1(R) restrictive strains. A sample of the results that were reproducibly obtained in three independent assays is shown in Fig. 4. In good correlation with the differential accumulation of the complementary rasiRNAs in the MG1(R) background (Fig. 3, lanes F versus G), all four pESpol+ transgenes were repressed (compare, for instance, Fig. 4A and C), whereas none of the three pESpol− transgenes was repressed (Fig. 4D and F). This correlation indicates that the antisense “pol” rasiRNAs are most likely the effectors which specifically guide presumptive repressor complex(es) toward the pESpol+ target.

Moreover, we observed an almost complete lack of repression of the pESpol+ target in the wRev(R) restrictive background (Fig. 4A and B). This observation is in good agreement with the low abundance of the complementary rasiRNAs in this strain (Fig. 3E). The ability of this strain to repress transcripts of the gypsy retrovirus may be explained by the accumulation of antisense rasiRNAs homologous to other parts of the element, such as, for instance, the 5′UTR (52).

So, the fact that both the strain- and strand-specific variations in the amount of gypsy rasiRNAs are correlated with the sensitivity of the complementary targets strongly suggests that they are bona fide effectors of this silencing phenomenon.

Unlike transgene-induced silencing, the efficiency of gypsy silencing is not copy number dependent.

In Drosophila, the posttranscriptional silencing of the endogenous adh gene depends both on the number and on the expression level of homologous transgenes, while the accumulation of complementary siRNAs varies accordingly (39). The high accumulation of gypsy rasiRNAs found in MG1 might result from the fact that this restrictive strain contains the highest copy number (about 30) of gypsy proviruses recently inserted into the euchromatic part of the genome.

To determine whether these proviruses do contribute to the production of gypsy rasiRNAs, we managed to get rid of them by inter- and intrachromosomal recombination (see Materials and Methods). Essentially, the pericentric region of the MG1 X chromosome, which contains a flam restrictive allele, was recombined into a permissive genome devoid of euchromatic proviruses. As a control, we then checked by FISH that the polytene chromosome arms of the newly obtained MG1(rec) strain did not contain any of the parental proviruses (data not shown). The fact that the amounts of gypsy rasiRNAs do not significantly differ between both strains (Fig. 2A, lanes 1 and 2, and data not shown) shows that they are not produced by euchromatic proviruses in a copy-dependent manner. This indicates that the bulk of gypsy rasiRNAs are produced by pericentromeric proviruses, even in a gypsy-rich strain such as MG1.

This mapping to the flam region of the genetic determinant of the wealth of MG1 rasiRNAs does not imply that the pericentromeric trigger(s) must also map there. We can imagine that some proviruse(s) located in the autosomal heterochromatin of the permissive parental strain could become very efficient triggers when combined with an exceptionally efficient MG1 flam restrictive allele. So, the reason why the production of gypsy rasiRNAs by some pericentromeric trigger(s) is so efficient in this particular genetic background is still unknown.

Effect of RNA silencing mutations on the production of gypsy rasiRNAs and on the repression of gypsy.

Accumulation of the GMR-wIR siRNAs is strongly reduced in dcr-2 mutant Drosophila heads because they lack the Dcr-2 RNase III activity (31). To test the involvement of this RNAi gene in the repression of gypsy, we introduced the dcr-2^L811fsX recessive mutant allele into the gypsy-containing restrictive background of the A237 strain. gypsy rasiRNAs could still be detected at the same level in both the wild-type and the mutant dcr-2 genotypes (Fig. 2A, lanes 4 and 5, respectively) showing that the Dcr-2 RNAi protein is not required for the production of these small RNAs. In order to check that the dcr-2 allele used in this experiment was actually lacking the Dcr-2 function, we tested it with an RNAi-based assay directed against the endogenous white gene. In a wild-type context, a single dose of the GMR-wIR transgene can almost completely repress the white gene, producing a yellowish eye phenotype (data not shown). In contrast, in homozygous dcr-2^L811fsX mutant flies, we obtained a red-eyed phenotype, indicating a defective RNAi pathway (see Fig. 6C).

In the RNAi silencing pathway, Dcr-2 is not only essential to dice dsRNAs into smaller siRNAs but also to assemble the silencing complex (43). That is why dcr-2 mutants are only partially rescued by the exogenous supply of siRNAs that should bypass the first deficient dicing step (31). To test whether Dcr-2 is similarly involved in the silencing of gypsy downstream of the rasiRNA production, we monitored the expression of the gypsy proviruses that were introduced into the mutant stock by the X restrictive chromosome of the A237 strain. Whole-mount in situ hybridization of an antisense gypsy riboprobe with either larval female gonads or adult ovaries revealed that these proviruses were still repressed in the dcr-2 mutant background (Fig. 5A2 and B2).

As reported previously, a gypsy-lacZ reporter is derepressed in the larval female gonads when introduced into a flam^R;piwi² genetic background (52). We checked that the piwi¹ and piwi³ mutant alleles also knocked down the repression of the endogenous gypsy proviruses in the same conditions (Fig. 5A4 and A6). In contrast, aub, the piwi paralog, is not required for gypsy silencing (Fig. 5A8 and B4).

Piwi, aub, and flam are not required for RNAi in vivo.

In the GMR-wIR RNAi assay described above, white was not derepressed in piwi¹ and piwi³ homozygous mutant flies (Fig. 6A and B), unlike in mutants for dcr-2, the typical RNAi gene (Fig. 6C). So, in the eye, Piwi is required for both cosuppression (39) and PcG-dependent repression (19) but not for hairpin-induced RNAi. Similarly, GMR-wIR could still repress white in aub^N11/aub^HN2 (Fig. 6D) and aub^N11/aub^QC42 eyes (data not shown), whereas these mutant genotypes are known to impair PcG-dependent repression in the eye (19), as well as RNAi induced by the injection of dsRNA into embryos (24).

To test the effect of flam on RNAi, we had to switch to another RNAi assay that could be performed in the follicle cells where flam is known to be expressed. The ES2 and ES5 constructs (Fig. 1D), known to be expressed in these somatic ovarian tissues (52), were used as RNAi targets. As previously reported, the ES2 transgene, which does not contain gypsy sequences, is not sensitive to flam-dependent repression (Fig. 7A and E), while ES5, which contains a 59-nt gypsy fragment, is repressed in the flam restrictive background (Fig. 7C and G). A lacZ hairpin was expressed in the somatic ovarian tissues by the UASt-IRlacZ transgene (23) under control of the Act5C-GAL4 driver. The hairpin could target the homologous lacZ region of the reporter constructs not only in the restrictive background (Fig. 7B) but also in the absence of any flam repressive activity (Fig. 7F and H). Thus, the Flamenco function that is involved in the RNA silencing of gypsy in the ovary is not required for RNAi in this tissue.

gypsy rasiRNAs appear to be the bona fide effectors of a RNA silencing mechanism directed against sense gypsy transcripts.

Expression of the Ste and roo repeated genetic elements is increased in RNA silencing mutant strains showing reduced accumulation of the corresponding rasiRNAs (5, 7, 62). Similarly, in the present study, we provide correlations between the silencing of another element, gypsy, and the presence of complementary rasiRNAs. First, the highest accumulation of these small RNAs was observed precisely where the strongest repression occurred, i.e., in the ovaries of the MG1 strong restrictive strain. Second, an asymmetry in rasiRNA strand polarity (in favor of antisense strands) was correlated with a preferential sensitivity of the (sense) complementary targets.

Thus, the fact that the yp3-lacZ heterologous transcription unit could be made sensitive to the repression when fused with the small “pol” fragment in the sense, but not antisense, orientation confirms previous hints (52) that the prevalent antisense gypsy rasiRNAs cannot target gypsy at the DNA but at the RNA level. Note, however, that this does not rule out a model of transcriptional silencing that would be targeted via the hybridization of rasiRNAs with nascent transcripts.

Genome integrity is maintained by similar rasiRNA-mediated silencing pathways operating either in the somatic or germinal lineages.

Most transposable elements have to be expressed in the germ line to increase their copy number by transposition. However, some endogenous retroviruses escape this germ line tissue specificity constraint because they can invade the germ line from the permissive follicle cells where they are expressed. Indeed, the amplification of gypsy and ZAM proviruses involves expression and assembly of particles in this somatic tissue, transfer to the germ line and integration in the progeny (9, 10). It is also known that transfer to the germ line (endogenization) of gypsy is highly efficient after infection (25, 58). The novel rasiRNA-mediated silencing phenomenon that we describe here definitely operates in these ovarian somatic tissues. By virtue of its tissue specificity, gypy silencing can therefore protect the Drosophila genome from both the expansion of endogenous proviruses and the endogenization process of viruses transmitted horizontally.

A similar RNA-silencing mechanism has just been shown to protect the Drosophila genome from the deleterious expression of repeated elements in the germ line (62). So, both the germinal and the somatic lineages appear to use very similar rasiRNA-mediated silencing mechanisms that display the following differences with the miRNA- and/or the siRNA-dependent silencing pathways. Unlike miRNAs, the 3′ ends of rasiRNAs and siRNAs have 3′-end modifications that make them resistant to the β-elimination reaction. RNAi involves the loading of siRNAs by the AGO2 slicer protein. In contrast, the AGO-like Piwi protein is required for gypsy silencing in the larval female gonads, while Piwi (as well as its paralog, the Aub protein) was reported to specifically interact with various rasiRNAs in the ovary (51, 62). Moreover, we show here that neither Piwi nor Aub are required for RNAi, at least in the eye, another imaginal somatic tissue.

Biogenesis of gypsy rasiRNAs.

The biogenesis of rasiRNAs also seems to be quite different from that of other effectors of RNA silencing. Like the roo and Su(Ste) rasiRNAs, the gypsy rasiRNAs are longer (24 to 30 nt) than miRNAs, siRNAs and any small RNA ever characterized biochemically as products of the three Drosophila RNase III enzymes (8, 33, 50). As expected, their accumulation was found to be independent of the RNAi-specific Dcr-2 RNase III activity. Moreover, the aforementioned strong asymmetry in strand polarity sheds further doubts about the existence of a dsRNA precursor.

The Su(Ste) rasiRNAs seem indeed to be exclusively of the antisense polarity (62). The complete absence of one strand is also typical of piRNAs, the small RNAs recently reported to interact with Piwi-like proteins in the testes of mammals (3, 18, 28). Concerning the roo retroelement, a few sense rasiRNAs were observed but, unlike antisense roo rasiRNAs, they were not affected in armi mutant ovaries, in which roo expression is increased, suggesting that they do not contribute to roo silencing (62). Similarly, a low level of sense gypsy small RNAs was detected in our Northern blots, but these RNAs were not able to silence the complementary target. So, the few roo and gypsy small sense RNAs might not be bona fide rasiRNAs and all Drosophila rasiRNAs would be exclusively of the antisense polarity. The mechanism of production (i.e., what precursor is used by what enzyme?) of such predominantly antisense rasiRNAs is nevertheless still elusive.

The rasiRNAs production is triggered by some pericentromeric gypsy provirus(es).

The pericentric flam region of the strongest restrictive MG1 strain was recombined into a permissive genome devoid of euchromatic proviruses, and the recombinant strain turned out to contain as many gypsy rasiRNAs as the parental MG1 strain. The results of this genetic recombination experiment indicate that the genetic determinant(s) responsible for the unusually high level of gypsy rasiRNAs in the MG1 strain map at the base of the X chromosome where flam is located. This provides additional evidence that flam is involved in the production and/or stabilization of gypsy rasiRNAs (52). However, since flam has not yet been molecularly characterized (46), we cannot choose between the two following hypotheses. (i) Putative gypsy provirus(es) inserted in this repetitive DNA-rich region of the X restrictive chromosomes might be expressed in such a way (i.e., aberrant transcription from flanking repetitive sequences?) as to trigger the production of rasiRNAs. This model is reminiscent of that of the 1A region, rich in telomeric associated sequences, apparently responsible for the trans-silencing, in the Drosophila female germ line, of any target homologous to elements inserted therein (48). (ii) Instead of gypsy provirus(es) inserted at the base of the X chromosome, flam might be a heterochromatic RNA silencing gene involved in the production or stabilization of rasiRNAs triggered by proviruses inserted elsewhere in the genome. Concerning other retroelements, it would be interesting to know whether flam is also able to control the accumulation of the corresponding rasiRNAs in the ovary. With regard to siRNAs, flam is probably not involved in their accumulation since we show here that hairpin-induced RNAi can be observed in the ovary independently of the flam genetic background.

Even in the second model, not all gypsy proviruses can drive the production of rasiRNAs. Indeed, the amount of rasiRNAs was not reduced in the recombinant strain, even though the permissive genome into which the strong MG1 flam restrictive allele was recombined is devoid of functional gypsy proviruses and only contains defective pericentromeric gypsy insertions. Thus, the many functional euchromatic proviruses present in the MG1 genome do not contribute significantly to the bulk of rasiRNAs. In this model, gypsy rasiRNAs would be triggered by some pericentromeric (not necessarily X-linked) defective provirus(es) shared by all Drosophila strains, including the permissive parental strain, under control of the flam restrictive function.

In both models, only pericentromeric endogenous provirus(es) would be able to trigger preprogrammed, RNA-mediated resistance of their host to the cognate functional retroviruses. Such an innate immunity system would compensate for the inability of retroviruses to trigger RNAi because of the lack of dsRNA during retroviral replication. Similarly, the repression by the host of many other transposable elements might also take advantage of the presence of defective copies of such elements in heterochromatin. These copies would be maintained not only because they are less deleterious in this gene-poor region of the genome but also because they would build up a repertoire of rasiRNA-producing sequences subsequently used by similar innate immunity systems to ensure genomic stability.

• 1.Alefelder, S., B. K. Patel, and F. Eckstein. 1998. Incorporation of terminal phosphorothioates into oligonucleotides. Nucleic Acids Res. 26:4983-4988. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 2.Ambros, V. 2004. The functions of animal microRNAs. Nature 431:350-355. [DOI] [PubMed] [Google Scholar]
• 3.Aravin, A., D. Gaidatzis, S. Pfeffer, M. Lagos-Quintana, P. Landgraf, N. Iovino, P. Morris, M. J. Brownstein, S. Kuramochi-Miyagawa, T. Nakano, M. Chien, J. J. Russo, J. Ju, R. Sheridan, C. Sander, M. Zavolan, and T. Tuschl. 2006. A novel class of small RNAs bind to MILI protein in mouse testes. Nature 442:203-207. [DOI] [PubMed] [Google Scholar]
• 4.Aravin, A., and T. Tuschl. 2005. Identification and characterization of small RNAs involved in RNA silencing. FEBS Lett. 579:5830. [DOI] [PubMed] [Google Scholar]
• 5.Aravin, A. A., M. S. Klenov, V. V. Vagin, F. Bantignies, G. Cavalli, and V. A. Gvozdev. 2004. Dissection of a natural RNA silencing process in the Drosophila melanogaster germ line. Mol. Cell. Biol. 24:6742-6750. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 6.Aravin, A. A., M. Lagos-Quintana, A. Yalcin, M. Zavolan, D. Marks, B. Snyder, T. Gaasterland, J. Meyer, and T. Tuschl. 2003. The small RNA profile during Drosophila melanogaster development. Dev. Cell 5:337-350. [DOI] [PubMed] [Google Scholar]
• 7.Aravin, A. A., N. M. Naumova, A. V. Tulin, V. V. Vagin, Y. M. Rozovsky, and V. A. Gvozdev. 2001. Double-stranded RNA-mediated silencing of genomic tandem repeats and transposable elements in the Drosophila melanogaster germline. Curr. Biol. 11:1017-1027. [DOI] [PubMed] [Google Scholar]
• 8.Bernstein, E., A. A. Caudy, S. M. Hammond, and G. J. Hannon. 2001. Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409:363-366. [DOI] [PubMed] [Google Scholar]
• 9.Brasset, E. E., A. R. Taddei, F. F. Arnaud, B. B. Faye, A. M. Fausto, M. M. Mazzini, F. F. Giorgi, and C. C. Vaury. 2006. Viral particles of the endogenous retrovirus ZAM from Drosophila melanogaster use a preexisting endosome/exosome pathway for transfer to the oocyte. Retrovirology 3:25. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 10.Chalvet, F., L. Teysset, C. Terzian, N. Prud'homme, P. Santamaria, A. Bucheton, and A. Pélisson. 1999. Proviral amplification of the gypsy endogenous retrovirus of Drosophila melanogaster involves env-independent invasion of the female germline. EMBO J. 18:2659-2669. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 11.Cogoni, C., and G. Macino. 1999. Gene silencing in Neurospora crassa requires a protein homologous to RNA-dependent RNA polymerase. Nature 399:166-169. [DOI] [PubMed] [Google Scholar]
• 12.Cox, D. N., A. Chao, J. Baker, L. Chang, D. Qiao, and H. Lin. 1998. A novel class of evolutionarily conserved genes defined by piwi are essential for stem cell self-renewal. Genes Dev. 12:3715-3727. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 13.Dalmay, T., A. Hamilton, S. Rudd, S. Angell, and D. C. Baulcombe. 2000. An RNA-dependent RNA polymerase gene in Arabidopsis is required for posttranscriptional gene silencing mediated by a transgene but not by a virus. Cell 101:543-553. [DOI] [PubMed] [Google Scholar]
• 14.Djikeng, A., H. Shi, C. Tschudi, and E. Ullu. 2001. RNA interference in Trypanosoma brucei: cloning of small interfering RNAs provides evidence for retroposon-derived 24-26-nucleotide RNAs. RNA 7:1522-1530. [PMC free article] [PubMed] [Google Scholar]
• 15.Ebhardt, H. A., E. P. Thi, M. B. Wang, and P. J. Unrau. 2005. Extensive 3′ modification of plant small RNAs is modulated by helper component-proteinase expression. Proc. Natl. Acad. Sci. USA 102:13398-13403. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 16.Fire, A., S. Xu, M. K. Montgomery, S. A. Kostas, S. E. Driver, and C. C. Mello. 1998. Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391:806-811. [DOI] [PubMed] [Google Scholar]
• 17.Gans, M., C. Audit, and M. Masson. 1975. Isolation and characterization of sex-linked female-sterile mutants in Drosophila melanogaster. Genetics 81:683-704. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 18.Girard, A., R. Sachidanandam, G. J. Hannon, and M. A. Carmell. 2006. A germline-specific class of small RNAs binds mammalian Piwi proteins. Nature 442:199-202. [DOI] [PubMed] [Google Scholar]
• 19.Grimaud, C., F. Bantignies, M. Pal-Bhadra, P. Ghana, U. Bhadra, and G. Cavalli. 2006. RNAi components are required for nuclear clustering of Polycomb group response elements. Cell 124:957-971. [DOI] [PubMed] [Google Scholar]
• 20.Hamilton, A. J., and D. C. Baulcombe. 1999. A species of small antisense RNA in posttranscriptional gene silencing in plants. Science 286:950-952. [DOI] [PubMed] [Google Scholar]
• 21.Ilyin, Y. V., V. G. Chmeliauskaite, and G. P. Georgiev. 1980. Double-stranded sequences in RNA of Drosophila melanogaster: relation to mobile dispersed genes. Nucleic Acids Res. 8:3439-3457. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 22.Kalmykova, A. I., M. S. Klenov, and V. A. Gvozdev. 2005. Argonaute protein PIWI controls mobilization of retrotransposons in the Drosophila male germline. Nucleic Acids Res. 33:2052-2059. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 23.Kennerdell, J. R., and R. W. Carthew. 2000. Heritable gene silencing in Drosophila using double-stranded RNA. Nat. Biotechnol. 18:896-898. [DOI] [PubMed] [Google Scholar]
• 24.Kennerdell, J. R., S. Yamaguchi, and R. W. Carthew. 2002. RNAi is activated during drosophila oocyte maturation in a manner dependent on aubergine and spindle-E. Genes Dev. 16:1884-1889. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 25.Kim, A., C. Terzian, P. Santamaria, A. Pélisson, N. Prud'homme, and A. Bucheton. 1994. Retroviruses in invertebrates: the gypsy retrotransposon is apparently an infectious retrovirus of Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 91:1285-1289. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 26.Kogan, G. L., A. V. Tulin, A. A. Aravin, Y. A. Abramov, A. I. Kalmykova, C. Maisonhaute, and V. A. Gvozdev. 2003. The GATE retrotransposon in Drosophila melanogaster: mobility in heterochromatin and aspects of its expression in germline tissues. Mol. Genet. Genomics 269:234-242. [DOI] [PubMed] [Google Scholar]
• 27.Lambertsson, A., S. Andersson, and T. Johansson. 1989. Cloning and characterization of variable-sized gypsy mobile elements in Drosophila melanogaster. Plasmid 22:22-31. [DOI] [PubMed] [Google Scholar]
• 28.Lau, N. C., A. G. Seto, J. Kim, S. Kuramochi-Miyagawa, T. Nakano, D. P. Bartel, and R. E. Kingston. 2006. Characterization of the piRNA complex from rat testes. Science 313:363-367. [DOI] [PubMed] [Google Scholar]
• 29.Lee, R. C., R. L. Feinbaum, and V. Ambros. 1993. The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell 75:843-854. [DOI] [PubMed] [Google Scholar]
• 30.Lee, Y. S., and R. W. Carthew. 2003. Making a better RNAi vector for Drosophila: use of intron spacers. Methods 30:322-329. [DOI] [PubMed] [Google Scholar]
• 31.Lee, Y. S., K. Nakahara, J. W. Pham, K. Kim, Z. He, E. J. Sontheimer, and R. W. Carthew. 2004. Distinct roles for Drosophila Dicer-1 and Dicer-2 in the siRNA/miRNA silencing pathways. Cell 117:69. [DOI] [PubMed] [Google Scholar]
• 32.Li, J., Z. Yang, B. Yu, J. Liu, and X. Chen. 2005. Methylation protects miRNAs and siRNAs from a 3′-end uridylation activity in Arabidopsis. Curr. Biol. 15:1501-1507. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 33.Liu, Q., T. A. Rand, S. Kalidas, F. Du, H. E. Kim, D. P. Smith, and X. Wang. 2003. R2D2, a bridge between the initiation and effector steps of the Drosophila RNAi pathway. Science 301:1921-1925. [DOI] [PubMed] [Google Scholar]
• 34.Martens, H., J. Novotny, J. Oberstrass, T. L. Steck, P. Postlethwait, and W. Nellen. 2002. RNAi in Dictyostelium: the role of RNA-directed RNA polymerases and double-stranded RNase. Mol. Biol. Cell 13:445-453. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 35.Mejlumian, L., A. Pélisson, A. Bucheton, and C. Terzian. 2002. Comparative and functional studies of Drosophila species invasion by the gypsy endogenous retrovirus. Genetics 160:201-209. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 36.Mochizuki, K., N. A. Fine, T. Fujisawa, and M. A. Gorovsky. 2002. Analysis of a piwi-related gene implicates small RNAs in genome rearrangement in Tetrahymena. Cell 110:689-699. [DOI] [PubMed] [Google Scholar]
• 37.Molnar, A., T. Csorba, L. Lakatos, E. Varallyay, C. Lacomme, and J. Burgyan. 2005. Plant virus-derived small interfering RNAs originate predominantly from highly structured single-stranded viral RNAs. J. Virol. 79:7812-7818. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 38.Mourrain, P., C. Beclin, T. Elmayan, F. Feuerbach, C. Godon, J. B. Morel, D. Jouette, A. M. Lacombe, S. Nikic, N. Picault, K. Remoue, M. Sanial, T. A. Vo, and H. Vaucheret. 2000. Arabidopsis SGS2 and SGS3 genes are required for posttranscriptional gene silencing and natural virus resistance. Cell 101:533-542. [DOI] [PubMed] [Google Scholar]
• 39.Pal-Bhadra, M., U. Bhadra, and J. A. Birchler. 2002. RNAi related mechanisms affect both transcriptional and posttranscriptional transgene silencing in Drosophila. Mol. Cell 9:315-327. [DOI] [PubMed] [Google Scholar]
• 40.Pasquinelli, A. E., B. J. Reinhart, F. Slack, M. Q. Martindale, M. I. Kuroda, B. Maller, D. C. Hayward, E. E. Ball, B. Degnan, P. Muller, J. Spring, A. Srinivasan, M. Fishman, J. Finnerty, J. Corbo, M. Levine, P. Leahy, E. Davidson, and G. Ruvkun. 2000. Conservation of the sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature 408:86-89. [DOI] [PubMed] [Google Scholar]
• 41.Pélisson, A., S. U. Song, N. Prud'homme, P. A. Smith, A. Bucheton, and V. G. Corces. 1994. Gypsy transposition correlates with the production of a retroviral envelope-like protein under the tissue-specific control of the Drosophila flamenco gene. EMBO J. 13:4401-4411. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 42.Pélisson, A., L. Teysset, F. Chalvet, A. Kim, N. Prud'homme, C. Terzian, and A. Bucheton. 1997. About the origin of retroviruses and the co-evolution of the gypsy retrovirus with the Drosophila flamenco host gene. Genetica 100:29-37. [PubMed] [Google Scholar]
• 43.Pham, J. W., J. L. Pellino, Y. S. Lee, R. W. Carthew, and E. J. Sontheimer. 2004. A Dicer-2-dependent 80s complex cleaves targeted mRNAs during RNAi in Drosophila. Cell 117:83-94. [DOI] [PubMed] [Google Scholar]
• 44.Prud'homme, N., M. Gans, M. Masson, C. Terzian, and A. Bucheton. 1995. flamenco, a gene controlling the gypsy retrovirus of Drosophila melanogaster. Genetics 139:697-711. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 45.Reinhart, B. J., and D. P. Bartel. 2002. Small RNAs correspond to centromere heterochromatic repeats. Science 297:1831. [DOI] [PubMed] [Google Scholar]
• 46.Robert, V., N. Prud'homme, A. Kim, A. Bucheton, and A. Pélisson. 2001. Characterization of the flamenco region of the Drosophila melanogaster genome. Genetics 158:701-713. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 47.Ronaldson, E., and M. Bownes. 1995. Two independent cis-acting elements regulate the sex- and tissue-specific expression of yp3 in Drosophila melanogaster. Genet. Res. 66:9-17. [DOI] [PubMed] [Google Scholar]
• 48.Ronsseray, S., T. Josse, A. Boivin, and D. Anxolabehere. 2003. Telomeric transgenes and trans-silencing in Drosophila. Genetica 117:327-335. [DOI] [PubMed] [Google Scholar]
• 49.Rubin, G. M., and A. C. Spradling. 1982. Genetic transformation of Drosophila with transposable element vectors. Science 218:348-353. [DOI] [PubMed] [Google Scholar]
• 50.Saito, K., A. Ishizuka, H. Siomi, and M. C. Siomi. 2005. Processing of pre-microRNAs by the Dicer-1-Loquacious complex in Drosophila cells. PLoS Biol. 3:e235. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 51.Saito, K., K. M. Nishida, T. Mori, Y. Kawamura, K. Miyoshi, T. Nagami, H. Siomi, and M. C. Siomi. 2006. Specific association of Piwi with rasiRNAs derived from retrotransposon and heterochromatic regions in the Drosophila genome. Genes Dev. 20:2214-2222. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 52.Sarot, E., G. Payen-Groschene, A. Bucheton, and A. Pélisson. 2004. Evidence for a piwi-dependent RNA silencing of the gypsy endogenous retrovirus by the Drosophila melanogaster flamenco gene. Genetics 166:1313-1321. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 53.Savitsky, M., D. Kwon, P. Georgiev, A. Kalmykova, and V. Gvozdev. 2006. Telomere elongation is under the control of the RNAi-based mechanism in the Drosophila germline. Genes Dev. 20:345-354. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 54.Schupbach, T., and E. Wieschaus. 1991. Female sterile mutations on the second chromosome of Drosophila melanogaster. II. Mutations blocking oogenesis or altering egg morphology. Genetics 129:1119-1136. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 55.Schwarz, D. S., G. Hutvagner, T. Du, Z. Xu, N. Aronin, and P. D. Zamore. 2003. Asymmetry in the assembly of the RNAi enzyme complex. Cell 115:199-208. [DOI] [PubMed] [Google Scholar]
• 56.Sijen, T., and R. H. A. Plasterk. 2003. Transposon silencing in the Caenorhabditis elegans germ line by natural RNAi. Nature 426:310-314. [DOI] [PubMed] [Google Scholar]
• 57.Smardon, A., J. M. Spoerke, S. C. Stacey, M. E. Klein, N. Mackin, and E. M. Maine. 2000. EGO-1 is related to RNA-directed RNA polymerase and functions in germ-line development and RNA interference in Caenorhabditis elegans. Curr. Biol. 10:169-178. [DOI] [PubMed] [Google Scholar]
• 58.Song, S. U., T. Gerasimova, M. Kurkulos, J. D. Boeke, and V. G. Corces. 1994. An env-like protein encoded by a Drosophila retroelement: evidence that gypsy is an infectious retrovirus. Genes Dev. 8:2046-2057. [DOI] [PubMed] [Google Scholar]
• 59.Sugiyama, T., H. Cam, A. Verdel, D. Moazed, and S. I. S. Grewal. 2005. From The Cover: RNA-dependent RNA polymerase is an essential component of a self-enforcing loop coupling heterochromatin assembly to siRNA production. Proc. Natl. Acad. Sci. USA 102:152-157. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 60.Taverna, S. D., R. S. Coyne, and C. D. Allis. 2002. Methylation of histone H3 at lysine 9 targets programmed DNA elimination in Tetrahymena. Cell 110:701-711. [DOI] [PubMed] [Google Scholar]
• 61.Ullu, E., H. D. Lujan, and C. Tschudi. 2005. Small sense and antisense RNAs derived from a telomeric retroposon family in Giardia intestinalis. Eukaryot. Cell 4:1155-1157. [DOI] [PMC free article] [PubMed] [Google Scholar]
• 62.Vagin, V. V., A. Sigova, C. Li, H. Seitz, V. Gvozdev, and P. D. Zamore. 2006. A distinct small RNA pathway silences selfish genetic elements in the germline. Science 313:320-324. [DOI] [PubMed] [Google Scholar]
• 63.Vaury, C., A. Bucheton, and A. Pélisson. 1989. The beta heterochromatic sequences flanking the I elements are themselves defective transposable elements. Chromosoma 98:215-224. [DOI] [PubMed] [Google Scholar]
• 64.Voinnet, O. 2005. Induction and suppression of RNA silencing: insights from viral infections. Nat. Rev. Genet. 6:206-220. [DOI] [PubMed] [Google Scholar]
• 65.Volpe, T. A., C. Kidner, I. M. Hall, G. Teng, S. I. Grewal, and R. A. Martienssen. 2002. Regulation of heterochromatic silencing and histone H3 lysine-9 methylation by RNAi. Science 297:1833-1837. [DOI] [PubMed] [Google Scholar]
• 66.Yu, B., Z. Yang, J. Li, S. Minakhina, M. Yang, R. W. Padgett, R. Steward, and X. Chen. 2005. Methylation as a crucial step in plant microRNA biogenesis. Science 307:932-935. [DOI] [PMC free article] [PubMed] [Google Scholar]