Expression of Drosophila virilis Retroelements and Role of Small RNAs in Their Intrastrain Transposition

Transposition of two retroelements (Ulysses and Penelope) mobilized in the course of hybrid dysgenesis in Drosophila virilis has been investigated by in situ hybridization on polytene chromosomes in two D. virilis strains of different cytotypes routinely used to get dysgenic progeny. The analysis has been repeatedly performed over the last two decades, and has revealed transpositions of Penelope in one of the strains, while, in the other strain, the LTR-containing element Ulysses was found to be transpositionally active. The gypsy retroelement, which has been previously shown to be transpositionally inactive in D. virilis strains, was also included in the analysis. Whole mount is situ hybridization with the ovaries revealed different subcellular distribution of the transposable elements transcripts in the strains studied. Ulysses transpositions occur only in the strain where antisense piRNAs homologous to this TE are virtually absent and the ping-pong amplification loop apparently does not take place. On the other hand small RNAs homologous to Penelope found in the other strain, belong predominantly to the siRNA category (21nt), and consist of sense and antisense species observed in approximately equal proportion. The number of Penelope copies in the latter strain has significantly increased during the last decades, probably because Penelope-derived siRNAs are not maternally inherited, while the low level of Penelope-piRNAs, which are faithfully transmitted from mother to the embryo, is not sufficient to silence this element completely. Therefore, we speculate that intrastrain transposition of the three retroelements studied is controlled predominantly at the post-transcriptional level.


Introduction
Transposable elements (TEs) are repetitive sequences capable of moving in genomes under certain conditions, and they are widely observed in practically all organisms studied so far. The diversity of TEs and the degree to which they burden eukaryotic genomes are highly variable. In mammals, including humans, mobile genetic elements constitute up to 50% of the genome [1], while only 15-20% of the comparatively small Drosophila genome is composed of TEs [2]. Different classes of transposons, such as LTR-containing retroelements, LINEs and DNA transposons, are also represented to different degrees in the genomes of various organisms. Host organisms employ multiple strategies to silence TEs and viruses to prevent them from amplifying in the genome, because the vast majority of parasite insertions are likely to be deleterious and impose a fitness cost on the rest of the genome [3,4]. Recent data accumulated from Ceanorharbditis elegans and Drosophila, strongly suggest that RNA interference represents one of the most efficient host processes for silencing transcription and uncontrolled movement of parasite DNA [5,6,7]. Even though eukaryotic genomes have developed multiple systems for silencing TEs, certain families of TEs sometimes go out of control and are able to amplify and jump throughout the chromosomes [8]. The hybrid dysgenesis (HD) syndrome, described in Drosophila melanogaster and Drosophila virilis, represents such a case, where multiple transpositions of TEs lead to harmful consequences [9,10].
In D. melanogaster the HD syndrome is usually observed in the progeny of interstrain crosses when the female parent does not carry active copies of a certain TE (P, I or hobo), while the male parent carries multiple copies of a given element. Briefly, in D. melanogaster the dysgenic traits in the F1 progeny from a dysgenic cross usually include high levels of sterility, gonadal atrophy, occurrence of multiple visible and chromosomal mutations, and other genetic abnormalities. Although in D. virilis we observed virtually the same abnormalities, HD syndrome in this species is unusual in the fact that several transposable elements belonging not only to different families but also to different classes of TE are mobilized by the dysgenic crosses [10,11,12]. In our earlier studies, we showed that in D. virilis, similar to D. melanogaster, there are strains of three cytotypes, namely, neutral, M-like and P-like strains, depending upon their roles in HD [11]. In D. melanogaster strains of M-cytotype do not contain functional P-elements and produce partially sterile progeny when crossed with males from Pstrains carrying multiples copies of full-size P-elements while neutral strains do not produce significant proportion of sterile progeny when crossed either with M-like or P-like strains [9]. In D. virilis strains named by analogy with D. melanogaster ''M-like strains'', including the wild-type strain 9 used in the present study, usually contain only heterochromatic, highly diverged copies of Penelope retroelements. Furthermore, such diverged copies of Penelope are located in such strains mainly in the pericentromeric heterochromatin [13]. These strains produce high levels of gonadal sterility and other manifestations of HD when crossed with males of strain 160, which represents the only strong P-like strain described in D. virilis so far and contains multiple copies of Penelope probably playing an important role in HD [10]. In situ hybridization on polytene chromosomes and Southern blot analysis revealed mobilization of several unrelated TEs in the progeny of dysgenic crosses. These elements include Helena, Paris, Tv1, Telemac, Ulysses and Penelope [12,14]. Among these, Ulysses which represents a typical retroelement with LTRs of 2 kb in size and two ORFs, was the first element described in D. virilis and subsequently found in several visible mutations, including white, obtained in the progeny of dysgenic crosses [10,14]. Furthermore, this element was found at the breakpoints of inversions detected in the progeny of dysgenic crosses and, hence, it was implicated in the formation of aberrations never before found in D. virilis [15]. In contrast to Ulysses, another well studied LTR-containing retroelement gypsy, previously described in D. virilis (gypsyDv) [16], was never found in mutations in the progeny of dysgenic crosses [12]. It has been shown by different methods that multiple active copies of Penelope are present in strain 160, while strain 9 does not carry full-size Penelope copies in the euchromatic chromosome arms [10,17]. Highly diverged and apparently ancient copies of Penelope, termed ''Omega'' (V), located mostly in the heterochromatic chromocenter, were, however, detected and investigated in both strains studied [13]. In situ hybridization with polytene chromosomes and Southern blotting analysis showed that contrary to Penelope, full-size Ulysses copies are found in all D. virilis strains studied so far, with an average of 10-15 copies per strain [18].
There is molecular and genetic evidence suggesting that the TE ''Penelope'' plays an important role in D. virilis HD [10,17]. The Penelope retroelement does not belong to one of the previously well studied classes of TE, but rather represents its own superfamily characterized by the presence of a reverse transcriptase (more closely related to telomerases than the those of other retrotransposons) and a very unusual endonuclease containing the GIY-YIG domain [13]. Penelope-like elements (''PLEs'') have been described in recent years in various animals from rotifers to fish and reptiles [19,20,21].
In our previous studies, the injection of Penelope-containing constructs into the embryos of a D. virilis strain 9 lacking active Penelope resulted in multiple mutations in the progeny. It was shown that almost half of all visible mutations isolated in these experiments were due to insertions of Ulysses [10], which, contrary to Penelope, has nearly symmetrical distribution in the parental strains [18].
Recently, we have monitored the biogenesis of small RNAs homologous to various D. virilis transposons and measured the transmission levels of corresponding siRNAs and piRNAs in various inter-strain crosses. Using P-like strain 160 and a few neutral D. virilis strains that contain multiple full-size and potentially functional Penelope copies, however, we detected no obvious correlation between dysgenic traits and maternally deposited Penelope-derived piRNA levels [22]. Therefore, we sought to expand these studies in order to reveal correlations between the levels of naturally occurring transposition in D. virilis laboratory strains and RNA production and/or the biogenesis of the TE-derived small RNAs in question.
Herein, we demonstrate asymmetric transposition of Penelope and Ulysses in the laboratory strains of D. virilis without performing dysgenic crosses. By RNA whole-mount in situ hybridization a different subcellular strain specific localization of the TEs transcripts was revealed. Furthermore, we show that processing of Penelope and Ulysses transcripts lead to the formation of different classes of small RNAs that may be implicated in transposition control of these TEs. For comparison, we have also investigated expression of gypsyDv, which is based upon previous studies lost transposition activity in D. virilis and is not mobilized by dysgenic crosses in this species [12].

Results
Analysis of transpositions of Penelope and Ulysses in two D. virilis strains by in situ hybridization on polytene chromosomes In the course of investigations performed over the past 20 years we detected asymmetric transpositions of Penelope and Ulysses in D. virilis strain 160 and strain 9. Using in situ hybridization with salivary gland polytene chromosomes, we failed to detect any transpositions of Ulysses in strain 160, which preserved stable pattern of the transposon distribution in the chromosomes. On the other hand, the number of Penelope copies in this strain increased from 37 to 53 since 1991. We detected 27 new sites of Penelope hybridization and the disappearance of 11 previously observed sites in the chromosomes of the strain 160. Interestingly, nearly half of the new sites-12-were found in the chromosome 2 ( Table 1). It is noteworthy that we failed to detect Penelope hybridization to chromosome 6 (microchromosome), not only in strain 160, but also in all other D. virilis strains studied in our laboratory [18].
While we did not find new sites for Ulysses in strain 160, we did reveal active transposition of this TE in M-like strain 9. It is noteworthy that all the chromosomes of strain 9 were involved in the transposition process by Ulysses (Table 1). It is necessary to note that even though transpositions of retroelements do not occur by a ''cut and paste'' mechanism, in strain 9 we detected six new sites of insertion in parallel with the disappearance of four ''old'' sites detected in 1991. Such a phenomenon was described in D. melanogaster, when certain copies of the retroelement gypsy or Ielement disappeared without a trace from a few cytological locations [23,24].
Characteristically, the presumably inactive gypsyDv taken for comparison exhibited practically identical preferentially heterochromatic distribution in the chromosomes of the D. virilis strains studied, which was preserved without any change during the whole period of observation (Table 1). It is noteworthy that a vast majority of the same, probably heterochromatic, cytological sites contain gypsyDv in all other laboratory and geographic D. virilis strains studied so far (data not shown), which implies that this TE has probably lost its transposition ability in this species. It is noteworthy that both Ulysses and gypsyDv are often found in nearcentromeric sites (i.e. 19D, 29F, 39F, 49F and 59F) while Penelope with one exception are not found in these presumably heterochromatic regions (Table 1).
Since we detected different transposition behavior of Penelope and Ulysses depending upon the strain, it was of significant interest to monitor the transcription of various TEs, including these retroelements, in the strains compared.

Transcription analyses of various TEs and transcripts subcellular localization
Previously it was shown by Northern analysis that Penelope transcription is significantly induced in the ovaries of dysgenic hybrids between females of strain 9 and males of strain 160 [10]. However, transcription of other TEs has not been analyzed either in parental strains 9 and 160 or in their hybrids. Therefore, we explored a transcription of several transposons by a set of complementary methods. In order to detect a presence of transcripts, semiquantitative RT-PCR was performed on cDNA from both strains with specific primers to the transposons. It was also of interest to investigate the transposon transcription separately in female gonads and carcasses. RT-PCR products were detected for all elements studied both in ovaries and in the carcasses, but the presence of some transcripts was shown to be strain specific ( Figure 1A). Thus, Penelope and Helena transcripts were amplified only in strain 160, what is not unexpected, because functional copies of these particular TEs are not present in strain 9 [12]. On the other hand, Ulysses and gypsyDv are transcribed in ovaries and in the carcasses of both strains 9 and 160 ( Figure 1A).
Surprisingly, quantitative RT-PCR (qRT-PCR) experiments revealed a comparatively low but significant level of Paris transcription in strain 9, while full-size copies of this TE were previously reported either to be lacking or represented by only one euchromatic copy in this strain [12,17]. The transcripts detected may either emerge from this single copy or result from read through transcription of Paris heterochromatic diverged sequences. qRT-PCR experiments demonstrated that the transcription levels of Ulysses and gypsyDv in strain 9 are similar to those in strain 160 ( Figure 1B).
Northern blot analysis corroborates the qRT-PCR data, and demonstrates approximately the same level of Ulysses and gypsyDv transcription in the ovaries of both strains. Moreover RNAs homologous to these TEs are represented in Northern blots by identical bands (presumably splicing forms) in both strains ( Figure 1C).
It is noteworthy that Ulysses probe did not reveal any significant hybridization with D. melanogaster RNA because the representatives of Ulysses family are absent in the genome of this species. D. virilis gypsy probe hybridized with D. melanogaster RNA and revealed fullsize transcript (7 kb) and several additional bands probably resulted from splicing ( Figure 1C).
At the next stage, in order to monitor subcellular localization of TE transcripts in the ovaries, we performed RNA in situ hybridizations with Penelope, Ulysses and gypsyDv sense and antisense probes. Localization of Penelope sense transcripts is shown in Figure 2A, B. It is evident that strong hybridization of Penelope in strain 160 is restricted to the cytoplasm of nurse cells, and, to a lesser extent, to the nuclei of nurse cells, while the ovaries of strain 9 do not contain any label, as expected. We observed only very weak hybridization with a probe revealing antisense Penelope transcript in strain 160 (data not shown). In contrast to Penelope, a probe revealing sense transcript of Ulysses detected multiple signals in the nuclei of nurse cells in both strains in the form of discrete putative nascent transcripts as well as single strong signals (one per nucleus) probably representing RNA processing sites (foci) ( Figure 2E). This hybridization pattern resembles I-element localization in D. melanogaster ovaries [25]. Interestingly, in strain 9 the labeling seems to be more  pronounced in the cytoplasm of nurse cells in comparison with strain 160 (Figure 2C, D). On the other hand, in approximately 90% of strain 160 ovarioles, quite distinct large foci close to the nuclear membrane are seen ( Figure 2E). These structures, where in D. melanogaster accumulation of I-element and other TEs transcripts takes place, probably represent the sites of processing of various TEs RNAs leading to their retention in the nuclei [25]. Characteristically, the foci are never seen in the ovarioles of strain 9, which correlates with the active transposition of Ulysses in strain 9. A significant signal is also observed in the cytoplasm of centripetal and squamous follicle cells in the ovaries of both strains ( Figure 2F). Whole mount detection of sense gypsyDv transcripts revealed rather weak hybridization in the cytoplasm of nurse cells in both strains ( Figure 2G), while the somatic follicle cells were practically free of label with a few specific exceptions (Fig. 2H, I). The pattern observed in D. virilis is strikingly different from subcellular localization of gypsy transcripts in ovaries of D. melanogaster permissive strains [26].
Overall, the analysis of the transcription of various TEs, including Penelope, Ulysses and gypsyDv performed by different complementary techniques in the strains compared, revealed characteristic differences in the TE's RNA levels and transcript localization in the cells of the ovaries. In order to further investigate the fate of TEs transcripts, we decided to perform Penelope and Ulysses produce strikingly different sets of of transposon-homologous small RNAs in gonads of strains 9 and 160 Since in contrast to Penelope we failed to reveal a clear-cut correlation between the expression levels of Ulysses and gypsyDv and their transposition behavior in both strains we investigated the biogenesis of small RNAs homologous to these TEs.
In order to evaluate the possible role of small RNAs, such as piand siRNAs, in controlling the detected transpositions of Penelope and Ulysses, we explored small RNA libraries obtained from the ovaries and testes of strains 9 and 160 [22]. The analysis of small RNA populations homologous to Penelope and Ulysses in these strains revealed drastic differences in the processing of their transcripts. While Ulysses-derived small RNAs in both strains are represented by predominantly piRNAs 23-29 nt in length (92% of total reads), Penelope-homologous small RNAs are virtually absent in strain 9, and in strain 160 they mostly belong to the siRNA species 21 nt in length (61% of total reads) [22]. It is well known that the phenomenon of RNA interference is based on homologydependent gene silencing, and since we detected full-length sense transcripts of Ulysses ( Figure 1C), it was logical to expect that Ulysses-derived small RNAs will have an antisense orientation. However, contrary to the expectation, up to 99% of the ovarian piRNAs homologous to Ulysses have sense orientation and apparently arise from processing transcripts originating from active euchromatic copies of the element [22]. Surprisingly, a high level of Ulysses-derived antisense piRNAs was found in the small RNAs libraries obtained from the testes of both strains. Moreover, the piRNAs are predominantly homologous to the sequences of the TE's huge LTRs ( Figure 3A, C). gypsyDv-piRNAs are represented predominantly by antisense population in both the ovaries and the testes. In this case the LTRs of this TE are not enriched in piRNAs, while general pattern of piRNA localization along the body of this TE is very similar in ovaries and testes but is different depending on a strain ( Figure 3B, D).
There is evidence suggesting an important role of maternally inherited small RNAs in repression of Iand P-elements or Penelope in HD syndrome [27,28]. However, in the latter case it was not shown which class of small RNAs is responsible for the effect. Here, we monitored separately the maternal deposition of siRNA and piRNA of Penelope in the strains compared. The experiments showed that although Penelope siRNAs are present at high level in the ovaries of both strains, this class of small RNA is practically absent in 0-2 hour embryos ( Figure 4A, B). On the contrary, Penelope-derived piRNAs as expected are effectively transmitted to the progeny ( Figure 4C, D). It is necessary to mention that Penelope transcripts are detected in ovaries of strain 160 and maternally inherited by the early embryos (0-2 hours) [29].
It was of significant interest to compare the data accumulated in the course of in situ hybridization studies on polytene chromosomes with the results of mapping sequenced small RNAs homologous to Penelope and Ulysses. The comparative analysis has shown that large part of Ulysses-derived piRNAs map to the D. virilis genome 17-21 times [22]. Similarly, a significant proportion of Penelope-derived piRNA sequences map 2, 22 and 39 times while we did not detect  sequences found in the genome 3-18 times ( Figure 5). Probably, rearranged or full-length transcribed Penelope elements as well as ancient diverged copies (Omega) located in heterochromatic clusters may serve as the source of the multiple piRNAs homologous to the element in strain 160.

Discussion
In our experiments, we showed that Penelope and Ulysses are able to asymmetrically transpose in D. virilis parental strains even without performing dysgenic crosses that drastically increase the frequency of unrelated TE transpositions in this species. It is necessary to mention that transpositions of various TEs were detected in laboratory strains of D. melanogaster and sometimes the intrastrain mobility of certain TEs correlates with their expression level [24,30].
In our study, we observed similar levels of Ulysses transcription in both D. virilis strains while this TE is transpositionally active only in strain 9. Furthermore, although gypsyDv full-size transcripts are present in both strains, this element is amazingly stable in terms of transposition.
Whole-mount in situ hybridization experiments demonstrated different subcellular localization of TEs transcripts, which in general correlates with their transposition behavior in the strains studied. Contrary to D. melanogaster permissive strains, gypsyDv is weakly expressed only in the nurse cell cytoplasm and specific groups of follicular cells in both strains what correlates with its stability in the genome of D. virilis. Abundant Penelope transcripts were observed in the cytoplasm of nurse cells of strain 160, where this TE is probably transpositionally active. Furthermore, while Ulysses expression in the form of strong nascent transcripts has been detected in the nurse cells of both strains, only in strain 160 well developed foci (presumptive sites of Ulysses RNA processing) were seen correlating with high stability of Ulysses localization in this strain.
The transposition behavior of the three studied TEs apparently depends upon many factors, and is controlled at the posttransciptional level. The retrotransposon gypsyDv does not transpose, apparently due to accumulation of mutations disturbing env or other domains of this TE [31]. However, other authors exploring the PCR technique and PTT analysis concluded that the genome of D. virilis may contain at least one copy of gypsyDv putatively encoding a complete envelope protein [32] and, hence, we can not exclude that gypsyDv may be active in some strains of D. virilis. On the other hand, Ulysses is transpositionally active in strain 9, probably because in this strain only sense Ulysses-piRNAs are present, and the ping-pong cycle is blocked. Along these lines, the ratio of Ulysses-derived primary to secondary piRNAs also differs strongly in the strains studied and the low level of Ulysses secondary piRNAs in strain 9 may reflect the absence of ping-pong amplification loop necessary for Ulysses silencing [22]. Interestingly, in testes a high level of Ulysses-derived antisense piRNAs was found, and, surprisingly, this fraction is predominantly homologous to LTRs of this TE ( Figure 3A, C). This phenomenon might resemble the different functional activities of Argonaute group proteins in the testes and ovaries [33]. Alternatively, LTRhomologous antisense piRNAs may be coming from a solo Ulysses LTRs located in a piRNA-producing cluster functioning only in testes.
Despite the fact, that Penelope is one of the most abundant transposon in the genome of D. virilis with more than 50 copies in strain 160, we did not detect transpositions of the element to chromosome 6 (microchromosome). This may result from either Penelope transposition preferences or from the recently described peculiar chromatin structure of chromosome 6 in D.virilis [34,35]. It is also tempting to speculate that such transposition preferences in avoiding of heterochromatic regions and perhaps piRNA loci might be a reason for a continuing transposition activity of this element in strain 160 of D. virilis as well as in transgenic D. melanogaster strains transformed with full-size Penelope [36].
Comparing the general localization of hybridization sites specific for the studied TEs in the D. virilis genome enables us to conclude that the observed distribution is not random, and there are sites where two or three TEs are found. Probably these sites (e.g., 19D and 49F) represent ''hot spots'' or ''nests'' of transposons previously described both in the D. virilis and D. melanogaster genomes [18,37]. In particular, we do not rule out that at least one of such hot spots, i.e. 49F that coincides with the coordinates of cluster #3 [22], might serve as a putative flamenco piRNA locus in D. virilis genome that produces the most abundant fraction of sense oriented transposon-homologous piRNAs in D. virilis genome.
In the present investigation we did not monitor intrastrain transposition of other TEs mobilized by dysgenic crosses which may represent another interesting avenue of future research, because there are at least two other elements, Paris and Helena, which are abundant in strain 160, but absent or found in small numbers in strain 9 [12,17].
Recently, based upon the analysis of maternal inheritance of small RNAs in various systems of D. melanogaster HD, it was suggested that piRNAs have an important role in the regulation of the syndrome by homology-dependent TE silencing [25,27]. In D. virilis Penelope-derived small RNAs were also implicated in HD syndrome regulation [22,28]. Moreover, we speculate that Penelope is transpositionally active in strain 160, because, for some reason in this particular quite exceptional strain, small RNAs are represented predominantly by siRNAs. The detected siRNAs probably originated from double stranded stem regions of Penelope transcripts containing the same regions in sense and antisense orientation (long inverted repeats). Although siRNAs represent the major class of small RNAs homologous to Penelope, it is evident that this class of small RNA is not efficiently transported from mother to embryo and probably does not play any role in Penelope silencing in the germ line [22]. Intriguingly, in whole mount experiments we were not able to detect Penelope transcription to somatic follicular cells of the ovaries of strain 160, and thereby a subcellular origin of the Penelope-siRNAs remains to be investigated.
Collectively, our studies show that two TEs mobilized in dysgenic crosses, namely Penelope and Ulysses, are drastically different, both in transposition behavior in the parental strains, subcellelar compartmentalization of the transposon transcripts and their processing into small RNAs. It is necessary to mention that we do not rule out the possibility that the causes of occasional transpositions of TEs taking place in the parental strains might be completely different from the causes of much greater mobilization observed in the progeny of dysgenic crosses between these strains.
Although, the investigation of transcription levels and cellular distribution of the transcripts do not provide in all cases a straightforward explanation for the observed interstrain specific transpositions of several transposons, the obtained results should be taken into account in further attempts to explain the molecular mechanisms underlying the behavior of various retroviruses (latent infection) and transposons in laboratory and geographical strains, as well as to shed light on D. virilis HD syndrome and the role of co-mobilization of unrelated TEs in this process.

D. virilis strains
D. virilis strain 160 and strain 9 were obtained from the Stock Center of the Institute of Developmental Biology, Moscow. Strain 160 represents an old laboratory strain carrying recessive mutations in all autosomes (b, gp, cd, pe, gl) while wild-type strain 9 was collected about thirty years ago in Batumi (Georgia, Caucasus).
All flies were reared at 25uC on standard resin-sugar-yeast-agar medium containing propionic acid and methylparaben as mold inhibitors.

Cytological analysis
Larvae were grown at 18uC on medium supplemented with live yeast solution for 2 days before dissection. Salivary glands from third instar larvae were dissected in 45% acetic acid and squashed. Procedures and labeling of DNA probes for in situ hybridization were as described [18].

Northern blotting
Total RNA was isolated from ovaries of 7-10 days old flies using TRIzol reagent (Sigma). PolyA-RNA was purified using OLIGO-TEX mRNA mini kit (QIAGENE) and fractionated as described [38]. High Range RNA Ladder (Promega) was used as marker. 32P-labeled single stranded RNA probes revealing sense transcripts were synthesized using MAXIscript T7 kit (Ambion).
qRT-PCR was done with 3 biological replicates and carried out using 5x SYBR Green PCR Master Mix (Evrogen) in accordance with the manufacturer's protocol. Quantification was normalized to the endogenous rp49 and calculation of relative expression levels was done using the 2 -ddCt method.