Abstract
Gypsy is an endogenous retrovirus of Drosophila melanogaster. Phylogenetic studies suggest that occasional horizontal transfer events of gypsy occur between Drosophila species. gypsy possesses infective properties associated with the products of the envelope gene that might be at the origin of these interspecies transfers. We report here the existence of DNA sequences putatively encoding full-length Env proteins in the genomes of Drosophila species other than D. melanogaster, suggesting that potentially infective gypsy copies able to spread between sexually isolated species can occur. The ability of gypsy to invade the genome of a new species is conditioned by its capacity to be expressed in the naive genome. The genetic basis for the regulation of gypsy activity in D. melanogaster is now well known, and it has been assigned to an X-linked gene called flamenco. We established an experimental simulation of the invasion of the D. melanogaster genome by gypsy elements derived from other Drosophila species, which demonstrates that these non-D. melanogaster gypsy elements escape the repression exerted by the D. melanogaster flamenco gene.
RETROELEMENTS form a large and diverse class of mobile elements that can be found in all eukaryotes. The structural organization of retroelements reflects their common mechanism of replication: They propagate by reverse transcription of RNA intermediates and integrate their genetic information into the genome of a host cell. Their presence in all organisms suggests that they are very ancient. Several lines of evidence, such as their regulation and putative domestication by the host, indicate that the transposable elements and particularly retroelements (REs) can co-evolve with the host genome. Comparison of the phylogenies of retroelements and of their hosts provides a reliable background for the study of their evolutionary history. In particular, the mechanisms responsible for their distribution and their impact on the host genome can be explored.
Gypsy (DmeGypV) is an endogenous retrovirus of Drosophila melanogaster, showing a genomic structure remarkably similar to the proviral form of vertebrate retroviruses (Figure 1) and exhibiting infectious properties in particular conditions (Kimet al. 1994; Bucheton 1995). Sequences similar to DmeGypV are largely distributed among Drosophila species. Previous phylogenetic studies suggested that horizontal transfer events of gypsy elements between Drosophila species can occur occasionally (Mizrokhi and Mazo 1991; Alberola and de Frutos 1996; Terzianet al. 2000). However, the mechanisms by which horizontal transfers occur remain obscure. The cross-species horizontal transfer events are well documented for DNA transposons, such as P and mariner (Clarket al. 1994; Robertson 1997) or REs such as copia (Jordanet al. 1999). For these noninfectious elements, a vector-mediated transfer mechanism was proposed to explain transfer between individuals. By contrast, horizontal transfers of infectious elements like gypsy do not require any vector in principle. It has been shown in experimental conditions that gypsy can be horizontally transmitted between D. melanogaster individuals (Kimet al. 1994; Songet al. 1994) and from D. melanogaster cultured cells to D. hydei cultured cells (Syominet al. 2001). The infectious properties of gypsy may result from the expression of the envelope (env) gene as illustrated in Drosophila cell culture using a Moloney murine leukemia virus-based retroviral vector pseudotyped with the gypsy envelope (Teyssetet al. 1998).
However, the ability of gypsy to invade the genome of a new species is conditioned by its capacity to be expressed in this species. The genetic basis for the regulation of gypsy activity is now well known, and it has been assigned to an X-linked gene called flamenco (Prud'hommeet al. 1995; Robertet al. 2001). The flamenco alleles belong to one of two categories: permissive, which allows the expression of gypsy in the somatic follicle cells surrounding the female germline, and restrictive, which represses this expression. The multiplication of vertically transmitted gypsy proviruses was shown to require the env-independent transfer of gypsy products from the permissive somatic cells to the female germline (Chalvetet al. 1999).
We analyzed two regions of gypsy, known to be critical for the tissue-specific tropism of retroviruses: (i) the 5′ regulatory region containing both the 5′ long terminal repeat (LTR) and the untranslated leader region (ULR), which drives the expression of gypsy and is controlled by flamenco (Pelissonet al. 1994), and (ii) the ORF3 coding for the envelope protein. We report here the existence of DNA sequences putatively encoding full-length and functional Env proteins in the genome of the closely related species D. simulans, D. erecta, D. orena, D. teissieri, D. yakuba, and the more distant species D. virilis and D. subobscura. The presence of potentially functional gypsy env genes not only in the genome of D. melanogaster but also in the genomes of distantly related species supports the hypothesis that there are potentially infectious gypsy copies able to spread between sexually isolated species. An experimental simulation of the invasion of the D. melanogaster genome by gypsy elements derived from other Drosophila species demonstrates that these gypsy elements are likely to be resistant to the control by flamenco.
MATERIALS AND METHODS
Fly stocks: w1118(R) is the laboratory white reference stock described in Lindsley and Zimm (1992); it contains the restrictive flam1118(R) allele. The same mutation was introduced by recombination into two different stocks, OR(P) and Rev(R), to give, respectively, the wOR(P) and wRev(R) stocks. (Chalvetet al. 1999 and our unpublished results). The Rev(R) stock is the one referred to as RevI (Dessetet al. 1999).
All non-melanogaster Drosophila species were kindly provided by Françoise Lemeunier (Centre National de la Recherche Scientifique, Gif-sur-Yvette, France). The OR(721)/FM3 D. melanogaster strain (Pelissonet al. 1994) is known to contain multiple copies of the active gypsy element and was used as a positive control in protein truncation tests (PTT; see below).
PCR primers and conditions: The primers used for env amplification from the melanogaster subgroup species were designed from the gypsy sequence (M12927). The upstream primer was positioned at the beginning of ORF3 and permitted to start transcription at the underlined ATG: 5′ GGATCCTAATACGACTCACTATAGGAACAGACCACCATGTTCATACCCTTGGTAG 3′. The upstream primers used for env amplification from D. subobscura (5′ GGATCCTAATACGACTCACTATAGGAACAGACCACCATGTTTGTACTCACTTTACT 3′) and D. virilis (5′ GGATCCTAATACGACTCACTATAGGAACAGACCACCATGTTCGTCCATTTCAAATT 3′) were based on the DsuGypV and DviGypV published sequences (GenBank accession nos. X72390 and M38438, respectively) and contained the trinucleotide ATG (underlined) upstream in the env sequence in order to create a start codon, which is normally generated by RNA splicing (Pelissonet al. 1994; Alberola and de Frutos 1996). All upstream primers were modified at their 5′ end with the T7 promoter sequence to generate PCR products suitable for subsequent PTT analysis.
The three downstream primers were positioned into the 3′ LTR of DmeGypV (5′ GGCGATAGCGATTTGATTGTAAA 3′) and DviGypV (5′ TTATCTTGGGTAAGTTGCGTTGA 3′), and at the 3′ end of the DsuGypV env gene (5′ TCAGAATGATGACCGTCC 3′) to amplify complete env sequences.
Reaction volumes were 50 μl and contained 50 ng template DNA and 2.5 units Taq DNA polymerase (Promega, Madison, WI). Reaction parameters were as follows: 3 cycles of 93° for 45 sec, 45° for 45 sec, 72° for 3 min; 25 cycles of 93° for 30 sec, 61° (56° for D. virilis) for 30 sec, 72° for 2 min, followed by 72° for 10 min.
A primer pair was designed (upper 5′ TCGCATGCCAACTACATT 3′, lower 5′ AGGCTCGTCTTCTCCTTA 3′) according to regions conserved between DmeGypV, DsuGypV, and DviGypV Env sequences to amplify and sequence the env gene derived from gypsy elements of D. simulans.
PTT: A T7-coupled reticulocyte lysate system (Promega) was used for the PTT analysis according to the protocol recommended by the manufacturer with minor modifications. The reactions were carried out in a volume of 20 μl, and half of the recommended volume of [35S]methionine was used. Purified full-length env gene PCR products as well as cloned env PCR product were transcribed and translated in a TNT T7-coupled reticulocyte lysate system at 30° for 1 hr. The translation products were separated by discontinuous SDS-PAGE through a 15% separating gel with Tris-glycine buffer. The signals were detected by autoradiography.
Cloning of the gypsy regulatory elements from various Drosophila species: pDsugyp and pDvigyp were constructed from PCR-amplified gypsy LTR-ULR regions derived from D. virilis and D. subobscura, fused transcriptionally to the lacZ reporter gene, and subcloned into the transformation vector pW7. Primers were designed according to the D. subobscura and D. virilis gypsy published sequences.
The gypsy LTR-ULR from D. subobscura was isolated by PCR amplification using 50 ng genomic DNA as template and 2.5 units Pfu DNA polymerase in the following conditions: 93°, 30 sec; 57°, 45 sec; 72°, 2 min for one cycle; annealing at 61°, 30 sec for 25 cycles; and 72°, 10 min as final extension. The upstream primer was 5′ GGAATTCCAGTTAAGAACTAAGTACATAAGTTATTCCC 3′ and the downstream primer was 5′ CGGGATCCCGCATTGGCTTATGGTTGGCAC 3′.
The DviGypV LTR-ULR was obtained by PCR amplification using as template 5 ng of cloned full-length DviGypV (supplied by M. Evgen'ev) and 2.5 units Pfu DNA polymerase in the following conditions: 93°, 30 sec; 55°, 30 sec; 72°, 1 min 30 sec for 25 cycles. The upstream primer was 5′ GGAATTCCAGTTAACAACTAAGCATAAATATATTGCCC 3′, and the downstream primer was 5′ CGGGATCCCGCTCATTGGCTTATGGTTGGC 3′. The underlined nucleotides correspond to the EcoRI- and BamHI-cleavable extensions used in the cloning strategy.
P-element-mediated germline transformation and genetic crosses: All transgenes were recovered as colored-eyed w+ individuals after P-mediated transformation (Spradling 1986) into the wOR(P) permissive stock that is devoid of functional gypsy element. Transgenic males were backcrossed to w1118(R)/w1118(R) and wRev(R)/wRev(R) restrictive females, on one hand, and to wOR(P)/wOR(P) females, on the other hand, to keep the transgenes in restrictive and permissive backgrounds, respectively. The same kind of crosses were performed during this experiment with the pgyp68.1 transgenic line, which contains the DmeGypV LTR-ULR (Figure 1) fused transcriptionally to lacZ (named pgyp in this article; Pelissonet al. 1994).
Histochemistry and quantitative measurements of β-galactosidase activity: Histochemical staining for β-galactosidase and quantitative measurements of β-galactosidase activity in the ovaries of the pgyp, pDsugyp, and pDvigyp transgenic flies were done according to Pelisson et al. (1994).
Sequencing: PCR products of interest were cloned into the pPCR-Script Amp SK(+) vector (Stratagene, La Jolla, CA) and both strands were sequenced using vector-specific T7 and T3 primers.
Nucleotide and amino acid sequence analyses: Multiple alignments of LTR-ULR nucleotide sequences were performed with the DIALIGN2 program (Morgenstern 1999). The MEME program was used in addition to visually adjust the initial alignments (Bailey and Elkan 1995). Multiple alignments of amino acid sequences were performed with CLUSTALX (Thompsonet al. 1997). Alignments were displayed using Bio-Edit (Hall 1999). Jukes-Cantor (JC) nucleotide distances were estimated using Distances from the University of Wisconsin Genetics Computer Group package (GCG, version 10.0). Diverge (GCG v10.0) was used to estimate the numbers of synonymous (Ks) and nonsynonymous (Ka) substitutions per site between two sequences coding for proteins. PlotSimilarity (GCG v10.0) was used to plot the average similarity between two aligned sequences at each position in the alignment, using a window of comparison of 50 nucleotides. The window of comparison was moved along all sequences, one position at a time, and the average similarity over the entire window was plotted at the middle position of the window. The average similarity across the entire alignment is plotted as a dotted line.
Structure of DmeGypV. (A) Organization of DmeGypV with the location of 5′ LTR and 3′ LTR, the ULR, and the genes corresponding to the three open reading frames. (B) Schematic drawing of the unspliced genomic and spliced subgenomic RNAs.
RESULTS
Sequences homologous to gypsy from D. melanogaster (DmeGypV) are widely distributed among Drosophila species. Complete gypsy elements have been previously cloned and sequenced from D. subobscura (DsuGypV) and D. virilis (DviGypV) species (Mizrokhi and Mazo 1991; Alberola and de Frutos 1996). Previous sequence comparisons suggested that the genetic organization of gypsy is similar in these three species and that only D. melanogaster contains elements possessing a complete functional env gene (Alberola and de Frutos 1996). However, as the genomes of these species contain many defective copies of gypsy in addition to complete elements, the complete and potentially functional proviruses might have escaped these analyses. For this reason, we searched in several Drosophila species, using a PCR-based approach, for the presence of not-yet-described env genes putatively coding full-length proteins.
The coding capacities for Env proteins are conserved in gypsy elements from various Drosophila species: The PTT, a mutation-detection method generally used to scan for premature termination mutations (Den Dunnen and Van Ommen 1999), was chosen to analyze the coding capacity of the gypsy env gene present in the genomes of different Drosophila species. gypsy env PCR amplifications were carried out with genomic DNAs extracted from six species of the melanogaster subgroup and the more distantly related D. virilis and D. subobscura species. In addition to blank reactions, several other controls were performed to guard against cross-contamination of genomic DNA. In this series of controls, primers amplifying the internal spacer regions of the multicopy Drosophila rDNA genes were used as previously described (Terzianet al. 2000). Each sample resulted in PCR products of a unique size distinct from that of D. melanogaster (data not shown), indicating that no cross-species contamination occurred.
A major 1.8-kb product and a few minor bands were observed in genomic DNA amplifications from the species of the melanogaster subgroup except D. simulans for which no product was obtained. The expected D. melanogaster PCR product size is 1.8 kb, corresponding to the 1.4-kb complete env gene plus 0.4 kb of the 3′ LTR. The specificity of the amplified fragments was checked by Southern blot hybridization at high stringency, using a radioactively marked internal fragment of DmeGypV env as a probe, revealing only the 1.8-kb fragment out of several PCR products obtained with the species of the melanogaster subgroup (data not shown). The 1.8-kb major PCR product from each species was subsequently gel purified and used for PTT analyses.
The 1.4- and 1.8-kb fragments observed with D. subobscura and D. virilis did not hybridize to the D. melanogaster probe in the Southern blot experiments, presumably because of the divergence between DmeGypV, DsuGypV, and DviGypV. In these two cases, we gel purified the 1.4- and 1.8-kb PCR products from DsuGypV and DviGypV, respectively.
The major protein products obtained in the PTT assay with D. teissieri, D. yakuba, D. erecta, and D. orena have molecular weights corresponding to the expected 54-kD Env protein of D. melanogaster (Figure 2). In the case of D. virilis and D. subobscura, we obtained PTT products of ~54 and 53.5 kD, respectively (Figure 2). These results show that the genomes of these species contain at least one copy of gypsy putatively encoding a complete envelope protein. One should note that protein products with smaller sizes are also present in this assay. They might correspond to the previously described truncated Env putative proteins of 52.8 and 42.7 kD for D. virilis and D. subobscura, respectively (Alberola and de Frutos 1996).
The PCR products of D. virilis, D. subobscura, D. erecta, and D. teissieri used in the PTT assays were cloned into the pCRScript cloning vector, and two positive clones for each species were randomly selected and sequenced. Although it was concluded from the published sequence of DsuGypV that D. subobscura does not contain gypsy elements encoding functional Env products (Alberola and de Frutos 1996), the sequences of the env gene that we obtained indicate that this species does contain such elements. This confirms the results of the PTT assays. The full-length coding capacity of the env sequences we obtained is restored due to the lack of a single T nucleotide present at position 5545 in the X72390 sequence eliminating a stop codon, and the addition of an A nucleotide at position 6639.
gypsy Env proteins encoded by different Drosophila species as revealed by PTT. D.m., D. melanogaster; D.t., D. teissieri; D.y., D. yakuba; D.e., D. erecta; D.o., D. orena; D.s., D. subobscura; D.v., D. virilis. C, control with no DNA added at the beginning of the reaction. Except for D.v. where a cloned env PCR product was transcribed and translated, proteins were directly in vitro transcribed and translated from purified env PCR products.
Similarly, the PCR products of D. virilis encode a complete Env protein. The deletion in our D. virilis env sequences of a C nucleotide present at position 5848 in the M38438 sequence eliminates a downstream stop codon present in the previously published sequence (Mizrokhi and Mazo 1991) and restores the N-terminal amino acid sequence highly similar to that of DmeGypV.
One of the two gypsy env sequences obtained from the D. teissieri genome encodes a full-length Env protein, whereas the other contains a deletion of 19 nucleotides that generates a frameshift mutation resulting in a stop codon.
Both gypsy env sequences of D. erecta show high levels of similarity, and none has a complete protein-coding capacity. Comparison with the DmeGypV env sequence reveals multiple point mutations and several significant deletions that create stop codons.
The env-based phylogeny is congruent with the int-based phylogeny of gypsy: We performed a phylogenetic analysis of gypsy based on the env gene sequences. A previous phylogenetic analysis based on the integrase gene showed that the distribution of gypsy among the eight melanogaster subgroup species does not follow the phylogeny of the host species, defining two main lineages: GypA and GypB (Terzianet al. 2000). The GypA lineage is restricted to D. simulans, D. sechellia, and D. mauritiana, whereas the presence of the GypB lineage is limited to D. melanogaster, D. teissieri, D. yakuba, D. erecta, and D. orena.
Since we could not amplify a gypsy homologous env gene from the D. simulans genome using the PTT assay primers, we carried out a PCR using a couple of primers conserved in DmeGypV, DsuGypV, and DviGypV Env sequences. A single fragment was obtained. It was cloned in pCRScript, and two independent clones were sequenced. The results indicate that D. simulans contains gypsy elements putatively encoding full-length Env proteins. As the percentages of nucleotide identity between both sequences from the same species were very high (>98%), only one sequence from each species was used to construct a multiple alignment with CLUSTALX (Figure 3). This alignment shows that the putative cellular endopeptidase cleavage site, the two putative N-linked glycosylation sites, the six cysteine residues, and the transmembrane domain described in Pelisson et al. (1994) were found at exactly the same positions in all sequences. Moreover, the rates of Ks are much higher than the rates of Ka(5 ≤ Ks/Ka ≤ 9), suggesting that the env sequences have evolved under purifying selection. Altogether, these data indicate the existence of a selective pressure for the conservation of the Env function in the gypsy elements from various Drosophila species.
We estimated the JC nucleotide distances between the aligned env sequences and compared them graphically with the int JC distances (Figure 4). The data demonstrate that there is a strong correlation between env and int distances (r = 0.87), indicating that the env and int phylogenetic trees are congruent. Moreover, the int:env distance ratios are very close to 1 (Figure 4), suggesting that int and env evolve at the same rate in the same gypsy genome, most likely because of the lack of recombination between int and env. These results provide significant information supporting the phylogeny of gypsy that we previously established (Terzianet al. 2000). Since the int/env-based phylogenies are different from those of the species, we propose that horizontal transmission of gypsy occurred between Drosophila species.
The expression of gypsy in D. subobscura and D. virilis is not repressed by the flamenco gene in D. melanogaster: We developed an experimental model of the invasion of the genome of D. melanogaster by a gypsy element from another Drosophila species. A series of experiments was carried out to study the ability of alien gypsy regulatory sequences to drive the expression of the bacterial lacZ reporter gene in D. melanogaster.
It has been previously shown that gypsy in D. melanogaster is controlled by the flamenco host gene (Prud'hommeet al. 1995) and that the 5′ LTR and ULR of DmeGypV contain cis-acting sequences responsible for the expression pattern of gypsy in the function of flamenco permissive and restrictive alleles (Pelissonet al. 1994). The LTR-ULR sequences of gypsy elements from D. subobscura and D. virilis were transcriptionally fused to lacZ and inserted into the pW7 transformation vector. The resulting constructs, pDsugyp and pDvigyp, were introduced into the genome of the wOR(P) by P-mediated transformation and the resulting transgenic, as well as the pgyp transgenic, individuals were crossed to w1118(R)/ w1118(R) and wRev(R)/wRev(R) restrictive females or to wOR(P)/wOR(P) permissive females in order to keep the transgenes in restrictive and permissive backgrounds (see materials and methods). Several transgenic lines were obtained for each construct and at least three independent transformed lines were analyzed to rule out possible chromosomal position effects of the insertion sites of the transgenes. All transgenic lines bearing the same construct gave identical results.
Alignment of partial amino acid of Env sequences from DviGypV, DsuGypV, DmeGypV, DteGypV, and DsiGypV. Identical residues have a black background, and similar residues are shaded. The first residue of the partial DmeGypV Env sequence corresponds to position 21 of the 483-amino-acid full-length Env sequence, whereas the last residue corresponds to position 454. Open arrowhead, cysteine residue; solid arrowhead, N-linked putative glycosylation site; arrow, putative endopeptidase cleavage site; box, transmembrane domain.
lacZ expression was detected in follicle cells of vitellogenic stages of oogenesis in the ovaries of flamenco permissive pgyp transgenic flies, as previously shown by Pelisson et al. (1994). Both squamous (nurse-cell-associated) and columnar (oocyte-associated) follicle cells were stained in permissive ovaries. The highest β-galactosidase activity was observed in the centripetal follicle cells. lacZ expression was strongly repressed in flamenco restrictive females. This is in agreement with the previous results reported by Pelisson et al. (1994).
Correlation between JC nucleotide pairwise distances among env and int sequences. me, DmeGypV; te, DteGypV; si, DsiGypV; su, DsuGypV; vi, DviGypV. The line marks the 1:1 ratio that would be expected if int and env genes had the same divergence rate.
Histochemical staining of dissected ovaries of permissive transgenic females homozygous for the pDsugyp and pDvigyp constructs show that reporter gene expression remains restricted to the somatic follicular epithelial cells surrounding egg chambers (Figure 5). However, some significant differences were observed: First, the β-galactosidase activity promoted by the LTR-ULR regions of DsuGypV and DviGypV is higher than that of the LTR-ULR of DmeGypV; second, the pDsugyp and pDvigyp transgenes are expressed at all stages of oogenesis; third, the expression of pDsugyp and pDvigyp is not much repressed by restrictive alleles of flamenco, namely, w1118(R) and wRev(R), in contrast to pgyp, which is completely repressed in the presence of these alleles. Quantitative measurements of β-galactosidase activity in the ovaries of transgenic flies confirmed the histochemical observations. High levels of β-galactosidase activity were observed in the ovaries of permissive and restrictive transgenic flies for pDsugyp and pDvigyp while pgyp was expressed only in permissive females (Figure 6). A significant repression was observed for pDsugyp in the repressive wRev(R)/wRev(R) genetic background. However, the level of β-galactosidase activity in this case is still much higher than that observed with pgyp in the same background. Hence, DviGypV and DsuGypV are both able to express themselves at a high level in D. melanogaster ovaries, whichever flamenco alleles are present in the females.
The extent of divergence between the gypsy LTR-ULR sequences varies along the nucleotide sequence: In an attempt to understand the evolutionary forces that shaped the regulatory sequences of gypsy leading to the expression profiles that we observed, we sequenced the LTR-ULR from pDsugyp and pDvigyp and compared these sequences with the sequence of the LTR-ULR of DmeGypV previously published (M12927). A multiple sequence alignment was generated using the DIALIGN segment-to-segment approach, which is particularly appropriate for detecting local similarities, and some editing was done manually according to the MEME results (Figure 7).
lacZ expression pattern of pgyp (D. melanogaster), pDsugyp, and pDvigyp homozygous transgenic flies in permissive [P: wOR(P)/wOR(P)] or restrictive [R1: w1118(R)/w1118(R)] ovaries. Age of females was 2–3 days.
Many differences are observed between the gypsy LTR-ULR sequences. They are not equally distributed along the LTR-ULR regions. The divergences among these sequences result from point mutations and large insertions and deletions (indels): In particular, the DsuGypV LTR region contains several insertions compared to the other two sequences.
We tried to characterize the differences in the nucleotide sequences that might be responsible for the differences in the patterns of expression observed previously among pgyp, pDsugyp, and pDvigyp in transgenic flies. Because pDsugyp and pDvigyp transgenic flies share a similar pattern of expression different from that of pgyp, we determined whether they also share similarities in the regions that diverge from DmeGypV. This was done by plotting DviGypV vs. DsuGypV, DviGypV vs. DmeGypV, and DmeGypV vs. DsuGypV divergences along the sequence (Figure 8) using PlotSimilarity. Distance values were calculated for a window of 50 nucleotides after removal of all gaps from the alignments. The plots showed clearly that the ULR region allows us to discriminate DmeGypV from DviGypV and DsuGypV. We found one domain where the extent of divergence between DviGypV and DsuGypV is minimal whereas the extent of divergence between DviGypV and DmeGypV or DmeGypV and DsuGypV is maximal (Figure 8). This 30-bp sequence lies within the ULR region, upstream of the first Su(Hw)-binding domain (Figure 7). It is also worthy of note that the similarity scores between the gypsy LTR-ULR sequences change suddenly at the U3/R boundary (Figure 8). The highest level of similarity observed in the U3 region concerns DmeGypV and DviGypV, whereas the highest similarity score observed in the ULR region concerns DsuGypV and DviGypV. This difference in the degree of similarity between the U3 and ULR regions suggests that DviGypV results from recombination events between DmeGypV and DsuGypV.
β-Galactosidase activity in ovaries of transgenic females homozygous for pDvigyp, pDsugyp, and pgyp, respectively. Except for pgyp, each value is the average of β-gal activity from three independent transgenic lines and the standard deviation is given. Concerning pgyp, each value is the average of β-gal activity from three independent protein extracts from the same transgenic line. Data are expressed as 1 OD per hour and per microgram of total protein. P, wOR(P)/wOR(P); R1, w1118(R)/w1118(R); R2, wRev(R)/wRev(R).
DISCUSSION
The Envelope proteins can be a major determinant of horizontal gypsy transfers: An increasing amount of experimental data indicates that many classes of transposable elements have been transferred horizontally between species (Flavell 1999). We previously reported that the int sequences of gypsy elements from D. subobscura and D. virilis, on one hand, and D. melanogaster and D. teissieri, on the other hand, are too close together to have diverged from a vertically inherited ancestor (Terzianet al. 2000). The phylogenetic analysis that we report here shows that the env-based and int-based phylogenies are congruent and that these two genes evolved at the same rate. These results also suggest that these two genes have evolved as parts of the same gypsy genomes during their horizontal and vertical transfers. Altogether, these data support our hypothesis that horizontal transfers have played a crucial role in the evolutionary history of these gypsy genomes (Terzianet al. 2000).
Alignment of the LTR (A) and ULR (B) sequences of DmeGypV, DviGypV, and DsuGypV. (A) The R region, which lies between the U3 and U5 regions, is shaded. (B) The binding sites for Su(Hw) are shaded (Parkhurstet al. 1988), and the region that discriminates DmeGypV from DviGypV and DsuGypV (see Figure 8) is boxed.
The mechanisms by which horizontal transfers of noninfectious elements could occur remain obscure and probably involve diverse vectors (Houcket al. 1991). Polytropic retroviruses are known to be capable of infecting new host species by horizontal transfer due to their infectious properties provided by the envelope protein. The gypsy endogenous retrovirus of D. melanogaster encodes an Env protein responsible for its infectious properties (Kimet al. 1994; Songet al. 1994; Teyssetet al. 1998). Therefore, the gypsy endogenous retrovirus can potentially jump from one individual to another without the need of a vector, increasing the probability of transfer compared to noninfectious retrotransposons. It has been proposed that horizontal transfer is a strategy that transposons use to escape the host-mediated mechanisms repressing transposition (Flavell 1999; Jordanet al. 1999). Since the mechanisms repressing gypsy activity, such as the repression of DmeGypV by the flamenco gene, appear to be very efficient, the maintenance of gypsy in Drosophila species might be ensured by horizontal transmissions. Our present results show that the ability to encode full-length Env proteins has been conserved in different species, supporting the hypothesis of the existence of selective forces that favor horizontal transfers between individuals from the same species or from different species. Moreover, the excess of synonymous to nonsynonymous substitutions indicates that purifying selection is acting on Env function. Preservation of Env functionality could then represent an advantage for gypsy maintenance in Drosophila species. The gypsy Env protein thus might be a major component of the mechanisms involved in the evolution of this endogenous retrovirus.
We cannot exclude that the gypsy env gene was instead co-opted by the host genome for some cellular functions and that the env genes described in this work are not part of full-length active gypsy copies. For instance, recent results suggest that the HERV-W Env protein might have a physiological role during pregnancy and placenta formation whereas the HERV-W gag and pol genes are defective (Blondet al. 2000). However, the fact that the gypsy int and env genes have the same evolutionary history of a complex pattern of horizontal and vertical transfers and are both under positive selective pressures, in addition to the fact that int sequences were issued from complete gypsy copies (Terzianet al. 2000), strongly suggest that both genes are members of full-length gypsy elements.
Sequence similarity between DmeGypV and DviGypV (solid line), DmeGypV and DsuGypV (shaded line), and DviGypV and DsuGypV (dotted line) using PlotSimilarity (see materials and methods). The x-axis indicates the nucleotide positions along the alignment (gaps were removed from the alignment). The U3, R, U5, and ULR regions are shown.
DsuGypV and DviGypV might invade flamenco restrictive D. melanogaster populations: Gypsy proviruses of the present-day populations of D. melanogaster (DmeGypV) are efficiently repressed by the restrictive alleles of the flamenco gene. Although the molecular mechanism of this repression is yet unknown, it appears that this regulation acts primarily on the accumulation of the transcripts of DmeGypV in the follicle cells of the ovaries (Pelissonet al. 1994). We show here that, in contrast to DmeGypV, DsuGypV and DviGypV are not repressed by the restrictive flamenco alleles and that the difference in this behavior lies in the LTR-ULR regulatory sequences of gypsy. Unless flamenco can downregulate them at a post-transcriptional level, DsuGypV and DviGypV should therefore be able to transpose in D. melanogaster. We should note that there is no experimental evidence that flamenco controls DmeGypV post-transcriptionally. The LTRs of DmeGypV, DsuGypV, and DviGypV all contain cis-acting sequences driving the expression of follicle cells. The follicle cells are assumed to be the somatic source of the soma-toward-germline transfer responsible for the DmeGypV amplification (Chalvetet al. 1999). This suggests that the various gypsy elements likely share a common mechanism of transposition.
A pairwise detailed comparison of the three aligned sequences revealed a stretch of 30 nucleotides that is shared by DviGypV and DsuGypV and is strongly divergent in DmeGypV. This sequence was previously described as part of a negative regulatory domain of DmeGypV transcription under the conditions of transient expression in Drosophila cultured cells (Mazoet al. 1989). We do not know if this is a coincidental correlation or if this divergence of sequence is responsible for the differences in lacZ expression driven by pgyp, pDsugyp, and pDvigyp in D. melanogaster transgenic restrictive females. Indeed, we cannot exclude the hypothesis that indels present in the LTR-ULR of DmeGypV and absent in the two other sequences are the major factors responsible for the differences of expression of the elements in the three species.
These data raise the question of the status of a naive genome and its potential invasion by retroelements. One simple hypothesis is that a naive genome is a priori permissive for a retroelement because acquisition of mechanisms of resistance have a fitness cost. Our results suggest that an “alien” gypsy element could propagate into a naive host genome. Indeed, flamenco seems to be a highly specific host defense system that could not control gypsy elements from other Drosophila species. It is probably one of the less “costly” mechanisms for D. melanogaster, but its inefficiency in the case of invasion by DsuGypV or DviGypV would make it necessary for the host to select new mechanisms to avoid the deleterious effect of gypsy activity. Several examples (reviewed in Boeke and Stoye 1997) indicate that the host/retroelement “arms race” results in highly specific interactions. We do not yet know the flamenco gene at the molecular level, but it will be very useful to determine the evolutionary history of this gene in Drosophila species in order to understand the relationships between endogenous retroviruses and their hosts.
Acknowledgments
We thank Dr. Michael Evgen'ev for generously providing the DviGypV clone. This work was supported by grants from the Association pour la Recherche sur le Cancer and the Fondation pour la Recherche Médicale (FRM). L.M. was the recipient of fellowships from the French Government and the FRM.
Footnotes
-
Sequence data from this article have been deposited with the EMBL/GenBank Data Libraries under accession nos. AJ308090–AJ308096.
-
Communicating editor: M. Veuille
- Received May 14, 2001.
- Accepted October 19, 2001.
- Copyright © 2002 by the Genetics Society of America
