Distribution of the copia transposable element in the repleta group of Drosophila

Summary - The occurrence of the copia transposable element in 18 species of the repleta group of Drosophila has been studied using the Southern technique. The homologous sequence of copia was detected, either with radioactive or non-radioactive nucleic acid detection systems, as a pattern of multiple bands in species of the mercatorum and mulleri subgroups. Nevertheless, this sequence was not detected in the hydei subgroup. The intraspecific polymorphism in the pattern of bands indicates that this sequence is likely to be mobile. Some of the results could suggest the existence of restriction polymorphism of the copia homologous sequence in D koepferae populations. The partial sequencing of 2 independent clones isolated from D buzzatii clearly establishes that these elements are related and are likely to be the same.


INTRODUCTION
Copia retrotransposon from D melanogaster is 1 of the best known retroviral type elements in the genus Drosophila (Mount and Rubin, 1985;Emori et at, 1985). Retrotransposons are recognized by structural and functional similarities to integrated retroviruses. They are bound by long terminal repeats (LTRs) at their termini and contain open reading frames resembling gag and pol genes from retroviruses (Finnegan 1989; see Bingham and Zachar, 1989 for review). There are 2 distinct lineages of retrotransposons based on the order of the gene complement and reverse transcriptase (RT) amino-acid sequence relationships Eickbush, 1988, 1990;McClure, 1992). More closely related to retroviruses and sharing a common ancestor with caulimoviruses, is a group including several retrotransposons of D melanogaster (gypsy, 17.6, 412, 297, micropia), S cerevisae (Ty3) and B mori (Mag). On the other hand, copia-like elements have a gene order which is different from all other retroid family members in that the integrase domains are located at the amino terminal of the RT domain. Retrotransposons from distantly related taxonomic groups such as D melanogaster (copia and 1731), S cerevisae (Tyl and Ty2), A thaliana (Tal) and N tabacum (Tntl) are clustered in this latter group Eickbush 1990, McClure 1992).
The presence of the copia element has been reported in the major Drosophila radiations, suggesting an ancient origin of this component in the genome (Martin et at, 1983, Stacey et at, 1986. Nevertheless, the distribution of copia is discontinuous within the different radiations analysed. In the virilis-repleta radiation, hybridizing sequences have been found in the mulleri and mercatorum subgroups (repleta group), but no detectable hybridization was observed in the hydei subgroup (repleta group) or in any of the representatives of the virilis group. However, even in closely related species, the relative abundance of the copia element can be highly variable. In the melanogaster subgroup the number of dispersed copies of copia ranges from 60 in D melanogaster (Finnegan et at, 1978) to 0 in D yakuba and D erecta (Dowsett, 1983). Similar differences were observed in the obscura group, with more than 30 copies of the homologous sequence in D pse!doobscura and no detectable copies in D subobscura (Martin et at, 1983).
A preliminary approach to the molecular evolution of the transposable elements is to investigate their presence (or absence) in a species group in which the biogeographic and phylogenetic relationships are known. The repleta group of Drosophila has been thoroughly studied and its phylogeny and biogeography have been deduced Fontdevila 1982;Ruiz et at, 1982). It is distantly related to the melanogaster group (Throckmorton, 1982), but copia homologous sequences have been detected in some of its species (Martin et at, 1983;Stacey et at, 1986).
Here we expand the survey to 18 species of this group comprising 3 different subgroups (mulleri, mercatorum and hydei). The 2 sibling species D buzzatii and D koepferae have been studied in more detail by analysing strains from different geographic origins. Moreover, partial sequencing of 2 independent clones isolated from D buzzatii demonstrates the presence of copia itself in this species.
Characterization of copia in different species is a tool to solve some questions, such as which molecular features act as functional determinants and the nature of the evolutionary dynamics of the element in the genus Drosophila.

Drosophila stocks
The strains used were originated from collections made by 1 of us (AF) and coworkers; there are some exceptions: D mulleri and D wheeleri were provided by W Heed; D buzzatii populations from Tunis and Chile were provided by J David and D Brncic, respectively; and D borborerrea and D serido were purchased from Bowling Green.

Probe
The pDmcopia was kindly provided by J Modolell. It is a full-length sequence of copia obtained from cDm5002 (Dunsmuir et al, 1980), cloned in pUC8.

Restriction enzymes
The enzymes were purchased from Boehringer Mannheim and used according to the supplier's instructions.
Genomic DNA extraction, agarose gel electrophoresis and Southern blotting Genomic DNA extraction was performed as described previously (Pinol et al, 1988). Digested genomic DNA was loaded on a 0.6% agarose gel (0.5 x 14 x 20 cm).
Electrophoresis was carried out at 20-25 V overnight. When using non-radioactive DNA detection methods, the amount of DNA loaded in each lane was adjusted by a correction factor obtained from the densitometric analysis of an electrophoresis previously carried out. Blotting on a nitrocellulose filter (Hybond C and Hybond C-EXTRA) was as described in Maniatis et al (1982).

Hybridization
The pDmcopia probe was labelled with either 32 P-ATP, biotin-11-dUTP (using nick-translation) or digoxigenin-11-dUTP (using a random primed reaction). When using 32 P-ATP-labelled probes the hybridization conditions were the same as those described in Maniatis et al (1982). The post-hybridization washes were always carried out at 65°C, twice in 2 x SSC for 15 min, and once in 2 x SSC 0.1% SDS for 30 min, which represents medium stringency wash conditions (Stacey et al, 1986). The autoradiography was exposed 24-36 h at -70°C with an intensifying screen. When using biotinor digoxigenin-labelled probes, the hybridization was performed at 42°C in 50% formamide and washes at rt twice in 2 x SSC, 0.1% SDS for 5 min, and then at 50°C twice in 0.1 x SSC, 0.1% SDS for 15 min (described in the non-radioactive nucleic acid detection systems from BRL and Boehringer Mannheim).

Cloning and sequencing
The genomic library from D buzzatii DNA was prepared as described by Pifiol et al (1988) and screened with pDmcopia probe. DNA from positive lambda clones was prepared, BamHi, EcoRI, HinDIII and Sail digested and hybridized with the same probe. Restriction fragments containing copia from independent lambda clones were subcloned into pTZ-18U (US, Biochemical) and partially sequenced by the dideoxy chain termination method using Sequenase (US Biochemical) or T7 DNA polymerase (Pharmacia). For sequence comparisons the FASTA program from the EMBL data bank was used.

Distribution of copia in the repleta group
In order to test the presence of copia in different species of the repleta group, an initial qualitative screening was carried out with species belonging to clusters 6uzzatii, martensi.s and mulleri (mulleri subgroug, . These clusters were chosen because the presence of copia in D mulleri has previously been described (Stacey et al, 1986). Southern blots of EcoRI-digested DNAs were hybridized with 32 P-labelled pDmcopia probe. Under medium stringency wash conditions, autoradiography shows patterns of multiple and discrete bands (fig 1-3). The time required to obtain a visible signal in the repleta group species clearly overexposes the band corresponding to D melanogaster. The patterns were different for each species tested, and indicate the presence of a repetitive sequence homologous to copia in the repleta group. Some of the bands detected are shorter than the copia element which is 5 kb long, suggesting that the homologous sequence has at least the element in the genome of D koepferae. The intensity of bands is greater for the lanes corresponding to Argentinian populations when the same amount of DNA is loaded (fig 4, bands 2-4).
We have also used a biotin-labelled pDmcopia probe to extend the survey of the presence of copia in the mulleri subgroup species. We included DNA from hydei and mercatorum subgroups as additional reference points, for it is known that copia is detected in D mercatorum but not in D hydei DNA (Martin et al, 1983;Stacey et al, 1986). The DNA loaded in each band was adjusted beforehand (see Material and methods) in order to obtain both qualitative and quantitative results. As it can be seen in figure 5, a sequence homologous to the copia element was detected with the biotin-labelled pDmcopia probe in all the mulleri subgroup species tested, but no detectable hybridization was observed in the hydei subgroup (represented here by D hydei and D hydeoides). The relative intensity of the bands was greater for the lanes corresponding to D mercatorum, D mulleri and D buzzatii.
Isolation of copia from D buzzatii As a preliminary step for the molecular characterization of the copia element in D buzzatii, a genomic library was screened with digoxigenin-labelled pDmcopia probe. Two independent clones were isolated and restriction fragments hybridizing with pDmcopia were subcloned and partially sequenced.
The alignment of the sequences with copia from D melanogaster (Dm copia) is shown in figure 6. The sequenced region of each of the clones aligns with Dm copia in different positions: Db 07X (A 5) aligns in the 3' region of integrase while Db 05TqE (A 12) corresponds to reverse transcriptase. The identity between D buzzatii subclones and DM copia is higher than 75% at the nucleotide level (77.4% for integrase and 76.5% for reverse transcriptase) and about 70% at the amino-acid level (74.2 and 68.9%, respectively). When considering similarities at the aminoacid level, the percentage increases to 95.2% for integrase region and to 88.8% for reverse transcriptase. It is noteworthy that copia from D melanogaster is the only Dro.sophila-transposable element sequence that aligns with our subclones at the nucleotide level when using the FASTA program. The sequence identity with other elements is not enough to allow their alignment with D buzzatii subclones.
On the other hand, amino-acid sequences obtained for putative ORFs of both the integrase and reverse transcriptase regions align with elements from distantly related taxonomic species, such as Nicotiana tabacum or Arabidopsis tltaliana, but with no other Drosophila-transposable element. Only 1731 from D melanogaster is aligned with Db 05TqE at the amino-acid sequence level (reverse transcriptase), but the percentage of identity changes from 68.9% between D buzzatii subclone and Dm copia to 32.2% between the same subclone and 1731.
In order to test the reliability of pDmcopia hybridization signals in the repleta group species, the Db 05TqE subclone from D buzzatii was used as a probe for D buzzatii and D I!oepferae EcoRI-digested DNA (fig 7). The hybridization patterns obtained for D koepferae were compared with those obtained with the pDmcopia probe when the same strains were used (see fig 4, bands 1,4; fig 7, bands 1, 2). The 3.4 kb EcoRI internal fragment is observed with both probes in the bands corresponding to Argentinian populations (fig 4, band 4; fig 7, band 2). The hybridization signal is greater for the Db 05TqE probe, since it contains a fragment of the element from a closely related species and a higher sequence conservation is expected. However, the relative intensity of the faint bands in relation to the internal fragment in each band is equivalent with both probes. Moreover, the signal is always more intense for the Argentinian than the Bolivian populations when the same amount of DNA is loaded. The coincidence of these results demonstrates the specificity of pDmcopia hybridization in the repleta group.

DISCUSSION
We have analysed the occurrence of copia in the repleta group. The results obtained are summarized in table I. It can be seen that a sequence homologous to copia from D melanogaster (Dm copia) is detected in all the tested species from the mulleri and mercatorum subgroups. Therefore, using both radioactive and non-radioactive detection methods, our results are in good agreement with those reported by Martin et al (1983) and Stacey et al (1986), where a sequence homologous to Dm copia was detected in the repleta group species D mulleri and D mercatorum.
The negative result obtained here for D hydei is also in agreement with the work of Martin et al (1983), where no complementary sequences were detected in this species. We have also shown that copia is not detected in D hydeoides. The negative result in both species of the hydei subgroup could be explained by either the absence of copia in this subgroup or a greater divergence rate of the element in these species, which would avoid detection by hybridization with the pDmcopia probe. Both alternatives suggest particular evolutionary events of the copia element in the hydei subgroup in relation to other repleta subgroups.
In the mulleri subgroup species tested, the similarity with the pDmcopia probe is enough to detect the homologous sequence in the repleta group species under medium stringency wash conditions (Stacey et al, 1986). The hybridization signal is heterogeneous between species, suggesting different degrees of similarity between the copia element from the repleta group and D melanogaster. However, similar patterns of hybridization are obtained with both the pDmcopia and D buzzatii probes for D buzzatii and D koepferae DNA, although the degree of divergence between them is nearly 30% at the DNA level. We therefore deduce that weak signals obtained with pDmcopia probe in Southern blots are due to sequence divergence or the low number of copies of the copia element rather than cross hybridization of the probe with other transposable elements present in these species.
Differences are also observed between closely related species such as D koepferae and D buzzatii. Populations from different geographic localities from both species were analysed for the genomic distribution of copia. Polymorphism in the genomic location of the elements is detested as heterogeneity in the patterns of the bands obtained for strains of the same species, according to the great variability in the reported chromosomal distribution of copia (Strobel et al, 1979;Montgomery and Langley, 1983;Bi6mont et al, 1985;Pasyukova et al, 1986;Ronsseray and Anxolab6h!re, 1986;Leigh-Brown and Moss, 1987).
The most striking differences in the pattern of bands are observed between Argentinian and Bolivian populations of D koepferae. The prominent band observed in the former could be due to the presence of an internal EcoRI fragment or a cluster where the copia element and its flanking regions would be regularly interspersed (Rubin 1983;Yamaguchi et al, 1987;Belyaeva et al, 1984, Crozatier et al, 1988Di Franco et al, 1989).
Using a second restriction enzyme, HindIII, a pattern of multiple bands is obtained with a pDmcopia probe in both Argentinian and Bolivian populations of D koepferae (fig 4b). It is noteworthy that hybridizing fragments are longer than 5 kb, suggesting the lack of a HindIII restriction target site in the copia sequence. The pattern of multiple bands obtained removes the possibility of a tandem arrangement of the element and suggests that the prominent band in the Argentinian populations is due to the presence of a 3.4 kb-long EcoRI internal restriction fragment.
It is interesting to note that a single change in an internal EcoRI site could explain the pattern observed. In the populations where only 1 EcoRI internal site is present, a pattern of multiple bands is expected, with the fragment lengths determined by the external flanking EcoRI sites. The presence of a second EcoRI internal site generates a pattern with a prominent band corresponding to the internal restriction fragment. Therefore, if copies of the element with either 1 or polymorphism in the copia element between Argentinian and Bolivian populations of D koepferae, in which a certain degree of genetic divergence has previously been described (Fontdevila et al, 1988).
On the other hand, patterns of bands obtained for South American and European populations of D buzzatii are rather similar, which means that polymorphism in the genomic distribution of copia in this species is very low. Such a regular distribution of the element could be due to the absence of recent transposition events or to genetic drift of a common ancestral set of inactive copies of the element.
Knowing the degree of divergence, we can expect that the homologous elements will remain unsolved until both active and inactive copies of the same element are characterized in closely and distantly related species. We have analysed 2 closely related species, D koepferae and D buzzatii in more detail, and different situations are observed. In one, an EcoRI restriction polymorphism is observed in the element. In the other a set of ancestral inactive copies is likely to be responsible for the observed patterns of hybridization. The partial sequencing of 2 independent clones isolated from D buzzatii reveals a 70-75% identity in both nucleotide and amino-acid sequences between D buzzatii and Dm copia (the similarity raises to 89-95% at the amino-acid level). It is interesting to note that no other transposable element from D melanogaster is similar enough to be aligned with D buzzatii sequences in the EMBL data bank at the nucleotide level with the FASTA program, and only the amino-acid sequence of the 1731 element from D rnelanogaster is aligned with the D buzzatii RT subclone. In this case the amino-acid identity percentage goes from 68.9 to 32.2% in relation to Dm copia.
It is well known that divergence rate between homologous retroviral proteins is faster than for structural genes and the high mutation rate is attributed to the low fidelity of RT (for a review, see Doolittle et al, 1989). Although RT is the slowest changing of the retroviral gene products, the amino-acid sequence divergence among different retrotransposons is greater than 60%. Moreover, retrotransposons are clustered in 2 different branches according to RT phylogenies. The copia element from D melanogaster is clustered with retrotransposons from very different organisms, such as yeasts (Tyi, S cerevisiae) or plants (Tnti, N tabacum; Tal, A. thaliana), and only with one other from Drosophila (1731, D melanoga.ster).
The nucleotide identity percentage between D B!zzatii isolated sequences and Dm copia is similar to that obtained for structural genes, such as Adh (72.3% at the nucleotide level). If we consider a higher rate of divergence for retrotransposons than for structural genes, we could postulate any mechanism accounting for the conservation of the element sequence between D melanogaster and D buzzatii, such as horizontal transmission of the element between these species. Evidence for , the horizontal transmission of other Drosophila elements between phylogenetically distant species has previously been described (Maruyama andHartl, 1991, Daniels et al, 1990). However, the copia element is detected in all the tested mercatorum and mulleri subgroup species, and the absence of any homologous sequence is confirmed in the hydei subgroup. In this case, we postulate transmission of the copia element into the mulleri subgroup after the separation of hydei subgroup and before the irradiation of the mulleri and mercatorum subgroups. From that moment, the copia element in these species would have changed in relation to D melanogaster according to the predicted rate of divergence for retrotransposons. On the other hand, if the copia element was present in ancestral species before the irradiation of the repleta group, we can postulate the loss of the element in the hydei subgroup genomes.
Other retrotransposons have been isolated and sequenced from the virilis-repleta radiation species such as micropia from D hydei (Lankenau et al, 1988(Lankenau et al, , 1990 and gypsy from D virilis (Mizrohki et al, 1991). Amino-acid identity percentage ranges from 70 to 90% between homologous retrotransposons from these species and D melanogaster, which agrees with our results. Therefore, the high levels of nucleotide and amino-acid sequences identity between the D 6uzzatii element and the copia from D melanogaster clearly establishes that the elements are related and are likely to be the same.

ACKNOWLEDGMENTS
This work was supported by grant PB86/0064 from the DGICYT (Ministerio de Educacion y Ciencia) awarded to AF, grant 113088 from Universitat Autonoma de Barcelona awarded to OC and by a fellowship from the Programa de Formació d'lnvestigadors (Universitat Autonoma de Barcelona) awarded to OF. We are very grateful to W Heed, J David and D Brncic, for provision of some of the strains used in this work, and J Pinol for help in obtaining genomic libraries.