Protein Determinants of Insertional Specificity for the Drosophila Gypsy Retrovirus

Corresponding author: Victor G. Corces, Department of Biology, The Johns Hopkins University, 3400 N. Charles St., Baltimore, MD 21218., corces{at}jhu.edu (E-mail)

Communicating editor: M. J. SIMMONS

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The gypsy retrovirus invades the germ line of Drosophila females, inserting with a high frequency into the ovo locus. Gypsy insertion sites in ovo are clustered within a region in the promoter of the ovo gene that contains multiple binding sites for the OvoA and OvoB proteins. We found that a 1.3-kb DNA fragment containing this region is able to confer gypsy insertional specificity independent of its genomic location. The frequency of gypsy insertions into the ovo gene is significantly lower in wild-type females than in ovo^D1 females. In addition, gypsy insertions in ovo^D1 females occur during most stages of germ-line development whereas insertions in wild-type females occur only in late stages. This pattern of temporally specific insertions, as well as the higher frequency of insertion in ovo^D1 females, correlates with the presence of the OvoA or OvoD1 proteins. The results suggest that gypsy insertional specificity might be determined by the binding of the OvoA repressor isoform to the promoter region of the gene.

ORIGINALLY considered as a long terminal repeat (LTR) retrotransposon, gypsy is currently classified as a retrovirus belonging to the family of errantivirus and is the first retrovirus described in insects (BOEKE et al. 1999 ). The expression of gypsy, and as a consequence its infectivity and transposition, are controlled by the flamenco (flam) gene (PRUD'HOMME et al. 1995 ). In the presence of a flam permissive allele, the amount of full-length gypsy RNA, as well as the amount of a spliced mRNA encoding the envelope protein, is increased significantly (PELISSON et al. 1994 ; SONG et al. 1994 ). Transcription of full-length gypsy mRNA and expression of the envelope protein in homozygous permissive flam females occur mainly in the follicle cells surrounding the developing oocyte during stages 9–10 of oogenesis (PELISSON et al. 1994 ; SONG et al. 1997 ). Mutations in the flam gene are necessary to produce significant levels of envelope proteins, suggesting that gypsy requires the envelope to infect the oocyte and subsequently integrate into the genome of next generation germ-line cells. It was shown that follicle cells produce infectious virus particles, capable of infecting larvae of a stock that lacks active gypsy elements (KIM et al. 1994 ; SONG et al. 1994 ). However, there is also evidence suggesting that gypsy is capable of transposing in flam-permissive female offspring, even when the envelope gene is mutated and therefore no envelope protein is made (CHALVET et al. 1999 ).

Characterization of the factors controlling gypsy infection and transposition has always been based on a genetic assay provided by the ability of gypsy to insert with very high frequency into the ovo gene (MEVEL-NINIO et al. 1989 ). However, the mechanisms determining such specificity remain largely unknown. Integration into ovo occurs with a frequency of 10% among the offspring of mutant flam females crossed to ovo^D1 males (PRUD'HOMME et al. 1995 ; MEVEL-NINIO et al. 1996 ). The function of the ovo gene product is cell autonomous and is necessary for the development of the female germ line and the normal progression of oogenesis (OLIVER et al. 1987 ). The ovo gene encodes two proteins, OvoA and OvoB, which bind the promoter region of the ovo gene to repress or activate its transcription, respectively. Adult females homozygous for a null mutation of the ovo gene do not develop germ-line cells. The ovo^D1 allele is caused by a point mutation that creates a new in-frame methionine codon in the 5' region of ovo, adding an extra amino terminus domain that in the wild type is present only in the ovoA protein (MEVEL-NINIO et al. 1996 ). The ovo^D1 allele is dominant negative and causes female sterility even when heterozygous. The sterility is due to the expression of Ovo^D1B protein, which is made at the same time of development as OvoB but has the repressor activity of OvoA; the presence of Ovo^D1B is sufficient to arrest oogenesis at stage 4 (LU et al. 1998 ). Insertion of gypsy into the ovo^D1 allele in a heterozygous female reverts the phenotype to fertility, although the reversion occurs only in those germ cells in which gypsy is inserted into the ovo^D1 sequence, preventing the expression of the Ovo^D1B protein. The ability of gypsy to integrate specifically into ovo sequences was recently analyzed by DEJ et al. 1998 . These studies concluded that gypsy integrates in at least seven different target sites localized within a 200-bp sequence present in the promoter region of the ovo gene. Close analysis of these sites reveals a very relaxed consensus sequence consisting of six alternating pyrimidines and purines. The weak conservation of the observed target sequence suggests that gypsy site-specific integration is not due to a direct interaction of the gypsy integrase with these sequences. Instead, insertional specificity is probably due to interactions with additional factors that remain so far undetermined.

Factors governing DNA integration into the genome are of particular importance to clarify the overall mechanisms of retrovirus infection. Understanding these mechanisms is also an invaluable tool for the development of genome manipulation and gene therapy technologies. The gypsy retrovirus of Drosophila shares properties with retroviruses and also with retrotransposons, since it is capable of transposing both vertically in the germ line of the females (CHALVET et al. 1999 ) and horizontally, using the envelope protein to produce infectious particles (KIM et al. 1994 ; SONG et al. 1994 , SONG et al. 1997 ). Such duality could be the evolutionary basis for the integration specificity into the ovo gene shown by gypsy and constitutes a unique feature among retroviruses and LTR retrotransposons from higher eukaryotes. This property provides a unique opportunity to study integration site specificity in vivo using the genetic and molecular tools available in Drosophila. Here we test whether the integration specificity of gypsy into ovo is related to the ability of gypsy preintegration complexes to interact with proteins that bind DNA sequences of the ovo gene. The results suggest a correlation between the developmental pattern of expression of OvoA and the timing of insertion of gypsy during germ cell differentiation. A role for OvoA in targeting gypsy to ovo is reinforced by the fact that gypsy insertions into ovo are not observed at the time when only OvoB protein is expressed in the germ line. Furthermore, a significant increase of gypsy insertions takes place in the presence of the Ovo^D1B protein, which is functionally equivalent to OvoA but is expressed with the same developmental pattern as OvoB.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

P-element-mediated germ-line transformation and gypsy insertion into the y^+ovo transgene:
The yellow-CaSpeR plasmid, containing all the coding and regulatory regions of the yellow (y) gene, was modified by adding a NotI site in the Eco47III site of y (GEYER et al. 1986 ). The NotI site was used to clone a PCR-generated fragment of the 5' region of the ovo gene spanning nucleotides 12–1298 (MEVEL-NINIO et al. 1996 ). The oligonucleotides used to amplify the ovo fragment were AGTTGGCCATGACCAACAGAGCGA at the 5' end and CTCCCGCTCTGCGGGCT TCTCTTT at the 3' end. NotI sites were added at the 5' ends of both primers to facilitate cloning into the yellow-CaSpeR plasmid. Males from the ovo^D1 stock were used to obtain the template genomic DNA for the PCR amplification. The resulting plasmid was called CasPeR y^+ovo and was used to microinject Drosophila embryos.

P-element-mediated transformation was performed as described by RUBIN and SPRADLING 1982 . The CasPeR y^+ovo plasmid was injected into y w; Sb [ 2-3]/TM6 embryos (ROBERTSON et al. 1988 ) at a concentration of 0.5 µg/µl. Transgenic flies were identified by the rescue of the white phenotype. Two lines, y w P[y^+ovo; w⁺]1.1 and y w ; P[y^+ovo; w⁺]2.1, were selected among a total of five and genetically mapped to the X and second chromosomes, respectively. The yellow phenotype was completely rescued in all five lines.

Females homozygous for y v f mal flam were crossed to y w P[y^+ovo; w⁺]1.1 and y w ; P[y^+ovo; w⁺]2.1 males. The heterozygous female offspring with genotypes y v f mal flam/y w P[y^+ovo; w⁺]1.1 and y v f mal flam/y w; P[y ^+ovo; w⁺]2.1/+ were individually crossed to y w^67c males. To detect gypsy insertions into the y^+ovo transgene, the offspring of this cross were systematically screened for individuals with a y²-like phenotype. The y²-like phenotype is expected because after insertion of gypsy the body and wing enhancers are blocked by the Su(Hw) insulator and cannot activate transcription of y in these tissues. Because the y^+ovo transgene is heterozygous in these females, only the offspring with a y⁺ or a y²-like phenotype were considered in all determinations of cluster size. As a positive control to test whether flam-permissive females were producing active virus particles, y v f mal flam females were crossed to ovo^D1 males and the offspring were screened for fertile females as described below.

Analysis of gypsy insertions into ovo by PCR:
Genomic DNA from single female crosses was extracted from 50–100 flies for the detection of gypsy insertions into ovo in wild-type females. DNA extraction was carried out using the potassium acetate quick prep as described in DEJ et al. 1998 . The primers used to amplify gypsy insertions were P3 CTTTGCCGAAAATATGCAATG and P1 CAACATGACCGAGGAGCGGTCATAAAC located in the 5' and 3' ends of gypsy, respectively, and P4 CGGCTTTTTCAGCGGCTAACC and P2 CTCCCGCTCTGCGGGCTTCTCTTT located in the ovo sequences flanking the gypsy insertion sites at the 5' and 3' sites, respectively (Fig 3). A combination of P1 with P2 or P4 will detect the insertion of gypsy in either orientation. The alternative combination of P3 with P4 or P2 can detect the same type of insertions and was used to confirm the results obtained with the first combination of primers. The same primers were used to clone and sequence the insertion sites of gypsy in the y^+ovo transgene. Approximately 100 ng of genomic DNA was used per PCR amplification. Conditions for this reaction were as follows: 91° (1 min), 65° (1 min), 72° (2–10 min) for 35 amplification cycles. Amplified DNA fragments containing gypsy insertion sites from the y^+ovo-gypsy transgenes were cloned into the PCR 2.1 TA cloning vector from Invitrogen (Carlsbad, CA) and sequenced using an ABI 377 automated DNA sequencer.

View larger version (45K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Structure of transgenes and fly phenotypes. (A) Organization of the y+ovo transgene showing the location of the 5'ovo region. Wing, body, and bristle enhancers are indicated as ovals. Putative Su(Hw) binding sites in the 5' ovo fragment would block the interactions between wing and body enhancers and the y promoter. (B) Body cuticle phenotype of a male of the genotype y w P[y+ovo; w+]1.1. The 5'ovo fragment does not bind Su(Hw) in vivo and, therefore, fails to block the effect of the wing and body cuticle enhancers, resulting in normal pigmentation of the abdomen. (C) Structure of the y+ovo-gypsy transgene after gypsy insertion into the 5'ovo region. After gypsy inserts into the 5'ovo fragment, the gypsy insulator precludes the wing and body enhancers from activating transcription by the y promoter. The bristle enhancer is not affected by the insertion of gypsy. (D) Body cuticle from a male of the genotype y w P[y+ovo-gypsy; w+]1.1 obtained by gypsy insertion, showing a y2-like phenotype.

View larger version (70K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Binding sites for Ovo proteins and gypsy insertion sites in the ovo promoter region contained within the 1.3-kb 5'ovo fragment. Highlighted sequences correspond to the Ovo recognition sequences as described by LU et al. 1998 . Bold underlined sequences indicate insertion sites A to H, as described by DEJ et al. 1998 . Gypsy insertion sites identified in this work are indicated with numbers above the previously described sites. All sites coincide with previous ones except 8.2 and 8.1, which define a new insertion site. Transcription start sites for RNAs encoding the OvoA and OvoB proteins are indicated. Numbers correspond to nucleotide positions starting at the zero nucleotide position of the ovo sequence (MEVEL-NINIO et al. 1996 ).

View larger version (19K): [in this window] [in a new window] [Download PPT slide]   Figure 3. PCR analysis of gypsy insertions in wild-type females. (A) Diagram representing a gypsy insertion in the 5'ovo region. The positions of the P1, P2, P3, and P4 PCR primers are indicated. The location of the ovoD1 mutation (D1) is also indicated, although ovoD1 mutants were not used in this experiment. The same primers were used for cloning and sequencing the insertion sites shown in Fig 2. All insertions observed in DEJ et al. 1998 , except for one, were found between P4 and P2. (B) Experimental protocol for PCR amplification using the primers described in A. Lane a is an example of a positive signal in an experimental sample. Lane b is a control in which the PCR product was obtained from the offspring of an ovoD1 revertant female.

Drosophila stocks, ovo reversion assay, and determination of gypsy insertion during development:
All strains used in this work were kept at 25°. The flam stock was kept as y v f mal flam/FM3. ovo^D1 males were maintained by crossing them to females carrying attached X chromosomes. The SS strain was used as a flam-permissive stock that does not carry active gypsy elements (PRUD'HOMME et al. 1995 ). We used the ovo^D1 reversion assay described in PRUD'HOMME et al. 1995 to determine the gypsy activity of the flam stock. Five to 10 y v f mal flam virgin females were crossed with ovo^D1 males and kept in the same vial for 3–5 days. After this time the flies were removed from the vials and the larvae were allowed to develop. Groups of 5–10 ovo^D1/ovo⁺ females from the offspring were transferred to new vials and crossed to wild-type males. After several days, vials were examined for the presence of eggs or larvae. All the mothers in vials with eggs were dissected and the females displaying functional ovaries were considered fertile.

To determine the timing of insertion of gypsy during development, a similar procedure was carried out with some modifications to ensure that ovaries were healthy and completely developed. Females homozygous for y v f mal flam were crossed to v ovo^D1 males and to SS males as a control. Groups of five female offspring were crossed to wild-type males in fresh food containing a few grains of dry yeast. Vials were examined daily for the presence of eggs and those vials containing eggs were separated. After 5 days, all the females in egg-containing vials were dissected to identify those carrying functional ovaries. Females from vials lacking eggs were transferred to fresh food and the same process was repeated for another 5 days. The ovarioles from each functional ovary in fertile females were separated using tungsten needles and counted using a dissecting microscope. Ovaries containing 11 or more ovarioles were considered fully functional and were included in a single category. After 10 days, all females in vials lacking eggs were discarded and considered sterile. Statistica for Windows release 4.0 was used for the statistical analysis of the results.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Integration of gypsy into ovo is determined by local sequences contained within the gene:
To understand the nature of the mechanisms responsible for the high rate of gypsy insertion into the ovo gene, we first asked whether the specificity is due to a local feature of ovo sequences or whether it depends on a general property of the gene or its flanking genomic sequences. A possible explanation for the insertional specificity of gypsy is that particular sequences in ovo attract proteins present in the preintegration complex. These proteins could be the gypsy integrase or any other protein component of the complex. A candidate for a targeting sequence is the gypsy insulator, which is located in the 5' untranslated region and contains 12 binding sites for the Su(Hw) protein (SPANA and CORCES 1990 ). Since several gypsy insulators present in different chromosomal locations appear to associate together during interphase (GERASIMOVA and CORCES 1998 ), it is possible that Su(Hw) proteins bound to the cDNA of gypsy in the preintegration complex could be targeted to other gypsy insulator sites in the genome. Interestingly, the analysis of DNA sequences adjacent to the insertion sites of gypsy in the ovo gene reveals the presence of putative Su(Hw) binding sites (data not shown). Thus it could be possible that gypsy insertion into ovo is due to the interaction of proteins binding the insulator sequences present in both the gypsy cDNA and the ovo gene. If this hypothesis were true, the putative binding sites in the ovo gene should bind Su(Hw) protein in vivo and the sequence containing such sites should act as an insulator.

To test this hypothesis, a genomic copy of the yellow (y) gene from Drosophila melanogaster, containing a 1.3-kb DNA fragment from the 5' region of ovo inserted between the yellow body and bristle enhancers (Fig 1A), was cloned into the CasPeR vector carrying white (w) as a reporter gene. This plasmid, named CasPeR y^+ovo, was microinjected into y w; Sb [ 2-3]/TM6 Drosophila embryos. The 1.3-kb fragment inserted into the regulatory region of y corresponds to ovo sequences spanning nucleotides 12–1298 in the 5' region of the gene (MEVEL-NINIO et al. 1996 ). We refer to this sequence as the 5'ovo region. This fragment was selected because it contains all the putative Su(Hw) binding sites together with all seven gypsy insertion sites found previously in the ovo gene (DEJ et al. 1998 ). Offspring with colored eyes were selected and examined for the pigmentation of body cuticle, bristles, and wing blades (Fig 1B). If the 1.3-kb fragment of ovo contains Su(Hw) binding sites, it will function as an insulator in vivo, causing a y²-like phenotype similar to that in the gypsy-induced y² allele (GEYER and CORCES 1992 ). The transgene conferred full expression of the y gene in all tissues, including body cuticle, wing blades, and bristles in five independent transgenic lines. This result indicates that functional Su(Hw) binding sites are not present in the 5' region of the ovo gene.

It is possible that Su(Hw)-binding sequences targeting gypsy to ovo could be present outside the 5'ovo region used in this experiment. Targeting gypsy by Su(Hw) binding sites could also be achieved by targeting the integration complex to the general genomic region where ovo is found and subsequent selection of the gypsy insertion site by a preference for the consensus gypsy insertion sequence YRYRYR, where Y = pyrimidine and R = purine (DEJ et al. 1998 ). To test this hypothesis, we examined whether gypsy can still insert with high frequency into the 5'ovo region in a genomic context different from that of the endogenous ovo gene. In this experiment, different genomic regions flanking the y^+ovo transgene provide such alternative genomic context. The rationale of the experiment is illustrated in Fig 1C. As mentioned before, flies containing the y^+ovo transgene show a wild-type y phenotype. It is well established that gypsy Su(Hw) binding sites located between body and wing enhancers and the promoter act as an insulator, blocking enhancer-promoter interactions and precluding transcription of y in the body cuticle and wing blades (GEYER and CORCES 1992 ). Therefore, we expect that, after insertion of gypsy into the 5'ovo region present in the y^+ovo transgene, y^+ovo-gypsy flies will display a y²-like phenotype (Fig 1D). To induce gypsy mobilization, we crossed y v f mal flam females to y w males carrying the P[y^+ovo; w⁺] 1.1 and P[y^+ovo; w⁺] 2.1 transgenes in the X and in the second chromosome, respectively. We selected y⁺ daughters of the genotype y v f mal flam/y w P[y^+ovo; w⁺] 1.1 and y v f mal flam/y w; P[y^+ovo; w⁺] 2.1/+ and crossed them individually to y w^67c males.

The offspring of individual females were screened for the y²-like phenotype as a marker for gypsy insertion into the 5'ovo region in the y^+ovo transgene (Fig 1D). The results of these experiments are shown in Table 1. Seven out of a total of 232 females carrying the P[y^+ovo; w⁺]1.1 transgene gave rise to one or more individuals with a y² phenotype and 3 females out of a total of 79 carrying the P[y^+ovo; w⁺]2.1 transgene produced offspring with a y² phenotype. The frequency of females producing y² in their offspring was 3% with the transgene in the X chromosome and 3.8% when the transgene was located in the second chromosome. To verify that the y²-like phenotypes were due to insertions of gypsy into the 5'ovo region of the transgene we established six independent y²-like strains starting with a single male or female offspring. All strains showed Mendelian segregation of the reporter gene (w⁺) located in the transgene, which was always associated with a y²-like phenotype. We followed the procedure of DEJ et al. 1998 , using PCR to clone and sequence the gypsy insertion site in each of the six strains. The results demonstrate that gypsy insertions occurred in the same gypsy target sites found previously in the endogenous ovo^D1 sequence (DEJ et al. 1998 ). Fig 2 shows the comparison of the gypsy target sites found previously and the six insertion sites identified in this work. Out of the six insertions, only one was found in a new site, very close to the other seven previously described. From these results we conclude that the signals necessary to determine the specificity of gypsy insertion into ovo reside within the 1.3-kb sequence present in the 5' region of this gene.

The frequency of gypsy integration into ovo is lower in wild-type than ovo^D1 females:
In the course of performing the experiments designed to measure the frequency of gypsy insertion into the y^+ovo transgene, we carried out control experiments to determine the activity of gypsy in the flam strain by measuring the reversion rate of the ovo^D1 allele. Fertile ovo^D1/ovo⁺ females arising in the offspring of a cross between ovo^D1 males and y f v mal flam females occur after insertion of gypsy into the 5' region of the ovo^D1 locus in germ-line cells (MEVEL-NINIO et al. 1989 ; PRUD'HOMME et al. 1995 ). In a typical experiment, 86 revertant fertile females were found out of 926 ovo^D1/ovo⁺ females analyzed (Table 1). This result indicates that gypsy is actively transposing in the germ-line cells of these females. However, the frequency of insertions in the controls (9.3%) was significantly higher (² = 10.45; P = 0.0012) than the frequency of gypsy insertions in the y^+ovo transgene (3 and 3.8%). An important distinction between control and experimental samples is that, in the former, gypsy insertions occur in ovo^D1 sterile females, whereas in the latter, insertions occur in wild-type ovo⁺ fertile females. To test whether the higher frequency observed in ovo^D1 flies depends on the ovo^D1/ovo⁺ genotype we performed an experiment to determine the frequency of gypsy insertions into the endogenous ovo gene in ovo⁺/ovo⁺ females.

Virgin females of the genotype y v f mal flam were crossed to ovo⁺ males. Offspring females from this cross should carry gypsy particles produced in the ovaries of the flam mutant mothers and, therefore, insertions into the ovo gene should occur in their germ line with the same frequency as they occur in ovo⁺/ovo^D1 females. However, because these females are ovo⁺ and perfectly fertile, insertions into the ovo gene of germ-line cells cannot be monitored using morphological or physiological traits. To detect these insertions we extracted genomic DNA from 50–100 progeny derived from the offspring of ovo⁺/ovo⁺ single female crosses and performed PCR amplifications using the primers indicated in MATERIALS AND METHODS (see also Fig 3). These primers allow detection of gypsy insertions into the ovo gene in either orientation (Fig 3). Among 134 females, two positive samples were found, indicating that only 1.6% of the females produced offspring bearing gypsy insertions into the ovo gene. The same crosses were carried out in parallel but using ovo^D1 males instead and detecting gypsy insertions by the reversion to fertility of ovo^D1/ovo⁺ females. The results indicate that the flam females that produced 1.6% of progeny carrying gypsy insertions into ovo in an ovo⁺/ovo⁺ genotype produced 10% of progeny carrying gypsy insertions into ovo when the genotype is ovo⁺/ovo^D1. This difference (1.6 vs. 10%) is similar to the difference found in the previous experiment, when comparing the frequency of gypsy insertions into the ovo gene in ovo⁺/ovo^D1 females (9.2%) with the frequency of insertions into the y^+ovo transgene (3 and 3.8%). Because the y^+ovo transgene is present in a strain with an ovo⁺/ovo⁺ genotype, the result suggests that the higher frequency of gypsy insertion into ovo might be due to the activity of the OvoD1 protein present only in ovo⁺/ovo^D1 females.

The timing of gypsy insertion during female germ-line development correlates with the expression pattern of Ovo isoforms:
The 5' region of the ovo gene contains several binding sites for Ovo proteins, suggesting an involvement of ovo in its own transcriptional regulation (LU et al. 1998 ). Comparison of the location of gypsy insertion sites with that of binding sites for Ovo proteins shows that all gypsy insertion sites are located in close proximity to Ovo protein binding sites and are contained within the 5'ovo region used to obtain the y^+ovo transgenic flies (Fig 2). The proximity between binding sites of Ovo proteins and insertion sites of gypsy in the ovo locus suggests the possibility of a functional correlation between the binding of Ovo proteins and gypsy insertion. Since the two protein products of ovo, OvoA and OvoB, bind to these sites and show distinct expression patterns and regulatory properties (MEVEL-NINIO et al. 1996 ; ANDREWS et al. 2000 ), we decided to examine whether there is also a correlation between the expression of these two proteins and the timing of gypsy insertion. The OvoB protein is expressed from early stages of embryogenesis and throughout development and is necessary for the normal development of the germ line. The OvoA protein is expressed only during the later stages of oogenesis and antagonizes the function of OvoB by repressing transcription of the same genes whose expression is activated by OvoB (MEVEL-NINIO et al. 1996 ; ANDREWS et al. 2000 ). To determine the time of gypsy insertion into ovo during female germ-line development we used the rationale depicted in Fig 4. The insertion of gypsy in a cell early during development of the female germ line will cause all or most stem cells in the germarium of each ovariole of the fully developed ovary to carry the same insertion (Fig 4A). An insertion at an intermediate time during development will result in only a few of the stem cells carrying the gypsy insertion (Fig 4B). Finally, an insertion at a very late stage of germ-line development will cause only a stem cell from a single ovariole in the whole ovary to produce oocytes carrying gypsy insertions (Fig 4C). An equivalent correlation was found when comparing the brood size of revertant ovo^D1 females after somatic recombination induced by X rays at different stages of development (PERRIMON 1984 ). Therefore, we can approximately determine the timing of gypsy insertion by counting the number of functional ovarioles in ovo^D1/ovo⁺ fertile females. Following a cross between y v f mal flam females and ovo^D1 males, fertile ovo^D1/ovo⁺ female offspring were selected. After several days in fresh medium and in the presence of males, the ovaries of these females were dissected. A total of 118 ovaries dissected from 111 revertant females out of a total sample of 1219 ovo^D1/ovo⁺ females were analyzed (Table 2). The number of ovarioles in a fertile ovary is considered to be 15 (SPRADLING 1993 ). We counted the number of functional ovarioles in ovaries from ovo^D1/ovo⁺ revertant females and found that it varies between 1 and 11 or more (we considered ovaries with 11 or more ovarioles as a single class corresponding to a fully functional ovary). Fig 5A shows a histogram of the frequencies of ovaries containing a different number of functional ovarioles. The most abundant class coincides with ovaries containing a single functional ovariole, which we considered as the result of gypsy insertion at a late stage. The least frequent class coincides with ovaries containing 11 or more functional ovarioles and corresponds to gypsy insertion at earlier stages of germ-line development. These results suggest that gypsy can insert into ovo at a wide range of stages of germ cell development in ovo^D1/ovo⁺ females. Interestingly, the probability of having a certain number of ovarioles increases significantly in a linear manner from P = 0.017 for 11 or more ovarioles to P = 0.14 for a single ovariole, with a regression coefficient of r = -0.96 and P < 0.0001 (Fig 5B).

View larger version (34K): [in this window] [in a new window] [Download PPT slide]   Figure 4. Model representing cluster size in relation to time of insertion. (A) Gypsy insertion occurs early in the development of the female germ line. (B) Intermediate insertion time with only half of the ovarioles carrying a gypsy insertion. (C) Insertion of gypsy occurs late during germ cell differentiation. Only one ovariole produces oocytes arising from the cell where gypsy originally inserted. In this case, a single stem cell produces oocytes containing an insertion of gypsy into ovo.

View larger version (28K): [in this window] [in a new window] [Download PPT slide]   Figure 5. Distribution of the number of ovarioles and cluster sizes after gypsy insertion into ovo. (A) Histogram showing the distribution of functional ovarioles per revertant ovary. (B) Correlation between the fraction of functional ovarioles and the frequency of each event. Note that the cluster size obtained for y2-like flies using the transgene roughly corresponds to 0.033, the same frequency obtained after dividing a single ovariole by a total of 30 per ovary.

To determine whether gypsy insertions into the ovo gene in wild-type females occur at similar stages and frequencies as in ovo^D1/ovo⁺ females, we used the same rationale as in the previous experiment. For this purpose, we monitored the proportion of flies carrying gypsy insertions in the y^+ovo transgene (y^+ovo-gypsy with a y²-like phenotype) in the offspring of single female crosses. As in the previous experiment, a large cluster of flies with a y²-like phenotype indicates an insertion into the transgene at an early stage (as in Fig 4A), whereas a small cluster indicates a later insertion during germ-line development (as in Fig 4C). Considering that each of the two ovaries of a healthy female contains 15 ovarioles, and assuming that each ovariole contributes equally to the pool of oocytes, each ovariole should be responsible for 1/30 of the eggs laid by a female. Table 3 shows the size of the clusters of flies with a y²-like phenotype observed from each female producing y²-like individuals in its offspring. We estimated the values for such clusters under the hypothesis that more than one ovariole was participating in the cluster (number of ovarioles divided by 30 times the observed number of offspring). Table 3 shows the expected numbers of y²-like individuals among the total offspring if 1, 2, or 3.9 ovarioles were present in one ovary (i.e., 1/30, 2/30, or 3.9/30 of the offspring; 3.9 is the average number of functional ovarioles obtained in the experiment using ovo^D1 females). The observed vs. expected value of ², considering all values together, shows that the y²-like flies arose with a frequency not significantly different from 1/30 (² = 7.97; P = 0.5372), whereas the differences are statistically significant when the expected values for 2 and 3.9 ovarioles are compared to the observed values (see Table 3). This result suggests that insertions of gypsy occur only during the latest stages of germ-line development in wild-type females, contrary to what is found in ovo^D1 heterozygous females, where gypsy inserts also at early stages of development. This conclusion can be rationalized in the context of the developmental expression of the ovo-encoded proteins. In particular, the results show an increasing amount and a broader distribution of gypsy insertion events during development in the presence of the dominant negative Ovo^D1B protein.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Site-specific integration of the retroviral genome can be mediated by protein-protein interactions between preintegration complexes and proteins that bind DNA or are associated with chromatin (BUSHMAN 1994 ; KIRCHNER et al. 1995 ; BUSHMAN and MILLER 1997 ; ZHU et al. 1999 ). Su(Hw) is a good candidate to mediate interactions leading to gypsy insertional specificity, but our results suggest that this protein is not directly involved in determining gypsy insertional specificity into ovo. This conclusion is based on the observation that a DNA fragment from the 5' region of the ovo gene, where gypsy inserts with high frequency, does not display the properties of an insulator and therefore does not bind Su(Hw) protein in vivo. Alternatively, the ability of gypsy to insert specifically into the ovo gene could be developmentally regulated by the expression of the Ovo proteins, since the frequency of germ-line insertion events of gypsy is affected by the ovo genotype of the female. Gypsy insertions are significantly more frequent and occur along different stages of germ-line development in ovo^D1 females, whereas they are less frequent and occur only at late stages in wild-type females. Since the number of gypsy integration complexes depends on maternal inheritance (CHALVET et al. 1999 ), and the flam mutant females used in experiments involving ovo^D1 and wild-type females are the same, we conclude that the insertion frequency is determined by the influence of Ovo proteins on the insertion process. This conclusion is also supported by the observation that Ovo protein binding sites are found in close proximity to gypsy insertion sites in the ovo gene.

The ovo gene encodes two protein products, OvoA and OvoB, both containing the same zinc finger domains and therefore recognizing exactly the same DNA-binding sites (GARFINKEL et al. 1994 ; LEE and GARFINKEL 2000 ). The OvoB protein is expressed during all stages of germ-line development, including oogenesis, and it is necessary for the normal development of the germ line (GARFINKEL et al. 1992 ; MEVEL-NINIO et al. 1996 ). OvoA protein probably antagonizes the function of OvoB, repressing transcription of the same genes activated by OvoB including the ovo gene itself (LU et al. 1998 ; ANDREWS et al. 2000 ). In the ovo^D1 allele, translation of the ovoB transcript starts prematurely and the resulting protein has an additional amino-terminal domain ordinarily present only in the OvoA protein and responsible for its repressive activity (MEVEL-NINIO et al. 1996 ; ANDREWS et al. 2000 ). Therefore, the OvoB product of the ovo^D1 mutation (Ovo^D1B) is functionally equivalent to the OvoA protein but it is expressed earlier in development. Expression of Ovo^D1B is equivalent to expression of OvoA and results in sterility of ovo^D1 females by arresting oogenesis at stage 4.

The low frequency of gypsy insertions into the ovo locus in wild-type females correlates with the absence of the OvoA protein during most of the female germ cell development. Our results suggest that gypsy insertion events in wild-type females take place only during late stages of gonadal development. This conclusion is supported by the observation that y²-like phenotypes due to gypsy insertion into the y^+ovo transgene occur at frequencies expected if the insertion takes place in a stem cell from a single ovariole, indicating that insertions never occur before this stage. In contrast, ovo^D1 revertant females show ovaries containing a wide distribution of functional ovarioles. The presence of a high number of ovarioles in an ovary is a consequence of an early insertion event, whereas a single ovariole represents an insertion during the very last division of the stem cells in the germarium (Fig 4). Our interpretation of these results is that ovo is an open target for gypsy insertion only when the OvoA protein, or the equivalent Ovo^D1B, is present in the germ cells. In wild-type females, nonmaternal OvoA protein is present only at the end of gonadal development, whereas the expression of Ovo^D1B occurs throughout germ cell development in ovo^D1 females.

An alternative explanation for the observed results could be that the germ cells carrying a gypsy insertion in the ovo locus of ovo^D1 mutants actually become phenotypically wild type and therefore outcompete the ovo^D1 mutant cells during the process leading to the population of the ovary. This competition could explain why most revertant ovaries in ovo^D1 females contain several ovarioles but fails to explain the large number of ovaries with a single ovariole. Also, the frequency distribution of the number of ovarioles per revertant ovary in ovo^D1 females contradicts such explanation. The smaller the number of ovarioles in a revertant ovary from an ovo^D1 female, the higher the frequency among revertant ovaries. If insertions took place only at a particular stage, the outcompetition would not produce a linear distribution of the frequencies. Rather, it would result in a normal distribution with an average that would correlate with the time at which insertions occur: the later the insertion the smaller the average. Additional observations supporting our interpretation are provided by experiments using X rays to induce somatic recombination in the germ line of Drosophila females during different developmental stages in wild-type as well as in ovo^D1 females (WIESCHAUS and SZABAD 1979 ; PERRIMON 1984 ). Irradiation during early embryonic development, including blastoderm, results in a low frequency of large clones of recombinant oocytes, whereas irradiation during third instar larva, pupa, or adult flies causes very small clones with a higher frequency (WIESCHAUS and SZABAD 1979 ). These results in wild-type females, where outcompetition is ruled out, are equivalent to the results obtained in experiments where gypsy inserts in ovo in an ovo^D1 mutant female.

Current models to explain how gypsy reaches the germ-line cells of the offspring of flam females suggest that gypsy virus particles are present in the posterior region of the embryo before cellularization, providing an additional explanation for the linear distribution mentioned above (SONG et al. 1997 ). Once the cytoplasmic membrane surrounds the nucleus of the pole cells, gypsy particles should be found in the cytoplasm of these cells. At this point the gypsy preintegration complexes must reach the nucleus to integrate into the genome. Among retroviruses, only lentiviruses have developed a mechanism to cross the nuclear envelope (ZENNOU et al. 2000 ). All other retroviruses depend on cell division to reach the nuclear DNA. Since it is very possible that gypsy also requires cell division to enter the nucleus, the probability of insertion into ovo will be higher at later stages of development, when more cell divisions have taken place. We should then expect a significant correlation between the size of the clusters of revertant ovarioles and the frequency of such clusters only if integration occurs with the same probability along all developmental stages. If insertion of gypsy depends on the activity of the OvoA protein this condition is met only in the case of the ovo^D1 mutant.

The specificity of gypsy integration into ovo might be mediated by interactions between the amino-terminal domain of the OvoA protein and a component of the gypsy integration complex. This interaction might involve the repressive domain of OvoA and the gypsy integrase (Fig 6) in a similar manner to the Ty5 integrase and Sir4 proteins (ZHU et al. 1999 ). Alternatively, gypsy integration might be mediated by the repressive activity of OvoA or Ovo^D1B on the chromatin structure of the ovo gene. Experiments to distinguish between these two alternatives are in progress.

View larger version (21K): [in this window] [in a new window] [Download PPT slide]   Figure 6. A model to explain gypsy insertion site specificity. The cDNA of the gypsy preintegration complex is bound at the ends by integrase. (A) Gypsy does not target the ovo gene early in development, when the OvoB protein is present in the 5' region of ovo. Gypsy is targeted to the promoter region of the gene only in the presence of OvoA at late stages of germ cell development. (B) Targeting of gypsy in ovoD1 mutants occurs in the same manner as in wild type, but earlier in development due to the presence of the OvoD1B protein.

	ACKNOWLEDGMENTS

We thank Kelly Baxter for her help during the cloning of the vector CaSpeR yellow. This work was supported by Public Health Service award GM56022 from the National Institutes of Health.

Manuscript received November 30, 2000; Accepted for publication March 29, 2001.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDREWS, J., D. GARCIA-ESTEFANIA, I. DELON, J. LU, and M. MEVEL-NINIO et al., 2000  OVO transcription factors function antagonistically in the Drosophila female germline. Development 127:881-892[Abstract].

BOEKE, J. D., T. H. EICKBUSH, S. B. SANDMEYER and D. F. VOYTAS, 1999 Metaviridae. Springer-Verlag, New York.

BUSHMAN, F. D., 1994  Tethering human immunodeficiency virus 1 integrase to a DNA site directs integration to nearby sequences. Proc. Natl. Acad. Sci. USA 91:9233-9237[Abstract/Free Full Text].

BUSHMAN, F. D. and M. D. MILLER, 1997  Tethering human immunodeficiency virus type 1 preintegration complexes to target DNA promotes integration at nearby sites. J. Virol. 71:458-464[Abstract].

CHALVET, F., L. TEYSSET, C. TERZIAN, N. PRUD'HOMME, and P. SANTAMARIA et al., 1999  Proviral amplification of the gypsy endogenous retrovirus of Drosophila melanogaster involves env-independent invasion of the female germline. EMBO J. 18:2659-2669[Medline].

DEJ, K. J., T. GERASIMOVA, V. G. CORCES, and J. D. BOEKE, 1998  A hotspot for the Drosophila gypsy retroelement in the ovo locus. Nucleic Acids Res. 26:4019-4025[Abstract/Free Full Text].

GARFINKEL, M. D., A. R. LOHE, and A. P. MAHOWALD, 1992  Molecular genetics of the Drosophila melanogaster ovo locus, a gene required for sex determination of germline cells. Genetics 130:791-803[Abstract].

GARFINKEL, M. D., J. WANG, Y. LIANG, and A. P. MAHOWALD, 1994  Multiple products from the shavenbaby-ovo gene region of Drosophila melanogaster: relationship to genetic complexity. Mol. Cell. Biol. 14:6809-6818[Abstract/Free Full Text].

GERASIMOVA, T. I. and V. G. CORCES, 1998  Polycomb and trithorax group proteins mediate the function of a chromatin insulator. Cell 92:511-521[Medline].

GEYER, P. K. and V. G. CORCES, 1992  DNA position-specific repression of transcription by a Drosophila zinc finger protein. Genes Dev. 6:1865-1873[Abstract/Free Full Text].

GEYER, P. K., C. SPANA, and V. G. CORCES, 1986  On the molecular mechanism of gypsy-induced mutations at the yellow locus of Drosophila melanogaster. EMBO J. 5:2657-2662[Medline].

KIM, A., C. TERZIAN, P. SANTAMARIA, A. PELISSON, and N. PRUD'HOMME et al., 1994  Retroviruses in invertebrates: the gypsy retrotransposon is apparently an infectious retrovirus of Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 91:1285-1289[Abstract/Free Full Text].

KIRCHNER, J., C. M. CONNOLLY, and S. B. SANDMEYER, 1995  Requirement of RNA polymerase III transcription factors for in vitro position-specific integration of a retrovirus like element. Science 267:1488-1491[Abstract/Free Full Text].

LEE, S. and M. D. GARFINKEL, 2000  Characterization of Drosophila OVO protein DNA binding specificity using random DNA oligomer selection suggests zinc finger degeneration. Nucleic Acids Res. 28:826-834[Abstract/Free Full Text].

LU, J., J. ANDREWS, D. PAULI, and B. OLIVER, 1998  Drosophila OVO zinc-finger protein regulates ovo and ovarian tumor target promoters. Dev. Genes Evol. 208:213-222[Medline].

MEVEL-NINIO, M., M-C. MARIOL, and M. GANS, 1989  Mobilization of the gypsy and copia retrotransposons in Drosophila melanogaster induces reversion of the ovo^D dominant female-sterile mutations: molecular analysis of revertant alleles. EMBO J. 8:1549-1558[Medline].

MEVEL-NINIO, M., E. FOUILLOUX, I. GUENAL, and A. VINCENT, 1996  The three dominant female-sterile mutations of the Drosophila ovo gene are point mutations that create new translation-initiator AUG codons. Development 122:4131-4138[Abstract].

OLIVER, B., N. PERRIMON, and A. P. MAHOWALD, 1987  The ovo locus is required for sex-specific germ line maintenance in Drosophila. Genes Dev. 1:913-923[Abstract/Free Full Text].

PELISSON, A., S. U. SONG, N. PRUD'HOMME, P. A. SMITH, and A. BUCHETON et al., 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[Medline].

PERRIMON, N., 1984  Clonal analysis of dominant female-sterile, germline-dependent mutations in Drosophila melanogaster. Genetics 108:927-939[Abstract/Free Full Text].

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[Abstract].

ROBERTSON, H. M., C. R. PRESTON, R. W. PHILLIS, D. M. JOHNSON-SCHLITZ, and W. K. BENZ et al., 1988  A stable genomic source of P element transposase in Drosophila melanogaster. Genetics 118:461-470[Abstract/Free Full Text].

RUBIN, G. M. and A. C. SPRADLING, 1982  Genetic transformation of Drosophila with transposable element vectors. Science 218:348-353[Abstract/Free Full Text].

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[Abstract/Free Full Text].

SONG, S. U., M. KURKULOS, J. D. BOEKE, and V. G. CORCES, 1997  Infection of the germ line by retroviral particles produced in the follicle cells: a possible mechanism for the mobilization of the gypsy retroelement of Drosophila. Development 124:2789-2798[Abstract].

SPANA, C. and V. G. CORCES, 1990  DNA bending is a determinant of binding specificity for a Drosophila zinc finger protein. Genes Dev. 4:1505-1515[Abstract/Free Full Text].

SPRADLING, A. C., 1993 Developmental genetics of oogenesis, pp. 1–70 in The Development of Drosophila melanogaster, edited by M. BATE and A. MARTINEZ ARIAS. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

WIESCHAUS, E. and J. SZABAD, 1979  The development and function of the female germ line in Drosophila melanogaster: a cell lineage study. Dev Biol. 68:29-46[Medline].

ZENNOU, V., C. PETIT, D. GUETARD, U. NERHBASS, and L. MONTAGNIER et al., 2000  HIV-1 genome nuclear import is mediated by a central DNA flap. Cell 101:173-185[Medline].

ZHU, Y., S. ZOU, D. A. WRIGHT, and D. F. VOYTAS, 1999  Tagging chromatin with retrotransposons: target specificity of the Saccharomyces Ty5 retrotransposon changes with the chromosomal localization of Sir3p and Sir4p. Genes Dev. 13:2738-2749[Abstract/Free Full Text].

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via Google Scholar Google Scholar Articles by Labrador, M. Articles by Corces, V. G. Search for Related Content PubMed PubMed Citation Articles by Labrador, M. Articles by Corces, V. G.