Restrictive flamenco Alleles Are Maintained in Drosophila melanogaster Population Cages, Despite the Absence of Their Endogenous gypsy Retroviral Targets (original) (raw)

Abstract

The flamenco (flam) locus, located at 20A1-3 in the centromeric heterochromatin of the Drosophila melanogaster X chromosome, is a major regulator of the gypsy/mdg4 endogenous retrovirus. In restrictive strains, functional flam alleles maintain gypsy proviruses in a repressed state. By contrast, in permissive strains, proviral amplification results from infection of the female germ line and subsequent insertions into the chromosomes of the progeny. A restrictive/permissive polymorphism prevails in natural and laboratory populations. This polymorphism was assumed to be maintained by the interplay of opposite selective forces; on one hand, the increase of genetic load caused by proviral insertions would favor restrictive flam alleles because they make flies resistant to these gypsy replicative transpositions and, on the other, a hypothetical resistance cost would select against such alleles in the absence of the retrovirus. However, the population cage data presented in this paper do not fit with this simple resistance cost hypothesis because restrictive alleles were not eliminated in the absence of functional gypsy proviruses; on the contrary, using 2 independent flam allelic pairs, the restrictive frequency rose to about 90% in every experimental population, whatever the pair of alleles and the allelic proportions in the initial inoculum. These data suggest that the flam polymorphism is maintained by some strong balancing selection, which would act either on flam itself, independently of the deleterious effect of gypsy, or on a hypothetical flanking gene, in linkage disequilibrium with flam. Alternatively, restrictive flam alleles might also be resistant to some other retroelements that would be still present in the cage populations, causing a positive selection for these alleles. Whatever selective forces that maintain high levels of restrictive alleles independently of gypsy, this unknown mechanism can set up an interesting kind of antiviral innate immunity, at the population level.

Introduction

The coevolution between parasites and their hosts often involves “arms races” that reinforce their mutual interactions. The purpose of the present study was to appreciate some of the evolutionary forces that drive the host–parasite interactions between the genome of Drosophila melanogaster and the gypsy retrovirus.

Gypsy/mdg4 is the first retrovirus that was recognized, as such, in invertebrates (Kim et al. 1994; Song et al. 1994). Like other retroviruses, gypsy is able to integrate its genetic material, as a provirus, in the host genomic DNA. In fact, gypsy is an endogenous retrovirus, meaning that the proviruses are integrated into the germ line chromosomes and can thereby be transmitted vertically, through fly generations, along with the proper Drosophila genetic material. The endogenous gypsy genetic material mainly consists of several dozen old defective proviruses that have accumulated many structural and functional alterations (Lambertsson et al. 1989). Each has generally been inserted at its own genomic location since the ancestor of all present-day flies and is therefore shared by all strains. As this is the case for the other Drosophila retroelements, most of these vestiges of a former gypsy activity are located in the heterochromatic pericentromeric regions of the genome (Vaury et al. 1989). In addition, the genome of some strains contain functional gypsy proviruses that recently integrated in what are now strain-specific euchromatic locations. The strains that are completely devoid of such potentially active proviruses are called “empty” strains.

An increase in gypsy copy number requires not only the functionality of the proviruses but also the permissive character of the host. The latter feature is genetically determined by an as-yet-uncloned X-linked heterochromatic locus (Prud'homme et al. 1995; Robert et al. 2001) called flamenco (flam). The restrictive function of flam represses gypsy expression in the somatic follicle cells of the ovary (Pelisson et al. 1994). As a result, proviral integration in the germ line chromosomes of the progeny is prevented (Chalvet et al. 1999). Deficiencies of the flam region behave like natural permissive alleles. Both are recessive because their combination with a single dose of a restrictive flam allele is sufficient to prevent gypsy replicative transposition (Prud'homme et al. 1995). Because these large deficiencies uncover vital flanking genes, they cannot be used to tell what the phenotype of the complete loss of flam function would look like. Whatever role this function plays in the biology of D. melanogaster, it does not seem to be affected in any of the natural flam alleles because, in our laboratory conditions, they do not show any visible phenotype other than the permissive character.

The flam genotype can be inferred from the expression level of ovarian β-galactosidase encoded by a gypsy-lacZ reporter transgene (Pelisson et al. 1994). This assay shows a semidominance effect because homozygosity of a restrictive allele can repress more efficiently than a heterozygote with a permissive allele (Pelisson et al. 1997).

According to previous genomic analyses (Robert et al. 2001), the flam locus spans at least 130 kb of a repeat-rich heterochromatic region that includes a sequencing gap of unknown size. Although the molecular characterization of flam is still in progress, we have some hints about the mechanism of gypsy repression in which flam is involved (Sarot et al. 2004). Available evidence supports a model whereby some defective gypsy proviruses can trigger the production of small interfering RNAs that would target the homologous transcripts for repression.

The results of a thorough genetic study of flam indicate that this locus displays a surprisingly high level of polymorphism both in the laboratory and in the nature (Pelisson et al. 1997). This is especially intriguing when one considers that it maps in the non–recombination-prone 20A1-3 pericentromeric region. Such regions are usually characterized by reduced levels of polymorphism as a result of genetic hitchhiking associated with the fixation of advantageous mutants (Begun and Aquadro 1992). This paradox could be explained by assuming that the flam polymorphism results from opposite selective forces. As suggested previously (Pelisson et al. 1997), selection could be driven, on one hand, by the positive impact of the enhanced fitness of the flam restrictive individuals in the presence of gypsy and, on the other hand, by a negative impact of the restrictive alleles on host fitness in the absence of gypsy, that is, a cost of resistance. However, the population cage study presented here does not support the resistance cost hypothesis because the flam polymorphism could be maintained in the absence of any functional gypsy provirus.

Materials and Methods

Laboratory D. melanogaster Strains

Flies were maintained on standard fly food (Gans et al. 1975) at 26 °C. All strains used in this study were “empty” strains, devoid of euchromatic gypsy proviruses (data not shown). The absence of active gypsy in these stocks was checked in Southern blots (data not shown) by the lack of the _Hin_dIII 1.6-kb band that has been reported to be diagnostic of functional gypsy copies (Chalvet et al. 1998). Finally, the absence of any _gypsy_-induced mutability in the permissive strains was also consistent with the lack of active provirus.

A70, an attached-X tester strain, is homozygous for the #_12a gypsy_-lacZ reporter transgene (Sarot et al. 2004). The X-linked Df(1)l11 deficiency carried by this strain uncovers the flam locus as well as vital flanking genes (Prud'homme et al. 1995). The recessive lethality of males is rescued by the Y-linked y+Ymal106 duplication.

OR(R), OR(P), SS(R), and SS(P) are the generic names of the 16 independent X chromosomal clones (4 replicates for each genotype) that were isolated from the previously described white-eyed wOR(P) and wSS(P) permissive empty stocks (Pelisson et al. 1997; Chalvet et al. 1999) as follows. Individual wOR(P) and wSS(P) males were successively mated with an attached-X stock and a FM3 balanced stock, both devoid of active gypsy proviruses. Each w/w homozygous chromosomal clone was established by crossing, respectively, the w/Y male with the w/FM3 female progenies and by getting rid of the FM3 balancer at the next generation. Females of each clone were crossed with A70 males to test its ability to repress the expression of the _gypsy_-lacZ reporter (see below), and 8 clones of each strain, 4 restrictive (R) and 4 permissive (P), were chosen for the present study. When genotyped with the ZAM-LacZ assay described below, all these clones were shown to be able to repress the ZAM retroelement.

As illustrated in figure 1, introduction of the gypsy-lacZ transgene into each of the 16 X chromosomal clones was achieved by crossing transgenic males to females of each clone, backcrossing the male progeny and then sibmating to make the transgene homozygous in each of these 16 genetic backgrounds.

FIG. 1.—

Mating scheme used to combine the gypsy-lacZ reporter (#12a) with each of the 16 X chromosomes cloned (w). The #12a transgene was introduced into 4 permissive and 4 restrictive X chromosomal clones extracted from each of the wOR and wSS polymorphic strains. Two successive backcrosses using females of the same chromosomal clone were followed by 2 successive sibmatings. In the w background, flies with the _miniwhite_-containing #12a transgene have colored eyes, 2 doses of the transgene being discriminated from a single dose by the corresponding darker eye color phenotype.

Natural Populations

In October 2000, at least 50 individuals were captured in each of 2 different cellars of the Montpellier area, Corconnes and Vendargues. DIS13-1 to DIS13-11 are 11 isofemale lines trapped by Dr Daniel Lachaise in Uganda in 2004.

Population Cages

Population cages were inoculated with samples of 200 fertilized females containing various proportions of either of the 2 allelic pairs. In each case, the 4 replicate clones were used to buffer the genetic background. So, for instance, the wSS population was inoculated with 10% of permissive alleles contained in the sample of 200 female founders corresponding to 5 and 45 females of each of the 4 SS(P); #12a and SS(R); #12a clones, respectively.

Cages were kindly provided by the Rouyer's laboratory. They consist of plastic boxes (30 × 20 × 10 cm), the bottom of which is fitted with 12 cups. Three times a week, they were supplied with 2 cups each containing about 15 g of standard fly food. Each cup therefore stayed 2 weeks in the cage, which is far enough for 1 generation to develop at 26 °C. Although limited, this regular supply of food could sustain continuous populations of several thousands of tiny starving individuals.

Determination of the allelic frequencies was performed as follows: 200–300 eggs were collected during several hours on a cup of fresh food and were allowed to develop outside of the cage, without food restriction. About 100 newly hatched females were fed for 2 days on fresh yeast, and the lacZ staining of one of their ovaries was performed as described below.

Genotyping

The flamP/flamR Polymorphism.

A flam allele is characterized as restrictive or permissive depending on whether or not it can repress the ovarian expression of the gypsy-lacZ reporter transgene. The β-galactosidase staining assay was performed essentially as previously described (Sarot et al. 2004). Briefly, ovaries were fixed for 5 min at room temperature in 2% formaldehyde and 0.2% glutaraldehyde, thoroughly rinsed, and then stained for 3 h at 23 °C with 3 mg/ml X-Gal. The genotype of females originating from cage populations could be directly determined because they already contained the gypsy-lacZ reporter. The 3 different female genotypes (flamP/flamP, flamP/flamR, and flamR/flamR) were distinguished owing to the semidominance effect. Indeed, the whole follicular epithelium of stage 10 homozygous permissive egg chambers was already homogeneously stained after 30 min, whereas no significant staining was visible in homozygous restrictive ovaries, even after the 180-min incubation time (fig. 2A). In heterozygous ovaries, β-galactosidase activity was only detected in the centripetal follicle cells, which cover the anterior part of the oocyte and in a few patches of cells randomly distributed throughout the epithelium. The same criteria were used to operationally define 3 levels of gypsy-lacZ expression in a tentative classification of the X chromosomes extracted from natural populations as permissive, “strongly” restrictive, and “weakly” restrictive, respectively. In this case, each tested X chromosome was combined with the Df(1)l11 flam deficiency in the presence of the #12a gypsy-lacZ reporter. Such hemizygous ovaries were dissected in the progeny of mass crosses involving at least 10 tested females and 10 males of the A70 tester stock.

FIG. 2.—

Validation of the β-galactosidase genotyping method by the _Xmn_I restriction site criterion. (A) Examples of each of the 3 different levels of gypsy-lacZ expression (as well as the 3 corresponding inferred genotypes), which could be typically distinguished by β-galactosidase assaying the ovaries of females extracted from the experimental populations. In fact, the samples of ovaries shown here were dissected either from a permissive (left) or a restrictive (right) wOR; #12a X chromosomal clone or from the heterozygous F1 progeny (middle) obtained by crossing them with each other. The dense blue β-galactosidase staining of flamP/flamP follicle cells appears black on this grayscale picture. In flamP/flamR ovaries, only the centripetal follicle cells of stages 10–12 show this strong level of staining. (B) Illustration of the corresponding _Xmn_I restriction site polymorphism. Heads of 3 individuals whose ovaries are shown in (A) were used to perform genomic DNA amplification in the flam region. The 1.27-kb PCR fragment produced by the homozygous permissive female (flamP/flamP) was missing the _Xmn_I diagnostic restriction site. By contrast, the flamR/flamR female genomic PCR product was cut into 2 fragments (0.67 and 0.6 kb) characteristic of _Xmn_I-containing X chromosomes. As expected, the heterozygous female produced the 3 possible bands in stoichiometric ratios (1.27 = 0.67 + 0.6). (C) The rare _Xmn_I-containing restrictive X chromosomes inoculated into one of the most permissive populations (seeded with 95% of homozygous permissive founders devoid of this _Xmn_I restriction site) had almost invaded the population at the 27th generation. Shown are the _Xmn_I-digested amplification products of 8 pools of 5 genomic DNAs extracted from the heads of females that had been individually sorted into the major flamR/flamR class, according to their ovarian lacZ staining. The absence of uncut DNA indicates that each of these 40 females was also homozygous for the _Xmn_I haplotype.

The XmnI Restriction Polymorphism.

Polymerase chain reaction (PCR) amplification of a DNA fragment in the flam region was performed with 1 μl of genomic DNA isolated as follows. Individual fly heads were crudely squashed with a pipet tip in 50 μl of squashing buffer (10 mM Tris pH 8.2, 1 mM ethylenediaminetetraacetic acid, 25 mM NaCl, and 0.2 mg/ml Proteinase K) and incubated for 30 min at 29 °C, then for 2 min at 85 °C. The following pair of primers was used at an annealing temperature of 56 °C: 5′-CCTCAGTTAAAAATATTGGACCACTT-3′ and 5′-TAGACATTTCACACTGGCCATAGTAA-3′. The PCR products were purified using the Qiaquick PCR spin columns (Qiagen, Courtaboeuf, France) and were then either sequenced or digested with _Xmn_I, according to the manufacturer's recommendations (Promega, Charbonnières, France). The digestion products were run on a 1.5% agarose gel in 1× tris-borate-EDTA.

The ZAM-LacZ Assay (Desset et al. 2003).

The ability of the 16 independent X chromosomal clones to repress the ZAM retroelement was tested using a similar assay as the aforementioned anti-gypsy resistance test. The reporter transgene consists of the ZAM promoter fused to the Escherichia coli lacZ gene. Tester females, containing this ZAM-lacZ transgene and unable to repress ZAM, were crossed to males under test and the level of presence/absence of β-galactosidase activity was monitored in the follicle cells of the heterozygous daughters.

Results

Further Evidence for a High Level of flam Polymorphism in Natural Populations

Table 1 shows that, when combined with a large deficiency of the flam locus (Df(1)l11), only half of the X chromosomes from the Uganda isofemale lines allowed complete expression of the gypsy-lacZ reporter transgene in the follicle cells of the ovary. About the same proportions were observed with 2 other natural populations, Vendargues and Corconnes, that were sampled and tested immediately after their capture en masse in the south of France.

Table 1.

Assay of the _gypsy_-Repressing Ability of flam Alleles Sampled from the DIS, Vendargues, and Corconnes Natural Populations

a

Flies of natural populations were tested as previously described (Sarot et al. 2004), by crossing sets of 10 females with males of the A70 tester stock and assaying the β-galactosidase activity of the gypsy-lacZ reporter in individual ovaries of the hemizygous progeny. Each tested X chromosome, which had been thus combined with the Df(1)l11 deficiency of the A70 stock, in the presence of the #12a gypsy-lacZ reporter was operationally classified as permissive or restrictive (strong or weak) according to the level of β-galactosidase expression observed in the corresponding hemizygous ovary (see Materials and Methods).

b

Pooled results of the tests performed with the 11 DIS isofemale lines collected in Uganda.

c

Pooled results of 6 and 3 tests, which were performed with females of both french populations, Vendargues and Corconnes, respectively.

Table 1.

Assay of the _gypsy_-Repressing Ability of flam Alleles Sampled from the DIS, Vendargues, and Corconnes Natural Populations

a

Flies of natural populations were tested as previously described (Sarot et al. 2004), by crossing sets of 10 females with males of the A70 tester stock and assaying the β-galactosidase activity of the gypsy-lacZ reporter in individual ovaries of the hemizygous progeny. Each tested X chromosome, which had been thus combined with the Df(1)l11 deficiency of the A70 stock, in the presence of the #12a gypsy-lacZ reporter was operationally classified as permissive or restrictive (strong or weak) according to the level of β-galactosidase expression observed in the corresponding hemizygous ovary (see Materials and Methods).

b

Pooled results of the tests performed with the 11 DIS isofemale lines collected in Uganda.

c

Pooled results of 6 and 3 tests, which were performed with females of both french populations, Vendargues and Corconnes, respectively.

The fact that 2 of the 3 populations exhibited what looks like different restrictivity levels is reminiscent of the whole range of variations due to the flam allelic series, previously described in the laboratory (Pelisson et al. 1997). However, because these intrapopulation variations were not further analyzed genetically, it is not known whether they actually reflect the presence of different types of flam restrictive alleles in the wild or if they result from the effect of modifiers on the restrictivity phenotype. Note, however, that such putative modifiers should be dominant unless they are either located in the small portion of the X chromosome uncovered by the Df(1)l11 deficiency or already present in the tester X chromosome.

At the time of the test, 4 months after their capture, 5 out of the 11 DIS lines were still containing each of these 3 phenotypic categories, suggesting that this polymorphism, whatever its genetic determinism, could be maintained in these lines, at least during this short period of time.

Isolation of 2 Pairs of flam Alleles from 2 Polymorphic Laboratory Strains

We then chose to study 2 simpler cases of flam polymorphism that each involved a single pair of alleles. As reported previously (Pelisson et al. 1997; Chalvet et al. 1999), wOR(P) and wSS(P) are 2 of the permissive strains that do not display any _gypsy_-induced mutability because they are devoid of functional gypsy copies. Surprisingly, some restrictive flies once happened to be serendipitously detected in a subline of each of these stocks. The frequencies of these intruders were high enough to enable their isolation, as described in Materials and Methods, in 4 restrictive X chromosomal clones that were called OR(R) and SS(R) to distinguish them from their 4 permissive siblings called OR(P) and SS(P), respectively.

A 1.27-kb fragment, located in the flam region, upstream of the DIP1 gene, was PCR amplified and sequenced in each of the 8 OR chromosomal clones (see Materials and Methods). All sequences were found to be identical except for a single T/C nucleotide change at position 677 (corresponding to position 21,441,509 in the genome assembly, release 4.2.1), which created a _Xmn_I restriction polymorphism and enabled to discriminate the 4 _Xmn_I-containing OR(R) clones from the 4 OR(P) permissive clones (data not shown). The fact that this single nucleotide polymorphism was linked to the restrictive/permissive character in the small sample of 4 OR(R) and 4 OR(P) clones suggested that the restrictive allele did not arise inside the OR(P) stock as a revertant, but more likely resulted from the casual contamination of this stock. Whatever the case, the putative contamination event has not introduced any wild-type white allele or any functional provirus that would otherwise have invaded the stock and caused its mutability. Southern blotting data (not shown) also confirmed the absence of effective contamination by functional proviruses containing the diagnostic _Hin_dIII site (Chalvet et al. 1998). The same putative contaminant chromosome cannot be at the origin of the SS(P)/SS(R) polymorphism because SS(R) chromosomes are missing the _Xmn_I restriction site (data not shown).

A locus controlling the activity of the ZAM and Idefix retroelements was mapped at the base of the X chromosome (Desset et al. 2003). However, this locus was not reported to be very polymorphic, and indeed, as described in Materials and Methods, we could check that the 16 independent X chromosomal clones all contained the “stable” allele that is able to repress ZAM expression in the ovary (data not shown). So, our population cage study could be performed in the absence of any interference with this additional host–parasite interaction.

The Permissive/Restrictive Polymorphisms of the wOR and wSS Strains Are Maintained in Experimental Populations Devoid of Active gypsy Proviruses

With the aim of an easy flam phenotyping, we wanted the #12a autosomal gypsy-lacZ reporter transgene (Sarot et al. 2004) to be present in every fly of the experimental populations. That is why this transgene was first introduced into each of the 16 X chromosomal clones used to seed the populations, as illustrated in figure 1 and described in Materials and Methods. These _gypsy-lacZ_–containing stocks were used to seed 6 independent experimental populations (see Materials and Methods).

As shown in figure 2A, we could take advantage of the semidominance effect to phenotypically discriminate between the 3 different genotypes. In most cases, any ovary could thus be unambiguously assigned to 1 of the 3 genotypic classes. The robustness of this phenotyping could be further validated by the 2 following criteria. First, the frequencies of the 3 genotypes were always in the Hardy–Weinberg proportions (data not shown). Second, at the 27th generation of a population seeded with 95% OR(P); #12a founders, females were simultaneously tested both for the level of lacZ expression in their ovary and for the _Xmn_I restriction site polymorphism described above. While most ovaries disclosed the absence of staining typical of the homozygous restrictive genotype, the corresponding genomic DNAs were all found to be homozygous for the _Xmn_I restriction site (fig. 2C).

This haplotype, which had been introduced into the population by only 5% of restrictive founders, had therefore already invaded 80% of the cage by generation 27 (fig. 3). This observation ruled out the possibility of an unusually high rate of mutation that would have transformed most of the permissive alleles into restrictive alleles. Instead, the decrease of the permissive allelic frequency in all 4 permissive populations (fig. 3) was more likely to result from selection. In fact, a balancing selection might have operated because every population eventually tended toward the same frequency of about 10% permissive alleles, whatever the initial frequency (fig. 3). However, because this frequency is as low as that of the lowest inoculation frequency, we could not observe a rise of the permissive frequency up to the plateau. So, we cannot discriminate between a stable equilibrium and a continuous decrease of permissive frequency. Although only studied in a single cage, the wSS flam polymorphism seemed to behave similarly and to be therefore driven by the same unknown selective forces as the wOR one.

FIG. 3.—

Evolution of the frequencies of 2 flam allelic pairs in experimental populations inoculated at various initial proportions. The data from the only population cage that was seeded with flies containing the wSS pair of flam alleles (10% of homozygous permissive and 90% of homozygous restrictive founders) can be distinguished from the others by the filled circle symbol. All populations were devoid of functional gypsy proviruses. One experiment (inoculated with 90% of wOR permissive alleles) had to be stopped prematurely at the 44th generation, before the equilibrium was reached, because of a contamination of the cage by flies devoid of the miniwhite transgene. The evolution of all the other populations slowed down after they had reached the frequency of about 10% permissive alleles.

Disscussion

Gypsy is controlled by a Drosophila gene called flam, the restrictive alleles of which, unlike permissive alleles, maintain this retrovirus in a repressed state. Permissive strains containing functional gypsy proviruses are unhealthy probably because they undergo frequent mutations that result from efficient gypsy replicative transposition. Stabilization of such mutable strains sometimes happens because of the fixation of restrictive alleles that stop the increase of genetic load (Pelisson et al. 1997). The persistence of permissive alleles, both in the laboratory and in the wild (Pelisson et al. 1997), might be explained by a low number of euchromatic, possibly functional, gypsy proviruses (Modolell et al. 1983; Bayev et al. 1984; Biemont et al. 1994) and a putative selective cost of the restrictive alleles in the absence of these parasites. To address these issues we set up cage populations and followed the evolution of a permissive–restrictive allelic pair that had been extracted from the wOR laboratory strain. In spite of the absence of functional gypsy proviruses in these populations, the frequency of the restrictive phenotype rose to a surprisingly high level, clearly demonstrating the absence of a cost of resistance.

The characteristics of this evolution were the following. 1) The dynamics of the restrictive/permissive phenotypes studied here truly resulted from selection, not mutation. The molecular nature of this allelic pair could not be assayed because the DNA sequence of flam is not yet available. However, we could take advantage of the fact that a polymorphic _Xmn_I restriction site was specifically linked to the 5% of restrictive alleles inoculated into one of the experimental populations. Twenty-seven generations later, when most chromosomes were phenotypically characterized as restrictive, this marker was still linked to each of them, providing strong evidence for an almost complete invasion of the population by the restrictive haplotype. 2) Indeed, whatever the composition of the initial inoculate (5, 10, 50, or 90% of restrictive alleles), every population tended to stabilize at the same restrictive frequency of 90%. However, we could not definitely discriminate between a real plateau due to some unknown mechanism of balancing selection, on one hand, and a continuous decrease of permissive allelic frequency due to directional selection, on the other hand. 3) The restrictive–permissive polymorphism originating from the unrelated SS strain also seemed to be stabilized at the same 90% restrictive level, suggesting that the same selective forces were operating on the restrictive and permissive chromosomes of both strains.

Even on the hypothesis of a true balancing selection, the effect of an additional selection, due to the presence of active gypsy proviruses in the nature, is expected to tend to the final elimination of permissive alleles from natural populations. However, as shown previously (Pelisson et al. 1997) and in the present study (table 1), such populations still exhibit high frequencies of permissive flam alleles. This might mean either that functional copies are too rare in the nature to have any significant impact on the flam polymorphism or that they are held in check by additional (copy-dependent, cosuppressionlike?) mechanisms able to take over the repression in permissive females.

Alternatively, because we could not perform the experiments for an infinite number of generations, we may have missed the ultimate elimination of the permissive alleles from the cages. Such a selection against homozygous permissive females could be explained, despite the absence of functional gypsy, if they were also sensitive to the deleterious presence of another hypothetical transposable element. The presence of this unsuspected cause of selection in the cages would have antagonized and eventually masked the selection cost of restrictive alleles. We are currently investigating the possibility that flam would confer resistance to other retroelements than gypsy.

Finally, it would be interesting to check that the probability of infecting an experimental population by gypsy virions, as previously described for a permissive genotype (Kim et al. 1994; Song et al. 1994), will be reduced by the presence of restrictive alleles in this population. Should this prediction be correct, then the unknown mechanism that maintains high levels of restrictive alleles independently of the presence of gypsy might be able to set up a sort of innate immunity, at the population level, against the horizontal transfer of an otherwise vertically transmitted retrovirus.

We are very grateful to François Rouyer and André Klarsfeld for generously sharing the population cage system and to Daniel Lachaise, our late colleague, who kindly gave us the DIS lines. We thank Nicolas Gilbert for critically reading the manuscript. Significant improvements of the manuscript are also due to insightful comments by unknown reviewers. This work was supported by grants from the Centre National pour la Recherche Scientifique, the Association pour la Recherche sur le Cancer, and the E.C.'s Research Training Network Silencing in Different Organisms.

References

Structural organization of transposable element mdg4 from Drosophila melanogaster and a nucleotide sequence of its long terminal repeats

,

Nucleic Acids Res

,

1984

, vol.

12

(pg.

3707

-

3723

)

Levels of naturally occurring DNA polymorphism correlate with recombination rates in D. melanogaster

,

Nature

,

1992

, vol.

356

(pg.

519

-

520

)

Population dynamics of the c_opia, mdg1, mdg3, gypsy_, and P transposable elements in a natural population of Drosophila melanogaster

,

Genet Res

,

1994

, vol.

63

(pg.

197

-

212

)

Potentially active copies of the gypsy retroelement are confined to the Y chromosome of some strains of Drosophila melanogaster possibly as the result of the female-specific effect of the flamenco gene

,

J Mol Evol

,

1998

, vol.

46

(pg.

437

-

441

)

Proviral amplification of the gypsy endogenous retrovirus of Drosophila melanogaster involves env-independent invasion of the female germline

,

Embo J

,

1999

, vol.

18

(pg.

2659

-

2669

)

COM, a heterochromatic locus governing the control of independent endogenous retroviruses from Drosophila melanogaster

,

Genetics

,

2003

, vol.

164

(pg.

501

-

509

)

Isolation and characterization of sex-linked female-sterile mutants in Drosophila melanogaster

,

Genetics

,

1975

, vol.

81

(pg.

683

-

704

)

Retroviruses in invertebrates: the gypsy retrotransposon is apparently an infectious retrovirus of Drosophila melanogaster

,

Proc Natl Acad Sci USA

,

1994

, vol.

91

(pg.

1285

-

1289

)

Cloning and characterization of variable-sized gypsy mobile elements in Drosophila melanogaster

,

Plasmid

,

1989

, vol.

22

(pg.

22

-

31

)

Drosophila melanogaster mutations suppressible by the suppressor of hairy-wing are insertions of a 7.3-kilobase mobile element

,

Proc Natl Acad Sci USA

,

1983

, vol.

80

(pg.

1678

-

1682

)

Gypsy transposition correlates with the production of a retroviral envelope-like protein under the tissue-specific control of the Drosophila flamenco gene

,

Embo J

,

1994

, vol.

13

(pg.

4401

-

4411

)

About the origin of retroviruses and the co-evolution of the gypsy retrovirus with the Drosophila flamenco host gene

,

Genetica

,

1997

, vol.

100

(pg.

29

-

37

)

Flamenco, a gene controlling the gypsy retrovirus of Drosophila melanogaster

,

Genetics

,

1995

, vol.

139

(pg.

697

-

711

)

Characterization of the flamenco region of the Drosophila melanogaster genome

,

Genetics

,

2001

, vol.

158

(pg.

701

-

713

)

Evidence for a _piwi_-dependent RNA silencing of the gypsy endogenous retrovirus by the Drosophila melanogaster flamenco gene

,

Genetics

,

2004

, vol.

166

(pg.

1313

-

1321

)

An env-like protein encoded by a Drosophila retroelement: evidence that gypsy is an infectious retrovirus

,

Genes Dev

,

1994

, vol.

8

(pg.

2046

-

2057

)

The beta heterochromatic sequences flanking the I elements are themselves defective transposable elements

,

Chromosoma

,

1989

, vol.

98

(pg.

215

-

224

)

Author notes

1

Present address: UMR 754, INRA-ENVL-UCBL-EPHE, Lyon, France.

Koichiro Tamura, Associate Editor

© The Author 2006. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, please e-mail: journals.permissions@oxfordjournals.org