The non-LTR retrotransposons R1 and R2 insert into the 28S rRNA genes of arthropods. Comparisons among Drosophila lineages have shown that these elements are vertically inherited, while studies within species have indicated a rapid turnover of individual copies (elimination of old copies and the insertion of new copies). To better understand the turnover of R1 and R2, 200 retrotranspositions and nearly 100 eliminations have been scored in the Harwich mutation-accumulation lines of Drosophila melanogaster. Because the rDNA arrays in D. melanogaster are present on the X and Y chromosomes and no exchanges were detected in these lines, it was possible to show that R1 retrotranspositions occur predominantly in the male germ line, while R2 retrotranspositions were more evenly divided between the germ lines of both sexes. The rate of elimination of elements from the Y rDNA array was twice that of the X rDNA array with both chromosomal loci containing regions where the rate of elimination was on average eight times higher. Most R1 and R2 eliminations appear to occur by large intrachromosomal events (i.e., loopout events) that involve multiple rDNA units. These findings are interpreted in light of the known abundance of R1 and R2 elements in the X and Y rDNA loci of D. melanogaster.
THE mobile elements R1 and R2 insert into specific sites a short distance apart in the 28S genes of arthropods (reviewed in Eickbush 2002). While both R1 and R2 are non-LTR retrotransposons, they are only distantly related (Maliket al. 1999). Phylogenetic analyses have suggested that these elements are highly stable in insect lineages and are likely to have been inherited via vertical transmission since the origin of the arthropods >600 million years ago (Latheet al. 1995; Burkeet al. 1998; Maliket al. 1999; Gentileet al. 2001). Their persistence is remarkable given that the insertion of either element renders that gene unavailable for the synthesis of functional 28S rRNA (Kidd and Glover 1981; Longet al. 1981; Jamrich and Miller 1984) and it is not unusual for the elements to occupy 30–60% of an insect's 28S genes (Jakubczaket al. 1991; Burkeet al. 1993; Latheet al. 1995).
The tandemly repeated rRNA genes located within an rDNA array undergo concerted evolution, a process that leads to sequence uniformity of all units within the array (Coenet al. 1982; Lyckegaard and Clark 1991; Schlötterer and Tautz 1994; Polancoet al. 1998). Several findings suggest that the abundance and persistence of R1 and R2 are also affected by these recombinational mechanisms. Analysis of a Drosophila simulans population revealed that each isofemale line had a unique set of R1 and R2 insertions (Pérez-González and Eickbush 2001), indicating that individual copies of R1 and R2 are rapidly eliminated from the rDNA locus. In a study of the Harwich mutation-accumulation lines of D. melanogaster (Pérez-González and Eickbush 2002), it was found that rates of R1 and R2 elimination were much higher than those observed for the elimination of transposons and retrotransposons that insert throughout the genome (Nuzhdinet al. 1997). R1 elements in most Harwich lines were retrotransposing at a rate higher than their elimination and thus were expanding in number, while R2 elements were not retrotransposing as frequently as they were being eliminated and thus their numbers were being reduced in these lines (Pérez-González and Eickbush 2002). This independence in the rates of R1 and R2 retrotransposition can explain the large differences in levels of R1 and R2 that have been observed in different species (Eickbush and Eickbush 1995; Latheet al. 1995) or in different strains of D. melanogaster (Jakubczaket al. 1992).
In D. melanogaster, the rDNA loci are located on both the X and Y chromosomes (Ritossa 1976). The X and Y rDNA units are highly similar in sequence but can differ in the total number of repeating units, the length of their intergenic spacers, the sequences of the internal transcribed spacer, the distribution of R1 and R2 elements, and even in the expression of the two loci (Tartof and Dawid 1976; Wellaueret al. 1978; Williamset al. 1987; Englandet al. 1988; Clarket al. 1991; Lyckegaard and Clark 1991; Polancoet al. 1998). Thus the X and Y rDNA arrays seem to be evolving predominantly in a haplotypic fashion with occasional exchange events to maintain the similarity of the two arrays (Coen and Dover 1983; Gillingset al. 1987; Williamset al. 1987; Polancoet al. 2000).
In this report we have characterized the chromosomal location of the many retrotranspositions and eliminations observed in our previous study of the Harwich accumulation lines of D. melanogaster (Pérez-González and Eickbush 2002). These studies provide insights into the frequency of exchange between the rDNA loci of the X and Y chromosomes, the germ line in which these elements are active, and whether their eliminations are uniform across the two rDNA loci.
MATERIALS AND METHODS
Fly stocks and DNA isolation: The Harwich lines were a gift of T. F. C. Mackay. Line designations were as in Mackay et al. (1992) and all flies were collected from the 353rd generation. Genomic DNA was isolated from individual males and females using procedures described previously (Pérez-González and Eickbush 2002).
PCR amplification: DNA fragments representing the 5′ junctions between the 28S rRNA gene and either R1 or R2 elements were generated by PCR amplification using an oligonucleotide primer that annealed to the 28S gene 73 bp (R1) or 80 bp (R2) upstream of each element's insertion site in combination with 11 different R1 and 6 different R2 oligonucleotide primers that annealed to specific locations within each element (listed in Pérez-González and Eickbush 2002 and Table 1). These primers were spaced along each element to give maximum resolution of the 5′-truncated elements present in each line on 5% native polyacrylamide gels. To detect R1 insertions that had an upstream R2 insertion, the 28S gene primer upstream of the R1 site was replaced with the primer 5′-TTGGCAGACCTAGTATCTTTC-3′, which was complementary to a sequence of the R2 element starting ∼60 bp from its 3′ end. This R2 3′-end primer was used in combination with the complete set of R1-specific primers. To resolve fulllength R1 and R2 junctions, as well as a number of 5′-truncated R1 elements ∼0.5 kb in length, the primers located 180 bp and 5.3 kb from the R1 5′ end, as well as the primer located 150 bp from the R2 5′ end, were 32P-end labeled and the PCR products were separated on high-voltage denaturing 8% polyacrylamide gels, followed by autoradiographic visualization of bands (Pérez-González and Eickbush 2002).
R1 and R2 copy number determination: The PCR-amplified bands corresponding to the 5′-truncated copies of R1 and R2 were of similar intensity, suggesting these bands represented elements present at one copy per haploid genome. The total numbers of 5′-truncated copies of R1 and R2 elements were therefore determined by counting the number of bands visible with the complete range of PCR primers. In the case of the full-length R1 and R2 elements and 5′-truncated elements ∼0.5 kb in length, certain of the individual-length PCR products were of higher intensities, indicating that multiple copies of R1 or R2 had given rise to products of the same length. Because one of the primers used for these full-length amplifications was 32P-end labeled, the number of full-length elements in each band could be estimated by exposure of the dried gel to a PhosphorImager cassette and the relative intensity of bands quantified using a Storm Analyzer (Molecular Dynamics, Sunnyvale, CA).
Assay for individual copies of R1 and R2 within the X and Y rDNA loci: The imprecise mechanism of non-LTR retrotransposition makes it possible to monitor the insertion of new R1 and R2 elements and the elimination of older elements, even though all copies are inserted into the same 28S sequence. During targetprimed reverse transcription (TPRT), the reverse transcriptase starts at the 3′ end of the RNA transcript but frequently does not reach the 5′ end of the transcript (Luanet al. 1993; Eickbush 2002). Integration of these truncated copies results in the formation of a 5′-truncated element. Even if the reverse transcriptase does reach the 5′ end of the element's RNA transcript, short duplications, deletions, and nontemplated additions frequently generate variable “full-length” 5′ junctions (Georgeet al. 1996; Eickbushet al. 2000; Pérez-González and Eickbush 2001, 2002). To score these 5′ truncations and full-length 5′ variants in the Harwich mutation-accumulation lines, a series of PCR amplifications was performed using oligonucleotide primers specific to 28S sequences upstream of either the R1 or R2 insertion site in combination with primers specific to sequences from throughout the length of each element (see materials and methods). The PCR products were separated on high-resolution native polyacrylamide gels and visualized after ethidium bromide staining. In areas where many similar-length R1 or R2 copies made it difficult to score individual differences on these gels, 32P-labeled primers were used in the amplifications and the PCR products separated on high-voltage denaturing polyacrylamide gels followed by autoradiography to provide 1-bp resolution.
PCR products of the same length that were present in a majority of the Harwich lines were scored as ancestral variants present in the original inbred stock from which the lines were generated. The absence of PCR products corresponding to these ancestral variants in specific lines was scored as elimination events, while PCR bands unique to a particular line were scored as new insertion events.
Our previous study of R1 and R2 elements in the Harwich lines of D. melanogaster involved the analysis of single males from each line. While this analysis revealed large numbers of retrotranspositions and eliminations, it did not reveal their chromosomal locations. To determine the distribution of these events on the X and Y chromosome we compared the patterns of R1 and R2 elements in males and females of each line. An example of the PCR assay comparing the R1 and R2 5′ variants between an individual male (lanes 1 and 3) and female (lanes 2 and 4) from line 7 is shown in Figure 1B. PCR bands present in both males and females were scored as insertions on the X chromosome, while bands present only in males were scored as insertions on the Y. Of the seven elements seen in Figure 1B, one R1 and three R2 copies were located on the Y chromosome, while two R1 and one R2 copies were located on the X chromosome.
During our analysis of females from each line, we noted 10 R1 insertions and one elimination that were not detected on the X chromosome of the male previously tested from that line. Indeed, more extensive sampling of males and females of several lines revealed additional examples of R1 insertions that were not fixed on all X or Y chromosomes in the population (data not shown). Because of the large number of R1 insertions already scored, we have limited our analysis to only those events that had occurred on the X and Y chromosomes of the males used in our original study (Pérez-González and Eickbush 2002).
A small fraction of the rDNA units of D. melanogaster contains both R1 and R2 insertions (Roihaet al. 1981; Jakubczaket al. 1992). To identify these “doubly inserted units” the R1 upstream 28S primer was replaced with a primer that annealed to sequences starting ∼60 bp from the 3′ end of the R2 element (Figure 2A). An example of the PCR products generated using the upstream 28S primer (lane 1) and the upstream R2 primer (lane 2) can be seen in Figure 2B. While three 5′-truncated R1 elements were identified with the up-stream 28S gene primer, only the R1 copy that generated the 290-bp fragment in lane 1 was part of a doubly inserted rDNA unit. The PCR approach summarized in Figure 2 did not enable the identification of which R2 copy was part of the doubly inserted rDNA units.
Summary of the insertion and elimination events: Figure 3, A and B, provides a detailed accounting of the R1 and R2 events identified in the 19 Harwich lines examined in this study. In each part the 19 horizontal lines represent full-length R1 or R2 elements in D. melanogaster. The vertical bars bisecting each horizontal line represent the ancestral R1 and R2 elements present in each line with the position of the bar indicating the approximate point of its 5′ truncation. The region at the left of each line corresponds to full-length elements and has been expanded in scale to allow small differences in the lengths of the elements to be more easily distinguished. The prominent shaded bar within these expanded regions represents the canonical full-length R1 or R2 element as defined by Jakubczak et al. (1990). Ancestral insertions that were present on the X chromosome are marked with an X at the top of Figure 3, A and B, while copies on the Y chromosome are unmarked. Ancestral R1 elements in the same rDNA units as an R2 element are marked with a D (for double) at the bottom of Figure 3A.
Shaded ovals replacing the vertical bars in a strain (Figure 3) represent the elimination of ancestral copies in that line, while solid triangles represent new R1 or R2 insertions. The location of the triangles again indicates the position of a 5′ junction along the length of the element. New R1 and R2 insertions located on the X chromosome are individually marked with an X. New insertions that generated a doubly inserted rDNA unit are marked with a D above the triangle or vertical bar.
A limitation of our PCR approach is that it did not score a new insertion when that event gave rise to PCR products of identical length to that of a preexisting band. Similarly, if an ancestral band corresponded to two or more copies of an element, the elimination of one of these copies was not scored. Multicopy R1 and R2 bands are marked in Figure 3, A and B, with a dot at the bottom of each figure. On the basis of the number of elements in these multicopy bands, relative to those in single-copy bands, we estimated that only ∼30% of the R1 elements in the Harwich lines are scored as individual 5′ variants. Most of the R1 elements that could not be scored were full-length insertions with a precise 23-bp deletion of the 28S gene (Jakubczaket al. 1990; Pérez-González and Eickbush 2002). Approximately 40–60 of these canonical full-length R1 elements are present in each line. In the case of the R2 elements a greater percentage of the copies were 5′ truncated, and the 5′ junctions of full-length elements contained more small deletions, duplications, and insertions, resulting in a larger percentage of single-copy bands. As a consequence, we estimate that ∼70% of the R2 insertions and deletions within the Harwich lines were scored by our PCR approach.
R1 and R2 elements and X-Y exchange: As shown in Table 1, 22 ancestral R1 copies were present in the Harwich lines and equally divided between the X and Y chromosome. Thirty-four ancestral R2 copies were present in these lines with two-thirds located on the Y chromosome. All ancestral R1 and R2 copies scored as being on the X or on the Y in one line were consistently associated with that chromosome across all lines. Thus we saw no evidence that an ancestral R1 or R2 element moved from one chromosome to the other. It should be noted that because we did not monitor the Y locus in the absence of the X locus, our approach scored all movement of elements from the Y chromosome to the X, but only exchanges from the X to the Y if these events were associated with the loss of that element from the X.
The confidence limit for the rate of X-Y exchange, assuming that these events are Poisson distributed, was calculated in a manner similar to that described for the insertion or deletion rates when no events were observed (Nuzhdin and MacKay 1995). The 95% upper confidence limit for the rate of exchange when no events are observed is 2.996 (the expected number of exchanges at that rate) divided by the number of opportunities (19 lines × 353 generations). This gives a rate of X-Y exchange of <4.5 × 10–4/generation. This rate is consistent with previously calculated rates of ∼10–4 X-Y exchanges per generation based on variants present in the intergenic spacer and internal transcribed spacer regions of the rDNA repeats (Gillingset al. 1987; Williamset al. 1989).
Patterns of R1 and R2 retrotransposition: A total of 184 R1 retrotransposition events were detected in the 19 Harwich lines (Table 1). As shown in Figure 4A, the number of new insertions in each line varied from 0 to 6 for the X chromosome and from 2 to 18 for the Y chromosome. The preference of R1 insertions on the Y chromosome was found for all lines except line 1. As we did not observe transfers of the R1 or R2 elements from the X to the Y chromosome (see above), the discovery of new insertions on the Y chromosome indicated that these retrotransposition events occurred in the male germ line. Indeed, the threefold greater number of new insertions on the Y (141 events) compared to the X (43 events) is consistent with a model in which most, if not all, R1 retrotranspositions occurred in males (see discussion).
In contrast to the many R1 insertions, only 16 new R2 insertions were scored (Table 1). As shown in Figure 4B these insertions were found at a low level in all lines with no chromosome having more than two new insertions. The discovery of R2 insertions on the Y chromosome (seven events) again indicated that R2 retrotransposition occurred in the male germ line. However, the somewhat greater number of new insertions on the X (nine events) instead of the threefold lower level seen with the R1 elements suggested that the R2 retrotranspositions also occurred in females.
Patterns of R1 and R2 elimination from the rDNA loci: A combined 23 R1 and R2 eliminations were scored on the X chromosome while over three times as many, 71 events, were scored on the Y chromosome (Table 1). The rate of element elimination can be calculated by dividing the number of elimination events by the number of opportunities (number of ancestral copies × 19 lines × 353 generations). The rate of elimination of elements from the Y chromosome (3.1 × 10–4/element/generation) was nearly twice that of the X chromosome (1.6 × 10–4/element/generation). The elimination rate on the Y was similar for both R1 and R2 elements, while on the X the elimination rate of R1 was higher than that of R2 (Table 1).
Figure 5 shows the total number of elimination events on the X and Y chromosomes for each of the 19 Harwich lines. A wide range in the number of elimination events was seen on the Y chromosome (Figure 5A) with 10 lines experiencing 0–2 eliminations and 9 lines undergoing 4–11 eliminations. As can be seen in Figure 3, A and B, those lines with the largest numbers of eliminations had in many instances eliminated the same R1 and R2 copies. Indeed, 68% of the eliminations from the Y chromosome (48/71 events) involved the six ancestral R2 elements labeled a–f at the bottom of Figure 3B and the three ancestral R1 elements labeled g–i at the bottom of Figure 3A.
One can construct a probable relative order in which the frequently deleted R1 and R2 elements were located on the Y chromosome array by assuming that their eliminations involved the deletion of multiple rDNA units in a single event (Figure 6). Three of the Harwich lines deleted all nine elements while another four lines eliminated at least three contiguous elements. Finally, four lines eliminated only one or two variants with only one, line 16, requiring two separate events to explain its deletion pattern. It thus appears that R1 and R2 elements inserted on the Y chromosome were being eliminated at two rates. The nine elements indicated in Figure 6 appear to be in a recombination hotspot and were eliminated at a rate of 8.1 × 10–4/element/generation, while the remaining 25 elements on the Y chromosomes were eliminated at the rate of only 1.4 × 10–4/element/generation.
The number of elimination events on the X chromosome for each of the 19 Harwich lines is shown in Figure 5B. In contrast to the Y chromosome, most lines had three or fewer eliminations. However, the X chromosome also showed a subset of R1 and R2 copies that were eliminated in multiple lines. A total of 70% of the eliminations (16/23) involved two R1 elements and one R2 element (these copies are labeled α to γ at the bottom of Figure 3, A and B). Elimination of the R2 copy was frequently linked to the elimination of one of the R1 elements (3 of 4 events), suggesting that the recombination events responsible for these eliminations again involved multiple rDNA units. The location of two R1 elements in this hotspot (a combined 12 events) accounted for the higher rate of R1 elimination compared to R2 elimination from the X chromosome (Table 1).
These data suggest that there are areas of high turnover on the X and Y chromosome. The rate of elimination of R1 and R2 elements from these hotspots in the rDNA loci averaged eight times faster (64 events involving 12 elements) than the rate of elimination of copies outside these hotspots (30 events involving 44 elements).
Formation and eliminations of double-inserted rDNA units: Finally, we also examined the rate of elimination and insertion of elements that were part of doubly inserted rDNA units. The PCR assay revealed nine ancestral R1 copies that had an upstream R2 (five on the X chromosome and four on the Y; Figure 3A). Four of these double insertions involved the set of R1 copies only 0.5 kb in length. Because the length of the D. melanogaster R2 element poly(A) tail can vary from 10 to 30 bp (Lathe and Eickbush 1997), it was not possible to assign a specific R1 element to this set of four double insertions. We observed 10 eliminations of double inserts, all involving the set of R1 copies only 0.5 kb in length. Thus the average rate of double-insert eliminations was 1.9 × 10–4, similar to the average rate of single-element eliminations (2.5 × 10–4, Table 1).
Only three instances were observed in which a new R1 or R2 insertion resulted in the formation of a double-inserted rDNA unit (R1 elements in lines 5, 17, and 20 that are marked with a D in Figure 3A). In two instances (lines 5 and 17) an ancestral R1-inserted unit became a double-inserted unit. These instances clearly involved the insertion of an R2 element upstream of an R1 element. In the third instance (line 20) the formation of the double insertion involved a new R1 insertion. Because this line also contained new R2 insertions, we were unable to distinguish the order of events in which this double insert formed.
It is highly significant that of the 184 new R1 insertions, at most, only one occurred in a unit already occupied by an R2 element. Because in these Harwich lines ∼30% of the rDNA units available for R1 insertions (i.e., units not already containing an R1 insertion) are inserted with R2 elements, we should have seen the formation of >50 double-inserted units if the R1 insertions were random. Clearly the presence of an R2 insertion in a 28S gene greatly inhibited R1's ability to insert into that gene. In contrast, of the 16 new R2 insertions, 2 and possibly 3 formed double insertions. Because ∼50% of the available rDNA units for R2 insertions are already inserted with R1 elements, we should have seen the formation of ∼8 double-inserted units if the R2 insertions were random. Thus R2 elements also appear inhibited from inserting upstream of an R1 element, but this inhibition is not as severe as the inhibition of R1 elements from inserting downstream of an R2 element.
Using 5′-junction variation generated during retrotransposition in a collection of laboratory-maintained D. melanogaster lines, we have been able to address a number of questions concerning the integration and elimination of R1 and R2 elements from the rDNA loci. First we demonstrated that copies of these elements were rarely transferred between the rDNA loci of the X and Y chromosome. Using either intergenic spacer repeats or internal transcribed spacer polymorphisms as markers for X-Y exchange, previous studies have suggested that the exchange of rDNA units between the X and Y chromosomes occurs at a rate of ∼10–4 events/generation (Coen and Dover 1983; Gillingset al. 1987: Williamset al. 1989; Polanco et al. 1998, 2000). In our study we observed no exchanges in 22 specific R1 and R2 insertions on the X chromosome and 34 specific insertions on the Y chromosome. We estimate that the rate of X-Y exchange is <4.5 × 10–4 events/generation, consistent with the low frequency of exchange seen in the previous studies.
The absence of exchanges between the X and Y rDNA loci involving the ancestral R1 and R2 insertions suggested that any new insertions seen on one chromosome originated on that chromosome. Therefore the observation of new R1 and R2 insertions on the Y chromosome indicated that these retrotransposition events occurred in the male germ line. Indeed, finding three times the number of new R1 insertions on the Y chromosome as compared to the X suggested that R1 retrotransposition occurs predominantly, if not exclusively, in the male germ line. An X chromosome in a population spends only one-third of its existence in the male germ line while the Y is restricted to the male germ line. Thus, assuming approximately equal numbers of rDNA units on the X and Y chromosomes and equal probability of insertion into rDNA units of each locus, one would expect a threefold higher rate of insertions on the Y chromosome. Predominant male germ-line retrotransposition has previously been seen for another retrotransposon, the copia element of D. melanogaster (Pasyukovaet al. 1997).
In the case of the R2 insertions, a comparable number of retrotransposition events occurred on the X and Y chromosomes, suggesting that R2 retrotransposition occurred in both the male and female germ lines. Unfortunately, the number of events is too low to determine the relative frequency of events in each germ line.
The high level of R1 retrotransposition events on the Y chromosome is in stark contrast to earlier observations, based originally on the Oregon-R strain of D. melanogaster, that there are no R1 insertions (known as type I insertions in these studies) on the Y chromosome (Tartof and Dawid 1976; Wellaueret al. 1978). Indeed, the absence of R1 elements on the Y chromosome has been an underlying assumption in various studies of the rDNA locus (Englandet al. 1988; Clarket al. 1991). However, studies that indicated that R1 insertions do exist on the Y chromosome have also appeared (Cantú and Gay 1984; Jakubczaket al. 1992; Kommaet al. 1993). In the most comprehensive study to date, Manzo (2000) found in a survey of 65 D. melanogaster strains that the level of R1 insertions varied between 4 and 35% of the Y rDNA units, compared to between 15 and 65% of the X rDNA units. Thus R1 elements are present on the Y chromosome of D. melanogaster but on average at lower levels.
If R1 elements retrotranspose in the male germ line, then what would explain the lower levels of R1 observed on this chromosome in most strains? This apparent paradox could be explained if there were a more rapid loss of elements from the Y chromosome. When averaged over all R1 and R2 elements on the X and Y chromosomes, the rate of elimination from the Y (3.1 × 10–4/element/generation) was nearly twice that of the X chromosome (1.6 × 10–4/element/generation). However, these rates are only crude estimates because the actual number is a combination of two rates. A fraction of the insertions on the X and Y chromosomes appear to be located within recombinational “hotspots” and were lost at a rate eight times faster than those located outside these hotspots. Thus the rate of element loss will vary for different strains and will depend upon the number of elements located within these hotspots. We propose that when averaged over the entire species, a faster rate of R1 loss from the Y chromosome will compensate for the higher rate of insertion on this chromosome and thus explain why Y chromosomes frequently possess fewer R1 elements. It will be important to determine the rates and patterns of R1 and R2 retrotranspositions for other D. melanogaster strains, especially for those strains with levels of R1 and R2 insertions on their X and Y chromosomes that differ from those in the Harwich lines.
Another important insight gained from the experiments in this report concerns the mechanism of R1 and R2 element elimination. Two types of recombination mechanisms could give rise to the elimination of these insertions from the rDNA locus: unequal crossovers and gene conversions. Both mechanisms have been postulated to be involved in the concerted evolution of the rDNA units (Williamset al. 1989; Schlötterer and Tautz 1994; Polanco et al. 1998, 2000). The data shown in Figure 6 suggest that many R1 and R2 copies were eliminated from the Y chromosome by single-crossover events that simultaneously removed multiple insertions. Seven of the Harwich lines underwent a recombination event that removed three to nine elements from the Y chromosome. In a similar manner, four lines may have had deletions that eliminated two elements from the hotspot located within the X chromosome. Evidence for large deletions also involving elements located outside the hotspots may be found in two Harwich lines. Three elements, in addition to two within the hotspot, were eliminated from the X chromosome in line 22, while all other lines had no more than one elimination outside the hotspot. Four elements, all located outside the recombination hotspot on the Y chromosome, were eliminated in line 4. In total, we can postulate 13 deletion events that would account for nearly two-thirds of all observed R1 and R2 eliminations (60 of 96 events).
In contrast to the many elimination events, we did not observe instances where the elements located within the X or Y chromosome hotspot underwent duplications (i.e., PCR bands that appear at twice the intensity of the other bands). This suggests that the crossovers that gave rise to the eliminations of R1 and R2 were probably not interchromosomal or sister-chromatid exchange because both products of such recombination events should have been observed. Our observations could be explained if the R1 and R2 eliminations occur by means of intrachromosomal events: crossovers involving chromosomal loops in which one recombination product is lost from the cell. Because such intrachromosomal events would delete only rDNA units, what is the mechanism that replenishes the number of units? One model is that there are regions of the rDNA locus (e.g., near the center) that contain few R1 and R2 insertions. If standard interchromosomal or sister-chromatid unequal crossover events occur predominantly in these “insertionfree” areas, then the number of uninserted units could be driven by selection to increase without a corresponding increase in the number of inserted units. We are in the process of constructing a bacterial artificial chromosome library of genomic DNA from one of the Harwich strains to recover large segments of the rDNA loci on a series of overlapping clones (W. D. Burke and T. H. Eickbush, unpublished data). Using these clones to localize the ancestral R1 and R2 copies to specific locations on the X and Y rDNA loci can provide direct support for the hypothesis that many R1 and R2 eliminations involve single events that remove multiple rDNA units, as well as identify potential locations within the loci that are free of R1 and R2 insertions.
A final aspect examined in this report is the formation of rDNA units that contain both R1 and R2 insertions. Of the 184 new R1 insertions observed, at most one insertion was into an R2-occupied rDNA unit. In this one instance, it seems more likely that the R1 insertion occurred first and then an R2 element inserted up-stream of it. In either case, the presence of an R2 insertion in a 28S gene greatly inhibits R1's ability to insert into that gene. In contrast, of the 16 new R2 insertions, 2 or 3 formed double insertions. Consistent with this frequency, R2 insertions upstream of an R1 element were also observed in 2 of 15 germ-line insertions generated by injection of the R2 retrotransposition machinery into D. melanogaster embryos (Eickbush and Eickbush 2003). These data provide an explanation for the varying levels of doubly inserted units seen in different strains of D. melanogaster (Jakubczaket al. 1992). Strains with high levels of doubly inserted rDNA units are predicted to have had more recent R2 activity, while strains with low levels of doubly inserted units have had more recent R1 activity.
This study combined with our two previous studies (Pérez-González and Eickbush 2001, 2002) has given us a better understanding of the forces that affect the maintenance of R1 and R2 elements in the rDNA arrays of Drosophila. The rDNA loci of Drosophila are highly variable in both the number of rDNA units and the fraction of these units that are inserted with R1 and R2 (Lyckegaard and Clark 1991; Jakubczaket al. 1992; Manzo 2000). The stable vertical descent of R1 and R2 in Drosophila suggests that these elements have evolved highly efficient means of integrating retrotransposition with the expression and recombinational processes of the rDNA loci. The continued study of R1 and R2 will provide insights into the regulation of both these elements and the complex rDNA loci they inhabit.
We thank X. Zhang, H. A. Orr, and J. Jaenike for helpful discussions. We thank T. F. C. Mackay for the Harwich lines. This research was supported by National Science Foundation grant MCB-9974606 to T.H.E.
Communicating editor: S. Henikoff
- Received May 21, 2003.
- Accepted June 24, 2003.
- Copyright © 2003 by the Genetics Society of America