• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of molcellbPermissionsJournals.ASM.orgJournalMCB ArticleJournal InfoAuthorsReviewers
Mol Cell Biol. Oct 2008; 28(20): 6452–6461.
Published online Aug 4, 2008. doi:  10.1128/MCB.01015-08
PMCID: PMC2577413

Epigenetic Regulation of Retrotransposons within the Nucleolus of Drosophila[down-pointing small open triangle]


R2 retrotransposable elements exclusively insert into a conserved region of the tandemly organized 28S rRNA genes. Despite inactivating a subset of these genes, R2 elements have persisted in the ribosomal DNA (rDNA) loci of insects for hundreds of millions of years. Controlling R2 proliferation was addressed in this study using lines of Drosophila simulans previously shown to have either active or inactive R2 retrotransposition. Lines with active retrotransposition were shown to have high R2 transcript levels, which nuclear run-on transcription experiments revealed were due to increased transcription of R2-inserted genes. Crosses between R2 active and inactive lines indicated that an important component of this transcriptional control is linked to or near the rDNA locus, with the R2 transcription level of the inactive parent being dominant. Pulsed-field gel analysis suggested that the R2 active and inactive states were determined by R2 distribution within the locus. Molecular and cytological analyses further suggested that the entire rDNA locus from the active line can be silenced in favor of the locus from the inactive line. This silencing of entire rDNA loci represents an example of the large-scale epigenetic control of transposable elements and shares features with the nucleolar dominance frequently seen in interspecies hybrids.

Eukaryotic genomes have evolved elaborate surveillance and regulatory mechanisms to control the spread of transposable elements (38). The success of a transposable element is thus dependent upon its ability to elude these cellular controls. One ingenious approach used by transposable elements is to target locations within the genome that cannot be completely silenced. The most successful known examples of this approach are the numerous mobile elements that insert specifically into the rRNA genes of animals (10). Eukaryotic genomes encode hundreds to thousands of rRNA genes organized in tandem arrays within one or more chromosomal loci. Transcription of the rRNA gene arrays, also termed nucleolar organizer regions (NORs), is tightly coupled to the growth status of the cell. Significant progress has been made in understanding the transcription of the DNA encoding the rRNA genes (ribosomal DNA [rDNA]) as well as the many subsequent steps involved in ribosome biogenesis (14, 16). The level of regulatory complexity added by the presence of transposable elements is largely unknown.

The nonlong terminal repeat (non-LTR) retrotransposable elements, R1 and R2, insert into a conserved central region of the 28S gene (Fig. (Fig.1A).1A). R1 and R2 are present in most lineages of arthropods (2), and R2 elements have been found in a number of other divergent animal groups (20, 21). Phylogenetic analyses suggest that R2 elements have been a stable component of genomes throughout the evolution of arthropods (2, 27) and possibly since the origin of multicellular animals (20, 21). As such, they represent the longest known stable relationship of a mobile element and its host. During their long history, R2 elements have undergone little apparent change in structure or insertion strategy. The extreme sequence specificity of R2 has enabled detailed studies of its retrotransposition mechanism (5, 26), which has become the general model for the insertion of non-LTR retrotransposons (32).

FIG. 1.
R2 transcript and retrotransposition levels are correlated in different D. simulans lines. (A) The rDNA locus of D. simulans is composed of tandem units, each composed of 18S, 5.8S, and 28S genes (black boxes), ITSs (white boxes), and an intergenic spacer ...

While the presence of R1 and R2 insertions in rRNA genes has long been documented, little is known about the transcription of these insertions or of the inserted rDNA units they occupy, other than the general suggestions that they are cotranscribed with the rRNA genes and that this cotranscription is rare (9, 13, 18, 19, 25, 43). In Drosophila melanogaster, 30 to 75% of the total rDNA units present in an individual are inserted by R1 or R2 and, therefore, can no longer synthesize functional 28S rRNA (11, 17). These high levels of 28S gene disruption do not appear detrimental to the fly, suggesting that only a small number of uninserted rDNA units are needed for viability. Consistent with this conclusion, only a fraction of the rDNA units in eukaryotes are transcribed at any one time (6, 7). This fraction may be as low as 10% in D. melanogaster (15, 29, 43).

Population surveys and long-term laboratory studies have suggested that R1- and R2-inserted units are continually lost from the rDNA locus by recombination, and thus, their abundance requires that the elements frequently retrotranspose at the population level (17, 33, 34). Unfortunately, surveys of laboratory stocks and geographical lines of D. melanogaster have not yielded stocks with sufficiently high R1 or R2 retrotransposition rates for short-term study (J. Zhou and T. Eickbush, unpublished data). However, recent analyses of a population of Drosophila simulans have revealed multiple isofemale lines in which R2 retrotransposes at rates as high as one new insertion for every two progeny (44, 45). Lines with active R2 retrotransposition contained, on average, twice the number of R2 insertions as those without retrotransposition, but R2 abundance alone was not sufficient to predict whether a line had active R2 elements.

Here, we show that R2 retrotransposition in the active lines is directly correlated with increased transcription of R2-inserted rDNA units. In crosses between R2 active and inactive lines, the entire rDNA locus with active R2 is silenced, a phenomenon reminiscent of the nucleolar dominance observed in interspecies hybrids of both plants and animals (30, 35). Dominance appears to be determined by the distribution of R2 within the rDNA locus, suggesting a model whereby R2 regulation is a factor in dosage compensation within the rDNA locus.


Fly stocks and the isolation of total genomic RNA.

All stocks of D. simulans were derived from isofemale lines derived from a population in California (34). RNA was isolated (9) from 3- to 20-h-old embryos (0.02 g) after dechorionation, from 25 female/40 male adult flies, and from ovaries/testes dissected from twice this number of adults, as well as from the remaining carcasses. Initial nucleic acid pellets were treated with DNase I (20 units; Promega), phenol extracted, and ethanol precipitated. Total RNA was resuspended in distilled water. The integrity of the RNA was checked on 1% agarose gels and the concentration estimated by an optical density at 260 nm.

Synthesis of antisense RNA probes.

PCR products containing promoter sequences for T7 polymerase were generated for the regions of interest. Labeled antisense RNA was generated using T7 polymerase under the conditions suggested by the supplier (Invitrogen). The primers used to generate the 5′ R2 transcript were 5′-GGGGATCTGGGGTAATTGCG-3′ and 5′-GTAATACGACTCACTATAGGGCGATTTTGTGTTCGTAGTTCCA-3′; the primers for the ADH transcript were 5′-ACCACCAAGCTGCTGAAGAC-3′ and 5′-GTAATACGACTCACTATAGGGCGAAGTTGACCACGGCGGCCT-3′.

Northern blot analyses.

Total RNA (10 μg) was suspended in 1× morpholinepropanesulfonic acid (MOPS) (0.04 M MOPS [pH 7.0], 0.01 M Na acetate, 0.001 M EDTA), 2.2 M formaldehyde, and 50% formamide, heated at 65°C for 15 min, placed on ice, and one-tenth volume loading buffer (1× MOPS, 50% formamide, 2.2 M formaldehyde, 4% Ficoll, 0.25% bromophenol blue) was added. RNAs were separated on a 1% agarose gel containing 0.5 M formaldehyde/1× MOPS at 40 V for 3 h. Standard RNA lanes were cut from the gel and stained with ethidium bromide to monitor electrophoresis. RNA was transferred to GeneScreen Plus (prewet in 5× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate]) by capillary action using 20× SSC. After transfer, the blot was rinsed in 4× SSC, dried, and baked for 2 h under vacuum at 80°C. The filter was prehybridized in 2× SSC, 5× Denhardt's solution, 1% sodium dodecyl sulfate (SDS), 10% polyethylene glycol (PEG) (molecular weight, 8,000), 25 mM sodium phosphate (pH 7.2), 0.1% sodium pyrophosphate, 1 mg/ml salmon sperm DNA, and 50% formamide for 3 h at 58°C. Approximately 1 × 106 cpm/ml antisense probe was added to the solution, and hybridization was continued for 18 h. The filters were washed twice at 68°C, 10 min each in 0.5% bovine serum albumin, 4 mM sodium phosphate (pH 6.8), 2 mM EDTA, and 5% SDS, and three times in 4 mM sodium phosphate (pH 6.8), 2 mM EDTA, and 1% SDS. Filters exposed to a phosphorimager screen were analyzed using ImageQuant.

Slot blot analyses.

RNA was isolated from individual adult males as described above, except volumes were reduced. The total nucleic acid pellet was resuspended in 11 μl 1× DNase I buffer (Promega), and 1 μl was removed for PCR analysis before the addition of DNase I (1 unit). After incubation at 37°C for 40 min, samples were made in 10 mM EDTA, 10× SSC, and 2.5 M formaldehyde and incubated at 60°C for 30 min. The samples were drawn through prewet (10 mM Tris [pH 7.5] and 1 mM EDTA) GeneScreen under a slight vacuum and hybridized as described for the Northern blot analyses.

Nuclear run-on transcription assays.

Nucleus isolations, run-on transcription assays, and hybridizations were as previously described (43). Briefly, 3- to 20-h-old embryos were collected and homogenized, and the debris was removed by centrifugation. Nuclei were purified by sedimentation through a 1.7 M sucrose cushion. Isolated nuclei were incubated in the run-on transcription buffer which included 150 μCi [α-32P]UTP (3,000 Ci/mmol) at 25°C for 15 min. The isolated RNAs were hybridized to nylon membranes on which denatured DNA fragments corresponding to the 5′ and 3′ regions of the R2 element and the internal transcribed space (ITS; 0.5 μg each) were immobilized. Generation of the DNA fragments for this experiment was previously described (43).


Reverse transcription-PCRs (RT-PCRs) were done under the conditions supplied with the M-MLV reverse transcriptase (Invitrogen). One microgram of total RNA was incubated at 65°C for 5 min with a 3′ R2 primer (1 μM) and deoxynucleotides (0.5 mM). After a quick chilling, first-strand buffer (1×), dithiothreitol (0.01 M), and RNaseOut (40 units; Invitrogen) were added, and the reaction mixtures were incubated at 37°C for 2 min. The M-MLV reverse transcriptase (200 units) was then added and the incubation continued for 50 min. Two microliters of the RT reaction mixture was used in standard PCRs. Parallel reactions were performed without the reverse transcriptase. The primers used were as follows: RT primer, 5′-TTTTGATCGCGGACGTATGG-3′; PCR primers, 5′-TGCCCAGTGCTCTGAATGTC-3′, 5′-TGCCAAGACAGCAAGCAAAT-3′, 5′-CCCCTTGTAGTACGAGACTTC-3′, and 5′-GGAAATCTTCGAAAGATACTAGGT-3′.


Brain tissue was dissected from female third-instar larvae, fixed, and squashed as previously described (22). Rehydrated tissue was stained with 0.4 μg/ml DAPI for 2 min, washed briefly in water, and mounted in SlowFade reagent. All observations and photography were conducted on a Nikon DEC512 fluorescence microscope.

Pulsed-field gel electrophoresis.

Nuclei were isolated from 0- to 23-h embryos as described above. Nuclei were suspended in an equal volume of 1% InCert agarose (Cambrex) at 37°C, placed into 80-μl molds, and allowed to gel on ice. Nucleus digestions were modifications of a method by Smith and Cantor (39). Embedded nuclei were lysed by soaking with gentle agitation in 2% filtered N-lauroyl sarcosine (Sigma), 0.4 M EDTA (pH 8.0), and 2 mg/ml proteinase K for 24 h at 50°C. The lysis solution was replaced, and the plugs were soaked for an additional 24 h. The proteinase K was removed by washing the plugs for 1 h, first in 0.5 M EDTA (pH 9) at 50°C and then in 0.5 M EDTA (pH 8) at 4°C. The plugs were stored in 50 mM EDTA (pH 8.0) at 4°C. Prior to AscI and/or NotI digestion, lysed nucleus plugs were soaked overnight at 4°C in 10 mM Tris and 5 mM ETDA (pH 7.5), followed by 1 h of incubation on ice in 0.5 ml of enzyme buffer. The buffer was replaced with 200 μl buffer containing 200 units of enzyme and incubated at 37°C. The buffer containing fresh enzyme was replaced every 1.5 h for either 3 or 6 h. Control plugs were treated with the 3-h incubation, except the restriction buffer without enzyme was used. Plugs were subjected to pulsed-field gel electrophoresis at 12°C on a CHEF-DRII system (Bio-Rad) in a 1% agarose gel for 22 h at 150 V with switch times of 10 to 30 s. The buffer was continuously recirculated. Before blotting, the gel was treated for 15 min in 25 N HCl (2×), 20 min in 0.5 M NaOH/1.5 M NaCl (2×), and 30 min in 0.5 M Tris (pH 7.5)/3 M NaCl (2×). The gel was then blotted onto nitrocellulose membranes and hybridized as previously described (17). The probe was an 18S fragment generated by PCR using the primers 5′-GTCTTGTACCGACGACAGATC-3′ and 5′-GCAGAACAGAGGTCTTATTTC-3′ and randomly labeled.


R2 transcript levels correlate with retrotransposition.

New R2 insertions in the 28S genes can be scored because the R2 retrotransposition machinery frequently generates unique 5′ truncated insertions that are readily differentiated by PCR from R2 copies already present in the rDNA locus (34). Shown in Fig. Fig.1B1B is a summary of our previous analysis of 15 D. simulans isofemale lines derived from one population. The number of R2 elements and an indication of the level of R2 activity as measured by differences in the R2 truncation profiles are presented for each line. Six lines lacked evidence of R2 retrotransposition activity (0% R2 profile variants), five lines showed low to moderate R2 activity (10 to 20% of the flies contained variants), and four lines showed high activity (>60% of the flies contained variants) (44). A resurvey of the 15 lines suggested that the R2 retrotransposition status of each line had not substantially changed from this earlier report (data not shown).

As a first step to study R2 regulation, total RNA was isolated from adult males, blotted and hybridized with sequences from the 5′ end of R2 (Fig. (Fig.1C).1C). The levels of R2 transcripts varied >100-fold among the 15 lines. The major hybridizing band was 3,600-nucleotides (nt) long, consistent with the size of full-length R2 elements. The presence of hybridizing RNA species longer than 3,600 nt, most readily seen in line 89, is consistent with suggestions that the R2 transcripts were processed from cotranscripts with the 28S gene (9, 13, 18, 19, 25, 43). Because the probe for this blot was generated from the 5′ end of R2, the shorter RNA bands correspond to degradation intermediates rather than to transcripts from 5′ truncated elements. As controls for RNA loading and integrity, the ethidium bromide-stained gel of the RNAs used, as well as the RNA blot probed with a fragment of the alcohol dehydrogenase gene, are also shown. The relative levels of R2 RNA among the different lines were also seen in germ line (ovaries, testes) and somatic (remaining carcass) tissues of adult females and males, as well as in 3- to 20-h-old embryos (data not shown).

A correlation was observed between the level of retrotransposition activity (Fig. (Fig.1B)1B) and the R2 transcript level (Fig. (Fig.1C).1C). Those lines in which no R2 retrotransposition was detected had low or no detectable R2 transcripts, while, with one exception, the lines with the most-active R2 retrotransposition had the highest levels of R2 transcripts. The exception was line 71, in which R2 transcripts were at an intermediate level yet the rate of R2 retrotransposition was high (discussed further below).

R2 RNA levels are under transcriptional control.

The high levels of R2 transcripts detected in some lines could be a result of the R2 elements being more frequently transcribed (transcriptional control) or the R2 transcripts being more stable (posttranscriptional control). To distinguish between these possibilities, nuclear run-on transcription experiments were performed with three low-, two intermediate-, and two high-transcript-level lines. Nuclei were isolated from 3- to 20-h-old embryos and incubated in the presence of labeled nucleotides under conditions that allowed bound RNA polymerase to finish the transcription of already initiated reactions (43). RNA was then isolated and hybridized to separate membrane-bound DNA fragments corresponding to the 5′ and 3′ ends of the R2 element as well as those from the ITS region of the rDNA unit. The ITS fragment represented the total level of rRNA transcription (i.e., from both inserted and uninserted units) and served as a control for the transcription efficiency of the isolated nuclei from each line.

R2 run-on transcription levels standardized to the ITS level for these six lines are shown in Fig. Fig.2,2, along with the relative levels of stable R2 transcripts for each line detected by Northern blot analysis. Run-on transcription measured from either the 5′ or 3′ end of the element gave similar results. The levels of R2 run-on transcription for the three low-level lines (lines 1, 96, and 34) were <1% of the transcription level seen for the ITS. These values are similar to the low levels of R2 transcripts previously reported for D. melanogaster lines (9, 43). Consistent with the levels of R2 RNA seen on the Northern blots, the levels of run-on transcription were three to six times higher in line 57 and 10 to 20 times higher in lines 58 and 89. Run-on transcript levels in line 71 were higher than those predicted for the full-length RNA, suggesting more-rapid RNA degradation. The high rate of R2 transcription in line 71 was consistent with the high rate of R2 retrotransposition observed in this line (Fig. (Fig.1B1B).

FIG. 2.
R2 transcript levels are under transcriptional control. Total RNAs synthesized from 3- to 20-h-old embryonic nuclei under run-on transcription conditions were used to probe filters containing immobilized denatured DNA fragments corresponding to the ITS ...

These findings suggest that the major control over R2 retrotransposition in these lines is at the level of transcription. In the following experiments, lines with low R2 transcript levels and no retrotransposition are called “R2 inactive,” while lines with intermediate to high levels of R2 transcript and detectable retrotransposition are called “R2 active.”

R2 transcript levels in crosses between R2 active and inactive lines.

To address the genetic and epigenetic components of R2 transcriptional control, reciprocal crosses were performed between R2 active and R2 inactive lines. In D. simulans, the 28S rRNA genes are found only on the X chromosome (24, 33). Therefore, X chromosomes from R2 active lines are denoted XA, while X chromosomes from R2 inactive lines are denoted XI. Examples of the RNA blots from different sets of crosses are shown in Fig. Fig.3A.3A. Each blot monitored R2 transcription in the original parents as well as in the male and female progeny in the F1 and F2 generations.

FIG. 3.
R2 transcript levels in crosses between active and inactive lines. (A) Diagram of the reciprocal crosses between lines with intermediate/high R2 transcript levels (lines 57, 58, 71, 89, and 100) and lines with very low levels of R2 transcripts (lines ...

A consistent pattern of R2 transcript levels was observed in these crosses, providing several clues to R2 transcriptional control. First, the heterozygous (XAXI) F1 females (Fig. (Fig.3,3, lanes 1 and 9) had R2 transcript levels similar to those of the R2 inactive parent, indicating that the “genotype” for low R2 transcript levels was dominant over high transcript levels. Second, unlike their female siblings, the F1 males (Fig. (Fig.3,3, lanes 2 and 10) had levels of R2 transcript similar to those of their mother, irrespective of whether she had high or low levels of R2 transcripts. This sex-linked inheritance suggested that the control over R2 transcript levels mapped to the X chromosome rather than to the autosomes. Third, high R2 transcript levels returned in F2 males (Fig. (Fig.3,3, lane 12), even though both F1 parents had low R2 transcript levels (Fig. (Fig.3,3, lanes 9 and 10). Thus, R2 transcription could be “turned off” and “turned back on” in a single generation.

In total, seven sets of reciprocal crosses were conducted between five R2 active lines (lines 57, 58, 71, 89, 100) and three R2 inactive lines (lines 1, 34, 96). Quantitations of the relative R2 transcription levels in the parents as well as in the F1 and F2 progeny were combined in Fig. Fig.3B.3B. To compensate for the different initial levels of transcription in the R2 active lines, the RNA signal was standardized to that of F0 R2 active males (Fig. (Fig.3,3, lane 6), which was defined as 1.0 for each set of crosses. The levels of R2 transcription in three of the four pools of F2 progeny (Fig. (Fig.3,3, lanes 3, 4, and 12) were found to be about half that of the original active lines presumably because only one of the two genotypes in each would give rise to transcripts (Fig. (Fig.3A).3A). To test this prediction, nucleic acid was isolated from individual F2 males from the reciprocal crosses involving lines 58 and 34. As controls, nucleic acid was also isolated from individual F0 and F1 males. Purified RNA from each male was slot blotted onto a membrane and probed with the 5′ end of the R2 element (Fig. (Fig.4A).4A). To determine the origin of the rDNA locus in each F2 male, an aliquot of DNA was subjected to PCR analysis. Those blots from F2 males that had 5′ truncated R2 elements diagnostic for the rDNA locus of the active line (line 58) have been boxed. The analysis revealed a correspondence between high R2 transcript levels and the presence of the XA rDNA locus.

FIG. 4.
Control over R2 transcript level segregates with the rDNA locus. (A) Total nucleic acid was isolated from individual F0, F1, and F2 males of crosses between lines 58 and 34, as diagrammed in Fig. Fig.3.3. RNAs were blotted and probed with the ...

All crosses between R2 active and inactive lines gave similar results through the F2 generation (Fig. (Fig.3B),3B), suggesting that although the absolute levels of transcription varied, the genetic/epigenetic factors controlling R2 transcription were similar in all lines. Consistent with a common control mechanism in all D. simulans lines, reciprocal crosses between inactive lines (lines 96 and 1) and those between active lines (lines 89 and 58) gave low and high levels of R2 transcripts, respectively, in all F1 and F2 progeny (data not shown). These findings and the absence of maternal or paternal effects associated with R2 transcription suggest the R2 expression pattern can be explained by the Mendelian inheritance of a “dominant/recessive” locus on the X chromosome.

The apparent correlation between the rDNA locus itself and the R2 transcript levels observed in Fig. Fig.33 and and4A4A was further examined by scoring individual flies multiple generations after the initial cross between R2 active and inactive lines. RNA isolated from individual F5 males, from an original cross between 58 males and 34 females, and F8 males, from an original cross between 89 males and 34 females, revealed elevated levels of R2 transcripts only in those males that contained the rDNA locus from the original active line (Fig. (Fig.4B).4B). The one-to-one correspondence observed in F5 and F8 flies suggested that R2 transcript levels were associated with or closely linked to the rDNA locus on the X chromosome. Flies with elevated transcript levels supported R2 retrotranspositions, as new insertions could again be monitored when stocks with only rDNA loci from the R2 active line were reestablished from the F4 progeny of crosses between R2 active and inactive lines (data not shown).

Cytological evidence for nucleolar dominance.

Nucleolar dominance is the preferential transcription of the rDNA locus from one parent in offspring derived from crosses between different species or different geographical isolates of one species (35). Nucleolar dominance can be monitored cytologically by means of secondary chromosomal constrictions. As cells approach metaphase, RNA polymerase 1 transcription is shut down, and the nucleolus disappears. However, a “fingerprint” temporarily remains as transcriptional activity at an NOR slows chromosome condensation in early metaphase, resulting in a constriction at the rDNA locus (30). In D. simulans and D. melanogaster females, two constrictions are typically observed, indicating that the rDNA arrays on both X chromosomes are usually transcribed (8). Nucleolar dominance could explain the R2 transcript levels observed in the crosses between R2 active and R2 inactive lines described above if, in heterozygous females (XIXA), there is preferential transcription of the rDNA locus from the R2 inactive parent.

Brain tissue was dissected from third-instar larvae and fixed, and the chromosomes were stained with DAPI. Genotypes examined for the presence of secondary constrictions included homozygous females from active (lines 89 and 58) and inactive (line 34) lines and heterozygous females resulting from crosses between these lines. Examples of early-metaphase chromosome spreads from heterozygous females (derived from crosses between lines 89 and 34) as well as from homozygous females from the parental lines are presented in Fig. Fig.5A.5A. For the homozygous (XAXA, XIXI) females, both X chromosomes routinely had secondary constrictions, indicating that both rDNA arrays were transcribed. In the case of the heterozygous (XIXA) females, only one secondary constriction was regularly observed, suggesting that a single ribosomal array was transcribed. Similar results were obtained for the parental line 58 (i.e., two secondary constrictions) and for the heterozygous females produced in the cross between lines 58 and 34 (i.e., one secondary constriction) (Fig. (Fig.5B).5B). In crosses between two active lines (lines 89 and 58), both X chromosomes in the F1 females typically had a secondary constriction, indicating the expression of both rDNA loci in this heterozygote (data not shown).

FIG. 5.
Cytological evidence for nucleolar dominance. Shown are DAPI-stained metaphase chromosomes isolated from brain tissue of female third-instar larvae. (A) Chromosomes from the parental lines 89 and 34 as well as from the F1 heterozygous females. For each ...

The secondary constrictions were frequently distinctive in appearance, suggesting that the composition of the rDNA loci dictated not only how many loci were transcribed but also how extensively. For example, the secondary constrictions of the rDNA loci in the parental line 89 were thin and extended, while the constrictions in the parental line 34 invariably appeared merely as an elongation of the chromosome (Fig. (Fig.5A).5A). In the heterozygous females produced in the cross between lines 89 and 34, the single secondary constriction was seldom as thin as the constrictions seen in the line 89 females but more extended than those seen in the line 34 females. This intermediate appearance is consistent with a greater number of rDNA units transcribed from the single line 34 rDNA locus in a heterozygote than those transcribed from each loci in a line 34 homozygote. While these cytological observations do not conclusively indicate which chromosome is transcribed, they are, in conjunction with the molecular data described below, in accord with a model in which the rDNA locus from the R2 active lines is silenced when paired with an rDNA locus from an R2 inactive line.

Molecular evidence for nucleolar dominance.

Molecular evidence for nucleolar dominance in interspecies hybrids is usually obtained using sequence differences in the 18S or 28S rRNA synthesized by the parental rDNA loci (30, 35). Unfortunately, such sequence variation in rRNA would not be expected among individuals from the same population. However, as an alternative, it is possible to monitor transcription of specific rDNA units that contain highly truncated R2 insertions. Previous analysis of these short R2 insertions in D. melanogaster revealed that the corresponding 28S/R2 cotranscripts were stable, reproducible, and independent of full-length R2 transcription (9). RT of adult RNA using a primer annealed to the 3′ end of the R2 element and the subsequent amplification by PCR revealed distinctive 28S/R2 cotranscripts in the D. simulans lines used in this study. These specific truncated R2 transcripts were found in all R2 active lines as well as in several R2 inactive lines.

Figure Figure66 shows the results of reciprocal crosses between R2 active and inactive lines similar to those conducted for Fig. Fig.3.3. In the upper panel, the XA chromosome from the R2 active line (line 100) had specific transcripts corresponding to units with 5′ truncated elements that are 900 and 350 bp in length (lanes 5 and 6), while the XI chromosome from the R2 inactive line (line 1) had transcripts arising from a 510-bp R2 insertion (lanes 7 and 8). In the F1 heterozygous (XIXA) females, transcripts typical of the XI chromosome but not the XA chromosome were detected (lanes 1 and 9). Transcripts arising from the truncated R2 elements of the XA chromosome reappeared in F2 females only in the direction in which homozygous XAXA females were generated (compare lanes 3 and 11).

FIG. 6.
Transcripts of specific 5′ truncated R2 elements also provide evidence of nucleolar dominance. RNAs from reciprocal crosses as described in Fig. Fig.33 were used to monitor transcription of specific 5′ truncated R2 elements. For ...

The lower panel in Fig. Fig.66 shows the results of a second cross between R2 active and inactive lines. The XI chromosome in line 90 has high levels of transcripts corresponding to a 160-bp R2 insertion, while the XA chromosome in line 58 has high levels of a transcript from a 45-bp R2 insertion. A low level of transcripts from a similar 45-bp R2 insertion can also be detected from the XI chromosome (line 90). Again, the heterozygous F1 females (lanes 1 and 9) have an expression pattern identical to that of the XI chromosome with transcripts from the XA chromosome reappearing only in the homozygous XAXA F2 females.

In three additional crosses (between lines 89 and 34, 58 and 34, and 57 and 96) specific transcripts detected in the R2 active lines were absent when the XA chromosome was paired with the XI chromosome (data not shown). While the number of truncated R2 elements monitored in each case is limited, combining the data from all crosses with the full-length R2 transcript patterns and the number of secondary constrictions support our conclusion that the entire rDNA locus with active R2 elements is silenced when paired with an rDNA locus with inactive R2 elements. Finally, it should also be noted that detecting R2 transcripts from 5′ truncated elements from only one of the two chromosomes in the heterozygotes argues against posttranscriptional models of R2 regulation.

Distribution of R2 elements in the rDNA locus.

How does a cell determine whether both or only one rDNA locus is to be used for the synthesis of rRNA? Two properties of rRNA transcription suggest a possible model. First, only a small fraction of the many rDNA units in eukaryotes are actively transcribed at any one time (6, 7). In the case of Drosophila, this number may be as few as 30 to 50 units (29, 43). Second, electron microscopic observations in Drosophila suggest that transcriptionally active rDNA units typically occur in large consecutive blocks (4, 18). To optimize the synthesis of functional rRNA, arthropods may have evolved, over the hundreds of millions of years that R1 and R2 have existed in their 28S rRNA genes (27), mechanisms to activate those regions of the rDNA locus that contain the lowest ratio of inserted to uninserted units.

To determine the distribution of R2 elements in the rDNA locus, high-molecular-weight DNA was obtained by embedding embryonic nuclei in agarose plugs. Lysis and restriction digestion with NotI, an enzyme that cleaves R2 elements but not R1 elements or the uninserted rDNA unit, were then conducted within these plugs. The DNA was separated by pulsed-field gel electrophoresis, transferred to a solid support, and hybridized with a fragment from the 18S gene. The size of the rDNA unit in D. simulans is not uniform because of variation in the intergenic region but is typically about 11 kb in length (40). Figure Figure77 (left side) shows the NotI digestion profiles of three active (lines 58, 89, and 100) and two inactive (1 and 34) lines. The active lines 58 and 89 have high levels of R2 insertions that are extensively distributed throughout the rDNA loci, as evidenced by the paucity of NotI fragments of >100 kb. In contrast the inactive lines 1 and 34 have lower levels of R2 insertions, and multiple NotI fragments of >400 kb are observed. The rDNA locus of the active line 100 is somewhat intermediate with many NotI fragments ranging from 150 to 400 kb.

FIG. 7.
Distribution of transposable elements in the rDNA locus. High-molecular-weight DNA isolated from 0- to 23-h-old embryo nuclei was obtained from three active lines (lines 58, 89, and 100) and two inactive lines (lines 1 and 34). Restriction enzyme-digested ...

Because organisms would be expected to limit the transcription of the R1-inserted units in a manner similar to that of the R2-inserted units, more revealing would be a determination of the degree to which all inserted and uninserted units are intermingled in the rDNA loci. To this end, genomic blots were conducted with DNA digested with both NotI and AscI, an enzyme that cleaves R1 elements but neither R2 elements nor uninserted rDNA units (Fig. (Fig.7,7, right side). In the R2 active lines 58 and 89, the double-digested DNA showed little difference from that the DNA digested by NotI alone, except that in line 58, the few NotI fragments longer than 100 kb have been cleaved. The addition of AscI in the case of the R2 inactive lines 1 and 34 did reduce the size of the large NotI fragments, but fragments of >300 kb remained. The double digest in the case of line 100 had the most striking effect, in that all fragments have been reduced to less than 200 kb, more similar to that of the other R2 active lines.

These data reveal that the R1 and R2 insertions in the rDNA locus of the R2 active lines (lines 58, 89, and 100) are distributed throughout the locus such that no large region is free of these elements. The R2 inactive lines (lines 1 and 34) on the other hand have segments of the locus >30 units in length that are free of R1 and R2 insertions. The ability of cells to recognize this difference in the distribution of inserted units within the locus could explain why the R2 elements in the inactive lines are seldom transcribed and why R2 inactive rDNA loci appear dominant over R2 active rDNA loci.


The experiments in this report provide an explanation for the strikingly different rates of R2 retrotransposition detected in various stocks of D. simulans (44). Stocks with active retrotransposition have high levels of R2 transcripts, which were found to result from increased rates of transcription. Crosses between R2 active and inactive lines suggested that control over transcription was linked to the rDNA locus, the site of all R2 elements. Surprisingly, this transcriptional control affected the entire rDNA locus. Analysis of the distribution of R2 elements in various rDNA loci suggests a simple model for R2 regulation (Fig. (Fig.8)8) which is based on electron microscopic observations that animals typically transcribe contiguous blocks of rDNA units (4, 18). In this model, the cell identifies inserted units and modifies their chromatin structure to serve as foci for the formation of heterochromatin. With heterochromatin spreading from such foci, the regions of the rDNA locus remaining active would be those with the lowest level of insertions. In many organisms, large regions of the locus without insertions would remain active, and no R2 transcription/retrotransposition would occur (example A). However, in organisms in which inserted units are distributed throughout the locus, the complete inactivation of all inserted units would not leave a region with enough uninserted units to supply the needed levels of rRNA. As a result, some inserted units fall within the R2 active region, which results in some level of transcription of R2-inserted units (example B). Therefore, while the abundance of R2 insertions is a factor, it is the distribution of full-length copies in the locus that determines whether they are transcribed and, as a consequence, whether retrotransposition occurs. Typically, in females, the transcribed regions encompass units on both X chromosomes. However, if the distributions of R2 elements in the two loci are radically different, as in crosses between R2 active and inactive lines, then rDNA transcription can be limited to a single rDNA array with large regions free of insertions (example C).

FIG. 8.
Domain model for the regulation of R2 elements. Diagrammed at the top are four possible examples of the rDNA locus found on the X chromosomes of D. simulans. Each rDNA unit, as described in Fig. Fig.1A,1A, is represented by a gray box. R2 and ...

This model can explain all our observations pertaining to the rDNA locus. First, selective transcription of one or a small number of domains within the rDNA loci can explain the continuous range of R2 transcript and retrotransposition levels seen in our stocks of D. simulans. These levels can reflect the number of R2-inserted units that reside in the transcriptionally active region of the rDNA. Second, the model can explain the presence of transcripts from specific 5′ truncated R2 elements, even in the R2 inactive chromosomes. These specific 5′ truncated copies are predicted to be in the actively transcribed area even if full-length R2 insertions are not. Indeed, the ability to follow the expression of 5′ truncated R2 copies provides the strongest evidence for control via transcription of a localized domain rather than posttranscriptional mechanisms. Finally, the model in Fig. Fig.88 serves equally well for the regulation of R1 elements and, thus, can readily explain our previous findings that the activities of R1 and R2 appear to be independent (17, 34).

The domain model is also consistent with our previous observations in D. melanogaster in which transcription of specific 5′ truncated insertions was readily detected only in those stocks in which large numbers of uninserted rDNA units had been deleted from the locus (9). These deletion studies also revealed a possible difference between D. simulans and D. melanogaster. The large number of rDNA locus deletions in D. melanogaster gave rise to increased transcription of full-length R2 elements, as detected in run-on transcription assays (43); however, this increased transcription did not give rise to stable full-length R2 transcripts (9). This finding suggests that degradation, perhaps like that seen in line 71 of D. simulans, prevented the accumulation of full-length R2 transcripts to high levels. A greater role for posttranscriptional regulation of R2 in D. melanogaster may explain why it has been difficult to detect stocks with high levels of R2 transcripts in this species.

The ability of D. simulans to epigenetically silence the units in one rDNA locus shares features with previous examples of the uniparental rRNA gene expression (nucleolar dominance) observed in certain interspecies hybrids (reviewed in reference 35). Consistent with interspecies nucleolar dominance, the silencing observed in D. simulans can be established or lost in a single generation. The silencing in D. simulans is also independent of maternal or paternal affects and thus, like nucleolar dominance, is not a result of gametic imprinting. Because the only difference between inserted and uninserted rDNA units is the presence of the R1 or R2 element over 6 kb downstream of the promoter, the dominance observed within D. simulans is inconsistent with models that postulate nucleolar-dominance results from differences in transcription factor binding (28, 36, 37). Other groups have also described the limitation of such models to explain nucleolar dominance and have suggested alternatives that focus on the differences in the structure of the rDNA arrays or in loci linked to the arrays (3, 12, 23).

Although the process by which an rDNA locus is chosen for inactivation in interspecies hybrids is generally not known, the resulting active and inactive rRNA genes are characterized by differences in epigenetic markers that distinguish between euchromatin and heterochromatin (14). In this study, the selection of the array which is to be inactivated can be readily explained if R1 and R2 insertions serve as foci for the formation of heterochromatin in Drosophila. With heterochromatin spreading from such foci, the locus that remains active is that with a large region of uninserted units. This selective transcription of rDNA units would represent yet another example of how transposable-element-silencing mechanisms play a critical role in an essential cellular function, in this case rRNA gene dosage compensation.

Accumulating evidence suggests that transposable elements are controlled by small RNA-based regulatory pathways (31). Some of these small RNA pathways have been linked directly to the formation of heterochromatin (42). Our preliminary attempts have not consistently detected small R2 RNAs in D. simulans; however, this may only reflect the extremely low levels of these RNAs (data not shown). Recent reports have indicated that small RNAs which interact with piwi proteins (piRNA) control transposable elements in the germ line of D. melanogaster at a posttranscriptional level (41). piRNAs to R2 were present only at extremely low levels in these small RNA libraries. The piRNAs used to control transposable elements appear to be derived from a few master regulator genomic loci (1). The remarkable sequence specificity of R2 means that copies of this family are much less likely to end up in master regulator sites. Indeed there is little evidence that R2 copies are present anywhere outside the rDNA locus (D. Stage and T. Eickbush, data in preparation). Further experiments are needed to determine whether small RNA regulatory pathways are involved in the inactivation of inserted rDNA units and thus the regulation of the rDNA locus.

Finally, one of the remarkable properties of R2 (and R1) elements is that even though they would seem to have a precarious existence in the rDNA locus, they have much greater stability within Drosophila (or any insect lineage) than transposable elements that insert throughout the genome (2, 27). Part of this stability can be associated with the ability of R2/R1 to insert into genes that cannot be turned off in the germ line. As a result some level of R2/R1 transcription is likely to occur due to the stochastic processes associated with inactivating large numbers of rDNA units. Perhaps more significantly, the selection by the cell of an rDNA locus domain for transcription, rather than individual rDNA units, presents another opportunity for the element to escape regulation. The structure of the rDNA locus changes over generations due to the recombinational mechanisms associated with concerted evolution (10). The unequal crossovers that occur in the locus will duplicate and delete blocks of rDNA units; thus, an organism that had the ability to effectively silence its R2 insertions will occasionally undergo rDNA recombinations that change the location of the transcribed domain, leading to the reactivation of formerly inactive R2 elements. Thus, the combination of a continually changing landscape and a constant need for high levels of transcription makes the nucleolus an excellent niche for transposable elements. Further studies of R2 should continue to provide a valuable set of tools for probing both the landscape and the regulation of this fascinating region of the genome.


This work was supported by National Institutes of Health grant GM42790 to T.H.E.

We thank Deb Stage and Jun Zhou for helpful comments on the manuscript and Julio Vazquez for assistance in the cytological work. We thank Steve Henikoff for allowing us to conduct some of the experiments in his laboratory.


[down-pointing small open triangle]Published ahead of print on 4 August 2008.


1. Brennecke, J., A. A. Aravin, A. Stark, M. Dus, M. Kellis, R. Sachidanandam, and G. J. Hannon. 2007. Discrete small RNA-generating loci as master regulators of transposon activity in Drosophila. Cell 1281089-1103. [PubMed]
2. Burke, W. D., H. S. Malik, W. C. Lathe III, and T. H. Eickbush. 1998. Are retrotransposons longterm hitchhikers? Nature 392141-142. [PubMed]
3. Chen, Z. J., L. Comai, and C. S. Pikaard. 1998. Gene dosage and stochastic effects determine the severity and direction of uniparental ribosomal RNA gene silencing (nucleolar dominance) in Arabidopsis allopolyploids. Proc. Natl. Acad. Sci. USA 9514891-14896. [PMC free article] [PubMed]
4. Chooi, W. Y. 1979. The occurrence of long transcription units among the X and Y ribosomal genes of Drosophila melanogaster: transcription of insertion sequences. Chromosoma 7457-74. [PubMed]
5. Christensen, S. M., J. Ye, and T. H. Eickbush. 2006. RNA from the 5′ end of the R2 retrotransposon controls R2 protein binding to and cleavage of its DNA target site. Proc. Natl. Acad. Sci. USA 10317602-17607. [PMC free article] [PubMed]
6. Conconi, A., R. M. Widmer, T. Koller, and J. M. Sogo. 1989. Two different chromatin structures coexist in ribosomal RNA genes throughout the cell cycle. Cell 57753-761. [PubMed]
7. Dammann, R., R. Lucchini, T. Koller, and J. M. Sogo. 1995. Transcription in the yeast rRNA gene locus: distribution of the active gene copies and chromatin structure of their flanking regulatory sequences. Mol. Cell. Biol. 155294-5303. [PMC free article] [PubMed]
8. Durica, D. S., and H. M. Krider. 1977. Studies on the ribosomal cistrons in interspecific Drosophila hybrids. I. Nucleolar dominance. Dev. Biol. 5962-74. [PubMed]
9. Eickbush, D. G., and T. H. Eickbush. 2003. Transcription of endogenous and exogenous R2 elements in the rRNA gene locus of Drosophila melanogaster. Mol. Cell. Biol. 233825-3836. [PMC free article] [PubMed]
10. Eickbush, D. G., and T. H. Eickbush. 2007. Finely orchestrated movements: evolution of the ribosomal genes. Genetics 175477-485. [PMC free article] [PubMed]
11. Eickbush, D. G., D. D. Luan, and T. H. Eickbush. 2000. Integration of Bombyx mori R2 sequences into the 28S ribosomal RNA genes of Drosophila melanogaster. Mol. Cell. Biol. 20213-223. [PMC free article] [PubMed]
12. Flowers, J. M., and R. S. Burton. 2006. Ribosomal RNA gene silencing in interpopulation hybrids of Tigriopus californicus: nucleolar dominance in the absence of intergenic spacer subrepeats. Genetics 1731479-1486. [PMC free article] [PubMed]
13. George, J. A., and T. H. Eickbush. 1999. Conserved features at the 5′ end of Drosophila R2 retrotransposable elements: implications for transcription and translation. Insect Mol. Biol. 83-10. [PubMed]
14. Grummt, I., and C. S. Pikaard. 2003. Epigenetic silencing of RNA polymerase I transcription. Nat. Rev. Mol. Cell Biol. 4641-649. [PubMed]
15. Hawley, R. S., and C. H. Marcus. 1989. Recombinational controls of rDNA redundancy in Drosophila. Annu. Rev. Genet. 2387-120. [PubMed]
16. Henras, A. K., J. Soudet, M. Gérus, S. LeBaron, M. Caizergues-Ferrer, A. Mougin, and Y. Henry. 2008. The post-transcriptional steps of eukaryotic ribosome biogenesis. Cell. Mol. Life Sci. DOI.10.1007/s00018-008-8027-0 [PubMed] [Cross Ref]
17. Jakubczak, J. L., M. K. Zenni, R. C. Woodruff, and T. H. Eickbush. 1992. Turnover of R1 (type I) and R2 (type II) retrotransposable elements in the ribosomal DNA of Drosophila melanogaster. Genetics 131129-142. [PMC free article] [PubMed]
18. Jamrich, M., and O. L. Miller, Jr. 1984. The rare transcripts of interrupted rDNA genes in Drosophila melanogaster are processed or degraded during synthesis. EMBO J. 31541-1545. [PMC free article] [PubMed]
19. Kidd, S. J., and D. M. Glover. 1981. Drosophila melanogaster ribosomal DNA containing type II insertions is variably transcribed in different strains and tissues. J. Mol. Biol. 151645-662. [PubMed]
20. Kojima, K. K., and H. Fujiwara. 2005. Long-term inheritance of the 28S rDNA-specific retrotransposon R2. Mol. Biol. Evol. 222157-2165. [PubMed]
21. Kojima, K. K., K. Kuma, H. Toh, and H. Fujiwara. 2006. Identification of rDNA-specific non-LTR retrotransposons in Cnidaria. Mol. Biol. Evol. 101984-1993. [PubMed]
22. Krider, H. M., and W. Plaut. 1972. Studies on nucleolar RNA synthesis in Drosophila melanogaster. I. The relationship between number of nucleolar organizers and rate of synthesis. J. Cell Sci. 11675-687. [PubMed]
23. Lewis, M. S., D. J. Pikaard, M. Nasrallah, J. H. Doelling, and C. S. Pikaard. 2007. Locus-specific ribosomal RNA gene silencing in nucleolar dominance. PloS ONE 2e815. [PMC free article] [PubMed]
24. Lohe, A. R., and P. A. Roberts. 1990. An unusual Y chromosome of Drosophila simulans carrying amplified rDNA spacer without rRNA genes. Genetics 125399-406. [PMC free article] [PubMed]
25. Long, E. O., and I. B Dawid. 1979. Expression of ribosomal DNA insertions in Drosophila melanogaster. Cell 181185-1196. [PubMed]
26. Luan, D. D., M. H. Korman, J. L. Jakubczak, and T. H. Eickbush. 1993. Reverse transcription of R2Bm is primed by a nick at the chromosomal target site: a mechanism for non-LTR retrotransposition. Cell 72595-605. [PubMed]
27. Malik, H. S., W. D. Burke, and T. H. Eickbush. 1999. The age and evolution of non-LTR retrotransposable elements. Mol. Biol. Evol. 16793-805. [PubMed]
28. Mayer, C., K.-M. Schmitz, J. Li, I. Grummt, and R. Santoro. 2006. Intergenic transcripts regulate the epigenetic state of rRNA genes. Mol. Cell 22351-361. [PubMed]
29. McKnight, S. L., and O. L. Miller, Jr. 1976. Ultrastructural patterns of RNA synthesis during early embryogenesis of Drosophila melanogaster. Cell 8305-319. [PubMed]
30. McStay, B. 2006. Nucleolar dominance: a model for rRNA gene silencing. Genes Dev. 201207-1214. [PubMed]
31. O'Donnell, K. A., and J. D. Boeke. 2007. Mighty piwis defend the germline against genome intruders. Cell 12937-44. [PMC free article] [PubMed]
32. Ostertag, E. M., and H. H. Kazazian, Jr. 2001. Biology of mammalian L1 retrotransposons. Annu. Rev. Genet. 35501-538. [PubMed]
33. Pérez-González, C. E., and T. H. Eickbush. 2001. Dynamics of R1 and R2 elements in the rDNA locus of Drosophila simulans. Genetics 1581557-1567. [PMC free article] [PubMed]
34. Pérez-González, C. E., and T. H. Eickbush. 2002. Rates of R1 and R2 retrotransposition and elimination from the rDNA locus of Drosophila melanogaster. Genetics 162799-811. [PMC free article] [PubMed]
35. Pikaard, C. S. 2000. Nucleolar dominance: uniparental gene silencing on a multi-megabase scale in genetic hybrids. Plant Mol. Biol. 43163-177. [PubMed]
36. Reeder, R. H. 1985. Mechanisms of nucleolar dominance in animals and plants. J. Cell Biol. 1012013-2016. [PMC free article] [PubMed]
37. Santoro, R., J. Li, and I. Grummt. 2002. The nucleolar remodeling complex NoRC mediates heterochromatin formation and silencing of ribosomal gene transcription. Nat. Genet. 32393-396. [PubMed]
38. Slotkin, R. K., and R. Martienssen. 2007. Transposable elements and the epigenetic regulation of the genome. Nat. Rev. Genet. 8272-285. [PubMed]
39. Smith, C. L., and C. R. Cantor. 1987. Purification, specific fragmentation, and separation of large DNA molecules. Methods Enzymol. 155449-467. [PubMed]
40. Stage, D. E., and T. H. Eickbush. 2007. Sequence variation within the rRNA gene loci of 12 Drosophila species. Genome Res. 171888-1897. [PMC free article] [PubMed]
41. Vagin, V. V., A. Sigova, C. Li, H. Seitz, V. Gvozdev, and P. D. Zamore. 2006. A distinct small RNA pathway silences selfish genetic elements in the germline. Science 313320-324. [PubMed]
42. Verdel, A., S. Jia, S. Gerber, T. Sugiyama, S. Gygi, S. I. Grewel, and D. Moazed. 2004. RNAi-mediated targeting of heterochromatin by the RITS complex. Science 303672-676. [PMC free article] [PubMed]
43. Ye, J., and T. H. Eickbush. 2006. Chromatin structure and transcription of the R1- and R2-inserted rRNA genes of Drosophila melanogaster. Mol. Cell. Biol. 268781-8790. [PMC free article] [PubMed]
44. Zhang, X., and T. H. Eickbush. 2005. Characterization of active R2 retrotransposition in the rDNA locus of Drosophila simulans. Genetics 170195-205. [PMC free article] [PubMed]
45. Zhang, X., J. Zhou, and T. H. Eickbush. 2008. Rapid R2 retrotransposition leads to the loss of previously inserted copies via large deletions of the rDNA locus. Mol. Biol. Evol. 25229-237. [PubMed]

Articles from Molecular and Cellular Biology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...