Heterochromatic Silencing and HP1 Localization in Drosophila Are Dependent on the RNAi Machinery

See allHide authors and affiliations

Science  30 Jan 2004:
Vol. 303, Issue 5658, pp. 669-672
DOI: 10.1126/science.1092653


Genes normally resident in euchromatic domains are silenced when packaged into heterochromatin, as exemplified in Drosophila melanogaster by position effect variegation (PEV). Loss-of-function mutations resulting in suppression of PEV have identified critical components of heterochromatin, including proteins HP1, HP2, and histone H3 lysine 9 methyltransferase. Here, we demonstrate that this silencing is dependent on the RNA interference machinery, using tandem mini-white arrays and white transgenes in heterochromatin to show loss of silencing as a result of mutations in piwi, aubergine, or spindle-E (homeless), which encode RNAi components. These mutations result in reduction of H3 Lys9 methylation and delocalization of HP1 and HP2, most dramatically in spindle-E mutants.

Small RNA molecules have been found to play multiple roles in regulating gene expression. These include targeted degradation of mRNAs by small interfering RNAs (siRNAs) (posttranscriptional gene silencing, PTGS) (1, 2), developmentally regulated sequence-specific translational repression of mRNA by micro-RNAs (miRNAs) (3), and targeted transcriptional gene silencing (TGS) (49). RNAi activity limits transposon mobilization and provides an antiviral defense (10). Recent work demonstrated that RNA interference (RNAi) is required to establish silencing at heterochromatic domains in fission yeast (8, 9); appearance of transcripts from centromeric repeats is accompanied by loss of histone H3 Lys9 methylation (8, 9).

Many components of the RNAi machinery have been identified in Drosophila melanogaster, where they have been implicated in PTGS of the tandemly repeated Stellate genes, several retrotransposons, and Alcohol dehydrogenase (Adh) transgenes (5, 11, 12). Mutations in aubergine (aub), encoding a PAZ domain/PIWI domain (PPD) protein, and in spindle-E (also known as homeless, hls), encoding a DEAD-motif RNA helicase, up-regulate Stellate expression. hls mutations also increase the expression of some retrotransposons and genomic repeats (11, 12). Mutations in piwi (also a member of the PAZ domain family) block PTGS of Adh transgenes (5). Embryos with mutations in aub and hls do not support RNAi in response to injection of double-stranded RNA (13).

These findings suggest that the RNAi system could also play a role in targeting heterochromatin formation in Drosophila. Components of the heterochromatin-silencing complex have been identified by screens for dominant suppressors of position effect variegation (PEV), the silencing that occurs when a normally euchromatic gene is juxtaposed with a heterochromatic domain. The above mutations were originally identified in screens for germline or embryonic abnormalities; we have tested their potential to impact heterochromatic silencing using two systems.

Tandem repeats of a Drosophila white transgene P[lacW] result in a variegating phenotype (14). Silencing is lost in Su(var)205/+ mutants [reduction of heterochromatin protein 1 (HP1)] and altered by changes in the number of Y chromosomes, as expected for heterochromatin-induced silencing. We examined mini-white lines 6-2 (mini-w, one copy), BX2 (seven tandem copies), and DX1 (seven copies, one inverted) (15). Mutations in piwi and homeless do indeed relieve silencing at the repeat loci (Fig. 1). Two alleles of piwi (piwi1 and piwi2) and their heteroallelic combination were tested with similar results. Three tested alleles of hls (125, E1, E616), as well as the heteroallelic combinations E1/E616 and E1/DE8, cause suppression of variegation at DX1, the mini-w array showing the strongest silencing. Similar results were obtained for BX2. In some instances, these mutations fail to show a dominant phenotype, but loss of silencing is consistently observed when the mutation is homozygous or present in a heteroallelic combination.

Fig. 1.

piwi and homeless are suppressors of repeat-induced silencing. Stocks homozygous for a P[lacW] at 50C in one copy (6-2 mini-w), seven tandem copies (BX2 mini-w), or seven copies with one inverted (DX1 mini-w) were tested for loss of silencing. Heterozygous, homozygous, or heteroallelic combinations of piwi or homeless mutations result in an increase in expression as shown in photos of male eyes (above) or by levels of eye pigment extracted from male heads of the noted genotypes, measured at 480 nm (right). Mean values (bar) of triplicate determinations are reported in comparison with the value for the respective +/+ control mini-w stock (dashed line), with the standard error indicated (thin line). Northern analysis of white mRNA from selected genotypes indicates a similar response (fig. S1).

Insertion of the P element P[hsp26-pt, hsp70-w] in a euchromatic domain results in a uniform red eye, whereas insertion in the pericentric heterochromatin or much of the small fourth chromosome results in a variegating phenotype. These variegating lines show loss of silencing on introduction of dominant suppressors of PEV and respond to changes in copy number of the sex chromosomes as anticipated for heterochromatic silencing (16). We have examined the impact of mutant alleles of piwi, aubergine, and homeless on two such lines (15), and we have observed that the functions of all three loci are required for heterochromatic silencing (Fig. 2). Homozygous or heteroallelic mutations in piwi result in a twofold increase in white expression in line 118E-10. The heteroallelic mutant combination of aubergine produces more than a fivefold increase in pigment. Five alleles of homeless were tested, and all show a dominant suppression, i.e., cause a loss of silencing. Similar results were obtained with line 39C-12.

Fig. 2.

Suppression of PEV by components of the RNAi system. Homozygous or heteroallelic mutations in piwi result in an increase in white gene expression (loss of silencing) in line 118E-10 (transgene in pericentric heterochromatin). The heteroallelic mutant combination of aubergine produces strong suppression. The homeless mutations have a dominant phenotype with a several-fold increase in expression, depending on the allele. Males were photographed 3 days after eclosion. Similar, but less dramatic, results were obtained by using stock 39C-12(transgene in the fourth chromosome). Pigment values are reported relative to the control y w67c23 stock carrying the respective transgene.

The RNAi machinery may function throughanRNAmoleculethatdirectssequence-specific targeting of heterochromatin formation. Known components of heterochromatin in Drosophila include histone H3 specifically modified by methylation at Lys9 (H3-mK9), HP1, and HP2, a partner of HP1 in Drosophila (17). Volpe et al. (8) have reported a loss of H3-mK9 and Swi6, the HP1 homolog, at centromeric repeats in S. pombe as a result of mutations in the RNAi system. We examined the effects of homozygous mutations in piwi, aubergine, and homeless on HP1 and H3-mK9 by immunofluorescent staining of the polytene chromosomes (15). homeless mutations have a dramatic effect on the distribution of HP1, normally concentrated at the pericentric heterochromatin and the fourth chromosome. In hls/hls, HP1 is distributed across the whole of the polytene chromosomes (Fig. 3A). At the same time, there is a significant reduction of histone H3-mK9 (Fig. 3B). This reduction was also observed on Western blots of adult fly extracts (Fig. 3C). However, the total amount of extractable HP1 in various hls genotypes appears similar (Fig. 3C), which suggests that expression of HP1 is not affected in hls mutant lines, but rather the distribution within the nucleus. Thus, the RNAi system must be intact to achieve targeted methylation of histone H3 at Lys9, and proper localization of HP1. The changes observed readily account for the loss of PEV.

Fig. 3.

Mutations in components of the RNAi system result in a delocalization of HP1 and a strong reduction in H3 methylated at Lys9. (A) Salivary glands from wild-type Canton S larvae and hls/hls larvae were fixed (3.7% formaldehyde), squashed together on the same slide, treated with mouse monoclonal antibodies specific for HP1 (18) and a secondary Cy5-conjugated goat antibody directed against a mouse antibody, and viewed by confocal microscopy. Simultaneous preparation and treatment of the glands on the same slide permits an assessment of the relative amounts and distribution of the antigen in the two lines. HP1 shows dramatic delocalization in hls/hls mutants. Scale bar, 10 μm. (B) Salivary glands from wild-type Canton S larvae and hls/hls; BX2/BX2 larvae were squashed together on the same slide and stained using antibodies against histone H3-mK9 (Upstate Biotechnology). Immunostaining was performed in a manner to maximize detection of modified H3. There is a strong reduction of methylated H3 in the mutant line. (C) Western blot analysis showing amounts of HP1 and H3-mK9 in normal and homeless mutant flies. The histogram displays the quantification of triplicate blots. HP1 does not vary significantly, but H3-mK9 is reduced, relative to the amount of tubulin in either heterozygous or homozygous hls mutants. Means significantly different from Canton S at the 95% level of confidence are marked with an asterisk. (D) Histone H3-mK9 is localized to the 50C mini-white arrays that show strong silencing [BX2(7), DX1(7)] in a wild-type genotype, but is not detectable above the normal level when only a single active transgene [6-2(1)] is present. The strong accumulation on the multiple arrays is lost in lines with heteroallelic mutations in piwi or homeless. Chromosomes were probed with antibodies specific for H3-mK9. Gray value images were pseudocolored and merged. Scale bar, 10 μm.

A single copy of the P[lacW] transgene at 50C [line 6-2], as well as the BX2 and DX1 seven-copy arrays, were examined for the presence of H3-mK9 (15). The location of this heterochromatic array away from the chromocenter allows a determination of whether there is an accumulation of modified H3 correlated with gene silencing. No detectable H3-mK9 above the normal level was found associated with the fully active single copy, but a strong band of labeling was present in the two seven-copy array lines (Fig. 3D). Previous work has shown a strong association of HP1 with the silenced copies (19). In the piwi mutant, the strong H3-mK9 labeling is no longer discernible. In the hls mutant, there is a general loss of labeling across the nucleus (see also Fig. 3B). Taken together, these results indicate that the hls gene product is required for the proper targeting of H3 modification by methylation of Lys9 at the mini-white array.

Mutations in piwi and aubergine result in partial loss of H3-mK9, most evident at minor sites within the euchromatic arms (Fig. 4; figs. S2 and S3). Mutation of homeless has a more pronounced effect, resulting in dramatic loss of H3-mK9 and redistribution of HP1 and HP2 away from the chromocenter, along the euchromatic arms (Fig. 4; fig. S4). Antibodies specific for H3-mK9 and for H3-mK27 (20) were used to confirm that the effect is specific to H3-mK9 (fig. S5). HP1 interacts with SU(VAR)3-9, a major histone H3 methyltransferase, and the normal localization of these proteins to pericentric heterochromatin has been shown to be mutually dependent (21). The general distribution of HP1 along the chromosome arms in the absence of targeted H3-mK9 is not surprising, as HP1 has been shown to bind nonspecifically to nucleosome core particles and naked DNA (22). HP2 interacts with HP1 through the HP1 chromo-shadow domain, and has previously been found to undergo a shift in distribution in the chromosome upon redistribution of HP1 (17).

Fig. 4.

Mutations in components of the RNAi system result in a loss of histone H3-mK9, and a delocalization of heterochromatin proteins HP1 and HP2. Polytene chromosomes (prepared as in Fig. 3) were treated with rabbit polyclonal primary antibodies specific to HP1, HP2, or histone H3-mK9, as specified, and with antibodies against the female specific protein, Sex-lethal, used to distinguish mutant from wild-type chromosomes. Antibodies were applied to mixtures of Canton S wild type with piwi1/piwi2, aubQC42/ΔP-3a, or hlsE1/hlsE616 glands; piwi1/piwi1, hlsE1/hlsDE8, and hlsE1/hlsΔ125 showed similar results. In the supporting online material, adjacent nuclei on the same slide, but of different genotype, are presented for each comparison (figs. S2 to S4). The level of H3 methylated at Lys9 is progressively reduced, both at heterochromatic and euchromatic sites, in the piwi/piwi, aub/aub, and hls/hls lines, with a progressive delocalization of HP1 and HP2. Scale bar, 10 μm.

Although the decrease in silencing of the P[lacW] array and of the white transgenes in pericentric and fourth chromosome heterochromatin is readily detected, it is a partial effect; one does not observe restoration of a uniform red eye phenotype. The fact that HP1 and HP2 retain nearly normal distribution in the presence of piwi or aubergine mutations, but not following loss of homeless gene product, suggests that homeless encodes a more central function than piwi and aubergine for heterochromatin formation. All three loci appear to be involved in targeting histone H3 methyltransferase activity and localization of HP1 and HP2, demonstrating an important role for the RNAi machinery in establishing this pattern of histone modification and concomitant gene silencing.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

References and Notes

References and Notes

View Abstract

Navigate This Article