In order to test the role of RNA Pol II on heterochromatin, we employed genetic tests using the inversion of white-mottled4h stock. The In(1)w [m4h] stock has a pericentric inversion between the white gene and the centric heterochromatin. This arrangement results in a variegated eye pattern. Many genes acting as chromatin modifiers suppress or enhance the position-effect variegation (PEV) effect. We used mutations in the second largest subunit of RNA Pol II 140. The mutant alleles used were RNA Pol II 140 (A5) and RNA Pol II 140 (wimp). The A5 allele is a null mutant with a five amino acid deletion while wimp is an antimorph [13, 14]. We observed that RNA Pol II mutations weakly suppressed PEV as a heterozygote. However, the trans-heterozygote of RNA Pol II 140(A5) and dicer-2 (dcr-2 G173E) showed a stronger suppression of PEV when compared with either the single heterozygotes or control normal male flies (Figure 1A and 1B). This experiment revealed genetic interaction between RNA Pol II and Dicer-2, which is a central processing enzyme in the RNAi pathway.
To further understand the nature of the interactions between RNA Pol II and the RNA silencing machinery, we tested the effect of an array of different RNA silencing machinery mutations on PEV including those implicated in piRNA formation, which acts independently of Dicer [15]. In each case, different mutant alleles of each gene were tested in order to rule out any linked gene effect on PEV suppression. The PEV analysis indicated genetic interactions between dcr-2(L811fsX), dcr-2(G173E), ago-2(414), piwi[1], piwi[2], hls [616], hls [125], aub [QC42 ], aub [P3a] and RNA Pol II alleles (Figure 1A and 1B and Additional file 1). In each case the trans heterozygotes of RNA Pol II and RNA silencing pathway mutations exhibited stronger suppression of PEV than single heterozygotes. However, the suppression of PEV was strongest in the trans-heterozygotes of RNA Pol II and dcr-2 alleles among all the other combinations tested as indicated by the eye pigment measurements.
The suppression of PEV is a reflection of the changes in the chromatin structure of heterochromatin. The effect of RNA Pol II and RNA silencing machinery trans-heterozygote mutants was not confined to the heterochromatin environment of the chromocentre. This was shown in an experiment employing transgenic flies that have seven tandem copies of mini-white, referred to as DX1 (Figure 1C), that are located in the euchromatin of chromosome 2. Flies that are homozygous for this transgene arrangement have a heterochromatin environment around the mini-white arrays, thus silencing the expression of mini-white transgenes in mosaic fashion [16]. We tested two combinations, namely RNA Pol II 140(A5) and piwi[1], as well as RNA Pol II 140 (A5) and hls [125] on DX1 homozygous flies. In each case the trans-heterozygotes reversed the silencing of mini-white to a much greater extent when compared to either single heterozygotes or control flies with no mutations. The exact molecular mechanism of DX1 silencing has not been elucidated but is believed that pairing sensitive silencing might be one of the contributing factors. The experiments performed with RNA Pol II mutants suggest that there is also involvement of a transcriptional silencing component.
The strong suppression of PEV in trans-heterozygotes of RNA Pol II and RNA silencing machinery components led us to investigate the heterochromatin structure at the chromocentre of polytene chromosomes. In Drosophila, H3K9me2 modification is concentrated at the centric heterochromatin. H3K9me2 is also interspersed along the euchromatin arms where it is accumulated on transposable elements [17]. We reasoned that, because the suppression of the white gene is relieved in the PEV analysis, H3K9me2 at the chromocentre would be reduced. We performed experiments using third instar larval polytene chromosomes probed with antibodies against H3K9me2. We combined the trans-heterozygotes and the control wild type chromosomes in the same preparation so that they could be observed in one microscopic field for direct comparisons under identical experimental conditions. As our PEV analysis indicated that the trans-heterozygote of RNA Pol II 140 (A5) and dcr-2 (G173E) suppressed PEV very strongly, we analysed this combination for reduction in H3K9me2 at the chromocentre of polytene chromosomes. Indeed, compared to the wild type nuclei, RNA Pol II 140 (A5)/+; dcr-2 (G173E)/+ showed a reduction of H3K9me2 as visualized by the immuno-fluorescence experiments (Figure 2). A similar pattern showing decreased H3K9me2 deposition at the chromocentre was observed using the RNA Pol II 140(wimp)/+; dcr-2 (G173E)/+ combination and RNA Pol II 140(A5)/+; dcr-2 (L811fsX)/+, which illustrates the generality with regard to different alleles at both loci. We then performed immunofluoresence experiments on polytene chromosomes using RNA Pol II 140 (A5)/+ as a control. In accordance with our PEV analysis, RNA Pol II 140 (A5)/+; dcr-2 (G173E)/+ showed decreased H3K9me2 deposition at the chromocentre compared with the single heterozygote of RNA Pol II (A5)/+. The experiments were repeated five times with about 75 pairs of mutant and control nuclei observed. In each case about 75%-80% of the mutant nuclei showed a reduction of H3K9me2 at the chromocentre compared to the wild type. All the experiments were performed by switching the sexes of mutant and normal using antibodies against Sex-lethal, which is only expressed in females, to distinguish male from female nuclei. This was done in order to ensure that the reduction in H3K9me2 at the chromocentre was not sex specific. We then analysed polytene chromosomes using wild type control and trans-heterozygotes of: (1) RNA Pol II 140 (A5)/+; hls [125]/+; (2)RNA Pol II 140(A5)/+; hls [E61]6/+; (3)RNA Pol II140 (A5)/+; piwi[1]/+; and (4)RNA Pol II 140(A5)/+; Lip [D]/+[18] (Figure 2 and Additional file 2). Lip is synonymous with Dmp68 [19], which have been shown to be necessary for RNAi in tissue culture cells [20]. In each case the trans-heterozygote mutants showed reduced H3K9me2 deposition at the chromocentre. The immunofluoresence analysis of polytene chromosomes using trans-heterozygotes complements the PEV phenotypic analysis. These experiments indicate that RNA Pol II exhibits genetic interaction with the RNA silencing machinery components and that the suppression of PEV is correlated with a reduction in H3K9me2 at the chromocentre.
In order to quantify the reduction of H3K9me2, western blot analysis was performed on acid extracted histones using H3K9me2 antibodies. Adult carcasses, with the gonads removed, were used to rule out any effect of the RNA silencing machinery in the germline [21]. The analysis revealed that single heterozygotes of RNA Pol II 140(A5)/+ and RNA silencing machinery mutants alone showed very modest to no change in H3K9me2 levels compared to the wild type. However, trans-heterozygotes for the combined mutants showed a strong reduction in H3K9me2 levels compared with wild type and single heterozygotes (Figure 3). The western blot analysis also corroborated the PEV analysis.
We next performed chromatin immunoprecipitation (ChIP) using H3K9me2 antibodies on adult flies. The combination of RNA Pol II and dcr-2 was selected because it gave the strongest suppression of PEV in the w [m4h] background. The ChIP analysis revealed significant enrichment of H3K9me2 at the white locus in the vicinity of centromeric heterochromatin (w [m4h] genetic background). The tubulin locus (in euchromatin) did not show any enrichment of H3K9me2 nor any significant difference in the amount of H3K9me2 between the control (In(1)w [m4h];+/+), single heterozygous mutants and the double heterozygotes of RNA Pol II 140 and dcr-2(L811fsX). However, at the white (which lies in the vicinity of centromeric heterochromatin due to the inversion) locus, the double heterozygotes of RNA Pol II 140 (A5)/+; dcr-2 (L811fsX) showed significant reduction (about fourfold) of H3K9me2 compared to the control as well as to single heterozygote mutants (Figure 4). Also, there was no significant change in H3K9me2 between the control and single heterozygotes of RNA Pol II 140 and dcr-2. The ChIP analysis at the white locus in the In(1)w [m4h] genetic background indicates the importance of H3K9me2 in suppressing the white locus. The ChIP results are consistent with the PEV analysis and Western blot results.
In addition to the second largest subunit of RNA Pol II, we also studied the effect of the RNA Pol II's largest subunit mutation on H3K9me2 levels in the adult carcass (Figure 3). As the largest subunit gene (RNA Pol II215 W81) is located on the X-chromosome, a PEV analysis of male flies was not possible and the fact that a translocation balancer chromosome between the X chromosome and the second chromosome was not available precluded any immunofluoresence analysis on larval polytene nuclei. The mutant allele used was W81, which has a truncated carboxyl terminal domain (CTD) due to the presence of a premature stop codon.
Trans-heterozygotes of RNA Pol II215(W81)/+; dcr-2(L811fsX) showed a significant reduction of H3K9me2 in Western blot analysis compared to the wild type as well as RNA Pol II215(W81)/+ alone. The reduction of H3K9me2 with two different subunits of RNA Pol II, in combination with dicer-2 mutations, provides further evidence of a role of RNA pol II in heterochromatin formation in conjunction with RNA silencing genes.
The chromocentre of Drosophila is characterized by strong deposition of heterochromatin protein-1 (HP1). With a reduction of H3K9me2, HP1 is deposited at various low affinity-binding sites along the chromosome arms [22]. The presence of H3K9me2 provides high affinity binding sites for the docking of HP1. We examined the polytene chromosomes of RNA Pol II 140 (A5)/+; dcr-2(G173E) trans-heterozygotes for any changes in HP1 deposition pattern (Figure 5). The gently squashed polytene nuclei from third instar larvae showed mislocalization of HP1 to the euchromatin arms compared to wild type nuclei, which showed a much more discrete HP1 deposition at the chromocentre.
With regard to piRNA genes, previous experiments involved examining HP1 mislocalization in piwi[1]/piwi[2] heteroallelic mutants. This combination did not show a major mislocalization of HP1 [23]. To test the impact of RNA pol II, we introduced the RNA Pol II140(A5)/+ mutation in this background. This combination caused an obvious mislocalization of HP1 (Figure 5).
To establish whether the cytological observations represented a mislocalization or a quantitative difference, we used a Western blot analysis which indicated that HP1 protein levels were the same in the wild type and RNA Pol II 140(A5)/+; dcr-2 (G173E)/+, thus confirming that HP1 is mislocalized and not upregulated in the trans-heterozygote mutants (Additional file 3). The mislocalization can be attributed to a reduced H2K9me2 deposition at the chromocentres of mutants, which allow HP1 to associate with various low affinity binding sites. The above experiments also highlight the role of small RNAs generated by the transcription of heterochromatic repeats in guiding heterochromatin modifications (H3K9me2 and HP1) at the chromocentre.
In order to gain a further insight into the mechanism by which RNA Pol II and RNA silencing machinery regulate heterochromatin structure, we performed co-immunoprecipitation using extracts from Drosophila wild type embryos (6-18 h old). The specificity of Dicer-2 antibody was confirmed by western blot analysis (Additional file 4). We found co-IP between Dicer-2 and RNA Pol II ser-2 phos CTD, which is a transcriptionally competent form (Figure 6). This result suggests that the genetic interactions described above have a basis in a biophysical interaction.
In plants, experiments have implied a role of WG/GW motifs as docking sites for Argonaute binding such as for AGO4 to the CTD of the largest subunit of Pol IVb subunit NRPD1B [24], which is a specialized RNA pol II involved with transcriptional silencing. The CTD of the Drosophila lacks any reiterative GW/WG motifs and it might be speculated that the absence of these domains could contribute to the lack of any physical interaction between dAGO2 and RNA Pol II. We could not detect any interaction between dAGO2 and RNA Pol II CTD, implying that the interaction is very weak or indirect. The analysis of the amino acid sequence of RNA Pol II 140 revealed the presence of a PxVxV site (residues 350-354) (Additional file 5). Similarly, AGO2 also contains the pentapeptide PxVxV (residues 486-490)(Additional file 5). The peptide sequence PxVxM/L/V represents the conserved sequence found in all HP1 interacting proteins [25]. Recent experiments performed in flies demonstrate that PIWI interacts physically with the HP1 protein by virtue of the presence of the PxVxV domain [26]. The amino acid replacement of the central valine residue of the pentapeptide abolished interaction between HP1 and PIWI, thus highlighting the importance of the pentapeptide domain for this interaction. As RNA Pol II 140 possesses a PxVxV domain, it is interesting to speculate that HP1 might bridge PIWI and AGO2 with the RNA Pol II 140 subunit. This might constitute a novel RNA Pol II complex in metazoans exclusively dedicated for silencing. We could not address this issue because of the unavailability of suitable RNA Pol II 140 antibodies for immunoprecipitation.
However, we found dAGO1, which typically binds miRNAs, co-immunoprecipitated with RNA Pol II (8WG16) CTD antibody (Figure 6), but not with antibodies against the activated CTD. The presence of dAGO1 in the pulldown fraction using 8WG16 antibodies prompted us to investigate the role of the miRNA machinery in heterochromatin modifications. dcr-1 and ago-1 are two genes that play a predominant role in miRNA metabolism in flies [27, 28]. There was no effect on H3K9me2 modification at the chromocentre of polytene chromosomes of RNA Pol II(A5)/+; dcr-1(Q1147X) compared to wild type (Additional file 6). Similarly this trans-heterozygote combination did not relieve the silencing of the mini-white array DX1 or the In(1)w [m4h] heterochromatin environment. When trans-heterozygotes of RNA Pol II140 (A5)/+; ago1(k04845)/+ were introduced into the In(1)w [m4h] background, there was moderate suppression of PEV (Additional file 7). Similarly, the chromocentre of polytene chromosomes in this background caused moderate reduction of H3K9me2 compared to wild type and there was a significant reduction in H3K9me2 levels in the Western blot analysis. The presence of AGO1 in the RNA Pol II (8WG16) pulldown fraction indicates that AGO1 might have an affinity for binding to small RNAs arising from heterochromatin. This may well be the case as there is evidence that AGO1 and AGO2 have somewhat overlapping functions and there is sharing of biochemical components among miRNA, endo-siRNA and siRNA pathways in Drosophila[29, 30].
In order to address the in vivo association between RNA Pol II and small RNA silencing machinery proteins, we examined possible co-localization patterns between them on polytene chromosomes. The immunofluorescence analysis on polytene chromosomes revealed a few sites of co-localization between RNA Pol II (8WG16 antibody) and dAGO1 (Figure 7 and Additional files 8 and 9). The overlapping positions between AGO1 and 8WG16 (RNA Pol II) might potentially represent sites where the small RNA machinery is involved with RNA Pol II in maintaining local chromatin structure and hence gene expression. The in vivo association between RNA Pol II and AGO1 at few sites on polytene chromosomes provides further evidence of a physical association between RNA Pol II and the small RNA silencing machinery. The association of RNA Pol II with PIWI, AGO-2 and Dicer-2 could not be addressed because of the non-availability of antibodies suitable for polytene chromosome staining.
The TAF-1/TFIID (TATA box binding protein associated factor 1) is the major component of the transcription initiation complex in eukaryotes. Trans-heterozygotes of RNA Pol II140(A5)/+; TAF-1/+ and the single heterozygote TAF-1/+ shows no effect on suppression of PEV (Additional file 10). This control shows that mutations affecting transcription factors (TAF-1 interacts with RNA Pol II as part of general transcription machinery) had no effect on suppression of PEV and implicates the specific interaction of pol II with the small RNA silencing components.
In order to test whether the double heterozygous combinations affect post-transcriptional functions of the RNA silencing machinery, selected genotypes were examined for an effect on white RNAi. The trans-heterozygotes of (1) RNA Pol II(A5)/+; piwi[1]/+; (2) RNA Pol II (A5)/+; hls [125]/+; and (3) RNA Pol II (A5)/+; dcr-2(L811fsX)/+ had no effect on w-IR RNAi (Additional file 11). This experiment indicates that in RNA Pol II 140(A5)/+ heterozygotes in which the dose of RNA Pol II is halved, the effect is more pronounced on the heterochromatin structure (transcriptional gene silencing (TGS), but the w-IR RNAi pathway is unaffected under these circumstances. To obtain further insight into the role of RNA Pol II in silencing, we tested for an effect on TGS silencing involving the interaction between Alcohol dehydrogenase-white hybrid transgenes [31]. In these flies the transgene copies of w-Adh bring about silencing of Adh-w at the transcriptional level; however, the silencing is eliminated in a piwi mutant background [31, 32]. When RNA Pol II 140(A5)/+ was introduced into this genotype, there was no apparent effect on silencing (Additional file 12). The trans-heterozygotes of RNA Pol II 140(A5)/+; hls [125]/+ also had no effect on relieving the silencing of Adh-w by w-Adh transgenes.