The HP1 homolog rhino anchors a nuclear complex that suppresses piRNA precursor splicing

Abstract

piRNAs guide an adaptive genome defense system that silences transposons during germline development. The Drosophila HP1 homolog Rhino is required for germline piRNA production. We show that Rhino binds specifically to the heterochromatic clusters that produce piRNA precursors, and that binding directly correlates with piRNA production. Rhino colocalizes to germline nuclear foci with Rai1/DXO-related protein Cuff and the DEAD box protein UAP56, which are also required for germline piRNA production. RNA sequencing indicates that most cluster transcripts are not spliced and that rhino, cuff, and uap56 mutations increase expression of spliced cluster transcripts over 100-fold. LacI::Rhino fusion protein binding suppresses splicing of a reporter transgene and is sufficient to trigger piRNA production from a trans combination of sense and antisense reporters. We therefore propose that Rhino anchors a nuclear complex that suppresses cluster transcript splicing and speculate that stalled splicing differentiates piRNA precursors from mRNAs.

Figure 1. Rhi binding correlates with piRNA production

(A) piRNA and Rhi ChIP-Seq signal across the right arm of chromosome 2. (B) piRNA and Rhi ChIP-Seq signal across the 42AB piRNA cluster, which produces ~ 30% of fly ovary piRNAs. (C) Boxplots showing Rhi enrichment at protein coding genes, transposons and piRNA clusters. Outliers are not shown. (D) Rhi binding correlates with Rhi-dependent piRNA production. Scatter plots showing Rhi binding enrichment (y-axis) relative to reduction in piRNA production in rhi^2/KG mutants (x-axis). Each point represents a distinct piRNA cluster. Pre-I.S. is Pre-Immune Serum. The major germline piRNA cluster at 42AB and the major somatic cluster (flam) are indicated in green. See also Figure S1 and S2.

Figure 2. Rhi and Cuff suppress splicing at the 42AB and sox102F clusters

A, D. Rhi ChIP-Seq and RNA-Seq profiles for the 42AB cluster (A) and sox102F locus (D) are shown. For ChIP-Seq, Rhi signal (red) is superimposed on input (blue). RNA-Seq is shown for Oregon R (Ore. R.) control, cuff and rhi mutants. In Ore. R. controls, signal is spread over both 42AB and sox102F. In rhi and cuff mutants, by contrast, signal shifts to distinct peaks (Note that data are scaled to avoid peak clipping). At the sox102F locus, the boundaries correspond to annotated splice sites in the mature somatic transcript, and de novo transcript assembly from these data yields the annotated gene structure (Trinity As.) (B) A high resolution expansion of the indicated region of 42AB shows that the peak is interrupted by a region with little signal, defined by very sharp boundaries characteristic of intron removal. (C, F) qRT-PCR quantify the splicing efficiency at 42AB (C) and sox102F (F) loci. The diagrams show the putative introns (blue) and the position of primers (arrow) used to assay unspliced and spliced transcripts at 42AB (C) and sox102F (F). Both spliced and unspliced transcripts are amplified using the same forward primers. Reverse Primers for unspliced transcripts span the splice site and for spliced transcripts span the mature junction. Bar graphs show the ratio of spliced to unspliced RNAs in ovary (ov.) and carcass (ca.) in two different control strains (w¹¹¹⁸ and Ore. R.) and in cuff and rhi mutants. In ovaries, splicing at 42AB and sox102F increases over 80 fold in both cuff and rhi mutants. The sox102F locus is expressed in somatic tissue present in the carcass, and the transcripts are spliced. Data are mean ± standard deviation for 3 independent biological samples. (E) piRNA production from sox102F locus in Ore.R., cuff and rhi mutants. See also Figure S3, S4 and S5.

Figure 3. rhi, cuff and uap56 mutations do not alter global splicing efficiency, but lead to splicing of novel cluster introns

A-D. Scatter plots showing splicing efficiency at introns shared between the two indicated genotypes. Each point is one intron. Cluster mapping introns are in red and introns mapping outside clusters in black. E-G. Bar graphs quantifying shared and genotype specific introns. Introns outside of piRNA clusters are in black, introns mapping to all clusters are in red, and introns mapping to the top 20 clusters are in purple. Introns shared by the cn,bw, w¹ and Oregon R control strains are indicated by the open bars. Mutant specific introns are in solid bars. For each set of bar graphs, the genotypes are ordered as in E, and the number of introns detected is above or within the bar.

Figure 4. Rhi “tethering” leads to spreading through the target transcription unit, but does not reduce Pol-II occupancy

(A) Schematic diagram of the experimental design. An inducible UASp promoter was used to drive expression of the DNA binding protein LacI fused to full length Rhi, in the presence of a reporter gene containing LacI binding sites (LacO) upstream of a promoter (truncated vasa promoter) that drives expression of the 84B alpha tubulin 5′UTR and first intron fused to EGFP with nuclear localization signal (NLS). PCR amplicons indicates positions assayed for Rhi and Pol-II binding in panel B and C. (B) Fold enrichment by Rhi ChIP across the reporter in the absence (grey) or presence of LacI∷Rhi. The LacI∷Rhi lead to Rhi binding through the transcription unit. (C) Fold enrichment by RNA pol-II ChIP across the reporter, bars as indicated for panel B. RNA polymerase binding across the transcription unit is not altered in the presence of the LacI∷Rhi. The 42AB locus is used as a positive control for Rhi binding. The mocs and suUR loci are located downstream of reporter construct in the genome. We do not detect Rhi spreading or changes in RNA Pol-II at these loci. Data are mean ± standard deviation for 3 independent biological samples.

Figure 5. Tethering Rhi suppresses EGFP expression and splicing

(A) Germline expression of LacI (red) does not suppress EGFP expression (green). By contrast, expression of LacI∷Rhi (red) suppresses EGFP accumulation in the nurse cell nuclei (green). Note that the EGFP reporter is expressed in both the germline and surrounding somatic follicle cells (arrows, FC). The fusion does not expressed in the follicle cells, and EGFP expression in these cells is not reduced. The bar in the up right panel is 10 μm, and applies to all panels. (B) Splicing at the target locus. The diagram shows the target transgene and indicates that position of LacI or LacI∷Rhi binding (LacO) and the primers used to assay both spliced and unspliced transcripts by qRT-PCR. (C) Bar graph showing the ratio of spliced to unspliced target in the presence of LacI (black) or LacI∷Rhi (grey). LacI∷Rhi binding lead to a significant reduction in splicing efficiency (p = 0.008), data are mean ± standard deviation for 3 independent biological samples. See also Figure S6.

Figure 6. Rhi binding to complementary transcription units triggers piRNA production

(A) Diagrams show sense strand reporter and combination of sense and antisense reporters, indicating positions of LacI or LacI∷Rhi binding and position and orientation of promoters (vasP). (B) Length distribution of the small RNAs mapping to the reporter constructs. Blue indicates sense strand species and red indicates antisense species. Z scores indicate the significance of the 10 nt overlap between sense and antisense piRNAs (Ping-Pong signature). Z-score = 1.96 corresponds to p-value = 0.05. Too few piRNAs were detected with the LacI control for the Z score to be determined (indicated as n.d.). piRNAs carry a 3′ end modified and therefore are resistant to oxidation. Both un-oxidized (Figure S7) and oxidized RNAs (shown) were used to prepare libraries for sequencing. LacI∷Rhi binding to the sense strand reporter did not lead to significant production of oxidation resistant species between 23 and 30 nt. By contrast, LacI∷Rhi binding to the combination of sense and anti-sense reporters trigger production of oxidation resistant species showing length distributions characteristic of mature piRNAs. piRNAs from opposite strands showed a weak bias toward a 10 nt overlap, which is typical of primary piRNAs produced by a ping-pong independent mechanism. (C) Distribution of Small RNA reads over EGFP and the LacO repeats in the presence of LacI or LacI∷Rhi. Sense signal is in blue and anti-sense signal is in red. See also Figure S7.

Figure 7. Predicted Cuff structure and a model for Rhi complex function in piRNA biogenesis

(A) Predicted overlap of homology modeled Drosophila Cuff (wheat) and mouse DXO (Green), a murine mRNA decapping enzyme (Jiao et al., 2013). Catalytic residues are not conserved in Cuff, but residues that interact with the RNA backbone and bases are preserved (yellow). (B) Top: Co-transcriptional binding by the cap binding complex (CBC) promotes efficient pre-mRNA splicing, through a process that requires transient binding by UAP56. Subsequent cap binding by EIF4e promotes translation. Bottom: Cuff localizes to clusters through Rhi, and we speculate that co-transcriptional loading on capped cluster transcripts competes with CPC and stalls splicing with UAP56 bound. Stable cluster transcript-UAP56 complexes are directed to the piRNA biogenesis machinery.