Co-dependent Assembly of Drosophila piRNA Precursor Complexes and piRNA Cluster Heterochromatin

Abstract

In Drosophila, the piRNAs that guide germline transposon silencing are produced from heterochromatic clusters marked by the HP1 homolog Rhino. We show that Rhino promotes cluster transcript association with UAP56 and the THO complex, forming RNA-protein assemblies that are unique to piRNA precursors. UAP56 and THO are ubiquitous RNA-processing factors, and null alleles of uap56 and the THO subunit gene tho2 are lethal. However, uap56^sz15 and mutations in the THO subunit genes thoc5 and thoc7 are viable but sterile and disrupt piRNA biogenesis. The uap56^sz15 allele reduces UAP56 binding to THO, and the thoc5 and thoc7 mutations disrupt interactions among the remaining THO subunits and UAP56 binding to the core THO subunit Hpr1. These mutations also reduce Rhino binding to clusters and trigger Rhino binding to ectopic sites across the genome. Rhino thus promotes assembly of piRNA precursor complexes, and these complexes restrict Rhino at cluster heterochromatin.

Keywords: Drosophilia; heterochromatin; piRNA; transposon silencing.

Figure 1.. UAP56-THO Complex Interactions Are Required for piRNA Biogenesis

The uap56^sz15 point mutation disrupts piRNA biogenesis. We used IP-mass spectrometry to identify proteins showing altered binding to UAP56^sz15 protein relative to wild-type. (A) Rank-order plot of fold change in protein abundance in UAP56^sz15venus precipitates relative to UAP56venus controls. Average fold change was calculated from three biological replicates. The inset shows the top 20 proteins with the highest reduction in binding to UAP56^sz15venus.THO subunits are labeled in red. (B) Western blot for the THO complex protein Hpr1 in UAP56venus and UAP56^sz15venus precipitates from wild-type Drosophila ovary lysates. UAP56venus and UAP56^sz15venus were detected using rabbit anti-GFP, and Hrp1 was detected with ratanti-Hrp1. The blot was sequentially probed for Hpr1 and GFP, which were detected with distinct fluorescent secondary antibodies. A series of dilutions of each precipitated sample was analyzed. (C) Plot showing Hpr1 signal normalized to the corresponding Venus tag from four biological replicates. The p values were calculated from t tests. (D) Schematic representation of TREX complex integrity in wild-type and mutants (uap56^28/sz15, thoc5^e/1, and thoc7^d/Df), on the basis of IP-mass spectrometry. See also Figure S1 and Tables S1–S3.

Figure 2.. UAP56 and THO Binding Is Unique to piRNA Cluster Transcripts

(A) Genome browser views of Thoc5GFP, Hpr1,and UAP56venus RIP-seq signal, with IP and input controls, over the 42AB piRNA cluster (class I), a large intron in the Doa gene (class II), and mature spliced RNA from the CG7747 gene (class III). Germline cluster transcripts fall into class I and bind to UAP56 and THO complex subunits. A subset of introns define class II and bind strongly to THO but show weak binding to UAP56. Class III includes mature mRNAs and somatic piRNA cluster transcripts. (B) Scatterplots showing normalized piRNA clustertranscripts abundance (log₁₀[RPKM + 0.1]) in precipitates of Thoc5GFP, Hpr1, and UAP56venus, relative to input. The last panel on the right shows the rank-ordered fold enrichment (RIP/ input) for each cluster in the experimental precipitates and in GFP and non-specific antibody controls (as indicated in panel). (C) Scatter and rank-order plot, as described in (B), for exon mapping transcript enrichment. Color-coded contour lines reflecting data point density are overlaid on each scatterplot. (D) Scatter with contour lines and rank-order plots for intron mapping RNA-seq read RIP enrichment. See also Figure S2.

Figure 3.. Rhi Promotes UAP56 Binding to Cluster Transcripts

(A and B) Genome browser view of the left end of the 42AB piRNA cluster. (A) RIP-seq signal for UAP56venus and Thoc5GFP from wild-type ovaries. (B) RIP-seq signal for UAP56venus and Thoc5GFP from rhi^2/KG ovaries. The dashed boxes in (A) and (B) indicate a partial gypsy12 element in 42AB that is spliced in rhi^2/KG. This region is expanded on the right side of each panel. In wild-type, the unspliced transcripts from this region bind to UAP56venus and Thoc5GFP. In rhi^2/KG, unspliced transcripts bind to Thoc5GFP but not to UAP56venus, while the spliced transcripts do not bind either UAP56venus or Thoc5GFP. (C) Scatterplots comparing cluster transcript abundance (log₁₀[RPKM + 0.1]) in Thoc5GFP RIP (top row) and UAP56venus RIP (bottom row) relative to the corresponding inputs, from rhi^2/KG, thoc7^d/Df, and uap56^28/sz15 ovaries. The box plot summarizes cluster transcript fold enrichment (RIP/input) in wild-type and mutants. The p values were calculated using Wilcoxon rank-sum tests. GFP RIP served as non-specific control. (D) Scatterplots of intronic transcript abundance(log₁₀[RPKM + 0.1]) in Thoc5GFP RIP relative to input, from rhi^2/KG, thoc7^d/Df, and uap56^28/sz15 ovaries. The box plot summarizes the fold enrichment (RIP/input) in the mutants for the 14% of introns that are enriched by more than 2-fold in both Hpr1 and Thoc5GFP in wild-type (Figure 3). The p values were calculated using Wilcoxon rank-sum tests. GFP RIP served as non-specific control. See also Figure S3.

Figure 4.. The TREX Complex Restricts Rhi to H3K9me3 Marks on piRNA Clusters

(A–C) Genome browser view of chromosome 2R (A), the piRNA cluster at 42AB (B), and ectopic Rhi peaks at CG6470 (C). Red tracks are Rhi ChIPseq signal from w¹, thoc7^d/Df, thoc5^e/1, and uap56^28/sz15. The blue track is H3K9me3 ChIP-seq signal from w¹. The annotated piRNA clusters are highlighted in orange. The genomic positions of computationally defined Rhi ChIP-seq peaks for each genotype are color highlighted in the Rhi domain track. In wild-type, Rhi is largely confined to piRNA clusters. In thoc7^d/Df, thoc5^e/1, and uap56^28/sz15, cluster binding is reduced, and new peaks are present across the chromosome arm. (D) Rhi ChIP-qPCR for 42AB and CG6470, with rp49, cluster2, and flam as controls. Arrowheads in the gene models in (B) and (C) indicate the location of qPCR primers. Significance of ChIP signal in the mutants relative to w¹ was determined using t test from four biological replicates. *p < 0.05, **p < 0.001. (E) Pie charts showing the number of Rhi domains that overlap with annotated piRNA clusters (orange charts) and Rhi domains that overlap with H3K9me3 domains in each genotype (blue charts). (F) A speculative feedforward mechanism for piRNA cluster heterochromatin assembly. We propose that the RDC, through the Rhi chromo domain, samples H3K9me3 marks throughout the genome, but binding at transcriptional silent chromatin is unstable (1). In contrast, RDC binding to H3K9me3 marks at transcribed piRNA clusters is followed by Cuff association with capped cluster transcripts, which blocks cap binding by the cap binding complex, stalling splicing and stabilizing UAP56 and THO binding (pre-piRNP) (3). Within this chromatin-bound protein-RNA complex, the RDC does not exchange with the soluble pool, driving the complex to H3K9me3 marks on clusters (4). Deadlock then recruits transcription factors (TFIIA-S, Moonshiner, and TRF2) (Andersen et al., 2017) to trigger capped non-canonical transcription on both strands (5), which enhances prepiRNP assembly and RDC localization (6). See also Figure S4 and Table S4.