Rhino forms nuclear foci and binds the genome in OSCs

(A) Confocal immunofluorescent images of FLAG-Rhino (red) in FLAG-Rhino inducible OSCs upon Dox induction (5 ng/mL). H3K9me3 and OSC nucleus (DAPI staining) are shown in green and blue. Scale bar represents 10 µm. (B) Genome snapshot showing ChIP-seq in major dual-strand piRNA clusters (42AB and 38C). The gray-shaded areas indicate the sites at which FLAG-Rhino is localized in each piRNA cluster in OSCs. piRNA cluster 38C has two piRNA-producing regions, labeled 38C1 and 38C2. (C) Heatmap showing FLAG-Rhino localization of OSCs in Rhino peaks of Drosophila wild-type Oregon-R ovary. The regions of 6 kb around the Rhino localization site were plotted based on macs2 callpeak results. (D) Metaplot showing the localization of FLAG-Rhino in OSCs at piRNA clusters. For piRNA clusters, broad peaks of Rhino localization in the ovary greater than 1.5 kb were used. (E) Venn diagram showing the overlap of Rhino, H3K9me3, and H3K9me2 peaks in OSCs. (F) Heatmap showing the signals of ChIP-seq using anti-FLAG, anti-H3K9me3, and anti-H3K9me2 in FLAG-Rhino peaks of OSCs. The regions of 2 kb around the FLAG-Rhino localization sites were plotted based on macs2 callpeak results. (G) Heatmap showing the signals of ChIP-seq using anti-H3K9me3, anti-FLAG, anti-HP1a antibodies at H3K9me3 broad peaks of OSCs. The regions of 10 kb around the FLAG-Rhino localization sites were plotted based on macs2 callpeak results.

ADMA-histones tend to co-localize with FLAG-Rhino in OSCs

(A) Heatmap showing key histone modifications and HP1a ChIP-seq results at FLAG-Rhino localization sites (828 regions) in OSCs. The regions of 2 kb around the FLAG-Rhino localization sites were plotted based on macs2 callpeak results. (B) Genome snapshot showing all histone modifications’ ChIP-seq data in major dual-strand piRNA clusters (38C). The gray-shaded areas indicate the localization sites of FLAG-Rhino in each piRNA cluster in OSCs. (C) Metaplot showing the localization of ADMA-histones and FLAG-Rhino in OSCs at piRNA clusters. For piRNA clusters, broad peaks of Rhino localization in the ovary greater than 1.5 kb were used. (D) Pearson correlation coefficient comparing the coverage per 100 bp bin size of all ChIP-seq mapping results to the dm6 genome and its hierarchical clustering. (E) Venn diagram showing sites of overlapping localization of FLAG-Rhino, H3K9me3, H3R2me2a, H3R8me2a, H3R17me2a, H3R26me2a, H4R3me2a, and all ADMA-histones in OSCs.

Arginine methyltransferases DART1 and DART4 play a role in Rhino genomic localization and foci formation in OSCs

(A) Confocal immunofluorescent images of FLAG-Rhino (red) and nuclei (blue; DAPI stained) in the EGFP-, HP1a-, DART1-, DART4-, DART8-, and CG17726-KD OSCs. Scale bar represents 10 µm. (B) Comparison of FLAG-Rhino foci sizes among the EGFP-, HP1a-, DART1-, DART4-, DART8-, and CG17726-KD OSCs in (A). Thirty OSCs from three biological replicates were analyzed. Statistical significance in differences of Rhino foci between the genotypes was examined using Kolmogorov–Smirnov test. *P < 0.05, **P < 0.01, and ***P < 0.001. HP1a KD: ***P = 8.49 × 10−13, DART1 KD: ***P = 1.83 × 10−7, DART4 KD: ***P = 2.83 × 10−4, DART8 KD: n.s. P = 0.655, CG17726 KD: n.s. P = 0.454. The box indicates the lower (25%) and upper (75%) quartiles, and the open circle indicates the median. Whiskers extend to the most extreme data points. (C) ChIP-seq profile showing the localization of FLAG-Rhino, H3R2me2a, H3R8me2a, H3R17me2a, H3R26me2a, and H4R3me2a in EGFP-, DART1-, and DART4-KD OSCs at FLAG-Rhino localization sites. The genomic regions used to draw the profile are the same as in Fig. 1G. (D) Heatmap showing the localization of FLAG-Rhino, H3R17me2a, and H3R8me2a in EGFP- and DART4-KD OSCs at the sites of decreased FLAG-Rhino localization in DART4 KD. The regions with decreased FLAG-Rhino were defined as those with an M value greater than 0.5 by manorm. macs2 callpeak p = 0.05 was used for manorm with input as a control. (E) FLAG-Rhino ChIP-qPCR results of one site at which Rhino initially localized. DART1 KD: †P = 5.88 × 10−2, DART4 KD: * P = 3.17 × 10−2. (F) Venn diagram showing the overlap of H3R17me2a localization decreased in DART4 KD and H4R3me2a localization decreased in DART1 KD. The regions with decreased ADMA-histones were defined as those with an M value greater than 1 in manorm. macs2 callpeak p = 0.01 was used for manorm. (G) Venn diagram showing the overlap of Rhino localizations that were decreased by DART1 KD and DART4 KD. The decreased genomic sites are identified as in (F).

ADMA-histones co-localize with Rhino prior to Rhino spread in the ovary

(A) Genome browser snapshot showing the localization of H3R2me2a, H3R8me2a, H3R17me2a, H3R26me2a, and H4R3me2a in piRNA clusters in OSCs and Oregon-R ovary. The gray-shaded areas indicate the localization sites of FLAG-Rhino in each piRNA cluster in OSCs. (B) Heatmap showing the localization of H3R2me2a, H3R8me2a, H3R17me2a, H3R26me2a, and H4R3me2a at Rhino peaks in Oregon-R ovary. The Rhino peak regions shown in Fig. 1C were used. (C) Metaplot showing the localization of ADMA-histones in the ovary at piRNA clusters. For piRNA clusters, broad peaks of Rhino localization in the ovary greater than 1.5 kb were used.

DART4 GLKD has impacts on Rhino foci and Rhino genomic localization in the ovary

(A) Confocal immunostaining images of Rhino (red) in ctrl- and DART4-GLKD ovaries. DAPI shows nuclei (blue). Scale bar represents 10 µm. (B) Comparison of sizes of Rhino foci between the ctrl- and DART4-GLKD ovaries in (A). Ten ovaries from two biological replicates were analyzed. Statistical significance in differences of Rhino foci between the two genotypes was examined as in Fig. 3B. DART4 GLKD: ***P = 1.26 × 10−10. (C) ChIP-seq profile showing the localization of Rhino and H3R17me2a in ctrl- and DART4-GLKD ovaries at the FLAG-Rhino localization sites in OSCs. The FLAG-Rhino peak regions, as in Fig. 1G, are used. (D) Rhino ChIP-qPCR results of one site at which Rhino initially localized and piRNA cluster 42AB. t-test was used for statistical hypothesis testing. Rhino initial localization point: * P = 4.03 × 10−2, piRNA cluster 42AB: n.s. P = 0.11. (E) Genome snapshot of ChIP-seq data in euchromatic regions. The gray-shaded areas indicate genomic locations where Rhino localization is decreased in a H3R17me2a-dependent manner. (F) Scatter plot comparing Rhino enrichment in ChIP-seq between ctrl GLKD and DART4 GLKD. Each dot represents the square root of the read count per million mapped reads (RPM) within the annotated piRNA clusters in proTRAC database (colored in black, 184 clusters, without clusters on chromosome Y and uni-strand piRNA clusters flamenco/20A) and Rhino initial localizations as in Fig. 5C (colored in red, 828 regions). The dotted lines represent an approximation by a linear regression model.

DART4-dependent piRNA clusters (DART4 clusters) are genomic loci where Rhino spreads from H3R17me2a

(A) Venn diagram showing the overlap between sites with reduced Rhino localization and sites where H3R17me2a was lost in DART4 GLKD. The regions where Rhino was decreased were defined as those with an M value greater than 0.5 by manorm, and the regions where H3R17me2a was lost were defined as those with an M value greater than 0. For manorm, the output results of macs2 callpeak p = 0.01 were used with input as a control. (B) Scatter plot comparing Rhino enrichment in ChIP-seq between ctrl GLKD and DART4 GLKD. Each dot represents the square root of the read count per million mapped reads (RPM) within the annotated piRNA clusters in proTRAC database (colored in black, 184 clusters, without clusters on chromosome Y and uni-strand piRNA clusters flamenco/20A) and Rhino reduced regions in DART4 GLKD (colored in green, 43 clusters). The dotted lines represent an approximation by a linear regression model. (C) Chromosomal locations of Rhino reduced regions in DART4 GLKD and the estimated direction of Rhino spread based on initial localization in OSCs. Depressed areas indicate pericentromeres. Arrows show the spread direction based on ADMA-histones. Bidirectional arrows indicate that ADMA-histones are present at both ends and the direction of spread cannot be determined. Horizontal bars indicate clusters without ADMA-histones at both ends. (D) Genome snapshot showing Rhino reduced regions in DART4 GLKD around the bx gene. Black lines mark Rhino reduced regions boundaries, and gray-shaded areas highlight potential ADMA-histone localization at cluster ends. (E) ChIP-seq profiles of Rhino and H3R17me2a in ctrl GLKD and DART4 GLKD across all 43 Rhino reduced regions in DART4 GLKD, and Rhino in OSCs. (F) Scatter plot comparing piRNA enrichment between ctrl GLKD and DART4 GLKD. Each dot represents the logged RPM value within piRNA clusters (black: 184 annotated clusters; green: 43 Rhino reduced regions in DART4 GLKD). Dotted lines indicate linear regression. (G) Box plot showing RPKM of piRNAs uniquely mapped to Rhino sites in OSCs, Rhino reduced regions in DART4 GLKD, all proTRAC-annotated clusters, and highly transcribed clusters (42AB, 38C, 80F, 102F). Clusters with RPKM below 400 are shown.

ADMA-histones are enriched at the ends of evolutionarily young piRNA clusters

(A) RPKM of each ChIP-seq dataset (Oregon-R) in the piRNA cluster dataset with eight levels of evolutionary novelty. RPKM was calculated for the 500 bp region surrounding the ends of the piRNA clusters. t-test was used for statistical hypothesis testing between clusters found only in 1 strain (cluster 1, newly emerged) and clusters found in all 8 strains (cluster 8, evolutionarily conserved). All ADMA-histones: ** P = 1.11 × 10−3, H3R2me2a * P = 1.45 × 10−2, H3R8me2a: *** P = 1.27 × 10−4, H3R17me2a: ** P = 7.28 × 10−3, H3R26me2a: n.s. P = 0.23, H4R3me2a: ** P = 1.91 × 10−3, input: * P = 0.53. (B) Same analysis of Fig. 7A for each ChIP-seq dataset (ctrl GLKD and DART4 GLKD). H3R17me2a (ctrl GLKD): ** P = 2.77 × 10−3, H3R17me2a (DART4 GLKD): n.s. P = 0.24, input (ctrl GLKD): n.s. P = 0.42, input (DART4 GLKD): n.s. P = 0.47. (C) Model for the piRNA cluster formation. In the cell conditions lacking Kipferl such as in OSCs, Rhino initially localizes to bivalent nucleosomes containing H3K9me3 and ADMA-histones. In the second step, Rhino spreading occurs, leading to the formation of DART4 clusters. Transcription of adjacent regions and/or H3K9me2 enrichment may play a role in this process. Rhino in DART4 clusters is lost if ADMA-histones are lost. It remains unclear whether DART4 clusters can eventually develop into highly transcribed piRNA clusters, such as 42AB, over the course of evolutionary time.