A-C. Mutations in mnk, which encodes the DNA damage signaling kinase Chk2, suppress the Gurken and Vasa protein localization defects in rhi mutants. (A) In a stage 9 wild type oocyte, Grk (blue) is localized at the dorsal anterior cortex near the oocyte nucleus and Vas (red) is localized at the posterior cortex. Actin filaments (green) mark the cell boundaries. (B) In rhi egg chambers, this localization pattern is lost, with Grk and Vas dispersed throughout the oocyte. (C) mnk suppresses the rhi phenotype, and rescues Grk and Vas localization during late oogenesis. Images were acquired under identical conditions. Projections of 2 serial 0.6 mm optical sections are shown. Scale bars are 20 mm. D-F. rhi mutants have increased DNA damage in the germline. (D) Foci of γ̃H2Av are observed in wild type ovaries in region 2a and 2b of the germarium and correspond to the DSBs induced during meiotic recombination. (E) In rhi mutants, much larger foci also appear in region 2a of the germarium but persist in region 3 and the developing egg chambers. Samples were labeled and images were acquired under identical conditions. Projections of 5 serial 1mm optical sections are shown. Posterior is oriented to the right. Scale bars are 10 mm. (F) A schematic representation of the regions of the germarium and a developing egg chamber.

A-C. Rhi localization appears similar in (A) wild type, (B) armi and (C) aub stage 2 to stage 4 egg chambers. Projections of 5 serial 1mm optical sections are shown. Scale bar is 20 mm. D. GFP-Rhi transgene shows localization pattern similar to endogenous Rhi detected with anti-Rhi antiserum in the germline nuclei of stage 4-5 egg chambers. Scale bar is 10 mm. E-G. Wild type ovaries immunostained with (E and G) anti-Rhi antiserum and (F and G) anti-CID antibody show that some Rhi foci localize adjacent to CID foci (arrows) consistent with binding to peri-centromeric heterochromatin in some but not all chromosomes. Scale bar is 5 mm. H-O. rhi mutation disrupts localization of PIWI class Argonautes. Stage 4-5 (H, I, L and M), stage 2-3 (J and N), and germarium to stage 3-4 egg chambers (K and O) of wild type and rhi mutant ovaries were immunostained with corresponding antibodies. Projection of 3 serial 1mm optical sections. Scale bars (H, I, J, L, M and N) 10 mm (J,K and O) 20 mm. Wild type localization of Ago3 and Aub proteins to peri-nuclear nuage is disrupted in rhi mutants. Piwi protein localizes to the nuclei of both germline and somatic cells in wild type egg chambers. Only germline nuclear localization of Piwi in early stages is disrupted by mutations in rhi.

Aa. Sense and antisense piRNA reads mapping to the consensus Het-A sequence. Ab. Length histograms for all HetA piRNAs. Ac. Frequency distribution of overlapping HetA piRNA. A statistically significant 10 nt bias (red bar) is characteristic of ping-pong amplification. Ba. Sense and antisense piRNAs reads mapping to the consensus jockey sequence. Bb. Length histograms for all jockey piRNAs. Bc. Frequency distribution of overlapping jockey piRNAs. Ca. Sense and antisense piRNA reads mapping to the consensus blood sequence. Cb. Length histograms for all blood piRNAs. Cc. Frequency distribution of overlapping blood piRNA. The for the majority of transposons, including HetA, rhi mutations dramatically reduce sense and anti-sense piRNAs and nearly elinimate piRNA that overlap by 10nt. For a subset of elements, represented by jockey,rhi leads to a loss of anti-sense piRNAs but no significant reduction in sense strand piRNAs. A very limited number of transposons, including blood, show no change or an increase in sense strand piRNAs in rhi mutants. Mutations in rhi reduce piRNAs with a 10 nt overlap, even for elements that show an increase in piRNAs from opposite strands (Cc, blood).

1 (black arrows). Rhino binds to dual-strand clusters and promotes production of RNAs from both genomic strands (1a), which are exported from the nucleus (1b) and processed into piRNAs by a ping-pong cycle driven by Ago3 (pink hexagon) and Aub (yellow hexagon), localized to the perinuclear nuage (1c). 2 (red arrows). Transposons carrying piRNA homology are transcribed (2a), exported from the nucleus (2b), and degraded as they encounter the “ping-pong surveillance machine” within the nuage. 3 (green arrows). Protein coding genes with intronic transposon insertions are spliced, which removes piRNA homology (3a). These transcripts are exported from the nucleus (3b) escape recognition by the surveillance system, and are translated (3c). Sequences matching piRNAs are indicated by blue. Other transcribed regions are in orange.

A. Genome Browser view of tiling array data in near jing, a protein-coding gene in pericentromeric heterochromatin on chromosome 2R. Expression of jing exons (green bars) is unaltered by rhi mutants. However, several intronic and extragenic transposons are significantly over-expressed (pink bars). B and C. Genome wide analysis of transposon family expression in rhi and armi mutants. Tiling arrays were used to quantify expression of 95 transposon families in rhi, armi, and w-1118 controls. Graphs show expression in rhi and armi plotted against expression in w-1118. The lines intercept the origin and have a slope of 1, and thus indicate equal expression in both genotypes. Significantly over-expressed transposon families are indicated by blue data points. D and E. Genome wide comparison of protein coding gene expression in rhi and armi mutants, plotted against expression in w1118. Heterochromatic genes are indicated by green data points and euchromatic genes are indicated by red data point. Both classes cluster around the diagonal, indicating similar expression levels in mutant and controls.

A. Length histogram of piRNAs expressed in wild type ovaries. B. Length histogram of piRNAs produced in rhi mutants. Sense and antisense piRNAs are reduced by approximately 80%, and peak length shifts from 25 nt to 26 nt. C. Histogram of overlapping sense and antisense piRNA in wild type ovaries, showing a pronounced peak at 10 nt, characteristic of ping-pong amplification. D. Histogram of overlapping sense and antisense piRNA in rhi mutant ovaries. The 10 nt peak is nearly eliminated, suggesting a breakdown in the ping-pong amplification cycle.

A and B. Chromosome 2R and X density profiles of uniquely mapping plus (blue) and minus (red) strand piRNAs in wild type and rhi mutants. piRNAs map to dispersed loci on the chromosome arms and prominent heterochromatic clusters. Pericentromeric piRNAs from chromosome 2R (A) and all other autosomes (Supplementary Figure 10) are dramatically reduced in rhi mutants. Pericentromeric piRNAs on the X chromosome show relatively little change. C and D. Higher resolution maps of clusters 1 and 2, which map to the indicated regions on 2R and X, respectively. Mutaitons in rhi nearly eliminated piRNAs encoded by cluster 1, which is the major dual-strand cluster, but have little impact on piRNAs from cluster 2, which is the major uni-strand cluster. Quantitative strand-specific RT-PCR for RNA derived from dual strand cluster 1 (E) and uni-strand clusters 2 and flam (F). In wild type ovaries, RNA is detected from both the plus (+) and minus (−) strands of cluster 1, at two independent locations (F cl1-A and cl1-32, blue bars). RNAs from both strands of cluster 1 are dramatically reduced in rhi mutants (red bars). Significant levels of RNA are only detected from the plus strand of cluster 2 and flam (F, blue bars), and rhi does not block expression of these RNAs (F, red bars). G. Chromatin immunprecipitation/quantitive PCR analysis of Rhino binding to cluster 1/42AB, the euchromatic genes rp49 and ry, and the herterochromatic genes pld and jing, which flank cluser 1. Rhino protein is highly enriched at cluster 1 relative to cluster 2 and the euchromatic genes. The protein coding genes flanking cluster 1 show intermediate levels of binding, suggesting that Rhino may spread to regions flanking the dual strand clusters. Anti-GFP antibodies were used to precipitate Rhino-GFP from cross-linked ovary chromatin fractions. The approximate positions of the Q-PCR primer pairs used in ChIP and RT-PCR reactions are indicated by the red bars in panels C and D. The pink bar in C indicates the approximate position of an additional primer pair used in RT-PCR.