Cation-exchange chromatography enriches for Argonaute- and Piwi-enclosed RNAs. (A) Phase contrast image of OSS cells illustrating their sheet-like morphology; under high density they also tend to form clumps. (B) pCp labeling of RNAs from total OSS RNA and various elutions from a HiTrap Q column. The mobility of the RNAs in the extract and elutions is slightly different from that of the synthetic markers owing to different salt concentrations of the loaded samples. Highly abundant 2S rRNA is visible in the input and ran slightly larger than its 30-nt sequence. In the flowthrough, ∼21–24-nt RNAs are visible, inferred to include miRNAs and siRNAs. In the 0.3 M salt elution, ∼24–28-nt RNAs are visible, inferred to represent the piRNA fraction. A heterogenous ladder of RNA fragments elutes with 1.0 M salt. Libraries were constructed from the flowthrough and 0.3 M salt elution.

TE-piRNAs and TE-siRNAs in OSS cells. (A) Reproducibility of overall read distribution in the four OSS sequencing reactions. Analysis of raw reads is shown; the normalized data are completely overlapping (Supplemental Fig. S1). (B) Combined OSS library data. Blue depicts the size distribution across all reads, orange depicts miRNA reads, and red depicts reads that mapped to transposons; the latter clearly segment into a siRNA population (peaking at 21 nt) and a piRNA population (∼24–30 nt). (C–G) Combined OSS read distribution mapped to various families of TEs (as annotated by RepeatMasker). (Red) Antisense reads; (blue) sense reads. Two graphs are shown for each TE family, depicting the size distribution of reads that map uniquely to the genome and the distribution of reads that map to the genome 10 or more times. Various patterns are observed across the TEs with respect to piRNAs vs. siRNAs, or the fraction of unique vs. multiply-matching reads. The complete analysis of all read bins across all TE families is available in Supplemental Figure S2.

OSS cells are related to ovarian follicle cells. (A) Schematic of an individual mature germarium. The germline consists of 15 nurse cells and the developing oocyte; the latter is ensheathed by follicle cells of somatic origin. The accumulation of the three Piwi-class proteins—PIWI, AUB, and AGO3—in these celltypes is indicated. (B) Quantitative RT–PCR analysis of Piwi-class transcripts in OSS cells; ovaries were used as a positive control and female heads as a negative control. Transcript levels were normalized to RpL32 as a control, and expressed as the fold change above the level in heads. OSS cells express high levels of piwi, but not aub or AGO3. (C–E) Immunostaining of OSS cells with antibodies against the three Piwi-class proteins (green) verifies sole expression of PIWI; DNA was counterstained with DAPI (blue).

miRNA expression in OSS cells. (A) OSS cells predominantly express a characteristic population of miRNAs (left) that overlaps only partially with the most abundant miRNAs annotated from across Drosophila development using 454 Life Sciences (Roche) sequencing (Ruby et al. 2007a). However, analysis of all reads indicates that the 454-annotated miRNAs and OSS miRNAs are very highly overlapping (right). Therefore, numerous ostensibly tissue-specific Drosophila miRNAs were captured at low levels in OSS cells by deep sequencing. (B) Five novel miRNA loci, expressed at a level of more than five reads/14 million library reads, were identified in OSS cells. Mature products were highlighted in green, and star strands in red; for some hairpins, the two small RNA products were cloned nearly equivalently.

Sequence properties of TE-siRNAs and TE-piRNAs in OSS cells. (A–D) Aggregate nucleotide composition of all TE-derived reads of length 21 nt (siRNA) or 27 nt (as a representative piRNA size); in all graphs, the x-axis represents nucleotide position along the piRNA, while the y-axis represents the percent nucleotide composition and each position. 5′ U bias is observed for bulk AS-TE-siRNAs as well as S- and AS-TE-piRNAs, but is lacking in bulk S-TE-siRNAs. Note also that S-TE-piRNAs lack any bias for adenine at position 10 (above the adenine bias of neighboring nucleotide positions), as would be expected for ping-pong pairs. Analysis of other TE read sizes are available in Supplemental Figure S3. (E–H) Specific analysis of roo reads shows similar trends. Analysis of all other individual TEs are available in Supplemental Figure S4. (I) Overlap analysis of piRNAs with three hypothetical configurations of distance between 5′ ends of a piRNA and its nearest neighbor on the opposite strand; +10 offset is typical of piRNA ping-pong, while −16 offset is consistent with a phased arrangement of ping-pong pairs (gray arrows depict “missing” piRNAs). (J) Analysis of ovary (black) and 0–2-h embryo (blue) piRNAs reveals strong ping-pong (+10 offset pairs) and modest evidence for a phased ping-pong pairs. No ping-pong is observed for OSS piRNAs (red).

PIWI protein localization in OSS cells appears nucleoplasmic and not specific for any chromatin state. Triple staining of PIWI and DAPI with SU(VAR)205 (also known as HP1a) (A), H3K9me3 (B), H3K27me3 (C), H3K4me3 (D), and Polycomb (PC) (E). Arrowheads mark the location of the chromocenter, a DAPI-dense congregation of heterochromatin. PIWI is specifically absent from the chromocenter and does not overlap appreciably with either markers of silent or active chromatin. Some PIWI foci are near Polycomb foci, but these appear to be incidental and are never overlapping (arrows, E).