A transcriptome-wide RNAi screen in the Drosophila ovary reveals factors of the germline piRNA pathway

Abstract

The Drosophila piRNA pathway provides an RNA-based immune system that defends the germline genome against selfish genetic elements. Two interrelated branches of the piRNA system exist: somatic cells that support oogenesis only employ Piwi, whereas germ cells utilize a more elaborate pathway centered on the three gonad-specific Argonaute proteins (Piwi, Aubergine, and Argonaute 3). While several key factors of each branch have been identified, our current knowledge is insufficient to explain the complex workings of the piRNA machinery. Here, we report a reverse genetic screen spanning the ovarian transcriptome in an attempt to uncover the full repertoire of genes required for piRNA-mediated transposon silencing in the female germline. Our screen reveals key factors of piRNA-mediated transposon silencing, including the piRNA biogenesis factors CG2183 (GASZ) and Deadlock. Our data uncover a previously unanticipated set of factors preferentially required for repression of different transposon types.

Figure 1. Screen workflow and summary of preliminary experiments

(A) Relative expression levels of protein-coding genes in Drosophila melanogaster are shown for ovarian RNAseq data as histogram. Green bars highlight ovary-expressed genes with FPKM > 1. Doughnut diagram shows screened genes where dsRNA line was available from the VDRC. (B) Histograms show the relative expression levels of indicated transposons detected in ovaries that express dsRNA against yb, armi or spn-E in germline cells. Fold changes are relative to knockdown of w (indicated by red line). Measurements were carried out on ovary-dissected total RNA. Error bars indicate standard deviation (n=3). (C) Relative expression levels of the indicated mobile elements upon germline-specific knockdown of armi and spn-E are shown. Depletion of Yb served as control. Fold changes relative to dsRNA against w (indicated by red line) were calculated. Measurements were carried out on RNA extracted from whole female flies. Error bars indicate standard deviation (n=3). (D) Scheme of the screen setup. A germline-specific driver, nos-GAL4, was used to express UAS-dsRNA constructs in germ cells of the developing oocyte. UAS-Dcr2 was co-expressed specifically in germ cells to enhance the RNAi response. 2.5 day old female offspring flies were collected and following RNA isolation and reverse transcription probed for de-repression of four transposons by qPCR. Crosses were carried out in trays of 96 that contained a positive (armi) and negative (w) control knockdown.

Figure 2. Summary of primary screen and determination of candidate hits

(A) Heat map displaying transposon de-repression (as z-scores) for all 8,171 investigated ovary-expressed genes in red-blue scale. The average of the four tested transposons is shown along the separate z-scores of HeTA, TAHRE, blood, and burdock. Negative z-scores indicate transposon de-repression (shown in red). (B) Close-up of the heat map for 216 candidate hits with z-scores < −1.5. (C) Box plots summarizing z-scores of all screened genes, positive armi controls, and negative w controls. Average data from four transposons was used for the analysis.

Figure 3. Identification and validation of strong candidate genes

Heat maps summarizing transposon de-repression, germline marker gene expression, and sterility phenotypes upon germline-specific knockdown of indicated genes. Data is presented relative to depletion of Armi. Yellow boxes highlight known piRNA pathway components.

Figure 4. Subcellular localization phenotypes upon depletion of the piRNA pathway candidate factors GASZ and Del

(A) Knockdown of armi and gasz in the germline using nos-GAL4 causes Piwi delocalization from nuclei and redistribution of Armi from nuage-like sites. The localization of Aub and AGO3 is not changed. Knockdown of w is shown as control. (B) Loss-of-function allelic combination del³/del^KG severely impairs the localization of secondary piRNA pathway components Aub and Ago3, while Piwi and Armi are unaffected.

Figure 5. Germline knockdown of gasz or del affects different steps of piRNA biogenesis

(A) Size distribution of 18- to 29-nt small RNAs derived from each strand of the flam and 42AB clusters are shown as histogram (blue sense; red antisense). The fraction of miRNAs (green) for the indicated libraries is highlighted in the pie charts. (B) Histograms showing the relative enrichment of piRNAs overlapping by the indicated number of nucleotides are plotted for 42AB-derived sequences in the indicated knockdowns. A peak at position 9 (arrow, the number corresponds to the z-score) is suggestive of a ping-pong signature. (C) A histogram showing relative piRNA levels of a series of soma- and germline-dominant clusters. Total reads were normalized across libraries to flam-unique piRNAs, which are unaffected by nos-GAL4-specific knockdowns. Changes in piRNA levels are shown with reference to depletion of w (set to 100%). (D) Histograms showing the abundance of piRNAs mapping to soma dominant (green), intermediate (grey), or germline dominant (orange) transposons in w knockdowns compared to depletion of the indicated genes (log₂ scale). (E) Histograms of piRNAs mapping to the consensus sequence of the germline dominant batumi LTR transposon, are shown for the indicated knockdowns.

Figure 6. Specific requirements for silencing of different transposon types

(A) Scatter plot comparing transposon de-repression (as z-scores) for HeTA and TAHRE (top 500 candidates from the primary screen are shown in red, median 500 candidates are indicated in blue). Known piRNA pathway components are highlighted in green. (B) Similar as in (A), but HeTA de-repression is compared to transposon up-regulation of the LTR element burdock. (C) Heat map probing transposon de-silencing for known and newly identified piRNA pathway components. U2A and Ars2 represent factors required for silencing of a subset of transposable elements only. Transposon de-silencing is displayed as normalized read count mapping to the canonical sequence for each element relative to a baseline calculated from the average of five independent negative controls (log₂ scale).

Figure 7. Protein complexes implicated in silencing of germline transposons

Primary screen data is summarized for six known Drosophila macromolecular complexes: the nuclear cap binding complex (CBC), the THO complex, the EJC, the NSL complex, components of the SUMOylation machinery, and the H3K4 methyltransferase dSet1/COMPASS complex. Color-coding of boxes indicates the transposon z-scores measured for each gene as exemplified in the key at upper right (for armi and w knockdowns).