Trafficking of siRNA precursors by the dsRBD protein Blanks in Drosophila

Abstract

RNA interference targets aberrant transcripts with cognate small interfering RNAs, which derive from double-stranded RNA precursors. Several functional screens have identified Drosophila blanks/lump (CG10630) as a facilitator of RNAi, yet its molecular function has remained unknown. The protein carries two dsRNA binding domains (dsRBD) and blanks mutant males have a spermatogenesis defect. We demonstrate that blanks selectively boosts RNAi triggered by dsRNA of nuclear origin. Blanks binds dsRNA via its second dsRBD in vitro, shuttles between nucleus and cytoplasm and the abundance of siRNAs arising at many sites of convergent transcription is reduced in blanks mutants. Since features of nascent RNAs - such as introns and transcription beyond the polyA site - contribute to the small RNA pool, we propose that Blanks binds dsRNA formed by cognate nascent RNAs in the nucleus and fosters its export to the cytoplasm for dicing. We refer to the resulting small RNAs as blanks exported siRNAs (bepsiRNAs). While bepsiRNAs were fully dependent on RNA binding to the second dsRBD of blanks in transgenic flies, male fertility was not. This is consistent with a previous report that linked fertility to the first dsRBD of Blanks. The role of blanks in spermatogenesis appears thus unrelated to its role in dsRNA export.

Figure 1.

Blanks is not required for cytoplasmic RNAi. (A) Cartoon drawing of the modified Flag-blanks locus (top); transcription of the gene is now controlled by the copper-inducible mtn promoter contained in our marker/tag cassette. If all alleles of the targeted gene are modified, then gene activity can be fully controlled by the addition or omission of copper in the culture medium (bottom, normalized to blanks levels in non-edited cells). (B) We used the inducible Flag-blanks cell line to test the efficiency of cytoplasmic dsRNA-triggered RNA interference as a function of the presence or absence of blanks. We chose the cas9 gene as a target because it is constitutively expressed in these cells as well as their progenitors and there should be no secondary effects resulting from the loss of this heterologous gene. A non-specific dsRNA targeting Renilla luciferase was used as a control (RLuc). (C) Quantification of the Western Blots from B; the Cas9 protein signal was normalized to the tubulin loading control, then the ratio between specific and control-knockdown was calculated. The graph shows the average of three independent biological replicates ± Standard Deviation (SD), n.s.: not significantly different (Student's t-test).

Figure 2.

Blanks and Dcr-2 do not form a stable complex. (A) Cartoon drawing of the Blanks-Flag locus; in this case, the tag is inserted at the C-terminus, hence transcription occurs at endogenous levels driven by the gene's natural promoter. (B) Co-immunoprecipitation experiments from whole cell lysates do not indicate a stable complex between Blanks and Dcr-2. We also generated knock-in cell lines expressing Flag-tagged versions of R2D2 and Actin5C as positive and negative controls, then prepared whole cell lysates and immunoprecipitated the Flag-tagged proteins. Association of Dcr-2 was detected via a monoclonal anti-Dcr-2 antibody (Blanks-Flag and Actin5C-Flag samples, mAB 8–59, a kind gift of M. Siomi) or via an anti-Strep tag antibody (the R2D2 cells also harbor a Dcr-2-Strep-tag knock-in). While the co-immunopreciptiation of Dcr-2 was readily detectable with R2D2 (filled arrowheads), no association was seen with Blanks or Actin5C (open arrowheads).

Figure 3.

blanks dependent siRNAs are generated at sites of convergent transcription. Small RNAs were isolated from S2-cells with copper-inducible blanks and Dcr-2 alleles (see Figure 1 for a detailed example), sequenced, filtered for transposon-matching reads and the respective read counts are depicted as parts (reads) per million (normalized to the total number of genome matching reads in each library). The term ‘induced’ refers to culturing in the presence of 200 μM CuSO₄ while ‘uninduced’ refers to culture in the absence of CuSO₄ (also referred to as ‘shut-down’ in the manuscript). (A) The majority of convergent gene loci give rise to a moderate amount of dcr-2 dependent siRNAs (4366 loci analyzed). (B) A subset of these loci also requires blanks (see also Supplementary Table S2). (C) Re-expression of blanks by adding copper to the culture medium of our Flag-Blanks cells (‘induced’) recovered the expression of the small RNAs. In essence, the blanks-dependent siRNA loci are the grey points that remain visible below the diagonal that is populated by the wt (black) and induced blanks (red) samples. (D) The salient features of a blanks-dependent siRNA generating locus are illustrated by this genome browser image. The coverage tracks in the upper part were adjusted to differences in sequencing depth and thus faithfully convey different siRNA levels, the corresponding libraries are indicated on the left. We also present the annotation of individual siRNAs for the wild-type sample below the coverage with color-coded orientation (pink: 5′→3′ left to right, blue: 5′→3′ right to left). The gene annotation on the bottom shows two pairs of convergent genes. (E) The amount of transposon-targeting endogenous RNAs depends on dcr-2 but not on blanks. (F)To determine whether the remaining blanks-dependent siRNAs (bepsiRNAs) are mis-loaded into Ago1 we sequenced oxidized small RNA libraries. While miRNAs do not carry the protective 2′-O-methyl modification at the 3′-end that is characteristic for Ago2-loading, the transposon-targeting and the remaining bepsiRNAs are protected and hence correctly loaded.

Figure 4.

Blanks shuttles between nucleus and cytoplasm. (A) Fluorescence microscopy images of cells with Blanks-GFP and H2Av-GFP knock-ins. An example of the control (DMSO) and the Importazol-treated cells is shown for each. The DNA was stained with Hoechst33342. (B) Quantification of protein localization after treatment with the nuclear import inhibitor Importazol for 16 h. Cells were classified into either showing a predominantly nuclear or a whole-cell staining (‘overall’). The slides were prepared, anonymized and then analyzed by visual inspection using a fluorescence microscope. While the distribution of H2Av-GFP did not change during the course of Importazol treatment, the localization of Blanks changed from predominantly nuclear to mostly a whole-cell-staining.

Figure 5.

Fertility of male flies is influenced by localization and RNA-binding ability of Blanks. The bars indicate the average number of pupae, error bars represent the standard deviation (n = 5). (A) Flowchart of the fertility assay. (B) Ubiquitous expression of our blanks-transgenes has no dominant-negative effect on male fertility in a wild-type background. (C) The blanks-transgenes can rescue fertility; however, an increased nuclear retention via the appended NLS is detrimental if the blanks protein is competent for RNA binding. For the no rescue condition, zero pupae developed in all replicates (none detected, n.d.).

Figure 6.

The effect of wild-type and mutant blanks transgenes on the small RNA profile in blanks^MI10901 mutant testes. (A) The relative proportion of miRNA-matching reads in among the small RNAs is not sensitive to the presence or absence of Blanks. However, expression of an RNA-binding proficient Blanks protein with an additional nuclear localization signal (NLS) reduced the abundance of miRNAs. (B) This reduction of miRNA matching reads is due to a generalized but moderate effect on all miRNAs (blue dots), rather than selectively affecting, e.g. only the highly expressed species. (C) If the Blanks transgene harbors an inactivating point mutation in the second dsRBD in addition to the NLS, the inhibition of miRNA biogenesis is not visible.

Figure 7.

Analysis of Blanks exported siRNAs (bepsiRNAs) in fly testes. (A) Unbiased analysis of 21-mer siRNA reads mapping to all annotated transcripts extended by 150 nt on each end. Depicted are the sequence reads (normalized to total genome matching) for blanks^MI10901 mutant animals and mutants rescued with a wild-type blanks protein (referred to as wt). A substantial number of loci generates blanks-dependent siRNAs (below the diagonal); the small group of genes with increased expression (above the diagonal) are members of the 825-oak gene family, a known source of hairpin RNAs. (B) We calculated the change between wt and blanks^MI10901 mutants for each annotated gene and then plotted this relative to the fraction of reads that map in sense orientation at the respective locus. Most blanks-dependent siRNA loci generate siRNAs with sense and antisense orientation in roughly equal amounts, indicating that they derive from a dsRNA precursor (central region, fold change >>1). Consistent with the S2-cell data in Figure 3, not all dsRNA-derived siRNAs are blanks-dependent (central region, fold change ∼1). In addition, several structured non-coding RNAs (tRNA and rRNA) give rise to blanks-dependent siRNAs, which match exclusively in sense-orientation. In this plot, the fraction of sense reads was calculated based on the read-counts of the wild-type library. The supplement contains an overlay with a calculation of the fraction of sense reads from the blanks^MI10901 library. (C–E) Genome browser views for CG8176 and by (C), Mitf and Dyrk3 (D), CG10508 and CG12975 (E). The coverage traces were adjusted to account for the differences in sequencing depth.

Figure 8.

We propose that Blanks serves as a link between nuclear dsRNA and the export machinery. While the nuclear export of miRNA precursors has been characterized (left), export of dsRNA to the cytoplasm for processing by Dcr-2 is less well understood. We interpret our results that one export route for endo-siRNA precursors is through association with Blanks, a protein that binds to dsRNA with its second double-stranded RNA binding domain (dsRBD) and shuttles between nucleus and cytoplasm (right). The dsRBDs of Loqs, R2D2 and Blanks are indicated as spheres and numbered according to their position within the protein.