Drosophila melanogaster retrotransposon and inverted repeat-derived endogenous siRNAs are differentially processed in distinct cellular locations

Abstract

Background: Endogenous small interfering (esi)RNAs repress mRNA levels and retrotransposon mobility in Drosophila somatic cells by poorly understood mechanisms. 21 nucleotide esiRNAs are primarily generated from retrotransposons and two inverted repeat (hairpin) loci in Drosophila culture cells in a Dicer2 dependent manner. Additionally, proteins involved in 3' end processing, such as Symplekin, CPSF73 and CPSR100, have been recently implicated in the esiRNA pathway.

Results: Here we present evidence of overlap between two essential RNA metabolic pathways: esiRNA biogenesis and mRNA 3' end processing. We have identified a nucleus-specific interaction between the essential esiRNA cleavage enzyme Dicer2 (Dcr2) and Symplekin, a component of the core cleavage complex (CCC) required for 3' end processing of all eukaryotic mRNAs. This interaction is mediated by the N-terminal 271 amino acids of Symplekin; CCC factors CPSF73 and CPSF100 do not contact Dcr2. While Dcr2 binds the CCC, Dcr2 knockdown does not affect mRNA 3' end formation. RNAi-depletion of CCC components Symplekin and CPSF73 causes perturbations in esiRNA abundance that correlate with fluctuations in retrotransposon and hairpin esiRNA precursor levels. We also discovered that esiRNAs generated from retrotransposons and hairpins have distinct physical characteristics including a higher predominance of 22 nucleotide hairpin-derived esiRNAs and differences in 3' and 5' base preference. Additionally, retrotransposon precursors and derived esiRNAs are highly enriched in the nucleus while hairpins and hairpin derived esiRNAs are predominantly cytoplasmic similar to canonical mRNAs. RNAi-depletion of either CPSF73 or Symplekin results in nuclear retention of both hairpin and retrotransposon precursors suggesting that polyadenylation indirectly affects cellular localization of Dcr2 substrates.

Conclusions: Together, these observations support a novel mechanism in which differences in localization of esiRNA precursors impacts esiRNA biogenesis. Hairpin-derived esiRNAs are generated in the cytoplasm independent of Dcr2-Symplekin interactions, while retrotransposons are processed in the nucleus.

Keywords: CPSF73; Core cleavage complex; Dicer2; Endogenous small interfering RNA biogenesis; Symplekin; mRNA 3’ end processing.

Fig. 1

Dcr2 interacts with the N-terminal region of Symplekin a Dcr2 co-immunoprecipitates (co-IP) with the core cleavage complex (CCC) and R2D2 in Drosophila culture cell crude nuclear extract. Antibodies used for immunoprecipitation (IP) are shown above. Antibodies used for western blot (WB) are listed to the left. ‘Beads’ and ‘α-Myc’ are negative controls. No sample was loaded in lanes labeled ‘None.’ b Dcr2 binds Symplekin amino acids 1-271 and not amino acids 272-1165. Exogenously expressed HA-tagged Symplekin deletions are defined above the blots. Other labels are as in a. WB of full length Symplekin and Symplekin mutant IPs are the top figure while co-IP of Symp(1-271) with Dcr2 is shown in the bottom WB. c Dcr2 binds exogenously expressed CCC components CPSF73 and CPSF100. HA-tagged, full-length CPSF73 and CPSF100 were IPed from Drosophila culture cells stably expressing these proteins. Dcr2 co-IPs with both CPSF73 and CPSF100 was identified by western blot (top). Other labels are as in a. Dcr2 was IPed from these cells. Co-IP of HA-CPSF73 and HA-CPSF100 was identified by western blot (bottom). Controls are as in a. d CCC components CPSF100 and CPSF73 do not interact with Dcr2 in the absence of full length Symplekin. WB of Symp(1-271) IP (top) and WB of Symp (272-1165) IP (bottom) from Symplekin RNAi-depleted samples are shown. WB are labeled as in a

Fig. 2

Dcr2 is not involved in mRNA 3’ end processing a Dcr2 depletion does not affect CCC component protein levels. RNAi-depleted proteins are listed above the blot. Antibodies used for WB are listed to the left. b Dcr2 RNAi-depletion does not cause mRNA 3’ end misprocessing. An S1 nuclease assay was used to map histone (H)2A 3’ ends (left). Knockdowns are shown at the top. Potential mRNA 3’ end products are shown to the left: RT is the read-through misprocessed product, the open arrow marks the region of other misprocessed products, and the black arrow defines the properly processed product. CPSF73 is shown as a positive control. c RT-qPCR using primers that amplify misprocessed sop mRNAs (right) reveals very little misprocessed sop in Dcr2 knockdown samples. Knockdowns are shown on the x-axis. Degree of Misprocessing = Log₁₀ (2^ΔΔCt(ORF-MP)). Error bars represent one standard deviation. d Dcr2 RNAi-depletion does not cause 3’ end misprocessing and read-through transcription. RNA-seq reads mapping to and downstream of the IP3K1 gene are shown in red. Maximal read counts for IP3K1 are shown on the y-axis

Fig. 3

Dcr2 only interacts with the CCC in the nucleus. a Dcr2 is present in the nucleus. WB of refined nuclear (NE) and cytoplasmic extracts (CE) reveals a nuclear pool of Dcr2 (top). b Endogenous Dcr2 co-IPs the CCC and R2D2 from refined NEs (top). No interaction between Dcr2 and the CCC is observed in CE (bottom). Antibodies used for IP are shown above. Antibodies used for WB are listed left. ‘Beads’ and ‘α-Myc’ are negative controls. A lighter exposure of the R2D2 WB is shown at the bottom of the top NE group

Fig. 4

CCC depletion differentially affects esiRNA biogenesis from retroTns and inverted repeat loci. a Ratios of percent miRNAs, transpsoson (Tn)-derived and hairpin (hp)-derived siRNAs from Symplekin (pink), CPSF73 (red), and Dcr2 (blue) knockdown samples are shown. Percentages are the total miRNA, Tn or hp normalized read count (reads per million mapped (RPMM)) divided by the total normalized read count (summed normalized miRNA, pre-miRNA, Tn, non-coding RNA, and hp RPMs). The percents of each smRNA group were normalized to the LacZ control. Error bars represent one standard deviation. b CPSF73 and Symplekin RNAi-depleted samples are represented as in a. RPMMs of esiRNAs mapping to hp loci Esi1 and Esi2 in Symp and CPSF73 depleted samples were normalized to corresponding esiRNAs in LacZ samples (left). RPMMs of RNA-seq reads mapping to sense (S) precursors of Esi1 and Esi2 RNA in these knockdowns were also normalized to corresponding RNA-seq reads in LacZ samples (right). Error bars are as in a. c RPMMs of esiRNAs mapping to Dm297, mdg1 and jockey retroTns in Symp and CPSF73 depleted samples were normalized to corresponding esiRNAs in LacZ samples (left). RPMMs of RNA-seq reads mapping to sense (S) and antisense (AS) precursors of Dm297, mdg1, jockey transcripts in these knockdowns were also normalized to corresponding RNA-seq reads in LacZ samples (right). Error bars are as in a

Fig. 5

CCC factor RNAi-depletion may alter hairpin secondary structure a Potential secondary structures for Tns (left) and hps (right). Complementary regions are shown in green and magenta. b Depletion of CCC components does not cause 3’ end misprocessing of retroTn transcripts. S and AS RNA-seq reads from LacZ (control), Symplekin, and CPSF73 RNAi-depleted samples were visualized using the UCSC genome browser and are overlayed. Schematic of a genomic region containing a mdg1 element is shown above the bedgraph. No unique reads are observed flanking the retroTn. c CG6903 read-through transcripts could hybridize to Esi2 RNAs in CCC factor knockdowns. Symplekin (blue), CPSF73 (green) depleted samples and LacZ (red) control RNA mapping to this region are shown. RNA-seq data is displayed in the top two panels, siRNA-seq data is displayed in the bottom panel

Fig. 6

RetroTn- and hp-derived esiRNAs have different physical characteristics. a Percentage of miRNAs (gray), Transposons (light purple), and hairpin (purple) mapping esiRNAs in the LacZ control sample that are 19-23 nts. Error bars represent one standard deviation. b Percentage of miRNAs, Transposons, and hairpin mapping esiRNAs in the LacZ control sample that have a 3’ G, A, T, or C. Colors and error bars are as in a. c Percentage of miRNAs, Transposons, and hairpin mapping esiRNAs in the LacZ control sample that have a 5’ G, A, T, or C. Colors and error bars are as in a

Fig. 7

RetroTns dsRNA precursors are nuclear retained. RT-qPCR of retroTn and Esi1/2 mRNAs isolated from refined nuclear and cytoplasmic fractions reveals nuclear retention of retroTn esiRNAs precursors. (a, left) RetroTn RT-qPCR targets are shown on the x-axis. Fold change is the average of three experiments and is calculated as 2^{(Ct(Nuclear)-Ct(Cytoplasm))}. (a, Right) CG44774 is the Esi1 precursor. CR18854 is the Esi2 precursor mRNA. GAPDH is the control transcript. Error bars represent one standard deviation. b Taqman qPCR of retroTn and Esi1/2 derived esiRNAs isolated from refined nuclear and cytoplasmic fractions shows nuclear retention of retroTn derived esiRNAs. Labels, calculations, and error bars are as in a. c RT-qPCR of esiRNA precursors in Symplekin, CPSF73 or Dcr2 knockdowns shows nuclear retention of both hp and retroTn dsRNA precursors. Labels, calculations, and error bars are as in a. RNAi-depletion is defined as in Fig. 4

Fig. 8

EsiRNAs are differnentially processed in D. melanogaster cells. Data support a model in which double stranded retroTn transcripts are retained and processed to esiRNAs in the nucleus while RNAs containing inverted repeats are exported and processed in the cytoplasm. Dcr2 interacts with the N-terminal 271 amino acids of Symplekin in the nucleus, but not in the cytoplasm