Translational repression of the Drosophila nanos mRNA involves the RNA helicase Belle and RNA coating by Me31B and Trailer hitch

Abstract

Translational repression of maternal mRNAs is an essential regulatory mechanism during early embryonic development. Repression of the Drosophila nanos mRNA, required for the formation of the anterior-posterior body axis, depends on the protein Smaug binding to two Smaug recognition elements (SREs) in the nanos 3' UTR. In a comprehensive mass spectrometric analysis of the SRE-dependent repressor complex, we identified Smaug, Cup, Me31B, Trailer hitch, eIF4E, and PABPC, in agreement with earlier data. As a novel component, the RNA-dependent ATPase Belle (DDX3) was found, and its involvement in deadenylation and repression of nanos was confirmed in vivo. Smaug, Cup, and Belle bound stoichiometrically to the SREs, independently of RNA length. Binding of Me31B and Tral was also SRE-dependent, but their amounts were proportional to the length of the RNA and equimolar to each other. We suggest that "coating" of the RNA by a Me31B•Tral complex may be at the core of repression.

Keywords: deadenylation; maternal RNA; translational repression.

FIGURE 1.

Reporter RNAs maintain their repressed state during gradient centrifugation. (A) Cartoon of luciferase reporter RNAs. (B) Scheme of the assay. Black triangles indicate addition of embryo extract, and drop symbols indicate samples withdrawn for translation and luciferase assays. Numbers refer to the data shown in D. (C) Repressor complexes formed on radiolabeled reporter RNAs were separated by sucrose-gradient sedimentation. Distributions of the two RNAs are overlaid. UV absorption indicates the positions of free RNPs and the 80S ribosome. (D) As shown in B, luciferase RNAs were tested for translational repression either directly or after preincubation in embryo extract (samples 1 and 2 in B). A second set of samples was preincubated in extract and then separated by gradient centrifugation. Aliquots from the peak fractions as in C were assayed for translation in embryo extract either with or without a second preincubation in fresh extract (samples 3 and 4 in B). Luciferase activities in these assays are listed in Supplemental Table S1. (E) RNAs were purified from equal volumes of the peak fractions of gradients as in C, and equal aliquots were assayed for translation in rabbit reticulocyte lysate, which does not exhibit SRE-dependent repression (Jeske et al. 2006). Thus, similar luciferase yields indicated similar RNA recoveries for both RNAs. Error bars represent the standard deviation of three independent experiments. (F) Radiolabeled luciferase RNA from the sucrose gradient shown in Figure 1C was purified and analyzed by denaturing gel electrophoresis and phosphorimaging. Numbers above the lanes indicate fraction numbers of the sucrose gradient. Note that the inclusion of “short RNA” (see Materials and Methods) strongly stabilized the RNA compared to earlier experiments (Jeske et al. 2011).

FIGURE 2.

Analysis of the SRE-dependent repressor complex. (A) Radiolabeled, biotinylated RNAs (1-AUG nos and 1-AUG nos SRE⁻) were incubated for assembly of a repressor complex and separated on a sucrose gradient as in B, but the volume loaded was smaller (0.2 mL versus 1 mL). (B) Radiolabeled, biotinylated RNAs (1-AUG nos and 1-AUG nos SRE⁻) were separated on a preparative sucrose gradient (see Materials and Methods). Fractions pooled for the analysis of the repressor complex are indicated by the bracket. Error bars represent the standard deviation (n = 4). (C) Corresponding fractions from a total of 12 gradients from four independent experiments each for the SRE⁺ RNA and the SRE⁻ control were pooled, and RNPs were purified on streptavidin beads. Equal amounts based on trace-labeling of the RNA were analyzed by SDS-PAGE and silver staining. Arrowheads indicate bands enriched in the SRE⁺ RNP that might correspond to Smaug (109 kDa), Trailer Hitch (69 kDa), and Me31B (52 kDa). (D) Specific association of proteins with the SRE⁺ RNA was confirmed by Western analysis in an independent pull-down assay. Smg, Cup, Me31B, and PABPC served as controls for Bel. (E) Proteins in the purified RNP fractions were analyzed by mass spectrometry and label-free quantification. Apparent protein abundance in SRE⁺ versus SRE⁻ RNP was plotted on a log₂ scale. Proteins enriched in the SRE⁺ RNP beyond P = 0.05 and NOT11 are labeled. The complete list of proteins represented in E is found in Supplemental Table S2. Different sets of proteins in the same data are highlighted in Supplemental Figure S1, and additional enriched proteins are listed in Supplemental Table S3.

FIGURE 3.

MS analysis of proteins coprecipitated with Smg. (A) Quantitative MS data were plotted for the Smg immunoprecipitation versus a preimmune control. Proteins that were also significantly enriched in the streptavidin pull-down of the SRE-dependent repressor complex are highlighted as in Figure 2E. P-value cutoffs are indicated as lines. (B) Venn diagram comparing proteins enriched beyond P = 0.05 in the Smg immunoprecipitation and in the streptavidin pull-down (Fig. 2E). The 21 proteins in the overlap are listed. Belle, NOT10, and the CTLH complex subunit CG3295 had P-values higher than 0.05. All proteins enriched in the Smg IP are listed in Supplemental Table S4.

FIGURE 4.

SRE-containing RNAs do not oligomerize. (A) A biotinylated RNA of 200 nt (SREonly; SRE⁺ or SRE⁻) and a nonbiotinylated RNA of 630 nt (AUGonly; SRE⁺ or SRE⁻) were incubated together in embryo extract under conditions permitting assembly of the repressor complex. Streptavidin pull-downs were performed to enrich the biotinylated RNA together with potentially associated RNAs. RNA was eluted in formamide loading buffer at 95°C. The lanes labeled “RNA” show the purified RNAs used, “input” shows the RNAs after incubation in extract, “FT” is the flow-through of the pull-down, and “elution” shows the bound fraction. The figure shows one experiment of two. (B) RNA was purified from affinity-purified SRE⁺ and SRE⁻ RNPs and deep-sequenced. Reads mapping to the nos gene are displayed. For the experiment, bait RNAs were used at 10 nM. The abundance of nos has been estimated as 2 nM (Trcek et al. 2015). With an approximately twofold dilution upon extract preparation and an additional twofold dilution in the assay, endogenous nos sequences should have been detectable if an association with the bait RNA had taken place.

FIGURE 5.

The SRE-dependent repressor complex sequesters the RNA through multiple copies of Me31B and Tral. (A) Three biotinylated RNAs of different lengths but each containing two SREs were used, together with matching SRE⁻ controls, for repressor complex formation in embryo extract and streptavidin pull-down. Bound proteins were analyzed by Western blotting. Known amounts of recombinant proteins were used as standards. Analyses of Smg and Tral are shown as representative examples. (B) Stoichiometries of bound proteins were estimated from experiments as in A. Signals for SRE⁺ and mutant controls are shown. The horizontal lines mark a 1:1 molar ratio of protein to RNA. Error bars represent the standard deviation from three to five independent experiments ([*] P ≤ 0.05; [**] P ≤ 0.01; [***] P ≤ 0.001). Additional data are presented in Supplemental Figure S2. (C) An RNase I protection experiment was carried out as described in Materials and Methods. (D) Quantification of experiments as shown in C (average of n = 4 with three independent batches of embryo extract). Error bars represent the standard deviation. Data were fitted to a first-order decay with the last time point of both RNAs omitted. The half-life of the SRE⁺ RNA was 1.7-fold longer than that of the SRE⁻ control. (E) The association of Caf1 and Not2 with the SRE⁺ RNA and SRE⁻ control was examined as in A. Three streptavidin pull-down experiments were carried out with the 630 nt RNA and independent batches of embryo extract. Western blotting and comparison to standard curves was carried out for the proteins indicated. The average amount of Smg recovered was 200 ± 100 fmol. Tral was recovered at 1000 ± 150 fmol in the SRE⁺ sample and at 500 ± 120 fmol in the SRE⁻ sample (data not shown). All three subunits of the CCR4–NOT complex were present below the smallest amount in the standard curves (50 fmol). In a separate Western blot, signals for Caf1, Not2 and Ccr4 were below 12.5 fmol (data not shown).

FIGURE 6.

Me31B and Tral form a complex. (A) Structure of a complex between the C-terminal domain of DDX6 and an EDC3 peptide containing the FDF motif (PDB 2WAX) (Tritschler et al. 2009). Tral uses the same motif to bind Me31B. The black line represents the cross-link identified (Supplemental Fig. S3), with the C_α − C_α distance indicated. (B) Sf21 cells were infected with baculoviruses expressing GST-Me31B, Tral, or both as indicated. “Total” refers to an SDS lysate. Purifications on glutathione beads were carried out from native lysates. Proteins were analyzed by Western blotting for Me31B (top) or Tral (bottom). Drosophila embryo extract (DEE) and noninfected SF21 cells served as controls. (C) Glutathione bead eluates were analyzed by SDS polyacrylamide gel electrophoresis and Coomassie staining.

FIGURE 7.

Bel is required for nos mRNA translational repression in vivo. (A) Phenotypic quantification of embryos coming from bel^neo30/6 or bel^neo30/L4740 mutant females crossed with wild-type males. Numbers refer to the embryos examined. (B) nos mRNA quantification using RT-qPCR in wild-type and bel mutant embryos spanning 1 h intervals up to 4 h of development. RpL32 was used as a control mRNA for normalization. Means are from three to four biological replicates. The error bars represent SEM. (*) P < 0.05; (**) P < 0.01 using the bilateral Student's t-test. (C) PAT assays measuring nos mRNA poly(A) tail lengths in wild-type and bel mutant embryos spanning 1 h intervals up to 4 h of development. PAT assay profiles using ImageJ are shown in Supplemental Figure S4A. sop encodes a ribosomal protein and was used as a control mRNA. (D) In situ hybridization of nos mRNA (top panels) and immunostaining with anti-Nos (bottom panels) of wild-type and bel mutant 0–2 h embryos. Quantification of immunostaining is indicated below the images. (E) Western blots of wild-type and bel mutant 0–2 h embryos probed with antibodies against six components of the nos repressor complex, including the CCR4–NOT complex. Anti-α-tubulin (Tub) was used as a loading control. (F) Confocal images of syncytial embryos co-stained with rabbit anti-Bel and guinea pig anti-Smg. Bottom panels show a higher magnification. Quantification using the Pearson correlation coefficient (PCC) indicated significant partial colocalization (PCC = 0.52). Anterior is to the left. The scale bars represent 30 and 10 µm in top and bottom panels, respectively. (G) Quantification of nos and Hsp83 mRNAs using RT-qPCR in anti-GFP immunoprecipitations from bel^CC00869 embryos that express GFP-Bel and control (wild type) embryos that do not express GFP. RpL32 mRNA was used for normalization. mRNA levels in control embryos were set to one. Means are from two biological replicates quantified in triplicates. The error bars represent SEM. (***) P < 0.001 using the bilateral Student's t-test.

FIGURE 8.

Model of the SRE-dependent repressor complex. The cartoon is based on the results of this paper and the references cited in the Discussion. Note that the accuracy of our Western blots is limited, thus the stoichiometry of protein binding depicted in the figure should not be interpreted narrowly.