MicroRNA-repressed mRNAs contain 40S but not 60S components

miRNA-Targeted mRNAs Contain Reduced Amounts of 60S Ribosome Components.

To investigate the mechanism of repressed translation initiation, we tested the ability of the 40S and 60S ribosome subunits to physically associate with miRNA-targeted mRNAs in the cell-free reactions described first in Wang et al. (6). To enable recovery of miRNA-targeted mRNAs, firefly luciferase reporter mRNAs were subjected to polyadenylation such that, on average, two biotinylated adenosines were incorporated into a polyA tail consisting of ≈200 adenosines. These reporter mRNAs contained six imperfectly complementary binding sites to the CXCR4 siRNA (FL6X). Consistent with previous observations (6), the FL6X reporter mRNA demonstrated translation repression upon addition of CXCR4 siRNAs when normalized to an untargeted renilla luciferase reporter mRNA lacking CXCR4 siRNA-binding sites (RL0X, Fig. 1A).

Upon completion of the translation repression reactions, reporter mRNAs were precipitated with streptavidin beads, and precipitates were subjected to Northern blot analysis for ribosomal RNAs (Fig. 1B). In all analyses in Fig. 1B, a fixed amount of RNA was loaded in each lane, and Northern blot analysis signals were normalized by calculating the ratio of 60S rRNA:40S rRNA and tRNAi-Met:40S rRNA. Northern blot analysis probes directed against 18S rRNA, a component of the 40S ribosome subunit, detected negligible differences in 40S content between reactions containing or lacking siRNAs. However, Northern blot analysis probes directed against 28S, 5.8S, and 5S rRNAs of the 60S ribosome subunit detected significant reductions in 60S ribosome components relative to the 18S rRNA in reactions containing mRNAs targeted by miRNAs (Fig. 1B). Additionally, the reduction in 28S, 5.8S, and 5S rRNAs associated with FL6X was approximately the same as the degree of translational repression observed in Fig. 1A (≈60%). These results indicate that miRNAs promote reduced 60S ribosome subunit loading on target mRNAs. Conversely, Northern blot analysis probes directed against tRNAi-Met detected no change relative to the 18S rRNA in reactions containing mRNAs targeted by miRNA compared with untargeted mRNAs. This result indicates that miRNAs permit 43S ribosome subunit loading on target mRNAs.

Several control reactions confirm the specificity of reduced 60S ribosome recruitment to miRNA-targeted mRNAs: (i) There was no significant change in the 40S or 60S ribosomes in Northern blots of total lysates without precipitation of miRNA-targeted mRNAs from translation repression reactions, suggesting that the ribosome subunits are not being lost because of degradation. Relative to reactions with siRNAs, there was no change in the amounts of 60S ribosome subunits between reactions using (ii) FL6X mRNAs with unphosphorylated CXCR4 siRNAs (Fig. 1B Middle), (iii) FL6X mRNAs containing point mutations in the 5′ seed region of the miRNA-binding sites in the 3′ UTR (Fig. 1B Right), (iv) FL6X mRNAs with nonspecific control siRNAs [supporting information (SI) Fig. S1C], and (v) FL0X mRNAs with CXCR4 siRNAs (Fig. S1C). The absence of significant changes in 40S and 60S ribosomal RNA content in these control reactions is consistent with the absence of translational repression in these control reactions (Fig. 1 A and Fig. S1B) and indicates that the reduction in 60S ribosome RNAs is specific to miRNA-repressed mRNAs.

To independently confirm these observations, precipitated reporter mRNAs were subjected to Western blot analysis for 40S- and 60S-associated proteins (Fig. 1B). Whereas slightly increased amounts of 40S-associated proteins were detected on miRNA-repressed mRNAs, strongly decreased amounts of 60S-associated proteins were detected on these same mRNAs. Together, these observations demonstrate that miRNA-targeted mRNAs have steady-state levels of 40S ribosome components but reduced levels of 60S ribosome components relative to untargeted mRNAs.

Chemical Inhibitors Identify High Molecular Mass Complex Formation on miRNA-Targeted mRNAs That Depends on 40S but Not 60S Ribosomes.

To define the stage of translation initiation affected by miRNAs more precisely, we used ribosome-binding assays to analyze the sedimentation profiles of radiolabeled miRNA-targeted mRNAs from translation repression reactions (Fig. 2). Similar to all reports of polysome profiling of miRNA targeted mRNAs (2–5, 15, 16), ribosome-binding assays require chemicals to stabilize intermediates of monosome (80S) assembly. In ribosome-binding assays, addition of CXCR4 siRNAs to translation repression reactions without chemical inhibitors is not sufficient to capture complexes at specific stages of ribosome assembly during translational repression (data not shown). Complexes stabilized at specific stages of ribosome assembly on radiolabeled mRNAs are sedimented through a glycerol gradient and detected by Cerenkov scintillation counting of individual glycerol gradient fractions (reviewed in ref. 17).

First, translation repression reactions were performed in the presence of the eIF4A inhibitor hippuristanol (18). Hippuristanol blocks 43S recruitment to mRNAs, resulting in the majority of the mRNA migrating at the top of the glycerol gradient as unbound mRNA (fraction 4) in ribosome-binding assays. Addition of CXCR4 siRNA to translation repression reactions did not alter the sedimentation of mRNAs in the presence of hippuristanol (Fig. 2A), indicating that any complexes that form on repressed mRNAs require 43S recruitment. Next, translation repression reactions were performed in the presence of the nonhydrolyzable GTP analog GMP-PNP, which blocks 60S ribosome recruitment, resulting in the capture of 48S initiation complexes (fraction 8) in ribosome-binding assays. When CXCR4 siRNAs were added to translation repression reactions containing GMP-PNP, the mRNA sedimentation peak was shifted from fraction 8 to fraction 14, indicating the formation of a high molecular mass complex (Fig. 2B). These results suggest that this high molecular mass complex is formed on miRNA-targeted mRNA after 43S recruitment but before 60S recruitment.

Then, translation repression reactions were performed in the presence of the translation elongation inhibitor cycloheximide, which traps fully assembled 80S monosomes at the initiation codon of mRNAs (fraction 12). Reactions without CXCR4 siRNA led to the expected mRNA sedimentation, consistent with captured 80S complexes (Fig. 2 C–F). Addition of CXCR4 siRNAs to translation repression reactions containing cycloheximide, however, generated an mRNA sedimentation profile identical to the profile observed with GMP-PNP (Fig. 2C). This observation indicates that 80S monosomes do not form in these reactions, and that the repression occurs at an earlier step of translation initiation.

Formation of the high molecular mass complex was specific to the CXCR4 siRNA and its interaction with its target site. Inclusion of unphosphorylated CXCR4 siRNAs (Fig. 2D), a triple point mutant in the 5′ seed region of the miRNA-binding site in the FL6X mRNA 3′ UTR (Fig. 2E), nonspecific control siRNA (Fig. S2C) or fully phosphorylated CXCR4 siRNAs and the FL0X mRNA (Fig. S2D) all resulted in the formation of an 80S complex in the presence of cycloheximide and had no effect in translation repression reactions (Fig. 2G and Fig. S1B). Consistent with the lack of translational repression of uncapped mRNAs (ref. 6; Fig. S3), FL6X lacking a 7-methyl guanosine cap did not result in formation of the high molecular mass complex in reactions containing CXCR4 siRNAs but instead resulted in formation of an 80S complex (Fig. 2F). These results demonstrate that the high molecular mass complex forms only on translationally repressed mRNAs and possesses the 40S ribosome subunit but lacks the 60S ribosome subunit, further supporting a model that miRNAs repress translation by preventing 60S ribosome subunit recruitment to target mRNAs. Additionally, the ability of the eIF4A inhibitor, hippuristanol, to prevent formation of the high molecular mass complex suggests that miRNA-directed repression of translation occurs after cap-facilitated, 40S ribosome subunit recruitment.

To rule out any effect of cycloheximide in the reactions in Fig. 2, we performed Northern blot analysis on precipitated complexes from translation repression reactions containing cycloheximide (Fig. 2H). Consistent with data presented in Fig. 1B, Northern blot analysis of precipitated FL6X demonstrated similar levels of 18S ribosomal RNA but reduced 28S, 5.8S, and 5S ribosomal RNAs (but no change in tRNAi-met) associated with target mRNAs in reactions with CXCR4 siRNA compared with reactions without CXCR4 siRNAs. Also consistent with data presented in Fig. 1B, 60S ribosome association was reduced to a similar degree (≈70%) as translational repression observed in Figs. 1A and 2G.

48S Complexes Are Positioned at AUG on miRNA-Repressed miRNAs.

To identify the position of 40S ribosomes assembled on miRNA-repressed mRNAs, primer extension analysis was performed (Fig. 3A). In this assay, a radiolabeled primer hybridizing to sequences 3′ relative to the AUG codon of miRNA-targeted mRNAs was used to initiate reverse transcription without extraction from associated proteins. A “toeprint” of bound protein complexes is generated when steric hindrance prevents reverse transcriptase from transcribing cDNA from regions of the mRNA. Translation repression reactions with CXCR4 siRNAs generated bands at 18 nt 3′ relative to the AUG codon (compare lanes 5 and 6). This toeprint was identical to translation repression reactions containing GMP-PNP (lanes 10–12), which marks 40S ribosomes positioned at the start codon after completion of scanning. Consistent with the ability of hippuristanol to block formation of the high molecular mass complex in ribosome-binding assays (Fig. 2A), the CXCR4 siRNA-induced toeprint was blocked in the presence of hippuristanol (compare lanes 6 and 9).

Toeprinting was quantified by using the ratio of the 3′ (specific) band protected in toeprinting relative to the 5′ (nonspecific) band relative to AUG. RNA secondary structure causes MMLV reverse transcriptase to dropoff of its template, thus generating a nonspecific 5′ band that can be used to quantify specific miRNA toeprint formation. By this measure, the CXCR4 siRNA-induced toeprint ratio was 2.3 (lane 6), the GMP-PNP-induced toeprint ratio was 6.9 (lanes 10 and 11), and the combined ratio was 7.1 (lane 12), 3-fold more than the toeprint ratio induced by CXCR4 siRNA alone. These data indicate that GMP-PNP more strongly stabilizes 40S complexes positioned at AUG compared with miRNAs alone and may help explain why the addition of miRNAs alone is not sufficient to capture complexes in ribosome-binding assays.

eIF2 and eIF3 Are Associated with miRNA-Targeted mRNAs.

To investigate the complement of translation initiation factors associated with mRNAs in fraction 4 (free mRNA), fraction 8 (48S complexes), fraction 12 (80S complexes), and fraction 14 (high molecular mass complex), mRNA precipitates from glycerol gradient fractions in ribosome-binding assays were subjected to Western blot analysis. Because many translation initiation and miRNA-interacting factors are highly conserved through evolution, antibodies against these human and mouse proteins crossreact with their rabbit homologs.

The translation initiation factors eIF2 and eIF3 are recruited to 43S ribosome complexes before joining mRNAs and dissociate from 48S ribosome complexes just before (or concomitant with) 60S ribosome subunit joining mRNAs (reviewed in ref. 19). Therefore, we probed ribosome-binding assay fractions for these factors known to assemble with the 40S subunit during translation initiation (Fig. 3B). The eIF2 subunit, eIF2α, and the eIF3 subunit, eIF3g, were significantly enriched in fractions 12 and 14 from reactions with CXCR4 siRNAs compared with reactions without CXCR4 siRNAs. Together with toeprinting analysis, Western blot analyses support a model in which miRNAs block translation after 43S subunit joining and scanning but before eIF2 and eIF3 release and 60S ribosome subunit joining.

To interrogate the cap dependency of miRNA-directed translational repression and high molecular mass complex formation, we performed Western blot analysis of glycerol gradient fractions from ribosome-binding assays for the cap-binding protein (eIF4E) and the RNA helicase that facilitates 40S recruitment (eIF4A). Increases in both of these factors were observed in reactions containing CXCR4 siRNAs relative to those lacking CXCR4 siRNAs (compare fractions 12 and 14 without CXCR4 siRNAs vs. with CXCR4 siRNAs). Together, these observations indicate that eIF4E and eIF4A are still bound to translationally repressed mRNAs after 40S subunit joining and suggest that interaction of these proteins with the cap is important for translational repression by miRNAs.

In all species capable of small RNA-directed gene silencing, microribonucleoprotein (miRNP) complexes possess a member of the Ago family of proteins (reviewed in ref. 20). To determine whether Ago proteins are recruited to miRNA-targeted mRNAs in the reactions reported here, Western blot analysis was performed with antibodies against Ago2 (Fig. 3B). Consistent with the notion that the high molecular mass complex formed on miRNA-repressed mRNAs is a bona fide miRNP, fractions 12 and 14 were significantly enriched for Ago2 in reactions containing CXCR4 siRNAs relative to reactions lacking CXCR4 siRNAs. To demonstrate that Ago proteins recruited to miRNAs preannealed to mRNAs are functional, Ago2-dependent RNA cleavage assays were performed. Our data indicate that Ago2-mediated cleavage of target RNAs in vitro maps to the exact position as reported for Ago2-dependent cleavage in cells (Fig. S4 and SI Materials and Methods). These data indicate that these reaction conditions permit formation of functional miRNP/RISC on miRNA-repressed mRNAs.

Translation Repression Reactions.

All mRNA reporters used in these studies were prepared and used as described in ref. 6. Plasmids expressing all of these mRNA reporters are available from (www.addgene.org). The sequences of the CXCR4 siRNA were 5′P-GUUUUCACUCCAGCUAACACA-3 (sense strand) and 5′P-UGUUAGCUGGAGUGAAAACUU-3′ (antisense strand). The sequences of the GFP siRNA were 5′P-GGCUACGUCCAGGAGCGCACC-3′ (sense strand) and 5′P-UGCGCUCCUGGACGUAGCCUU-3′ (antisense strand). The control mRNA reporter used in Supporting Online Fig. 3 was human CD3 (kind gift of Chenqi Xu, Dana-Farber Cancer Institute).

Translation repression reactions were performed as described in ref. 6. Briefly, preannealed mRNA reporter (0.025 pmol) and CXCR4 siRNA (0.15 pmol) were incubated with a master mix containing 7 μl of nuclease-treated rabbit reticulocyte lysate (RRL, Promega), 4–8 units RNase Out (Invitrogen), 20 μM amino acid mixture (complete or minus methionine and cysteine, Promega), and 0.4 μl (5.7 μCi) Promix l-[³⁵S] in vitro cell labeling mix (Amersham Biosciences) at 30°C for 10 min. Reaction products were separated on 12% SDS/PAGE and transferred onto PVDF (BioRad) or subjected to dual luciferase assay.

Dual Luciferase Assay.

Dual luciferase assays were performed according to the manufacturer's protocols (Promega). Firefly luciferase activity was measured by adding 2 μl of each reaction with LAR I (20 μl) into one well of a 96-well plate and read in Victor³ V (PerkinElmer) for 5 sec. Renilla luciferase activity was measured by adding Stop & Glo (20 μl, Promega) to each well and reread for 5 sec.

Ribosome-Binding Assay.

Ribosome-binding assays were performed as described in ref. 18. In vitro translation repression reactions supplemented with cycloheximide (600 μM), GMP-PNP (1 mM), or hippuristanol (50 μM) were loaded onto 10–30% glycerol gradient containing 1× HSB (500 mM NaCl; 20 mM Hepes-KOH, pH 7.5; 30 mM MgOAc; and 2 mM DTT). Glycerol gradients were ultracentrifuged by using an SW41 rotor (Beckman) at 39,000 rpm for 3.5 h, sequentially fractionated (500 μl) from the top, and subjected to Cerenkov scintillation counting.

Precipitation of Biotinylated mRNAs.

Streptavidin agarose (SAA) beads (Invitrogen) were used to precipitate biotinylated mRNA reporters. SAA beads (100 μl) were washed in 1× HSB buffer three times and incubated with glycerol gradient fractions (200 μl) or whole lysate reactions at 4°C for 60 min. Reactions were centrifuged, and supernatants were removed. Precipitates were washed twice in 1× HSB and subjected to RNA extraction and precipitation or to Western blot analysis.

Northern Blot Analysis.

The Northern blot analysis was performed by PAGE as described in ref. 6. RNAs extracted from SAA precipitates were separated on 8% PAGE containing urea (7 M) and transferred to Hybond n + membranes (Amersham Biosciences) for 2.5 h at 350 mA. After UV cross-linking, membranes were hybridized with 5′ end-labeled primers for 5S rRNA, 5′-TTAGCTTCCGAGATCA-3′; 5.8S rRNA, 5′-GCTAGCGCTGCGTTCTTCATCGACGC-3′; 28S rRNA5′-AACGATCAGAGTAGTGGTATTTCACC-3′; 18S rRNA, 5′-CGGAACTACGACGGTATCTG-3′; and tRNAi-Met, 5′-GGTAGCAGAGGATGGTTTCGATCC-3′. Membranes were washed, visualized, and analyzed by PhosphorImager (Molecular Dynamics).

Western Blot Analysis.

The SAA precipitates were resuspended in 1× SDS loading buffer, boiled at 95°C for 5 min, and centrifuged. Supernatants were resolved on SDS–10% PAGE and transferred onto PVDF membranes (BioRad). Membranes were blocked in 5% nonfat milk powder in PBST (10 mM phosphate buffer, pH 7.2; 150 mM NaCl; and 0.1% Tween 20) for 60 min, washed twice with PBST, and incubated with antibodies in 1% nonfat milk powder–PBST at 4°C overnight. The anti-Ago2 antibody (Upstate) recognizes residues 7–48 of human Ago2, which are conserved amino acids between human, mouse, cattle, dog, and frog. Anti-eIF2α, -eIF4A, and -eIF4E were kind gifts from Jerry Pelletier. The anti-eIF3g antibody was a kind gift from Hiroaki Imataka, Riken Genomic Sciences Center, Wako, Japan. Anti-RPS7 and -RPL18 antibodies were used according to the manufacturer's protocol (Abnova). Membranes were washed three times with PBST, incubated with horseradish peroxidase-conjugated secondary antibodies (Jackson ImmunoResearch) at 1:5,000 in 1% nonfat milk powder–PBST, and developed by ECL (Pierce).

To strip Western blots of antibody complexes, membranes were incubated in stripping buffer (100 mM 2-mercaptoethanol; 2% SDS; and 62.5 mM Tris·HCl, pH 6.7) at 50°C for 30 min. These membranes were washed with PBST for 2 × 10 min, blocked in 5% milk–PBST, and reprobed with appropriate antibodies.

Toeprinting Assay.

Translation repression reactions containing mRNA (0.1 pmol) and CXCR4 siRNA (0.6 pmol), RRL (7 μl), and MgOAc (2 mM) with or without GMP-PNP (1 mM) or Hisppuristanol (50 μM) proceeded for 5 min at 30°C. Then, reverse transcription (RT) mix containing dNTPs (5 mM), 1× reconstitution buffer (20 mM Tris·HCl, pH 7.5; 100 mM KCl; and 1 mM DTT), 5′ end-labeled primer (0.2 pmol, 5′-TTATGCAGTTGCTCTCCAGCG-3′), and M-MLV RT (1 μl, Invitrogen) was added to translation repression reactions. These mixtures were incubated for 15 min at 30°C and subjected to deproteinization and ethanol precipitation. RNAs were resolved on a 10% sequencing gel (National Diagnotics) and visualized by PhosphorImager analysis (Molecular Dynamics).

See SI Materials and Methods for additional details.