Control of translation and miRNA-dependent repression by a novel poly(A) binding protein, hnRNP-Q

Abstract

Translation control often operates via remodeling of messenger ribonucleoprotein particles. The poly(A) binding protein (PABP) simultaneously interacts with the 3' poly(A) tail of the mRNA and the eukaryotic translation initiation factor 4G (eIF4G) to stimulate translation. PABP also promotes miRNA-dependent deadenylation and translational repression of target mRNAs. We demonstrate that isoform 2 of the mouse heterogeneous nuclear protein Q (hnRNP-Q2/SYNCRIP) binds poly(A) by default when PABP binding is inhibited. In addition, hnRNP-Q2 competes with PABP for binding to poly(A) in vitro. Depleting hnRNP-Q2 from translation extracts stimulates cap-dependent and IRES-mediated translation that is dependent on the PABP/poly(A) complex. Adding recombinant hnRNP-Q2 to the extracts inhibited translation in a poly(A) tail-dependent manner. The displacement of PABP from the poly(A) tail by hnRNP-Q2 impaired the association of eIF4E with the 5' m(7)G cap structure of mRNA, resulting in the inhibition of 48S and 80S ribosome initiation complex formation. In mouse fibroblasts, silencing of hnRNP-Q2 stimulated translation. In addition, hnRNP-Q2 impeded let-7a miRNA-mediated deadenylation and repression of target mRNAs, which require PABP. Thus, by competing with PABP, hnRNP-Q2 plays important roles in the regulation of global translation and miRNA-mediated repression of specific mRNAs.

Figure 1. Detection of poly(A) interacting proteins using UV crosslinking.

RRL (A), HeLa (B), or Krebs (C) S10 cytoplasmic extracts were subjected to UV-induced crosslinking with the ³²P-poly(A) tail. The extracts that were either mock-depleted (Control S10) or depleted of PABP were incubated with ³²P poly(A) tail-labeled globin mRNA at 32°C for 10 min. Prior to adding mRNA, the reaction mixtures were pre-incubated at 32°C for 2 min with GST-Paip2 (20 µg/ml) or PABP (5 µg/ml) as indicated. After UV irradiation and RNase treatment, labeled proteins were analyzed by SDS-PAGE and autoradiography. The positions of molecular mass markers are indicated on the right. Western blotting with an anti-PABP antibody was used to confirm sufficient depletion of PABP from the extracts (for representative analyses, see [54]). Of note, in this and other UV crosslinking analyses, PABP appeared as a fussy band, most probably because RNase digestion does not produce homogenous cross-linked RNA fragments.

Figure 2. Specificity of binding of hnRNP-Q2 to the poly(A) tail.

(A) Proteins of control or PABP-depleted RRL UV crosslinked with the ³²P-poly(A) tail were analyzed by SDS-PAGE as described for Figure 1. RNA competitors, 18S rRNA, poly(A), poly(G), poly(U), and poly(C), were included in the reaction mixtures at 5 µg/ml concentrations. (B) Recombinant hnRNP-Q2 (2 µg) was analyzed by SDS-PAGE and Coomassie blue staining. Prestained molecular mass markers (MBI Fermentas) are shown in the right lane. (C) High-affinity binding of hnRNP-Q2 to oligo(A). EMSA was performed as described in Materials and Methods. For each lane, a constant small amount of ³²P-oligo(A)₃₀ RNA was incubated with the indicated concentrations of hnRNP-Q2. The Kd value of ∼30 nM was calculated from three independent experiments.

Figure 3. Competition between hnRNP-Q2 and PABP for binding to the poly(A) tail in RRL.

(A) Western blot analysis of RRL depleted with either anti-FLAG (Control RRL) or anti-hnRNP-Q antibody. The blot was probed for anti-hnRNP-Q or anti-β-actin (loading control). (B) Proteins of control or hnRNP-Q2-depleted RRL that crosslink with the ³²P-poly(A) tail in the presence of the indicated concentrations of recombinant hnRNP-Q2. (C) Proteins of control or PABP-depleted RRL that crosslink to the ³²P-poly(A) tail in the presence of the indicated concentrations of recombinant PABP.

Figure 4. HnRNP-Q2-mediated inhibition of translation in Krebs extracts.

(A) Western blot analysis of Krebs S10 extracts depleted with either anti-FLAG (Control S10) or anti-hnRNP-Q antibody. The blot was probed for hnRNP-Q, YB-1, or β-actin (loading control). (B) Mock-depleted (control) or hnRNP-Q2-depleted S10 Krebs extracts were programmed with Cap-Luc-p(A)₉₈ mRNA in the absence or presence of the indicated concentrations of hnRNP-Q2. (C) Stability of Cap-Luc-p(A)₉₈ mRNA. ³²P-labeled Cap-Luc-p(A)₉₈ was used to program mock-depleted (control) or hnRNP-Q2-depleted S10 Krebs extracts not supplemented or supplemented with recombinant hnRNP-Q2 (30 µg/ml), as indicated. Total RNA was isolated at the indicated times from the aliquots of the reaction mixture, resolved by formaldehyde-agarose gel electrophoresis, and transferred to a membrane. Cap-Luc-p(A)₉₈ mRNA and 28S rRNA were detected by autoradiography (left panel, top) and staining (left panel, bottom), respectively. Cap-Luc-p(A)₉₈ mRNA band intensities were determined and corrected for loading by 28S rRNA (right panel). The values for Cap-Luc-p(A)₉₈ mRNA recovered at the beginning of incubation were set as 100%. (D and E) Control (mock depleted) and hnRNP-Q2-depleted Krebs extracts were programmed with PV IRES-Luc-p(A)₉₈ (D) or HCV IRES-Luc-p(A)₉₈ (E) mRNAs. hnRNP-Q2 was added to the reaction mixtures at the indicated concentrations. On panels B, D, and E, the data are averages of three assays with standard deviations from the mean.

Figure 5. HnRNP-Q2 inhibits m⁷G cap structure recognition by translation initiation factors.

(A–D) Inhibition of 80S and 48S initiation complex formation by hnRNP-Q2 in nuclease-treated RRL. 80S ribosome binding to 3′ end labeled globin mRNA was assayed in a cycloheximide (0.6 mM)-supplemented RRL, normal (A) or hnRNP-Q2-depleted (B), in the presence of control buffer (squares) or recombinant hnRNP-Q2 (15 µg/ml) (triangles). (C) Validation of the 48S pre-initiation complex formation in the presence of GMPPNP. GTP or GMPPNP were added to the reaction mixtures at 2 mM final concentration as indicated. Other conditions were similar to those described for panel B. (D) 48S pre-initiation complex formation in hnRNP-Q2-depleted RRL in the presence of GMPPNP and either control buffer (squares) or hnRNP-Q2 (25 µg/ml) (triangles). The reaction mixtures were analyzed on 5-ml 15%–30% (A and B) or 11-ml 10%–30% (C and D) sucrose gradients. (E) HnRNP-Q2 dose-dependent inhibition of eIF4E binding to the m⁷G cap structure in RRL as analyzed by chemical crosslinking. Control and hnRNP-Q2-depleted RRL were incubated with oxidized ³²P-cap-labeled poly(A)-extended Luc mRNA in the absence or presence of the indicated concentrations of recombinant hnRNP-Q2. The positions of eIF4E and eIF4A are indicated. Relative efficiencies of eIF4E crosslinking are indicated at the bottom (the value obtained for control RRL was set as 100%).

Figure 6. Poly(A) tail length and PABP-dependent inhibition of translation by hnRNP-Q2 in Krebs extract.

(A) Krebs extract that was not nuclease-treated was programmed with Cap-Luc mRNAs (0.2 µg/ml) bearing poly(A) tails of the indicated length. Control buffer of hnRNP-Q2 (20 µg/ml) was added to the reaction mixtures as indicated. (B) Sequestering of PABP by Paip2 renders translation insensitive to the poly(A) tail length and inhibition by hnRNP-Q2. Cap-Luc mRNA with increasing poly(A) tails was translated in the untreated extract in the presence of Paip2 (15 µg/ml) as described for panel A. hnRNP-Q2 (20 µg/ml) was added to the reaction mixtures where indicated. Inhibition of translation by hnRNP-Q2 is shown on the top of the panels. (C) Endogenous [³⁵S]methionine incorporation in the untreated extract in the presence of the indicated concentrations of hnRNP-Q2 or 10 µM hippuristanol (Hipp). Incubation was at 32°C for 2 h. Average values for trichloroacetic acid-insoluble radioactivity in 1-µl aliquots of the samples from three assays with standard deviations are shown.

Figure 7. HnRNP-Q2 silencing stimulates global protein synthesis.

L929 cells were infected with lentiviruses expressing the nontarget control shRNA or shRNAs (1 and 2) against hnRNP-Q. (A) Cytoplasmic extracts of control and hnRNP-Q knockdown cells equalized for protein content were subjected to Western blotting for hnRNP-Q2, PABP, eIF4GI, eIF4A, eIF4E, and β-actin, as indicated. (B) Quantitative analysis of hnRNP-Q2 bands in panel A using NIH ImageJ software. The values were normalized by those of β-actin. The value in control was set as 100%. The data are means with standard deviations from three experiments. (C) Protein synthesis in control and hnRNP-Q2-knockdown L929 cells as analyzed by [³⁵S]methionine/cysteine labeling. The mean values for ³⁵S incorporation into proteins from three independent assays with standard deviations are shown as percentages of the value in control (* p<0.025). (D) Representative patterns of ³⁵S-labeled proteins from panel C. Proteins were resolved by SDS 12% PAGE and detected by autoradiography.

Figure 8. HnRNP-Q2 counteracts PABP function in let-7a miRNA dependent deadenylation.

Kinetics of deadenylation of 6xB-3′UTR RNA in untreated (A) or nuclease-treated (B) S10 Krebs extracts. Recombinant hnRNP-Q2 (50 µg/ml) was added to the reaction mixtures where indicated. (C) Kinetics of deadenylation of 6xB-3′UTR RNA in hnRNP-Q2-depleted Krebs extract. Recombinant hnRNP-Q2 (36 µg/ml) or GST-Paip2 (16 µg/ml) were included in the reaction mixtures where indicated. 6xB-3′UTR RNA treated with RNase H in the presence of oligo(dT) is shown in lane 1. (D) Kinetics of deadenylation of 6xB-3′UTR RNA in hnRNP-Q2 and PABP double-depleted Krebs extract. Recombinant PABP (6 µg/ml) and hnRNP-Q2 (36 µg/ml) were included in the reaction mixtures either alone or in combination as indicated. The positions of intact and deadenylated RNAs are indicated on the right. The data are the representative of three independent experiments.

Figure 9. HnRNP-Q2 depletion augments miRNA-mediated repression in cultured cells.

(A) Activities of RL and RL-6xB reporters in control and hnRNP-Q2 knockdown L929 cells. Cells expressing control shRNA or hnRNP-Q shRNA2 were transfected with RL reporters either possessing or lacking the 6xB sequence in parallel with a Firefly luciferase (FL) reporter. Anti-let-7a or anti-miR-122a (negative control) 2′-O-Me oligonucleotides were cotransfected where indicated. Two days after transfection, activities of RL and FL were measured and their ratio was determined. (B) The expression of RL-6xB relative to RL (which was set as 100% in both control and hnRNP-Q2-knockdown cells). (C) Equal amounts (10 µg) of RNA from transfected cells were analyzed by Northern blotting using probes specific for RL (top) and FL (bottom) reporters. The bands were quantified using a Typhoon PhosphorImager (GE Healthcare). To correct for loading and transfection efficiency, the values for RL and RL-6xB mRNAs were normalized to those for FL mRNA. The RL-6xB/RL ratio for each condition is given under the top panel. Note that RL-6xB mRNA migrates slightly slower than RL mRNA due to the presence of the 6xB sequence. (D) The levels of RL-6xB mRNA in control and hnRNP-Q KD cells were normalized to those of RL mRNA (which were set as 100% for both conditions). In (A), (B), and (D) the values are means from three transfections with standard deviations (* p<0.025, ** p<0.001).