CPEB2-eEF2 interaction impedes HIF-1α RNA translation

Abstract

Translation of mRNA into protein proceeds in three phases: initiation, elongation, and termination. Regulated translation allows the prompt production of selective proteins in response to physiological needs and is often controlled by sequence-specific RNA-binding proteins that function at initiation. Whether the elongation phase of translation can be modulated individually by trans-acting factors to synthesize polypeptides at variable rates remains to be determined. Here, we demonstrate that the RNA-binding protein, cytoplasmic polyadenylation element binding protein (CPEB)2, interacts with the elongation factor, eEF2, to reduce eEF2/ribosome-triggered GTP hydrolysis in vitro and slow down peptide elongation of CPEB2-bound RNA in vivo. The interaction of CPEB2 with eEF2 downregulates HIF-1α RNA translation under normoxic conditions; however, when cells encounter oxidative stress, CPEB2 dissociates from HIF-1α RNA, leading to rapid synthesis of HIF-1α for hypoxic adaptation. This study delineates the molecular mechanism of CPEB2-repressed translation and presents a unique model for controlling transcript-selective translation at elongation.

Figure 1

Identification and expression analysis of CPEB2 isoforms. (A) CPEB2 proteins in the control (siCtrl) and CPEB2 knockdown (siCPEB2) neurons were detected at a size of around 100 and 135 kDa (see also Supplementary Figure S1 for antibody specificity). (B) Two alternatively spliced forms of CPEB2, rCPEB2a, and rCPEB2b were identified from rat hippocampal neuron cDNA that encoded proteins with additional amino acids (a.a.) at the N-terminus compared with the original mouse CPEB2 clone (NP_787951). The light- and dark-gray boxes indicate unique regions in rCPEB2a and rCPEB2b, respectively. RBD, RNA-binding domain; RRM, RNA recognition motif; Zif, zinc finger. The areas used for antibody production and siRNA knockdown are underlined. (C) CPEB2 expression in siCtrl, siCPEB2, untransfected (mock), and overexpressed (myc–CPEB2a+2b) Neuro-2a cells. The amount of proteins loaded from untransfected and overexpressed cells was 1/50th of that from siCtrl and siCPEB2 cells. (D) Genomic organization of three CPEB2 isoforms. The asterisk denotes the originally reported start codon in the NP_787951 clone. (E) Tissue distribution of CPEB2 in the western blot. The eEF2 signal served as a loading control. Figure source data can be found in Supplementary data.

Figure 2

CPEB2 interacts with eEF2. (A) Using the N-terminal 456 a.a. of CPEB2a as the bait, a yeast two-hybrid (Y2H) screen identified a clone containing a.a. 717–803 of eEF2. The various truncated eEF2 mutants were tested for positive (+) or negative (−) association with the CPEB2a N-terminus. (B) The N-termini of CPEB2a, CPEB2b (486 a.a.), and the common region (269 a.a.) were tested for their interaction with domain V of eEF2 in the Y2H system. (C) Co-immunoprecipitation assay. Using 293T cells expressing flag–eEF2 along with myc–CPEB2a or myc–CPEB2b, cell lysates were precipitated with myc antibody (Ab) and immunoblotted with flag Ab. IP: immunoprecipitation, IB: immunoblotting. (D) Using 293T cells expressing flag–eEF2N (domains I, G′, and II) or flag–eEF2C (domains III, IV, and V) along with myc–CPEB2a, cell lysates were precipitated with flag Ab and immunoblotted with myc Ab. (E) Reciprocal immunoprecipitation. Neuro-2a cell lysates, with or without RNase A treatment, were precipitated with control, CPEB2, or eEF2 IgG, and immunoblotted with CPEB2 and eEF2 antibodies. Figure source data can be found in Supplementary data.

Figure 3

CPEB2 decreases eEF2/ribosome-activated GTP hydrolysis in vitro. (A) The rate of ribosome-promoted GTP hydrolysis of eEF2 was determined using purified eEF2 and the 80S ribosome in the absence (none) or presence of recombinant (His)₆-sumo-tagged CPEB2a or a control, EGFP–Ms2CP (see also Supplementary Figure S2). Error bars indicate s.e.m. (n=3). One and two asterisks denote significant differences in the amount of hydrolysed GTP at each time point between CPEB2 and EGFP–Ms2CP, with P<0.05 and P<0.01, respectively (Student's t-test). (B) CPEB2–eEF2 interaction may or may not prevent eEF2 from docking to ribosomes, as illustrated. The reactions containing different combinations of CPEB2a, 80S ribosome, and/or eEF2 were precipitated with CPEB2 Ab and analysed for 28S and 18S ribosomal RNAs (rRNAs) by reverse transcription-PCR (RT–PCR). (C) Polysomal distribution of CPEB2. Neuro-2a lysates were treated with or without 50 mM EDTA before sucrose density gradient centrifugation. The proteins from gradient fractions were immunoblotted with CPEB2 and ribosomal protein S6 (RPS6) antibodies. Figure source data can be found in Supplementary data.

Figure 4

CPEB2 represses target RNA translation at elongation. (A) The reporter constructs used in the tethered function assay. The firefly luciferase was appended with 3′-UTR containing two Ms2CP-binding sites in sense (Luc) or antisense (LucR) orientation. CrPV-Luc reporter contains hairpin (hp) and internal ribosomal entry site (IRES) sequence in the 5′-UTR. Renilla luciferase was used to normalize variation in transfection efficiency. (B) The RNA-binding domain of CPEB2a (or CPEB2b) was replaced with the dimeric Ms2 coat protein (CPEB2aN–Ms2CP) and EGFP–Ms2CP was used as a control. The 293T cells transfected with the reporters and Ms2CP fusions were analysed for luciferase RNA levels by quantitative RT–PCR or (C) protein levels by dual luciferase assay (normalized: firefly/Renilla). (D) Similarly to (C), except transfected cells were treated with 2 μM cycloheximide (CHX) or 100 μM 4EGI-1 for 12 h before the assay. Error bars indicate s.e.m. (B, D) n=3; (C) n=5. One and two asterisks denote significant differences when compared with the EGFP–Ms2CP control, P<0.05 and P<0.01, respectively (Student's t-test).

Figure 5

CPEB2-inhibited translation requires an association with eEF2. (A) Schemes of the various truncated myc–CPEB2aN–Ms2CP constructs (see also Supplementary Figure S3 for subcellular distribution of these mutants). (B) The 293T cells transfected with luciferase reporters and Ms2CP fusions were analysed by the luciferase assay (normalized: firefly/Renilla). Error bars indicate s.e.m. (n=3). Two asterisks denote a significant difference, P<0.01 (Student's t-test). (C) The 293T lysates expressing EGFP–Ms2CP or various myc-tagged CPEB2aN–Ms2CP mutants were immunoprecipitated with myc Ab and probed with eEF2 and myc Abs. Figure source data can be found in Supplementary data.

Figure 6

CPEB2 downregulates HIF-1α RNA translation. (A) Western blot analysis of HIF-1α–c-Myc, and actin using lysates from HeLa cells treated with ±20 μM of MG132 and the indicated concentrations of CHX or 4EGI-1. The signals of HIF-1α and c-Myc were quantified and displayed as relative ratios (fold). (B) HeLa cells overexpressing myc–CPEB2a or its eEF2 binding-defective mutants (myc–CPEB2C and Δ188–364) were treated with 20 μM MG132 for 4 h and then used for western blotting (see also Supplementary Figure S5 for the interaction with eEF2), or (C) RNA isolation for quantitative RT–PCR (normalized with the GAPDH RNA level). (D) Two representative polysome profiles from HeLa cells with or without myc–CPEB2a expression. The polysomal distribution of HIF-1α and GAPDH RNAs in HeLa cells expressing myc, myc–CPEB2a, myc–CPEB2C, or the Δ188–364 mutant was determined by quantitative RT–PCR using RNAs isolated from each fraction. (E) HeLa cells transfected with plasmids encoding the two EGFP reporters with or without the HIF-1α 3′-UTR along with myc or myc–CPEB2a plasmid were metabolically labelled with AHA to tag de-novo synthesized polypeptides. EGFP and EGFP–Ms2CP from total cell lysate and the streptavidin-precipitated AHA-labelled proteins were detected by western blotting using EGFP Ab. The newly translated EGFP signals were quantified, expressed as a relative ratio and plotted against the time. (F) The polysome profiles of HeLa cells with or without myc–CPEB2a expression and ±200 μM 4EGI-1 treatment. The polysomal distribution of HIF-1α and GAPDH RNA was determined by quantitative RT–PCR. (G) The amounts of HIF-1α and GAPDH RNAs in the heavy density polysome fractions (#8–11) in (F) were summed and plotted against the treatment time of 4EGI-1. The levels of HIF-1α and GAPDH RNAs at time zero were arbitrarily set to 1. Figure source data can be found in Supplementary data.

Figure 7

Arsenite reduces CPEB2–HIF-1α RNA association and elevates HIF-1α synthesis. (A) The siCtrl and siCPEB2 Neuro-2a cells were treated with or without arsenite for 30 min before detection of HIF-1α protein or (B) RNA levels (normalized with the GAPDH RNA level). (C) Polysomal distribution of HIF-1α and GAPDH RNAs in siCtrl and siCPEB2 Neuro-2a cells. (D) Neuro-2a cells treated with ± 500 μM arsenite for 30 min were used for RNA immunoprecipitation (RNA-IP). The control and CPEB2 IgG-precipitated substances were analysed for HIF-1α and HIF-1β RNAs by quantitative RT–PCR (normalized with the non-specific bound GAPDH RNA level). (E) The 293T cells transfected with the firefly luciferase reporter containing HIF-1α 3′-UTR±CrPV IRES and Renilla luciferase along with myc or myc–CPEB2a were treated with ±500 μM arsenite for 1 h and harvested for luciferase assay (normalized: firefly/Renilla) or (F) RNA-IP with myc Ab. The precipitated HIF-1α and GAPDH RNAs were analysed by RT–PCR. Error bars indicate s.e.m. (n=3). One and two asterisks denote significant difference, P<0.05 and P<0.01, respectively (Student's t-test). (G) Schematic model of CPEB2-governed HIF-1α synthesis. In the well-oxygenated environment, the binding of CPEB2 to HIF-1α 3′-UTR reduces the HIF-1α peptide elongation through its interaction with eEF2. Arsenite-induced HIF-1α RNA translation is in part caused by the release of CPEB2 from HIF-1α RNA, which presumably allows eEF2 to resume its maximal GTPase activity and enhances translation elongation of HIF-1α RNA. Figure source data can be found in Supplementary data.