mRNA translation regulation by the Gly-Ala repeat of Epstein-Barr virus nuclear antigen 1

Abstract

The glycine-alanine repeat (GAr) sequence of the Epstein-Barr virus-encoded EBNA-1 prevents presentation of antigenic peptides to major histocompatibility complex class I molecules. This has been attributed to its capacity to suppress mRNA translation in cis. However, the underlying mechanism of this function remains largely unknown. Here, we have further investigated the effect of the GAr as a regulator of mRNA translation. Introduction of silent mutations in each codon of a 30-amino-acid GAr sequence does not significantly affect the translation-inhibitory capacity, whereas minimal alterations in the amino acid composition have strong effects, which underscores the observation that the amino acid sequence and not the mRNA sequence mediates GAr-dependent translation suppression. The capacity of the GAr to repress translation is dose and position dependent and leads to a relative accumulation of preinitiation complexes on the mRNA. Taken together with the surprising observation that fusion of the 5' untranslated region (UTR) of the c-myc mRNA to the 5' UTR of GAr-carrying mRNAs specifically inactivates the effect of the GAr, these results indicate that the GAr targets components of the translation initiation process. We propose a model in which the nascent GAr peptide delays the assembly of the initiation complex on its own mRNA.

FIG. 1.

The effect of the GAr on mRNA translation is position and dose dependent. (A) Autoradiograph showing in vitro translation of increasing amounts of chicken Ova mRNA using RRL results in a steady linear increase in the amount of synthesized protein. Fusion of the full-length GAr sequence to the 5′ encoding sequence of Ova (FL.GAr-Ova) results in a rate of synthesis that is similar to that of Ova alone up to approximately 1 nM mRNA. At higher mRNA concentrations, the GAr blocks translation. Fusion of the GAr to the 3′ encoded region (Ova-FL.GAr) weakens its translation-inhibitory effect. The results in the graph are presented as relative translation efficacy of the different mRNAs. (B) [³⁵S]methionine metabolic cell labeling for 20 min in the presence of proteasome inhibitors (upper) and Western blotting (WB) of GAr fused to p53 following immunoprecipitation against p53. There is a 2.4-fold lower rate of p53 synthesis when the GAr is fused to its C terminus than with the N terminus, showing that the position-dependent effect of the GAr is not substrate dependent and is observed in vitro as well as in vivo. qRT-PCR analysis ensures that the transfection efficiency is similar and that the GAr sequence does not affect RNA levels (data not shown). The results are representative of at least three independent experiments, and data presented in graphs are derived from phosphorimager analysis. conc, concentration.

FIG. 2.

Short GAr sequence retards translation in cis in vivo and in vitro. (A) Constructs with identical 5′ and 3′ UTRs carrying the HA tag (HA) in the 5′ encoded sequence with or without a 43-aa GAr sequence. (B) The autoradiograph of in vivo pulse labeling shows suppression of translation by the 43GAr at 15 and 30 min using the Ova reporter gene. (C) The effect of different lengths of GAr constructs on translation rate in vivo illustrates that inhibitory effect of the 43GAr is comparable to the FL.GAr. (D) A kinetic in vitro translation assay shows that the GAr acts from the very onset of the translation process. Quantification of pulse-labeling experiments was carried using phosphorimager analysis. The figures are representative of at least three independent experiments, and error bars show standard deviations.

FIG. 3.

The GAr changes a classical ribosomal profile by time-dependent accumulation of a GAr-dependent peak. (A) In vitro translations assays using mRNAs carrying FL.GAr fused to the C or N terminus of Ova and the control Ova mRNA followed by analysis of polysomal profiles reveal a new peak (GP). This peak is more pronounced when GAr is fused to the N terminus than when it is fused to the C terminus of Ova, indicating that the GAr is more potent when located in the N-terminal region of a protein. rRNA analysis reveals a relative increase of 18S rRNA compared to 28S rRNA in the GP, indicating that this peak is composed of a relative increase in 40S and represents a “ribosome and a half” (Table 1). (B) The fusion of a 43-aa GAr sequence preceded by the HA tag to the N terminus of Ova shows the accumulation of the GP between 15 and 30 min of the translation reaction. In the non-GAr carrying Ova message, there is a small peak at this location after 15 min of translation, which disappears after 30 min. The experiment is representative of more than three independent tests. (C) Capped mRNAs were synthesized and labeled in vitro with [α-³²P]UTP. Labeled mRNA (10 mg/ml) was added to a 400-μl RRL-containing translation reaction for 10 min. The reactions were stopped by adding 20 mg/ml cycloheximide and loaded on an Optiprep gradient at 36,000 rpm (SW41) for 2 h. Fractions (0.5 ml) were taken from the top, and the amount of mRNA in each fraction was indirectly estimated by measuring the amount of ³²P using a scintillation counter. Introduction of a hairpin structure (−65 kcal/mol) that prevents scanning of the preinitiation complex leads to a similar distribution profile of the EBΔGAr mRNA as that of the EBNA-1 WT mRNA. 18S and 28S rRNAs from the corresponding fractions are indicated. OD, optical density.

FIG. 4.

Changes in the GAr amino acid sequence but not in the RNA sequence affects the translation-inhibitory capacity of the GAr. (A) An autoradiograph of a 20-min pulse-labeling in vivo experiment in the presence of proteasome inhibitors. Fusion of a sequence corresponding to 31 aa of the GAr to the N terminus of p53 (31GAr-p53) or a sequence in which the alanine and glycine codons have been changed from GCA and GGC/GGG to GCU and GGU, respectively, (31GArT-p53) results in a similar suppression of translation. (B) An autoradiograph of an in vitro labeling using capped mRNAs reveals similar results even though the effect of the short GAr on p53 synthesis is more potent in vivo. (C) An autoradiograph of in vivo cell labeling using different GAr sequences fused to Ova shows that positioning two alanine residues next to each other on three locations using the GCC codon (3A) disrupts the translation-inhibitory function of the GAr. Inserting serine residues in the N- and C-terminal parts increases the rate of translation further (2S). Insertion of a serine in the N terminus (1S-N) has a stronger effect than changes in the C terminus (1S-C). A GAr sequence derived from the Papio virus carries a single serine residue inserted at every seventh amino acid. (D) Western blotting (WB) in the presence of proteasome inhibitors shows that inserting an 80-aa-long polyQ sequence in the N terminus of p53 has no effect on the rate of synthesis compared with the FL.GAr. The figures are representative of three independent experiments, and the graph shows average data obtained from phosphorimager analysis. Error bars represent standard deviations.

FIG. 5.

The c-myc IRES overrides the translation-inhibitory effect of the GAr. (A) Autoradiograph of an in vivo pulse-labeling experiment shows that fusion of the c-myc IRES in the 5′ UTR of the FL.GAr-Ova message restores mRNA translation. This is not observed using the EMCV IRES, indicating that this effect is not mediated by cap-independent translation per se. Introduction of a hairpin structure 5′ of the c-myc IRES (HPc-myc-IRES-FL.GAr-Ova) blocks its capacity to stimulate synthesis of the FL.GAr-Ova mRNA. The c-myc 5′ UTR has no effect when fused to the Ova by itself (c-myc-IRES-Ova). (B) Northern blot analysis shows the integrity of the c-myc-IRES-FL.GAr-Ova mRNA and indicates that fusion of the GAr to a message does not increase its levels in the cells. (C) Fusion of the c-myc IRES in the 5′ UTR of the EBNA-1 message leads to an approximately twofold increase in EBNA-1 steady-state levels similar to those of the EBΔGAr. The data are representative of more than three separate experiments. WB, Western blotting.

FIG. 6.

Model for GAr-dependent inhibition of mRNA translation in cis. In panels A and C, GAr suppresses translation, and in panels B and D translation can proceed. The observation that the GAr is more potent in blocking translation at high mRNA concentrations suggests that it acts on blocking the effect of a positive-acting translation factor and that the ribosome has no difficulties per se to read through the GAr sequence. Its position and dose-dependent effects might represent two sides of the same coin and can be explained by a common model. More GAr accumulates on the polysome in proximity to the initiation site when the GAr is located in the N terminus (A) than when it is in the C terminus (B), making the GAr more potent in preventing a putative positive translation initiation factor to assist in translation initiation. Similarly, when the concentration of GAr-carrying mRNA is high in the in vitro assays (C), translation is more efficiently suppressed since the ratio of GAr to the positive translation initiation factor is high. Translation at low mRNA concentrations (D), on the other hand, can proceed relatively well since the ratio of GAr to the positive translation initiation factor is low. This model is further supported by the observations that altering the GAr amino acid sequence has a stronger effect in the N terminus than in the C terminus, indicating that the nascent amino acids are more important for its efficacy and that the GAr causes an accumulation of a GP, independent of its location in the message. The fact that changing the 5′ UTR of the GAr while leaving the coding sequence intact restores translation further demonstrates that the GAr is not acting on translation elongation. In this scenario, the GAr is modeled to directly squelch the function of a positive-acting translation initiation factor, but it is also possible that it mediates its effect via an intermediate factor.