Mechanisms governing the selection of translation initiation sites on foot-and-mouth disease virus RNA

Abstract

Translation initiation dependent on the foot-and-mouth disease virus (FMDV) internal ribosome entry site (IRES) occurs at two sites (Lab and Lb), 84 nucleotides (nt) apart. In vitro translation of an mRNA comprising the IRES and Lab-Lb intervening segment fused to a chloramphenicol acetyltransferase (CAT) reporter has been used to study the parameters influencing the ratio of the two products and the combined product yield as measures of relative initiation site usage and productive ribosome recruitment, respectively. With wild-type mRNA, ∼40% of initiation occurred at the Lab site, which was increased to 90% by optimization of its context, but decreased to 20% by mutations that reduced downstream secondary structure, with no change in recruitment in either case. Inserting 5 nt into the pyrimidine-rich tract located just upstream of the Lab site increased initiation at this site by 75% and ribosome recruitment by 50%. Mutating the Lab site to RCG or RUN codons decreased recruitment by 20 to 30% but stimulated Lb initiation by 20 to 40%. An antisense oligodeoxynucleotide annealing across the Lab site inhibited initiation at both sites. These and related results lead to the following conclusions. Recruitment by the wild-type IRES is limited by its short oligopyrimidine tract. At least 90% of internal ribosome entry occurs at the Lab AUG, but initiation at this site is restricted by its poor context, despite a counteracting effect of downstream secondary structure. Initiation at the Lb site is by ribosomes that access it by linear scanning from the original entry site, and not by an independent entry process.

Fig. 1.

Effects of mutating the Lab and Lb initiation codons to non-AUG codons. (A) The FMDV O1K sequence between the Lab and Lb sites, showing the position where an additional AUG codon (indicated by an asterisk) was introduced for the assays shown in panel D. (B) Autoradiograph of translation products resulting from translation of mRNAs in which the Lab site had been mutated to the designated non-AUG codons. (C) Comparison of the effects of mutating the Lab and Lb sites to ACG or GUA. The rightmost lane (GUA+ AUGinrep) was loaded with the translation assay of a construct with a GUA at the Lb site and an in-frame AUG introduced into the reporter CAT ORF. (D) Effect of mutating the Lab site to ACG or GUA and also introducing an additional AUG between the Lab and Lb sites (lanes with an asterisk). All translation assays were carried out under standard conditions. The band intensities were determined by densitometry and quantitated relative to the yield (i.e., band intensity) of the product initiated at the Lb site of the wild-type (wt) RNA, which was set at 1.0 (underlined).

Fig. 2.

Effects of antisense deoxyoligonucleotides targeted to the Lab and Lb sites. (A) Schematic diagram showing the annealing sites of antisense oligodeoxynucleotides as-Lab (as stands for antisense) and as-Lb. (B) Titration of each antisense oligonucleotide in translation assays. The assays had 10 μg/ml mRNA (25 nM), and the oligonucleotides were added at concentrations ranging from 2.5- to 40-fold molar excess. The amount of oligonucleotide is indicated by the height of triangle above the lane. −, negative control. (C) Effect of exogenous RNase H. The antisense oligonucleotides (at 40-fold molar excess over mRNA) were added at zero time, and where indicated, exogenous RNase H was present at 20 U/ml. The asterisk highlights a faint band that is presumed to arise from initiation at the first in-frame AUG codon downstream of the Lb site in the uncapped cleavage RNA product arising from RNase H cutting in the region where as-Lb anneals.

Fig. 3.

Effect of Mg²⁺ concentration on utilization of the Lab and Lb initiation sites. (A) Titration of added Mg²⁺ (mM) in translation assays. Autoradiograph of translation products resulting from translation of mRNA in which the Lb had been mutated to GUA (top panel), wild-type (wt) mRNA (middle panel), and mRNA in which the Lab site had been mutated to GUA (bottom panel). The band intensities were determined by densitometry and quantitated relative to the yield of the product initiated at the Lb site of the wild-type construct using the standard Mg²⁺ concentration (0.5 mM) routinely used in all assays, which was set at 1.0 (underlined). (B) Graph showing the product yield as a percentage of the maximum (max) yield of either the Lab product (top panel) or the Lb product (bottom panel), plotted against the Mg²⁺ concentration. For these assays, a batch of reticulocyte lysate was used, which likely had a slightly higher endogenous Mg²⁺ concentration than other batches used in this work. Consequently, the maximum yield of product from the Lb site of wild-type mRNA was seen at 0.25 mM added Mg²⁺, as against 0.5 mM for other batches, and the typical Lab/Lb product ratio of ∼40:60 observed using wild-type mRNA at 0.5 mM added Mg²⁺ with other batches occurred at 0.25 mM with this particular batch.

Fig. 4.

Effect of improving the context of the Lab site and changing the distance between the start of the oligopyrimidine tract and the Lab site. (A) Nucleotide sequence of the wild type and the Lab context mutant, in which the mutated nucleotides are underlined, and the sequence alignment between the wild-type and the oligopyrimidine tract mutants, in which the inserted nucleotides are underlined. Gaps introduced to maximize alignment are indicated by dashes. (B) Autoradiograph showing the effect of improving the context of the Lab site. (C) Autoradiograph of the translation products of the mutants in which the distance between the start of the oligopyrimidine and the Lab site have been changed. The analysis of translation products is shown with the relative yields of the different products expressed as described in the legend to Fig. 1.

Fig. 5.

The region between the Lab and Lb sites is not critical for productive ribosome recruitment but does influence initiation frequency at the Lab site. (A) Nucleotide sequence of the wild-type and mutants in which the CAT reporter sequence is fused directly to the Lab site. The reporter sequence is shown in italic type, and mutated nucleotides are underlined. (B) Translation products of the mutants in which the reporter was fused directly to the Lab site, without or with an in-frame AUG introduced at 54 nt downstream from the Lab site (designated +AUGif) and also with a mutation of the out-of-frame AUG (designated +AUGif-m). The analysis of translation products is shown with the relative yields of the different products expressed as described in the legend to Fig. 1.

Fig. 6.

The sequence downstream of the Lab site has an effect on start site selection. (A) Nucleotide sequences of deletion mutants lacking various lengths of FMDV sequences downstream of the Lab site and also with additional mutations eliminating G residues. The 3′ deletion mutants retained 0, 12, 24, or 48 nt of wild-type FMDV sequence after the Lab site, and the single 5′ deletion was missing the first 21 nt of the downstream FMDV sequence. These were fused to a CAT reporter with an optimal context initiation codon (shown in italics). The 3′ deletion mutants with 24 and 48 nt of retained FMDV sequence were further mutated (at the underlined positions) to generate AC-rich sequences lacking G residues. (B) Translation products of the mutants with different lengths of wild-type sequence after the Lab site. (C) Translation products of the mutants in which the sequence immediately after the Lab site has been mutated to generate an AC-rich sequence and eliminate all G residues. The analysis of translation products is shown with the relative yields of the different products expressed as described in the legend to Fig. 1. (D) Sequences between the two AUGs of selected constructs, showing potential base pairing. The underlined sequences labeled A and A′ are complementary, as is also the case for B and B′. For Lab-d21, the italicized sequences show an alternative complementarity, pairing B with B" (which could also occur with the wild-type sequence).

Fig. 7.

Effect of inserting a stem-loop structure at various positions downstream of the Lab site. The sequences of the constructs are summarized in linear format, and the secondary structure and sequence of the inserted stem-loops is also depicted. Lab+12/sl and Lab+21/sl have three in-frame AUG triplets, whereas Lab/sl has only two. Note that the Lab AUG in Lab/sl will be expected to base pair with the downstream AUG triplet, in the same way as shown for stem-loop 2. The autoradiograph of the analysis of translation products is shown with the relative yields of the different products expressed as described in the legend to Fig. 1.