Control of reovirus messenger RNA translation efficiency by the regions upstream of initiation codons

Abstract

The 10 species of reovirus messenger RNA are translated in vivo with efficiencies/frequencies that differ by as much as 100-fold. The sl mRNA, which is translated 10 times less efficiently than the s4 mRNA but 10 times more efficiently than the /1 and ml mRNAs, has a unique BamH1 cleavage site located immediately downstream of its initiation codon. Because the reovirus mRNAs have been cloned, this provides the opportunity for placing modified and altered sequences upstream of its coding sequence. The translation efficiencies of the variant mRNAs, transcribed via the SP6 in vitro transcription system, can then be measured in the rabbit reticuloycte lysate in vitro translation system. Using this system it was found that replacing the 5′-upstream sequence of the s4 mRNA with that of the s4 mRNA increases its in vitro translation efficiency by 4-fold; that the trinucleotide immediately upstream of the s1 initiation codon renders it very weak, and that it is only slightly superior to the weakest Kozak consensus sequence; that the nature of the nucleotides further upstream than position −3 can profoundly affect translation efficiency; that the nature of this effect is in turn markedly modified by the nature of nucleotides in positions −1 to −3; and that there is a minimum optimal 5′-upstream sequence length of about 14 nucleotides. We also investigated the effect of secondary structure involvement on the ability of 5′-upstream sequences to promote translation. Two effects were noted. First, being part of moderately stable stem loops (ΔG, −18 kcal/mol) decreased translation efficiency about 3-fold; second, mRNA in which only three 5′-terminal nucleotides were unpaired were translated five times less efficiently than mRNA in which six nucleotides were unpaired. Accessibility of the 5′-cap as well as secondary structure of the 5′-upstream sequences are therefore factors that affect translation efficiency. Finally, we showed that the ml mRNA, which is transcribed very poorly in vivo, is translated very efficiently in vitro; and that its 5′-upstream sequence is as effective in increasing protein σ1 formation as that of s4 mRNA. Since both ml mRNA and protein μ2 are stable in infected cells, the reason why ml mRNA is translated so inefficiently in vivo therefore remains unexplained.