SELECTION ON TRANSLATION ACCURACY AFFECTS LOWLY EXPRESSED GENES IN BACTERIA

The genetic code is redundant, meaning that more than a single codon encodes most amino acids. Synonymous codons, encoding the same amino acids are not used equally due to a combination of background substitution biases and natural selection. Natural selection is known to prefer certain codons to others due to the fact that specific codons are translated more accurately and more efficiently. Such selection on the usage of specific codons is thought to affect mostly highly expressed genes, since the inaccurate or inefficient translation of such highly expressed genes should confer global costs on cell fitness. At the same time, we hypothesized that the inaccurate translation of lowly expressed genes could carry rather dramatic effects on the expression of the genes themselves by insufficient protein production. To test this we have carried out a computational analysis aimed at examining whether lowly expressed genes are also subject to selection on translation accuracy, affecting their codon usage. Using three different gene expression datasets, combined with the genome sequence data of the bacterium Escherichia coli, we were able to demonstrate that while highly expressed genes appear to be subject to stronger selection on translation accuracy, lowly expressed genes are also subject to such selection. We are now carrying out an experimental analysis of the manner in which codon usage affects the function of a specific lowly expressed gene.