ABSTRACT

One hypothesis for the higher frequency of non-optimal codons at the 5’-end of genes is selection for codons that result in weaker mRNA secondary structures. Most of these studies have focused on cytoplasmic genes, and have excluded periplasmic genes from their analysis. Here we re-analysed 46 unique signal sequence variants we have generated from our previously published work on the role of non-optimal codons in the signal sequences of Escherichia coli secreted genes.

In this data-set, we compared the signal sequence variants to the wild-type sequence in terms of mRNA secondary structure, codon usage and relative expression levels. Using tRNA adaptation index (tAI) as a measure of codon usage, our analysis revealed that codon usage changes did not result in significant changes in predicted mRNA secondary structure. Nor were secondary structure changes correlated with changes in expression levels. However, there was a significant negative correlation with changes in tAI and their relative expression levels. These results indicate that increasing the tAI value of a signal peptide (going from non-optimal to optimal codons) results in lower expression of the signal sequence variant, and that change cannot be explained by changes in overall mRNA secondary structure. These results imply there are differing selection pressures are at present at the 5’ end of secretory genes compared to non-secretory genes. In light of these differences, whole genome analysis of codon usage should differentiate between secreted and non-secreted genes in future.