Codon usage in large modular proteins in bacteria

Modular polyketide synthases (PKS) and non-ribosomal peptide synthetases (NRPS) consist of large polypeptides, which can have more than 10000 amino acids. The polypeptides usually contain several modules each divided into domains. The fundamental structure of each module is similar, but they differ in detail according to the nature of the extender unit that they incorporate. A large number of PKS and NRPS clusters are known, which provides a lot of dana for testing evolutionary hypotheses about codon usage. Both types of clusters are present in actinomycetes, which have a high G+C-content and a highly biased codon usage. Many NRPS clusters are also present in Bacillus species and PKS clusters are common in myxobacteria. Comparison of the clusters in these distantly related bacteria with the similar clusters in actinomycetes shows whether the observed codon usage phenomena are determined by the evolutionary pressures on the proteins or caused by the host organisms. Use of common codons for an amino acid is associated with high efficiencies of translation, which can be estimated using the codon adaptation index (CAI). The CAI was calculated in sliding windows for the modular proteins and showed characteristic short regions of slow translation at conserved positions in modules, which may be associated with allowing time for protein folding. Selective pressures acting on the modules and domains were estimated using the ratio of non-synonymous/synonymous codons in sliding windows. This confirmed that most of the sequences had low ratios suggesting strong purifying selection.