Inferring gene function from evolutionary change in signatures of translation efficiency
1 Mediterranean Institute for Life Sciences (MedILS), 21000 Split, Croatia
2 Bioinformatics and Genomics Programme, Centre for Genomic Regulation (CRG), Dr. Aiguader 88, 08003 Barcelona, Spain
3 Universitat Pompeu Fabra (UPF), 08003 Barcelona, Spain
4 EMBL/CRG Systems Biology Research Unit, Centre for Genomic Regulation (CRG), Dr. Aiguader 88, 08003 Barcelona, Spain
5 Institució Catalana de Recerca i Estudis Avançats (ICREA), Pg. Lluis Companys 23, 08010 Barcelona, Spain
Genome Biology 2014, 15:R44 doi:10.1186/gb-2014-15-3-r44Published: 3 March 2014
The genetic code is redundant, meaning that most amino acids can be encoded by more than one codon. Highly expressed genes tend to use optimal codons to increase the accuracy and speed of translation. Thus, codon usage biases provide a signature of the relative expression levels of genes, which can, uniquely, be quantified across the domains of life.
Here we describe a general statistical framework to exploit this phenomenon and to systematically associate genes with environments and phenotypic traits through changes in codon adaptation. By inferring evolutionary signatures of translation efficiency in 911 bacterial and archaeal genomes while controlling for confounding effects of phylogeny and inter-correlated phenotypes, we linked 187 gene families to 24 diverse phenotypic traits. A series of experiments in Escherichia coli revealed that 13 of 15, 19 of 23, and 3 of 6 gene families with changes in codon adaptation in aerotolerant, thermophilic, or halophilic microbes. Respectively, confer specific resistance to, respectively, hydrogen peroxide, heat, and high salinity. Further, we demonstrate experimentally that changes in codon optimality alone are sufficient to enhance stress resistance. Finally, we present evidence that multiple genes with altered codon optimality in aerobes confer oxidative stress resistance by controlling the levels of iron and NAD(P)H.
Taken together, these results provide experimental evidence for a widespread connection between changes in translation efficiency and phenotypic adaptation. As the number of sequenced genomes increases, this novel genomic context method for linking genes to phenotypes based on sequence alone will become increasingly useful.