Massive factorial design untangles coding sequences determinants and phenotypic consequences of translation efficacy
Résumé
Translation plays a crucial role in the cellular economy. The translation rate of particular transcripts obviously determines the eventual production rate of corresponding proteins. Being very costly, translation is both deeply contributing and affected by the global physiological state of the cell. A better understanding of the factors governing translation efficiency is therefore of tremendous fundamental and applied importance.
Although plethora of sequence determinants of translation efficiency have been proposed, most are highly confounded properties arising from the same underlying sequence. No systemic study has purposely sought to unravel the relative contributions of such intertwined factors to eventual protein production and their effect on the cell.
We have applied a novel sequence design platform and scaling DNA synthesis capacities to systematically explore combinations of various compositional biases (nucleotide, codon and amino acid) and mRNA secondary structures, thereby implementing a thorough molecular Design of Experiment in different focal regions of the sequence space. After cloning in a standard reporter system, we used multiplexed deep sequencing approaches to quantify the consequences of such sequence variations in a high-throughput manner. We thus measured mRNA abundance and decay, ribosomal densities, protein production and growth rates.
Functional analysis of 244,000 precisely designed coding sequences in E. coli uncovers the overwhelming dominance of secondary structures in controlling translation initiation and elongation, as well as transcript stability. We identify a moderate role for codon usage in modulating elongation and ribosome loadings when the impact of secondary structures in limiting initiation is lifted by way of translational coupling. Beyond the cost of protein biosynthesis, we observed little effect of codon usage on cellular growth rates. Instead, we find that highly structured, slow initiating transcripts are exceedingly stabilized and can poison cells through unproductive ribosome entrapment.