Selection-driven cost-efficiency optimization of transcripts modulates gene evolutionary rate in bacteria

Abstract

BackgroundMost amino acids are encoded by multiple synonymous codons. However, synonymous codons are not used equally, and this biased codon use varies between different organisms. It has previously been shown that both selection acting to increase codon translational efficiency and selection acting to decrease codon biosynthetic cost contribute to differences in codon bias. However, it is unknown how these two factors interact or how they affect molecular sequence evolution.ResultsThrough analysis of 1320 bacterial genomes, we show that bacterial genes are subject to multi-objective selection-driven optimization of codon use. Here, selection acts to simultaneously decrease transcript biosynthetic cost and increase transcript translational efficiency, with highly expressed genes under the greatest selection. This optimization is not simply a consequence of the more translationally efficient codons being less expensive to synthesize. Instead, we show that transfer RNA gene copy number alters the cost-efficiency trade-off of synonymous codons such that, for many species, selection acting on transcript biosynthetic cost and translational efficiency act in opposition. Finally, we show that genes highly optimized to reduce cost and increase efficiency show reduced rates of synonymous and non-synonymous mutation.ConclusionsThis analysis provides a simple mechanistic explanation for variation in evolutionary rate between genes that depends on selection-driven cost-efficiency optimization of the transcript. These findings reveal how optimization of resource allocation to messenger RNA synthesis is a critical factor that determines both the evolution and composition of genes.Electronic supplementary materialThe online version of this article (10.1186/s13059-018-1480-7) contains supplementary material, which is available to authorized users.

The unique coding sequence ofpmoCABoperon from type Ia methanotrophs simultaneously optimizes transcription and translation

“…Past reports on the molecular optimization of genetic sequences have shown that transcribed regions of genes can have fine-tuned nucleotide compositions that reduce their biosynthetic cost 20 . For example, DNA coding sequences are found to reduce nitrogen atoms per codon when nitrogen is scarce in the environment 41 , and a large proportion of bacterial genomes present a dual optimization strategy to reduce per-codon nitrogen demand while increasing translation efficiency 42 . Extending these findings, we show that type Ia pmoCAB coding sequences possess an adaptation that minimizes not only nitrogen but also carbon and hydrogen per transcribed codon, while reducing the metabolic burden of erroneous biosynthesis 25 through optimal translation efficiency and accuracy.…”

Codon usage bias creates a ramp of hydrogen bonding at the 5′-end in prokaryotic ORFeomes

“…Similar to translation, codon usage bias has been associated with transcriptional selection 15 and optimization of transcription efficiency 16 . Recent reports support the idea that codon variants also define the energy and cellular resources required for transcript biosynthesis [17][18][19][20] . However, in contrast to translation, the potential links between position-dependent codon usage bias at the 5 -end of CDSs and transcription optimization have yet to be investigated.…”

“…The commands used to compile the ORFeomes were "Latest RefSeq" and "Exclude anomalous". A smaller representative dataset of Bacteria ORFeomes was compiled based on a previously curated list that has even representation across phyla 18 . The Archaea and Fungi ORFeomes were relatively small so all were included in all analyses.…”

Protein Abundance Prediction Through Machine Learning Methods

“…Orange3 data mining software (35). Next, we calculated the selection on transcript translational efficiency (St) and selection on transcript biosynthetic cost (Sc), as described by Seward & Kelly (36,37). Subsequently, we performed the Akashi Test (38) to calculate the selection on translation accuracy using the software Seforta (39).…”

“…We selected codon usage metrics that are calculated individually for each gene. For each CDS, we calculated various codon usage metrics (Table S1) by using EMBOSS (40), CodonW (41), CAIcal (42), coRdon (43), stAIcalc (44), and scripts included in the original manuscripts, such as CodonMuSe (37) and iCUB (45). For codon usage metrics that require highly expressed genes as a reference, we employed the CDSs of the top 10% proteins with the highest overall median abundance.…”