Selection for Reducing Energy Cost of Protein Production Drives the GC Content and Amino Acid Composition Bias in Gene Transfer Agents

Abstract

Gene transfer agents (GTAs) are virus-like elements integrated into bacterial genomes, particularly, those of Alphaproteobacteria The GTAs can be induced under conditions of nutritional stress, incorporate random fragments of bacterial DNA into miniphage particles, lyse the host cells, and infect neighboring bacteria, thus enhancing horizontal gene transfer. We show that GTA genes evolve under conditions of pronounced positive selection for the reduction of the energy cost of protein production as shown by comparison of the amino acid compositions with those of both homologous viral genes and host genes. The energy saving in GTA genes is comparable to or even more pronounced than that in the genes encoding the most abundant, essential bacterial proteins. In cases in which viruses acquire genes from GTAs, the bias in amino acid composition disappears in the course of evolution, showing that reduction of the energy cost of protein production is an important factor of evolution of GTAs but not bacterial viruses. These findings strongly suggest that GTAs represent bacterial adaptations rather than selfish, virus-like elements. Because GTA production kills the host cell and does not propagate the GTA genome, it appears likely that the GTAs are retained in the course of evolution via kin or group selection. Therefore, we hypothesize that GTAs facilitate the survival of bacterial populations under energy-limiting conditions through the spread of metabolic and transport capabilities via horizontal gene transfer and increases in nutrient availability resulting from the altruistic suicide of GTA-producing cells.IMPORTANCE Kin selection and group selection remain controversial topics in evolutionary biology. We argue that these types of selection are likely to operate in bacterial populations by showing that bacterial gene transfer agents (GTAs), but not related viruses, evolve under conditions of positive selection for the reduction of the energy cost of GTA particle production. We hypothesize that GTAs are dedicated devices mediating the survival of bacteria under conditions of nutrient limitation. The benefits conferred by GTAs under nutritional stress conditions appear to include horizontal dissemination of genes that could provide bacteria with enhanced capabilities for nutrient utilization and increases of nutrient availability occurring through the lysis of GTA-producing bacteria.

Keywords: GTA; alphaproteobacteria; bacteriophages; energy saving; gene transfer agents; metabolic efficiency; nutrient depletion; positive selection; virus exaptation.

FIG 1

GC1, GC2, and GC3 content of GTA regions and their immediate neighborhoods and all protein-coding genes in 212 alphaproteobacterial genomes. The neighborhoods immediately upstream and downstream of a GTA region consist of 17 genes each. Box plots represent median values bounded by the first and third quartiles. Whiskers show the values that lie in the range of the 1.5 × interquartile rule. Dots outside the whiskers represent the outliers.

FIG 2

Carbon content (A) and biosynthetic cost (B) of proteins encoded by GTA genes in 212 alphaproteobacterial genomes and their viral homologs. Box plots represent median values that are bounded by the first and third quartiles. Whiskers show the values that lie in the range of the 1.5 × interquartile rule. Dots outside the whiskers represent the outliers. The number of data points in each box plot is listed in Table S1.

FIG 3

The number of carbons (A) and number of high-energy phosphates (B) in proteins encoded by all protein-coding genes in 212 genomes, by highly expressed genes, and by GTA genes. Box plots represent median values that are bounded by the first and third quartiles. Whiskers show the values that lie in the range of the 1.5 × interquartile rule. Dots outside the whiskers represent outliers.

FIG 4

Carbon content of GTA proteins for four orders of the class Alphaproteobacteria. For each GTA protein, the heat map visualizes the number of carbons per side chain in amino acid averaged across taxonomic order. The numbers are shown either as raw values (A) or as values normalized by the carbon content of proteins encoded by 26 single-copy genes (B). The asterisks mark GTA proteins with significantly lower numbers of carbons per amino acid in the Sphingomonadales order than in the other three orders combined (α of 0.01; Mann-Whitney U test, all P values < 0.01). Box plots summarize the distribution of carbon content within each alphaproteobacterial order averaged across the examined GTA genes. Median values are bounded by the first and third quartiles. Whiskers show the values that lie in the range of the 1.5 × interquartile rule, and dots outside the whiskers represent the outliers. The phylogenetic tree is the reference alphaproteobacterial phylogeny (see Materials and Methods for details), in which branches are collapsed at the taxonomic rank of order. Numbers at the tree nodes represent bootstrap support values. Scale bar, number of substitutions per site.