Genome sequence of Blochmannia pennsylvanicus indicates parallel evolutionary trends among bacterial mutualists of insects

Abstract

The distinct lifestyle of obligately intracellular bacteria can alter fundamental forces that drive and constrain genome change. In this study, sequencing the 792-kb genome of Blochmannia pennsylvanicus, an obligate endosymbiont of Camponotus pennsylvanicus, enabled us to trace evolutionary changes that occurred in the context of a bacterial-ant association. Comparison to the genome of Blochmannia floridanus reveals differential loss of genes involved in cofactor biosynthesis, the composition and structure of the cell wall and membrane, gene regulation, and DNA replication. However, the two Blochmannia species show complete conservation in the order and strand orientation of shared genes. This finding of extreme stasis in genome architecture, also reported previously for the aphid endosymbiont Buchnera, suggests that genome stability characterizes long-term bacterial mutualists of insects and constrains their evolutionary potential. Genome-wide analyses of protein divergences reveal 10- to 50-fold faster amino acid substitution rates in Blochmannia compared to related bacteria. Despite these varying features of genome evolution, a striking correlation in the relative divergences of proteins indicates parallel functional constraints on gene functions across ecologically distinct bacterial groups. Furthermore, the increased rates of amino acid substitution and gene loss in Blochmannia have occurred in a lineage-specific fashion, which may reflect life history differences of their ant hosts.

Figure 1.

Circular map of the B. pennsylvanicus genome and genome features. The origin of replication was putatively set upstream of gidA, in accordance with the shift of GC skew. The concentric rings denote the following features: (1) numbered base pair coordinates beginning with base one of the gidA open reading frame (ORF); (2) GC skew as calculated by (G–C)/(G+C) using a 2.5-kb sliding window; (3) ORFS present on the leading strand (+); (4) ORFS present on the lagging strand (–); (5) pseudogenes of B. pennsylvanicus that are present in (orange) or absent from (black) B. floridanus; (6) ORFS of B. pennsylvanicus that are shared with (red), pseudogenes in (yellow), or absent in (dark blue) the genome of B. floridanus; (7) transfer RNAs (tRNAs) that are shared with (gray) or absent in (pink) B. floridanus; (8) ribosomal RNAs (rRNAs) 5S, 16S, and 23S (purple). Circular plot and GC skew analysis were generated by the JENA Prokaryotic Genome Viewer–Export Version v3.06.04.

Figure 2.

Comparison of B. pennsylvanicus and B. floridanus gene contents. Genes in the outer section of each circle are unique to one genome, while those listed in the intersection of the two circles are shared. Note that seven shared genes are apparently functional in one genome, but pseudogenes in the other. The truncation of B. floridanus dnaX (see text) is not included as a pseudogene here. The fusion of yidCD in B. pennsylvanicus is counted as two intact genes. (*) The 628 shared intact genes include genes that have a single frameshift within long poly(A) tracts but are otherwise intact (see text). These loci include B. pennsylvanicus hisH, ytfM, ubiF, and ybiS and B. floridanus ytfM, ybiS, and gmhB.

Figure 3.

Sequence context of homopolymeric frameshift mutations within five Blochmannia genes. The single frameshifts in Blochmannia genes (A) hisH, (B) ubiF, (C) ybiS, (D) ytfM, and (E) gmhB were “corrected” manually, based on comparisons with sequences that lack the frameshift. Genes were then aligned by their inferred amino acid sequence against E. coli. Underlined nucleotides indicate poly(A) and poly(T) tracts in which frameshifts occur. Regions highlighted in gray indicate the putative sites of transcriptional or translational slippage that may restore the proper reading frame. (•) Amino acid identical to the top sequence (B. pennsylvanicus).

Figure 4.

Distinct biosynthetic capabilities of B. pennsylvanicus and B. floridanus. The two mutualist genomes differ in genes encoded for the biosynthesis of (A) coenzyme A, and (B) isoprenoids. Gene names in gray are missing or pseudogenes (noted by ψ).

Figure 5.

Comparison of intergenic spacers (IGSs) in B. pennsylvanicus and B. floridanus. (A) A significant relationship in the lengths of homologous spacers implies a certain conservation of spacer length (Spearman's ρ = 0.7895; p < 0.0001), but B. pennsylvanicus IGSs are generally longer (average 290 bp) than those of B. floridanus (average 180 bp) (Wilcoxon Rank sum test, p < 0.0001). Homologous spacers were identified based on their positions in the genomes. IGSs that lacked a homolog were excluded. (B) B. pennsylvanicus IGSs (open circles) generally have lower %GC and shorter lengths than ORFs (filled squares). Horizontal lines mark the average %GC for ORFS (32.1%; solid line) and IGSs (20.0%; dashed line). Additionally the base composition of all homologous, nonzero IGSs were compared and show no strong correlation, yet cluster around ∼20% GC in both genomes (unpublished data).

Figure 6.

Correlated protein divergences suggest parallel selective constraints in endosymbionts and free-living bacteria. Protein divergence in Blochmannia shows a strong correlation with divergence at orthologous genes of (A) Buchnera (–A. pisum versus –S. graminum) and (B) E. coli versus P. luminescens. Lines that best fit the data and intercept zero are shown, but significance was tested using the nonparametric test of association (JMP 4.0; SAS Institute Inc.). For all bacterial pairs, the most conserved genes include certain chaperonins and translation functions (mostly ribosomal proteins), while the most divergent genes include surface structures. Substitution rates above 2 are prone to saturation but are included here for comparison. Supplemental Table S3 lists all of the pairwise divergences.

Figure 7.

Accelerated rates of evolution in the lineage leading to B. floridanus compared to B. pennsylvanicus, since their divergence from a common ancestor. Rates of protein divergence were compared using a relative rates test with E. coli as an outgroup (see Methods). The analysis included 516 proteins for which RSD identified orthologs in the two genomes, and for which divergences between the two Blochmannia or between either endosymbiont and E. coli did not exceed 2.0. On average, proteins evolved 1.88 times faster in B. floridanus compared to in B. pennsylvanicus. Only 50 of the proteins tested evolved more slowly in B. pennsylvanicus than in B. floridanus or at the same rate, with values ≤1on the histogram, while 467 genes evolved faster in B. floridanus. Supplemental Table S4 lists proteins with particularly accelerated evolutionary rates in B. floridanus.