Genome-wide analyses of Geraniaceae plastid DNA reveal unprecedented patterns of increased nucleotide substitutions

Abstract

Angiosperm plastid genomes are generally conserved in gene content and order with rates of nucleotide substitutions for protein-coding genes lower than for nuclear protein-coding genes. A few groups have experienced genomic change, and extreme changes in gene content and order are found within the flowering plant family Geraniaceae. The complete plastid genome sequence of Pelargonium X hortorum (Geraniaceae) reveals the largest and most rearranged plastid genome identified to date. Highly elevated rates of sequence evolution in Geraniaceae mitochondrial genomes have been reported, but rates in Geraniaceae plastid genomes have not been characterized. Analysis of nucleotide substitution rates for 72 plastid genes for 47 angiosperm taxa, including nine Geraniaceae, show that values of dN are highly accelerated in ribosomal protein and RNA polymerase genes throughout the family. Furthermore, dN/dS is significantly elevated in the same two classes of plastid genes as well as in ATPase genes. A relatively high dN/dS ratio could be interpreted as evidence of two phenomena, namely positive or relaxed selection, neither of which is consistent with our current understanding of plastid genome evolution in photosynthetic plants. These analyses are the first to use protein-coding sequences from complete plastid genomes to characterize rates and patterns of sequence evolution for a broad sampling of photosynthetic angiosperms, and they reveal unprecedented accumulation of nucleotide substitutions in Geraniaceae. To explain these remarkable substitution patterns in the highly rearranged Geraniaceae plastid genomes, we propose a model of aberrant DNA repair coupled with altered gene expression.

Fig. 1.

ML tree of Geraniaceae taken from 72-gene, 47-taxa analysis (Fig. S1). All nodes had 100% ML and MP bootstrap support.

Fig. 2.

Boxplots of the number of (A) nonsynonymous (dN) and (B) synonymous (dS) substitutions for individual Geraniaceae genes or groups of genes (see Materials and Methods). Wilcoxon rank sum tests were used to show that values for Geraniaceae dN were significantly higher than for other angiosperms (P < 0.001), whereas values of dS were not significantly different (P = 0.085). The horizontal line is the mean value of dN or dS in Geraniaceae. For each gene group, the box represents values between quartiles, dotted lines extend to minimum and maximum values, outliers are shown as circles, and the thick, black lines show median values. Asterisks show values for gene groups that are statistically different (P < 0.05 after Bonferroni correction; data not shown) than the values for same gene group in other angiosperms. See Tables S4 and S5 for post hoc statistics.

Fig. 3.

Correlation between dN and dS for the Geraniaceae (red triangles) and other angiosperms (black circles). Spearman's rank correlation rho (r_S) is significant (P < 0.001) for both Geraniaceae and angiosperms. Correlation coefficients were compared by using Fisher's Z transformation to show that there is a stronger correlation between dN and dS in the other angiosperms compared with Geraniaceae (P < 0.001).

Fig. 4.

ML trees for the fastest evolving large subunit ribosomal (A and B), small subunit ribosomal (C and D), and photosystem I genes (E and F). Labeled trees are shown in Fig. S3. Branch length is defined as the number of nucleotide substitutions per codon. The Geraniaceae clade is highlighted within each tree to show locus- and lineage-specific rate acceleration. Asterisks show the branch leading to grasses, a group known for lineage-specific rate acceleration (7).

Fig. 5.

Plot of ranked log P-values for each gene from likelihood ratio tests (LRT). P-values were determined by comparing two models H₀ and H_A. Corrected P-values were ranked and plotted. The vertical line shows a P value cut-off, and genes appearing to the right of the vertical line have a significantly higher dN/dS in Geraniaceae relative to other angiosperms. Genes that encode subunits of a functional complex were grouped according to Matsuoka et al. (56) and Chang et al. (57).