Are substitution rates and RNA editing correlated?

Abstract

Background: RNA editing is a post-transcriptional process that, in seed plants, involves a cytosine to uracil change in messenger RNA, causing the translated protein to differ from that predicted by the DNA sequence. RNA editing occurs extensively in plant mitochondria, but large differences in editing frequencies are found in some groups. The underlying processes responsible for the distribution of edited sites are largely unknown, but gene function, substitution rate, and gene conversion have been proposed to influence editing frequencies.

Results: We studied five mitochondrial genes in the monocot order Alismatales, all showing marked differences in editing frequencies among taxa. A general tendency to lose edited sites was observed in all taxa, but this tendency was particularly strong in two clades, with most of the edited sites lost in parallel in two different areas of the phylogeny. This pattern is observed in at least four of the five genes analyzed. Except in the groups that show an unusually low editing frequency, the rate of C-to-T changes in edited sites was not significantly higher that in non-edited 3rd codon positions. This may indicate that selection is not actively removing edited sites in nine of the 12 families of the core Alismatales. In all genes but ccmB, a significant correlation was found between frequency of change in edited sites and synonymous substitution rate. In general, taxa with higher substitution rates tend to have fewer edited sites, as indicated by the phylogenetically independent correlation analyses. The elimination of edited sites in groups that lack or have reduced levels of editing could be a result of gene conversion involving a cDNA copy (retroprocessing). If so, this phenomenon could be relatively common in the Alismatales, and may have affected some groups recurrently. Indirect evidence of retroprocessing without a necessary correlation with substitution rate was found mostly in families Alismataceae and Hydrocharitaceae (e.g., groups that suffered a rapid elimination of all their edited sites, without a change in substitution rate).

Conclusions: The effects of substitution rate, selection, and/or gene conversion on the dynamics of edited sites in plant mitochondria remain poorly understood. Although we found an inverse correlation between substitution rate and editing frequency, this correlation is partially obscured by gene retroprocessing in lineages that have lost most of their edited sites. The presence of processed paralogs in plant mitochondria deserves further study, since most evidence of their occurrence is circumstantial.

Figure 1

Mapping of changes in edited sites of five mitochondrial genes. Mapping of edited sites of five mitochondrial genes (atp1, cob, nad5, ccmB, and mtt2) on one of the two most parsimonious trees recovered from the combined analysis of all genes (Acorus was removed from the figure to save space). Branch lengths reflect the number of changes. For each gene, losses of edited sites are indicated above the branch where the change occurred. Gains of edited sites are indicated in a box below the branch where the change was optimized. Colors represent different genes, according to the box at the right. Family names are indicated next to the taxon name, with the following abbreviations: ALI = Alismataceae, LIM = Limnocharitaceae, APO = Aponogetaceae, CYM = Cymodoceae, POS = Posidoniaceae, RUP = Ruppiaceae, ZOS = Zosteraceae, POT = Potamogetaceae, JUN = Juncaginaceae, SCH = Scheuchzeriaceae, and HYD = Hydrocharitaceae.

Figure 2

Synonymous substitution rates (d_S) for five mitochondrial genes in Alismatales. Synonymous substitution rates (d_S) calculated using the program HyPhy ver. 0.9 with the MG94×HKY85_3 × 4 codon model and using one of the two trees obtained by the parsimony analysis for the combined data set. All trees are drawn at the same scale. The outgroup was removed from the figure to save space. Clades marked as HH, EE, and ALB correspond to likely insertions of processed paralogs in nad5 (see text).

Figure 3

Changes in the distribution of edited sites of three mitochondrial genes. Squares represent losses of edited sites from one clade with respect to its sister clade and/or its most recent ancestor.