Comparison of insertional RNA editing in Myxomycetes

Abstract

RNA editing describes the process in which individual or short stretches of nucleotides in a messenger or structural RNA are inserted, deleted, or substituted. A high level of RNA editing has been observed in the mitochondrial genome of Physarum polycephalum. The most frequent editing type in Physarum is the insertion of individual Cs. RNA editing is extremely accurate in Physarum; however, little is known about its mechanism. Here, we demonstrate how analyzing two organisms from the Myxomycetes, namely Physarum polycephalum and Didymium iridis, allows us to test hypotheses about the editing mechanism that can not be tested from a single organism alone. First, we show that using the recently determined full transcriptome information of Physarum dramatically improves the accuracy of computational editing site prediction in Didymium. We use this approach to predict genes in the mitochondrial genome of Didymium and identify six new edited genes as well as one new gene that appears unedited. Next we investigate sequence conservation in the vicinity of editing sites between the two organisms in order to identify sites that harbor the information for the location of editing sites based on increased conservation. Our results imply that the information contained within only nine or ten nucleotides on either side of the editing site (a distance previously suggested through experiments) is not enough to locate the editing sites. Finally, we show that the codon position bias in C insertional RNA editing of these two organisms is correlated with the selection pressure on the respective genes thereby directly testing an evolutionary theory on the origin of this codon bias. Beyond revealing interesting properties of insertional RNA editing in Myxomycetes, our work suggests possible approaches to be used when finding sequence motifs for any biological process fails.

Figure 1. Accuracy of different prediction methods of insertional RNA editing sites in Didymium.

Each graph shows the percentage of editing sites which are correctly predicted, predicted by one, two, or at least three positions away from the experimentally known correct editing site. (a) shows results for all 15 genes studied, (b) for the more conserved genes, and (c) for the less conserved genes.

Figure 2. Graphical representation of the positions of predicted editing sites.

These predicted editing sites are in the seven newly identified Didymium mitochondrial genes as well as in nad3 for which it is experimentally known that it is unedited in Didymium . The predictions for the nad2 gene are incomplete due to a lack of genomic sequence, indicated by the dashed lines for that gene.

Figure 3. Comparison of observed and expected conservation.

The observed conservation and background conservation for all 16 genes are compared for editing site at the (a) first and (b) third codon position.

Figure 4. -values for the differences between the observed and the background conservation.

These -values are calculated for shared editing sites in all 16 genes at the (a) first and (b) third codon position. The threshold for statistical significance ( as the -value cut off) is not indicated in the figure as it is far above the top of the graphs.

Figure 5. Relationship between codon bias () and the conservation at the second codon position.

and are the number of second and third codon position editing sites. Based on the conservation at the second codon position, the genes are separated into (a) four groups and (b) two groups. For the case of 16 known genes, we counted all unambiguous C insertional editing sites in Physarum and Didymium). For the case of 16 known genes + 8 genes in Physarum, we counted all unambiguous C insertional editing sites in Physarum and Didymium for the 16 known genes and unambiguous C insertional editing sites only in Physarum for the additional 8 genes. For 16 known genes + 8 genes in Physarum and Didymium, we counted all unambiguous C insertional editing sites in Physarum and Didymium for all 24 genes.

Figure 6. Comparison of (a) overall conservation and (b) codon bias for real and predicted mRNA sequences.

The data is close to the diagonal in both cases indicating that predicted sequences can be used to estimate these quanitities in cases where the true sequences are not known.