Plastid Transcript Editing across Dinoflagellate Lineages Shows Lineage-Specific Application but Conserved Trends

Abstract

Dinoflagellates are a group of unicellular protists with immense ecological and evolutionary significance and cell biological diversity. Of the photosynthetic dinoflagellates, the majority possess a plastid containing the pigment peridinin, whereas some lineages have replaced this plastid by serial endosymbiosis with plastids of distinct evolutionary affiliations, including a fucoxanthin pigment-containing plastid of haptophyte origin. Previous studies have described the presence of widespread substitutional RNA editing in peridinin and fucoxanthin plastid genes. Because reports of this process have been limited to manual assessment of individual lineages, global trends concerning this RNA editing and its effect on the biological function of the plastid are largely unknown. Using novel bioinformatic methods, we examine the dynamics and evolution of RNA editing over a large multispecies data set of dinoflagellates, including novel sequence data from the peridinin dinoflagellate Pyrocystis lunula and the fucoxanthin dinoflagellate Karenia mikimotoi. We demonstrate that while most individual RNA editing events in dinoflagellate plastids are restricted to single species, global patterns, and functional consequences of editing are broadly conserved. We find that editing is biased toward specific codon positions and regions of genes, and generally corrects otherwise deleterious changes in the genome prior to translation, though this effect is more prevalent in peridinin than fucoxanthin lineages. Our results support a model for promiscuous editing application subsequently shaped by purifying selection, and suggest the presence of an underlying editing mechanism transferred from the peridinin-containing ancestor into fucoxanthin plastids postendosymbiosis, with remarkably conserved functional consequences in the new lineage.

Fig. 1.

—Summary of plastid transcript editing in dinoflagellates. This diagram denotes the relationship between taxa under study, plastid affiliation, and presence of plastid biological features including minicircles, transcript polyuridylation, and transcript editing. Symbols are denoted in figure inset. Dashed line surrounding editing symbol between Amphidinium spp. and fucoxanthin lineages represents uncertainty regarding editing in basal peridinin dinoflagellates and related taxa. Placement of dashed boxes for both C-to-G and G-to-U base conversions in two places represents two alternate evolutionary hypotheses, as discussed in the main text. The single polytomy represents uncertainty in peridinin dinoflagellate branching order.

Fig. 2.

—Effect of editing on GC content. This figure shows the inherent GC content bias among codon positions in dinoflagellates, with higher GC content in positions one and two, and the effect of editing to increase GC content significantly in positions one and two, but not three. Bars are color-coded by codon position. Net GC increase defined as the difference between GC-enriching (i.e., A or T to G or C) and GC-depleting edits. Error bars denote 95% confidence intervals.

Fig. 3.

—Editing events occur in close proximity to each other. This figure shows the significant propensity of edits to occur in the same, or adjacent, codon as other edits. For each codon position, the number of edits expected to occur in the same or adjacent codon was determined based on a random distribution of edits with the same codon position preferences as observed in real data. Comparison of the actual distribution to the expected distribution showed highly significant clustering when considering edits in all three positions. Error bars represent SD of the simulation (n=100).

Fig. 4.

—Effect of editing on protein size and hydrophobicity. This figure shows scatter plots between molecular weight difference and GRAVY score before and after editing for fucoxanthin and peridinin dinoflagellate plastid sequences (A), and peridinin dinoflagellate data sets, Pyrocystis lunula and Symbiodinium minutum (B). Spearman’s rank correlation tests (rho) are reported along with their significance. The trend previously reported for S. minutum, that editing results in proteins that have lower molecular weight and are more hydrophobic (Mungpakdee et al. 2014), is observed here, but not in any other data set.

Fig. 5.

—Sliding window editing analysis and protein sequence conservation. This figure outlines the rationale behind correlating local editing rate to sequence conservation with reference sequences, and the overall effect that editing associates with more divergent regions of genes. (A) Exemplar graph showing the result of correlative sliding window analysis in the Karlodinium veneficum psaB gene. The x axis denotes the position along the gene, whereas the left hand axis denotes the percentage of edited residues in a given window, and is indicated by a blue line. The right hand axis denotes the percentage identity to a reference sequence (Emiliania huxleyi) for both the amino acid sequence (green line) and the nucleotide sequence (magenta line) in a given window. The horizontal dotted line denotes the average editing rate across the entire gene sequence. (B–E) Boxplots quantifying sliding window analysis, as in (A), as the Pearson’s correlation coefficient between per window amino acid similarity and editing rate across the whole sequence. Separate plots illustrate these values between reference sequences (B), between plastid types (C), by gene family (D), and by organism (E). Correlations with P value >0.05 were not included; n denotes sample size (number of genes). Significance is denoted: *P value<0.05, **P value<0.01, ***P value<0.001. Abbreviations: Ac, Amphidinium spp.; Ct, Chrysochromulina tobin; Eh, Emiliania huxleyi; Pt, Phaeodactylum tricornutum; Vb, Vitrella brassicaformis; Km, Karenia mikimotoi; Kv, Karlodinium veneficum; Pl, Pyrocystis lunula; Sm, Symbiodinium minutum.

Fig. 6.

—Positional entropy analysis of sequences. This figure shows that, although orthologues from all lineages under study are overall well-conserved compared with plastid sequences without editing, amino acids changed by RNA editing are present in significantly less well conserved positions than those unchanged by RNA editing. Separate boxplots show positional entropy scores of all sites between plastid types (A) and by organism (B), and between edited and nonedited sites (C). Positions for which residues were absent in more than half of reference sequences were not scored; n denotes sample size (number of positions). Significance is denoted: *P value<0.05, **P value<0.01, ***P value<0.001. Abbreviations: Km, Karenia mikimotoi; Kv, Karlodinium veneficum; Pl, Pyrocystis lunula; Sm, Symbiodinium minutum.

Fig. 7.

—Summary of corrective editing. This figure highlights the overall corrective effect of editing to result in amino acids that are more similar to their homologues in orthologues from organisms that do not undergo RNA editing than those originally encoded in the genomic sequence. (A) Scatter plot of the relationship between the number of editing events and the average editing score for each gene. (B) Boxplots showing that the corrective effect is independent of choice of reference organism; n denotes sample size (number of events). Abbreviations: Ac, Amphidinium spp.; Ct, Chrysochromulina tobin; Eh, Emiliania huxleyi; Pt, Phaeodactylum tricornutum; Vb, Vitrella brassicaformis; Km, Karenia mikimotoi; Kv, Karlodinium veneficum; Pl, Pyrocystis lunula; Sm, Symbiodinium minutum.

Fig. 8.

—Editing functions vary across genes and lineages. This figure shows how, despite that editing is corrective overall, this corrective effect varies across organisms and that noncorrective events occur in variable and conserved positions. Separate boxplots show editing scores of all sites between genes (A), plastid types (B), and by organism (C); n denotes sample size (number of events). (D) Scatter plot showing the relationship between positional entropy of edited positions and the effect of editing events, as assessed by average editing score. Lines represent density of values for clarity. Note not only that the plot is skewed toward positive editing score values but also that a number of noncorrective events exist, including in highly (i.e., entropy > 0.90) conserved positions. (E) Plot as in (D), but focussing on individual organisms. Significance is denoted: *P value < 0.05, **P value < 0.01, ***P value < 0.001. Abbreviations: Km, Karenia mikimotoi; Kv, Karlodinium veneficum; Pl, Pyrocystis lunula; Sm, Symbiodinium minutum.