Widespread and Extensive Editing of MitochondrialmRNAS in Dinoflagellates

Senjie Lin1*†, Huan Zhang1†, David F. Spencer2, John E. Norman2 andMichael W. Gray2

1Department of MarineSciences, University ofConnecticut, GrotonCT 06340, USA

2Program in EvolutionaryBiology, Canadian Institute forAdvanced ResearchDepartment of Biochemistryand Molecular BiologyDalhousie University, HalifaxNS B3H 4H7, Canada

We report evidence of extensive substitutional editing of mitochondrialmRNAs in the dinoflagellate species Pfiesteria piscicida, Prorocentrumminimum and Crypthecodinium cohnii, based on a comparison of genomicand corresponding cDNA sequences determined for two mitochondrialDNA-encoded genes, cox1 (cytochrome oxidase subunit 1) and cob(apocytochrome b ). In the cox1 mRNA, we identify 72 substitutions at 40sites in 39 codons, whereas in cob mRNA, we infer 86 editing events at 51sites in 48 codons. Editing, which takes place in distinct clusters, changes,2% of the total sequence, occurs predominantly at first and secondpositions of codons, and involves mostly (but not exclusively) A ! G(47%), U ! C (23%) and C ! U (17%) substitutions. In all but four of the158 cases, editing changes the identity of the specified amino acid. At 21(cox1 ) and 26 (cob ) sites, the same nucleotide change is observed at thesame position in at least two of the species investigated. At about one-third of the sites, editing results in an amino acid change that increasessimilarity between the dinoflagellate Cox1 and Cob sequences and theirhomologs in other organisms; presumably editing at these sites is of par-ticular functional significance. Overall, about half of the editing eventseither maintain or increase similarity between the dinoflagellate proteinsequences and their non-dinoflagellate homologs, while a further one-third of the alterations are “dinoflagellate-specific” (i.e. they involve achange to an amino acid residue selectively conserved in at least two ofthe dinoflagellate species at a given position). The nature, pattern andphylogenetic distribution of the inferred edits implies either that morethan one type of previously described editing process operates on agiven transcript in dinoflagellate mitochondria, or that a mechanisticallyunique type of mitochondrial mRNA editing has evolved within the dino-flagellate lineage.

q 2002 Elsevier Science Ltd. All rights reserved

Keywords: dinoflagellate; RNA editing; Pfiesteria piscicida; Prorocentrumminimum; Crypthecodinium cohnii*Corresponding author

Introduction

Dinoflagellates (phylum Dinozoa1) are amorphologically and ecologically diverse andimportant group of unicellular eukaryotes(protists) consisting of both photosynthetic and

non-photosynthetic members.2 These organismsare among the most important marine primaryproducers, are essential for the growth of coralreefs, form red tides, and elaborate toxins. Dino-flagellates constitute a sister group toapicomplexans,3 a phylum of obligate parasites(Sporozoa1) that includes Plasmodium, the causativeagent of malaria. Together with the ciliates(phylum Ciliophora1), dinoflagellates and api-complexans comprise a monophyletic lineage,3 thealveolates (Alveolata1).

Compared to the situation in apicomplexans4,5

and ciliates,6 – 8 relatively little is currently knownabout the overall size, physical form, organization

0022-2836/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved

† These authors contributed equally to this work.

E-mail address of the corresponding author:senjie.lin@uconn.edu

Abbreviations used: aa, amino acid; bp, base-pair; Cco,Crypthecodinium cohnii; nt, nucleotide; ORF, open readingframe; PCR, polymerase chain reaction; Pmi,Prorocentrum minimum; Ppi, Pfiesteria piscicida; PSU,practical salinity unit; RT, reverse transcription.

doi:10.1016/S0022-2836(02)00468-0 available online at http://www.idealibrary.com onBw

J. Mol. Biol. (2002) 320, 727–739

Figure 1. Alignment of deduced cytochrome c oxidase subunit I (Cox1) sequences. Pfiesteria-g, sequence deducedfrom genomic sequence of P. piscicida cox1 (acc. no. AF463412); Pfiesteria-c, sequence deduced from cDNA of P. piscicidacox1 (AF463413; only residues different from Pfiesteria-g are shown); Prorocentrum-g, sequence deduced fromP. minimum cox1 (AF463414); Prorocentrum-c, sequence deduced from cDNA of P. minimum cox1 (AF463415; only

728 Mitochondrial mRNA Editing in Dinoflagellates

and gene content of the dinoflagellate mito-chondrial genome, primarily because of the con-siderable technical difficulties that have impededattempts to isolate organelles and organelle DNAfrom these protists. Only two dinoflagellatemitochondrial genes (cox1, encoding subunit 1 ofcytochrome c oxidase, and cob, specifying apocyto-chrome b ) have been sequenced so far, from thedinoflagellates Crypthecodinium cohnii9– 12 andPfiesteria piscicida.13 In C. cohnii, the cox1 codingsequence is flanked by long stretches of non-coding DNA populated by a complex array ofsmall repetitive elements.11,12 The cob coding regionin P. piscicida is similarly flanked by varyinglengths of non-coding DNA in the three copiesdetected.13 The organization of these genes is verydifferent from that described for cox1 and cob in themitochondrial genome of both apicomplexans4,5 andciliates,6–8 raising many questions about theevolution of mtDNA within the alveolates.

We have continued to investigate the structureand expression of cox1 and cob in C. cohnii andP. piscicida as well as in a third dinoflagellate,Prorocentrum minimum. These three species exhibitvery different lifestyles. P. piscicida is a hetero-

trophic plankton that putatively secretes a potenttoxin associated with major fish kills in estuariesas well as with neurocognitive deficits in humans;as such, it has attracted considerable interestrecently.14 – 16 P. minimum and C. cohnii, which arealso potentially harmful, represent two other eco-logical niches, planktonic photoautotroph andbenthic heterotroph, respectively. Our studieshave led to the discovery of extensive post-transcriptional re-tailoring (RNA editing) ofmRNAs in dinoflagellate mitochondria.

RNA editing, particularly prominent in mito-chondria, is recognized by changes in an RNAsequence compared to that of its encoding DNA.Known RNA editing systems, which appearmechanistically and evolutionarily unrelated,17 areof two general types: insertional/deletional, inwhich nucleotides are added to (and in somecases also removed from) a transcript,18 – 20 and sub-stitutional, where one kind of nucleotide (nt) isreplaced by another at a given location.21 – 24 Wedocument here a novel type of substitutionalmRNA editing that operates in dinoflagellate mito-chondria and that appears to be widely distributedwithin this protist lineage.

residues different from Prorocentrum-g are shown); Crypthecodinium-g, sequence deduced from C. cohnii cox1(AF186994); Crypthecodinium-c, sequence deduced from cDNA of C. cohnii cox1 (AF487783; only differences withCrypthecodinium-g are indicated). Possible editing within the extreme N and the C-terminal portions of the cox1ORFs of P. piscicida, P. minimum and C. cohnii was not assessed in these experiments. Lowercase letters in sequencealignments indicate primer target regions. Non-dinoflagellate sequences are: Plasmodium, Plasmodium vivax (apicom-plexan, Y17721); Rhodomonas, Rhodomonas salina (cryptophyte alga, AF288090); Prototheca, Prototheca wickerhamii(green alga, U02970); Allomyces, Allomyces macrogynus (fungus, U41288); Homo, Homo sapiens (human, NC_001807);Bradyrhizobium, Bradyrhizobium japonicum (a-proteobacterium, AJ242592). At editing sites, red denotes a predictedchange to a highly conserved consensus (identical residue in at least five of the compared sequences); green indicatesa conservative replacement (i.e. to a residue that is chemically similar to the consensus); blue denotes a change to aresidue that is specifically conserved in at least two of the dinoflagellate species. At non-edited sites, a colon or aperiod indicates a high level of sequence identity, either 100% (:) or .75% (.), respectively; a caret (^) denotes con-servative replacement in all compared sequences at this position. Six invariant His residues that are ligands for hemea, CuB and heme a3 are indicated by #. X marks the position of an unedited UGA in the P. minimum cox1 mRNA(aa residue 284). Numbers above edited residues denote editing sites as listed in Table 2 (first column). Overliningdelineates three well-defined clusters of editing sites (I to III) discussed in the text (see Figure 4).

Table 1. Primers used for amplification of cox1 and cob sequences

Primer name Sequences (50 –30) Application

PPCOBF TTGATTCCAAATTTTTCTTTCTATTG P. piscicida cob forwardPPCOBR TTTTTTGGGAACAAGATACACATAGG P. piscicida cob reversePPCOB1F ATGGGTTATGTCTTACCTTTTGG P. piscicida cob forwardPPCOBcDNA1 CTCCAAAGAGTAAGAGAATGAG P. piscicida cob cDNA reversePPCOB2F TTCATTATCAATTCCCATTTCTTG P. piscicida cob common forwardPPCOB1R CTTGAAGGAATTGATAACCTATCC P. piscicida cob common reversePMCOBF ATGAAATCTCATTTACAAACATATCC P. minimum cob forwardPMCOBR CTTGAGGGAATTGAGCACCTATCC P. minimum cob reverseDINOCOX1F AAAAATTGTAATCATAAACGCTTAGG Dinoflagellate cox1 forwardDINOCOX1R TGTTGAGCCACCTATAGTAAACATTA Dinoflagellate cox1 reverseCOB-L1 ATGAAATCTCATTTACATAC C. cohnii cob forwardCOB-L2 TTACATACATATCCTTGTCC C. cohnii cob forwardCOB-R1 AGATATAAACATCTCTTGAG C. cohnii cob reverseCOB-R2 CATCTCTTGAGGTAATTGTG C. cohnii cob reverseCOX-L1 GCCAGACATTTGTTTGTTGG C. cohnii cox1 forwardCOX-L2 GTTTGTTGGACTTTATGTCC C. cohnii cox1 forwardCOX-R3 GTTATTCCTGATCCAATAGATG C. cohnii cox1 reverseCOX-R4 GTTCCAATAGATGACAGAAAATTCC C. cohnii cox1 reverse

Mitochondrial mRNA Editing in Dinoflagellates 729

Results

Comparison of gene and cDNA sequences

The genomic and cDNA sequences reported here(Figures 1 and 2) allowed us to survey almost all ofthe cox1 and cob coding regions in the three dino-

flagellate species studied. No indels had to beassumed in aligning the dinoflagellate cox1 andcob gene sequences except at the extremeC-terminal end. Gene and corresponding cDNAsequences were completely colinear.

From the P. piscicida gene and cDNA sequences,primers specific for the cob cDNA sequence were

Figure 2. Alignment of deduced cytochrome b (Cob) sequences. Pfiesteria-g, sequence inferred from Ppi-g (acc. no.AF357519); Pfiesteria-c, sequence based on Ppi-c (AF357518; only residues different from Pfiesteria-g are shown); Pro-rocentrum-g, sequence deduced from Pmi-g (AY030286); Prorocentrum-c, sequence deduced from Pmi-c (AY030285;only residues different from Prorocentrum-g are shown); Crypthecodinium-g, sequence deduced from Cco-g(AF403220); Crypthecodinium-c, sequence based on Cco-c (AF403221; only differences with Crypthecodinium-g areindicated). Note that the Cco-g reading frame is open for an additional 82 amino acid residues upstream of theindicated translation start (which is actually corresponds to the third N-terminal AUG in the C. cohnii cob ORF).Possible editing within the first 82 codons of the Cco Cob ORF was not examined here. Non-dinoflagellate sequencesare: Plasmodium, Plasmodium vivax (apicomplexan, Y17721); Venturia, Venturia inaequalis (fungus, AF004559); Homo,Homo sapiens (human, J01415); Rhodomonas, Rhodomonas salina (cryptophyte alga, AF288090); Prototheca, Protothecawickerhamii (green alga, U02970); Bradyrhizobium, Bradyrhizobium japonicum (a-proteobacterium, J03176). Color codingat editing sites and symbols are as indicated in the legend to Figure 1. Four invariant His residues that are ligands forthe heme b group in apocytochrome b are indicated by #. Numbers above edited residues denote editing sites as listedin Table 4 (first column). Overlining delineates four well-defined clusters of editing sites (I to IV) discussed in the text(see Figure 4).

730 Mitochondrial mRNA Editing in Dinoflagellates

designed (Table 1) and polymerase chain reaction(PCR) was performed using P. piscicida cDNA andgenomic DNA as templates (Figure 3). No PCRproduct was generated from the genomic template,confirming that in P. piscicida, the cob cDNAoriginates from transcribed sequences and doesnot represent a second (mitochondrial or nuclear)copy of cob that might have been derivedevolutionarily from an edited mRNA, as in thecase of the nucleus-encoded cox2 gene inlegumes.25

Base substitutions constituted the only differ-ences observed between gene and corresponding

cDNA sequences, with such comparisons revealingnumerous reproducible single-nt changes (bothtransitions and transversions), indicative of anovel type of RNA editing. The distribution andnature of these substitutions is summarized inFigure 4. In total, we infer 72 substitutions at 40sites in 39 codons in cox1 mRNA and 86 editingevents at 51 sites in 48 codons in cob mRNA.Notably, editing sites are not randomly distributedwithin the coding regions but are found in three(cox1 ) or four (cob ) distinct clusters (I–IV, Figure4), separated by relatively long stretches ofsequence that are devoid of editing sites in allthree species. In the cox1 case, a prominent non-edited stretch extends for 73 codons between edit-ing clusters II and III (Figure 4). In the cob case,clusters II (33 codons; ten editing sites) and III (30codons; nine editing sites) are flanked by stretchesof 30 codons (upstream of II), 37 codons (betweenII and III) and 31 codons (downstream of III) thatcontain no editing sites in any of the three species.

Inferred editing of dinoflagellate cox1 mRNA

Table 2 summarizes the substitutions foundbetween cox1 gene and cDNA sequences for thethree dinoflagellate species. Within a 1.0-kb regionrepresenting codons 27 to 363 of the 536-codon11

C. cohnii cox1 ORF, potential edits could beassessed in all three species (Figure 1). A total of23 (P. piscicida ), 25 (P. minimum ) and 18 (C. cohnii )editing events were identified within this region at34 different sites, indicating a similar density of

Figure 3. PCR results showing that Pfiesteria cobcDNA-specific primers (PPCOB1F, PPCOBcDNA1) didnot produce any amplicon from genomic DNA. Thequality of DNA used was verified by controls in whichPCR products were generated by P. piscicida commoncob primers (primers designed to amplify both genomic(G) and cDNA (C) cob; PPCOB2F, PPCOB1R) as well as18 S rDNA primers.13 Genomic DNA (G) and cDNA (C)from two P. piscicida strains (1, CCMP1831; 2, NCSU113-3) were used as templates with each set of primers. M,100-bp DNA ladder. See Table 1 for primer sequencesand Materials and Methods for more details.

Figure 4. Distribution of editing sites within the cox1 and cob mRNA sequences from P. piscicida (Ppi), P. minimum(Pmi) and C. cohnii (Cco). Linear length is proportional to the extent of sequence information obtained (cob sequenceis essentially complete for Ppi and Cco; cox1 sequence is complete for Cco). Roman numerals (I–IV) with braces denoteclustered editing sites. Vertical colored bars indicate type of edit: C ! U, blue; A ! G, lime, G ! C, aqua; U ! C, red;G ! A, pink; U ! A brown; U ! G, black. Symbols: p , UGA codon edited to sense codon; X, unedited UGA codon;open circle, silent third-position editing (no change in predicted aa). Sites at which editing improves similarity betweendinoflagellate and non-dinoflagellate species are designated by either a filled circle (editing to aa identical to consen-sus) or a pentagon (editing to aa chemically similar to consensus). A triangle denotes dinoflagellate-specific editing(see the text).

Mitochondrial mRNA Editing in Dinoflagellates 731

editing (1.8–2.5% of nucleotides substituted) in thethree species. Additional upstream and down-stream sequence was examined in the case ofC. cohnii, revealing a further six editing events.

More than 80% of the observed cox1 changescomprise just two types, A ! G and U ! C tran-sitions, with lesser proportions of C ! U andG ! C (Table 3). Both U ! C and C ! Usubstitutions have previously been reported inother cases of mitochondrial mRNA editing,19,21

but A ! I editing (operationally equivalent to anA ! G change) has so far been documented only

in mRNAs that are nucleus-encoded.23,24 G ! Ctransversions have not previously been reportedfor any editing system.17 At one site (35), the sameG ! C change is seen in all three species, whereasat another site (21), this edit occurs in bothP. piscicida and P. minimum (Table 2). That the samechange occurs at the same site in two or all threespecies strongly supports the authenticity of theinferred minor edits.

Although many of these editing events areunique to a single species (W, Table 2), identicalediting (same nt change at the same position,

Table 2. Inferred editing of cox1 mRNA in dinoflagellate mitochondria

Site P. piscicida P. minimum C. cohnii 1 2 3

1 Gua ! Cua V ! L W – –2 Ggc ! Cgc G ! R W – –3 Auc ! Guc I ! V Auc ! Guc I ! V Auu I X – –4 auA ! auG I ! M auA ! auG I ! M auA I – – X5 Aua ! Gua I ! V Aua ! Gua I ! V Aua ! Gua I ! V B – –6 Uuu ! Cuu F ! L Uuu ! Cuu F ! L Uuu ! Cuu F ! L B – –7 Uuu F Uuu ! Cuu F ! L Uuu F W – –

8 Guu V L Auu ! Guu I ! V W – –

9 Uuu F Uuu ! Cuu F ! L Uuu F W – –10 uCu ! uUu S ! F uUu F uCu ! uUu S ! F – X –11 Uuc ! Cuc F ! L Uuc ! Cuc F ! L Uua ! Cua L ! La B – –12 gGa G gGa ! gCa G ! A gGa G – W –13 Uuu F Uuu ! Cuu F ! L Uuu F W – –14 Auc ! Guc I ! V Auc ! Guc I ! V Aua ! Gua I ! V B – –

15 ucA S ucA ! ucG S ! Sa S – – W

16a Aca ! Gua T ! V Aca ! Gua T ! V Aua ! Gua I ! V B – –16b aCa ! gUa T ! V aCa ! gUa T ! V aUa gUa I V – X –17 uAu ! uGu Y ! C uAu ! uGu Y ! C gGa G – X –18 auG M auG M auA ! auG I ! M – – W19 ugG W ugA ! ugG p ! Wb ugG W – – W20 gGa ! gCa G ! A gCa A uCu S – W –21 gGa ! gCa G ! A gGa ! gCa G ! A gGa G – X –22 Auc ! Guc I ! V Auc ! Guc I ! V Aua ! Gua I ! V B – –

23 S S uAu ! uGu Y ! C – W –

24 Auc ! Guc I ! V Aug ! Gug M ! V Aua I X – –25 Aua ! Gua I ! V Aua ! Gua I ! V Aua ! Gua I ! V B – –26 Uuc ! Cuc F ! L Uuc ! Cuc F ! L Uuc ! Cuc F ! L B – –27 Auc ! Guc I ! V Auc I Aua ! Gua I ! V X – –28 Aua I Aua I Aua ! Gua I ! V W – –29 Auu I Auu ! Guu I ! V Auu I W – –30 uUa ! uCa L ! S uUa ! uCa L ! S uUa ! uCa L ! S – B –31 aAa ! aGa K ! R aAa ! aGa K ! R aAa ! aGa K ! R – B –32 aAu ! aGu N ! S aGu S aAu ! aGu N ! S – X –

33 Cca ! Uca P ! S A Uca S W – –

34 Uuc ! Cuc F ! L Uuc ! Cuc F ! L Uua L X – –35 gGa ! gCa G ! A gGu ! gCu G ! A gGa ! gCa G ! A – B –36 Aua ! Gua I ! V W – –37 Aua ! Gua I ! V W – –38 aAu ! aGu N ! S – W –39 Auu ! Guu I ! V W – –

Codon and predicted aa changes are inferred from differences between gene and corresponding cDNA sequences. Editing sitenucleotides are set in uppercase letters (e.g. site 8, Auu ! Guu in a first-position A ! G edit) except where the editing site nucleotidediffers from both the “pre-edited” and “edited” states (A and G, respectively, in this example). In such cases, the variant nucleotide is

shown in a lowercase letter and is boxed and shaded (e.g. in P. minimum in the example given). Codon 16 contains two different

editing sites (a and b). Amino acids set in italics differ at a given editing site between dinoflagellates prior to editing but yield thesame predicted aa after editing. Blanks at editing sites represent missing data for that species. Symbols in the final column indicatethe degree of conservation of editing: B, same nucleotide and aa change in all three dinoflagellate species; X, both nucleotide and aachange conserved in two species; W, unique nucleotide and aa change in one species. Vertical positioning of each symbol under thenumerals 1, 2 and 3 indicates where within the codon (first, second or third position) editing occurs.

a Silent editing (no aa change).b Editing of a potential termination codon (UGA ¼ p ) to a sense (UGG) codon.

732 Mitochondrial mRNA Editing in Dinoflagellates

yielding the same amino acid (aa) change) wasobserved in two of the species at ten sites (X,Table 2), most often between P. piscicida and P. mini-mum. A further 11 edits are found at the same sitein all three species (B, Table 2). These observationsindicate a strong tendency toward conservation ofediting sites during evolutionary diversification ofcox1 in dinoflagellates.

In both P. piscicida and P. minimum, two editingsites (16a, 16b) occur immediately adjacent to oneanother in the same codon (Table 2). C. cohniishares editing site 16a, but at site 16b the latteralready contains the nucleotide (U) that is gener-ated by editing in the other two species.

Inferred editing of dinoflagellate cob mRNA

A pattern of substitutions very similar to thatsummarized above for cox1 is revealed by compari-sons of cob gene and cDNA sequences obtainedfrom the same three dinoflagellate species (Figure2, Table 4). In this instance, regions encompassing370 codons (P. piscicida ), 328 codons (P. minimum )and 332 codons (C. cohnii ) of cob sequence wereanalyzed at both the gene and cDNA levels. Withinthe region for which sequence from all three dino-flagellates is available (corresponding to aa pos-itions 21–346 of P. piscicida cob and encompassingediting sites 3–48), identical editing was observedin two of the species at 16 sites and in all threespecies at a further eight sites. At an additionaltwo sites (10 and 28), an identical nt change occursat the same position, but within different codonsspecifying a different aa both before and after edit-ing; notably, the nt 30 to the edit is a U in thesecases. Considering all 84 editing events within thisregion, 23 (27%) are unique to one species, 34(40%) are conserved between two species, and afurther 27 (32%) are shared among all threespecies.

As in the cox1 case, A ! G changes predominatein cob mRNA, but C ! U changes are compara-tively more prominent in the latter mRNA (Table3). Low levels of G ! C, U ! G and U ! A trans-versions and G ! A transitions are evident in thecob mRNA. As with cox1, the authenticity of these“minor” edits (five in P. piscicida, five in P. minimumand one in C. cohnii ) has been verified, and is

further supported by conservation between speciesin three cases: site 10, a second-position G ! C editin all three dinoflagellates; site 41, a second-position G ! C edit in P. piscicida and P. minimum;and site 47, a first-position G ! A edit in bothP. piscicida and P. minimum (Table 4). A uniqueU ! A transversion in P. piscicida changesUUG(Leu) to AUG(Met) at editing site 1, creatingwhat may be the initiation codon for this particularmRNA.

Markedly non-random distribution in the typeof codon edited and in the position of theediting site within codons

Although the range of codon and resultant aachanges is broad for both the cox1 and cobmRNAs, changes are highly skewed toward a fewcodons in each case. The most frequent changesare AUN(Ile) ! GUN(Val) (N ¼ A, U, C), with afew AUA(Ile) ! AUG(Met), as well. UUY(Phe) !CUY(Leu) is the next most frequently observedchange. Together, these six codons account for.40% of the total inferred editing for both cox1and cob mRNAs.

As is also the case with the mRNA editing thatoccurs in plant organelles,21 substitutions occuralmost exclusively at first or second positions ofaffected codons in dinoflagellate cox1 mRNA(Table 5), resulting in all of these cases in an aachange relative to the gene-predicted aa. Only fivethird-position edits were observed (Table 2), fourof which also predict an aa change: AUA(Ile) !AUG(Met) at site 4 (both P. piscicida and P. mini-mum ) and at site 18 (C. cohnii ), and UGA(stop) !UGG(Trp) at site 19 in P. minimum. Of the 72 edit-ing events identified in dinoflagellate cox1 mRNA,only a single one, a third-position A ! G transitionat site 15 in P. minimum (Table 2), is silent.

Similarly, almost all (.90%) of the dinoflagellatecob edits are at first or second positions of affectedcodons (Table 5). One significant exception is anA ! G transition at the third position of a UGAcodon in both P. piscicida and C. cohnii (site 4,Table 4); this event converts a standard terminationcodon to UGG at the position of a highly conservedTrp. (A second UGA codon appears within theP. piscicida cob coding sequence; however, this

Table 3. Types of substitution inferred in dinoflagellate cox1 and cob mRNAs

cox1 cobPfi Pmi Cco Total (%) Pfi Pmi Cco Total (%) cox1 þ cob (overall total %)

A ! G 12 13 16 41 (56.9) 9 12 12 33 (37.6) 74 (46.8)G ! A 0 0 0 0 1 2 0 3 (3.5) 3 (1.9)C ! U 3 1 1 5 (6.9) 12 7 3 22 (25.6) 27 (17.1)U ! C 5 8 4 17 (23.6) 9 7 4 20 (23.3) 37 (23.4)G ! C 3 3 3 9 (12.5) 3 2 1 6 (7.0) 15 (9.5)U ! A 0 0 0 0 1 0 0 1 (1.2) 1 (0.6)U ! G 0 0 0 0 0 1 0 1 (1.2) 1 (0.6)Total 23 25 24 72 35 31 20 86 158No. of A, U to G, C 17 21 20 58 18 20 16 54 112% of A, U to G, C 73.9 84.0 83.3 80.6 51.4 64.5 80.0 62.8 70.9

Mitochondrial mRNA Editing in Dinoflagellates 733

Table 4. Inferred editing of cob mRNA in dinoflagellate mitochondria

Site P. piscicida P. minimum C. cohnii 1 2 3

1 Uug ! Aug L ! M W – –2 uCu ! uUu S ! F – W –3 uCu ! uUu S ! F uCu ! uUu S ! F uCu ! uUu S ! F – B –4 ugA ! ugG p ! Wa ugG W ugA ! ugG p ! Wa – – X

5 Cuu ! Uuu L ! F Cug L I W – –

6 Auu ! Guu I ! V Auu ! Guu I ! V Auu I X – –7 Acu ! Gcu T ! A Acu ! Gcu T ! A Acu ! Gcu T ! A B – –8 auA I auA ! auG I ! M uuA L – – W9 Auc I Auc ! Guc I ! V Auu I W – –10 gGu ! gCu G ! A gGu ! gCu G ! A uGu ! uCu C ! S – V –

11 uCu ! uUu S ! F Y N – W –

12 Uuu ! Cuu F ! L Uuu ! Cuu F ! L I X – –

13 Aua ! Gua I ! V Auc ! Guc I ! V Aua ! Gua I ! V B – –14 Cuu ! Uuu L ! F Cuu ! Uuu L ! F Cuu ! Uuu L ! F B – –15a Agu S Acu ! Gcg T ! A Aau N W – –15b agU S acU ! gcG T ! A aaU N – – W

16 L L Aua ! Gua I ! V W – –

17 auA ! auG I ! M auA I auA ! auG I ! M – – X18 uCu ! uUu S ! F uCu ! uUu S ! F uUu F – X –

19 Auu I L Auu ! Guu I ! V W – –

20 Uuc F Uuc ! Cuc F ! L Uuc F W – –21 Uuc ! Cuc F ! L Uuc F Uua L W – –22 aCa ! aUa T ! I aCa ! aUa T ! I aUa I – X –23 Aua I Aua ! Gua I ! V Aua I W – –24 Guu V Guu ! Auu V ! I Aua I W – –25 Cau ! Uau H ! Y Cau ! Uau H ! Y Uau Y X – –26 uAu ! uGu Y ! C uAu ! uGu Y ! C uAu ! uGu Y ! C – B –27 Uac ! Cac Y ! H Uac ! Cac Y ! H Ucu S X – –28 aAu ! aGu N ! S uAu ! uGu Y ! C aAu N – P –29 Uuc ! Cuc F ! L Uuc F Uua ! Cuab L ! Lb X – –

30 L L Aua ! Gua I ! V W – –

31 Guu V L Auu ! Guu I ! V W – –

32a uGa ! uCaa p ! Sa L gua I V – W –

32b uca p S L Aua ! Gua I ! V W – –

33a Uuc ! Cuu F ! L Uuc ! Cuu F ! L Uuc ! Cuc F ! L B – –33b uuC ! cuU F ! L uuC ! cuU F ! L Uuc cuC F L – – X34 Auu ! Guu I ! V Auu I Auu ! Guu I ! V X – –35 Cuu ! Uuu L ! F Cuu ! Uuu L ! F Cuu ! Uuu L ! F B – –36 Cuu ! Uuu L ! F Cuu L Cuu L W – –37 cAu H cAu H aAu ! aGu N ! S – W –38 Auc ! Guc I ! V Auc ! Guc I ! V Auc I X – –39 Uua L Uuc ! Cuc F ! L Uua L W – –40 gUu ! gCu V !A gUu ! gCu V !A gUu ! gCu V !A – B –

41 gGa ! gCa G ! A gGa ! gCa G ! A I – X –

42 Auc I Aug ! Gug M ! V Auu I W – –43 Uuc ! Cuc F ! L Uuu F Uuu ! Cuu F ! L X – –44 Auu I Acu ! Gcu T ! A Auu I W – –45 Uuc ! Cuc F ! L Uuc ! Cuc F ! L Uuc F X – –

46 Uuc ! Cuc F ! L Uuc F I W – –

47 Ggu ! Agu G ! S Ggu ! Agu G ! S Ggu G X – –48 uCu ! uUu S ! F uCu S uCa S – W –

Codon and predicted amino changes are inferred from differences between gene and corresponding cDNA sequences. Editing sitenucleotides are set in uppercase letters (e.g. site 5, Cuu ! Uuu in a first-position C ! U edit) except where the editing site nucleotidediffers from both the “pre-edited” and “edited” states (C and U, respectively, in this example). In such cases, the variant nucleotide is

shown in a lowercase letter and is boxed and shaded (e.g. in C. cohnii in the example given). Codons 15, 32 and 33 contain two

different editing sites (a and b). Amino acids set in italics differ at a given editing site between dinoflagellates prior to editing but yieldthe same predicted amino acid after editing. Blanks at editing sites represent missing data for that species. Symbols in the finalcolumn indicate the degree of conservation of editing: B, same nucleotide and amino acid change in all three dinoflagellate species;V, identical nucleotide change in all three dinoflagellate species in non-identical codons that specify different amino acids; X, bothnucleotide and amino acid change conserved in two species; P, identical nucleotide change in non-identical codons that specify differ-ent amino acids in two dinoflagellate species; W, unique nucleotide and amino acid change in one species. Vertical positioning of eachcolored square under the numerals 1, 2 and 3 indicates where within the codon (first, second or third position) editing occurs.

a Editing of a potential termination codon (UGA ¼ p ) to a sense codon.b Silent editing (no amino acid change).

734 Mitochondrial mRNA Editing in Dinoflagellates

UGA is edited to UCA, a Ser codon (site 32a).) Theother exceptions involve A ! G (first position) andU ! G (third position) substitutions in the samecodon (sites 15a and 15b, respectively) in P. mini-mum, and U ! C (first position) and C ! U (thirdposition) substitutions in the same codon (sites33a and 33b, respectively) in both P. piscicida andP. minimum (Table 4). At sites 15a and 15b, the twosubstitutions together change the encoded aa fromThr(ACU) to Ala(GCG). At sites 33a and 33b, thechange is from Phe(UUC) to Leu(CUU); only thefirst-position edit is seen at the correspondingcodon in C. cohnii, again resulting in the identicalaa change, from Phe(UUC) to Leu(CUC).Curiously, at the site-29 codon in C. cohnii, aU ! C edit changes a UUA codon to CUA, bothof which specify Leu. Excepting the effectivelysilent, third-position U ! G edit at site 15b(P. minimum ) and C ! U edits at site 33b(P. piscicida and P. minimum ), the site-29 substi-tution is the only example we encountered whereediting of dinoflagellate cob mRNA does notchange aa identity.

In addition to the markedly non-random posi-tioning of editing within codons, a notable asym-metry is apparent in the nature of the nucleosidegenerated by editing (Table 3). In the case of cox1,74–84% of editing events in each species are fromA or U to a G or C. The figure is somewhat lowerin the case of cob (51–80%), due to a lower pro-portion of A ! G and higher proportion of C ! Uediting in cob mRNA compared to cox1 mRNA.Considering all 158 editing events inferred in thetwo mRNAs in the three species, .80% of thechanges are to a G or C, and most of these arefrom an A or U. Thus, editing has a pronouncedtendency to increase the G þ C content of editedcodons in these dinoflagellate mRNAs.

Consequences of editing

In other systems, mRNA editing serves anobvious role in genetic information transfer: edit-ing generates continuous ORFs, creates translationinitiation and termination codons, and/or restorescritical aa residues at functional sites within theprotein.21 Alignment of Cox1 sequences (Figure 1)shows that about one-third of the editing eventspredict an obvious increase in aa sequence simi-larity between the dinoflagellate Cox1 proteinsand their non-dinoflagellate homologs. At sevensites (5, 11, 19, 22, 26, 30 and 35), editing changes

the dinoflagellate residue to one identical to thenon-dinoflagellate consensus residue. Except atsites 11 and 19, where one or two of the dinoflagel-late sequences already encode(s) the consensus aa,the predicted change occurs in all three species(Table 2). Two additional sites (9 and 34, Figure 1)predict a radical aa change to a residue chemicallysimilar to the consensus. At still other sites, editingpredicts a conservative aa change that does notobviously increase similarity to the non-dinoflagel-late consensus (e.g. at sites 8, 14, 25, 27, 28, 36 and39, all of which are Ile ! Val changes). Overall, amajority (almost 60%) of the cox1 edits result eitherin maintenance of or an increase in aa similarity/identity when the dinoflagellate Cox1 sequencesare compared with their non-dinoflagellatecounterparts.

At ten sites (3, 4, 6, 10, 16, 17, 21, 24, 31 and 32),editing changes the encoded aa to one that differsfrom the non-dinoflagellate consensus but is iden-tical between two or all three of the dinoflagellatetaxa (Figure 1). In one notable case (site 16), thedinoflagellate-specific edit actually decreases simi-larity with the non-dinoflagellate consensus. At allother sites, including most of the unique ones, theinferred editing does not obviously increase simi-larity, either with non-dinoflagellate sequences oramong the dinoflagellate ones.

A very similar pattern is seen when a Cob align-ment (Figure 2) is examined. Again, about one-third of the editing events predict an increase inaa similarity compared with non-dinoflagellateCob sequences. At these particular sites, theincrease occurs because editing generates an aathat is either identical (sites 4, 10, 13, 23, 29, 39, 40,45) or chemically similar (sites 3, 5, 12, 18, 21, 25,43) to the non-dinoflagellate consensus (Figure 2).Other sites (6, 7, 14, 17, 22, 26, 27, 33, 34, 35, 38, 41,47) display dinoflagellate-specific editing, whichin one case (site 22) results in a decrease in simi-larity compared to the highly conserved consensus.Still other sites show conservative aa changes thatmaintain sequence similarity; among thesechanges, Ile ! Val substitutions are again promi-nent (e.g. sites 16, 19, 30, 31).

UGA codons in dinoflagellatemitochondrial ORFs

In C. cohnii, no UGA codons are present withinthe cox1 ORF,9 – 12 initially leading us to surmisethat UGA may be a termination codon, not a Trp

Table 5. Position of editing sites within codons in dinoflagellate cox1 and cob mRNAs

cox1 cobPfi Pmi Cco Total (%) Pfi Pmi Cco Total (%) cox1 þ cob (overall total %)

1st 13 15 16 44 (61.1) 20 20 13 53 (61.6) 97 (61.4)2nd 9 7 7 23 (31.9) 12 8 5 25 (29.1) 48 (30.4)3rd 1 3 1 5 (6.9) 3 3 2 8 (9.3) 13 (8.2)Total 23 25 24 72 35 31 20 86 158No. 1st or 2nd 22 22 23 67 32 28 18 78 145% 1st or 2nd 95.7 88.0 95.8 93.1 91.4 90.3 90.0 90.7 91.8

Mitochondrial mRNA Editing in Dinoflagellates 735

codon, in dinoflagellate mitochondria.9 Sub-sequently, a UGA codon was found at the sameposition in both the C. cohnii12 and P. piscicida13 cobORFs; however, in both cases, this UGA is editedto a UGG(Trp) codon (Table 4; editing site 4). TheP. piscicida cob gene contains a second UGA, butthis is changed to UCA(Ser) via a second-positionG ! C editing event (Table 4; editing site 32a).Although these observations would seem to bolsterthe argument that UGA is not decoded as Trp indinoflagellate mitochondria, an unedited UGA is,in fact, present in the P. minimum cox1 mRNA(denoted by X in Figures 1 and 4). This UGAoccurs at the position of an otherwise invariantTrp in the Cox1 sequence of other organisms,including P. piscicida and C. cohnii, falling withinthe long non-edited stretch between editing clus-ters II and III in the cox1 mRNA. This observationargues that the dinoflagellate mitochondrial trans-lation system is able to recognize UGA codons asTrp, even though most UGA codons appear to beedited to sense codons prior to translation. UGAalso codes for Trp in both apicomplexans4,5 andciliates,6 so the change in specificity (from termin-ation to Trp) in these three phyla may reflect asingle mutational event occurring before thediversification of the alveolates.

Discussion

The inferred editing reported here was bothunexpected and surprising in its nature and extent.For both the C. cohnii Cox1 and Cob sequences,alignment with a selection of non-dinoflagellatehomologs provided little indication that mito-chondrial mRNA editing might be occurring inthis organism.12 Even when all three dinoflagellatesequences are included in the Cox1 and Cob align-ments (Figures 1 and 2), inspection of aa align-ments alone is not particularly compelling insuggesting editing. For example, to the extent thatthe dinoflagellate cox1 gene sequences have beendetermined, they already encode the expectedinvariant His residues that bind heme a, CuB andheme a3 (reviewed in Ref. 26; see Figure 1). Like-wise, the dinoflagellate cob sequences alreadyspecify, at the gene level, the four invariant Hisresidues that are ligands for the heme b group inapocytochrome b (26; see Figure 2).

As noted in Results, at a number of sites editingincreases aa identity or similarity; nevertheless,almost all of the editing sites occur outside of themost highly conserved blocks of identical sequencedefined on the basis of the alignments. The crypticcharacter of the editing reported here reinforcesthe need for caution when inferring proteinsequence directly from gene sequence in othersystems, particularly mitochondrial ones.

The aa alignments demonstrate that about half ofthe editing events have the effect of maintaining orincreasing similarity between the dinoflagellateand non-dinoflagellate taxa; the remaining editing

events are either dinoflagellate-specific, or occur atvariable sites and appear to change the aa residuemore or less randomly. A small number of editsactually represent changes from otherwise highlyconserved residues to unique ones (e.g. Thr ! Valat cox1 site 16 (Figure 1) and Thr ! Ile at cob site22 (Figure 2), in both P. piscicida and P. minimum ).The variety of nt changes documented here, andthe different consequences they imply for functionat individual sites, are somewhat at variance withthe characteristics of mRNA editing in plantorganelles. In the latter case, editing at individualsites typically serves to increase aa conservation incomparison with homologous sequences in non-plants.21 Our observations raise the question ofwhy dinoflagellate mitochondria should engage ina substantial proportion of editing whose func-tional purpose is not immediately obvious (e.g. aconservative change, such as AUA(Ile) !GUA(Val), at a site where either the unedited (Ile)or edited (Val) residue is present in other, non-dinoflagellate species).

In this regard, the patchy distribution of editingsites (Figure 4) provides a possible clue. Themarked clustering of editing sites suggests thatspecific regions of the cox1 and cob mRNA areselectively targeted for editing. Depending on themechanism by which sites are recognized and edit-ing is carried out, it is conceivable that functionallyneutral editing might occur within a region whereonly a sub-set of the observed edits is actuallyessential. In plant organelles, for example, editing(although evolutionarily conserved) appears to beessential at only a few sites.27

On the other hand, the editing reported heremight have functional consequences that are moresubtle than is immediately apparent, consequencesthat are related more to changes in the base compo-sition of a codon than to the aa specified by thatcodon. Notably, the majority of inferred edits areto a G or C, particularly in the case of cox1 mRNA(Table 3). If, as in other alveolates, mitochondria indinoflagellates need to import most or even all ofthe tRNAs required for translation in the organelle,editing may have the effect of re-tailoring anA þ U-rich mRNA (the result of AT drive at thelevel of the mitochondrial genome11) so that it isable to be translated more efficiently by a set ofnucleus-encoded tRNAs. This re-tailoring mightinvolve a reduction in the number of certain AU-rich codons whose presence is rate limiting formitochondrial translation. In this regard, it may besignificant that the most frequently edited codonsare AUN and UUY. In such a situation, it may bethe nucleotide change per se rather than theresultant amino acid change that is essential forfunction; nevertheless, the editing may be con-strained by a necessity to maintain a chemicallysimilar amino acid at the position in question, giv-ing the impression of “superfluous” editing at thelevel of protein sequence.

Considering the nature of the observed edits, it isdifficult at this point to suggest how editing might

736 Mitochondrial mRNA Editing in Dinoflagellates

occur in this system. The overwhelming majorityof edits are transitions, which might implicateenzyme activities (such as adenosine and cytidinedeaminases) that are known to carry out editing inother organisms. If this were so, more than onedistinct activity would presumably have to operatein dinoflagellate mitochondria. In Physarum mito-chondria, for instance, both insertional/deletionaland substitutional types of editing have beendocumented.19 On the other hand, we do observea small number of authentic G ! C, G ! A,U ! A and U ! G substitutions that must beaccounted for in any editing mechanism. Thatmay be a reason to favor a single system. Forexample, an excision–replacement type of editingcould serve to remove a specific patch of mRNAsequence and replace the sequence with the editedversion, with the information for editing perhapscoming from a guide-type (antisense) sequence.Such a mechanism could mediate any of thetransitions and transversions reported here. Anexcision–replacement type of editing has beenshown to operate at the 50-ends of mitochondrialtRNAs in certain amoeboid protozoa andchytridiomycete fungi,28 and the U insertion/deletion type of editing that occurs in trypanosomemitochondria employs many of the same activitiesthat we would predict would be required for anexcision–replacement type of editing in dinoflagel-late mitochondria: endonuclease, exonuclease,nucleotidyltransferase and ligase.18 Clearly,development of an in vitro editing system fromdinoflagellate mitochondria will be important tounderstanding the mechanism of this intriguingnew form of mRNA modification. Considering thetechnical difficulties in working at the biochemicallevel with dinoflagellate mitochondria, this will bea challenging undertaking.

Finally, the novel type of RNA editing that wereport here appears to be another example of aderived trait arising within a specific evolutionarylineage17. Examination of mRNA sequence fromboth apicomplexans29 – 32 and ciliates8,33 has so farfailed to reveal any evidence of RNA editing inthese two groups. On the other hand, our dataimply that the type of editing documented heremay be widespread within the Dinozoa. Inphylogenetic trees based on nuclear 18 SrRNA sequences, Crypthecodinium, Pfiesteria andProrocentrum are widely separated.34 Thus, thisunique type of mRNA editing appears to haveemerged early within the dinoflagellate lineage,after its separation from a common ancestor withapicomplexans.

Materials and Methods

Culture conditions and isolation of nucleic acids

Five strains of P. piscicida (CCMP 1830, 1831, 1834,1921, 1928) from the Provasoli–Guillard National Centerfor Culture of Marine Phytoplankton at Bigelow Labora-tory and one strain (NCSU113-3) provided by Dr JoAnne

Burkholder were grown in autoclaved seawater (15 PSU(practical salinity unit); 19(^1) 8C) with a supply ofthe cryptophyte alga, Rhodomonas sp. CCMP 768(also grown in 15 PSU seawater amended with f/2nutrients35 as food source). The identity of theseP. piscicida strains was verified by 18 S rRNA sequencing.P. minimum (CCMP696) was cultured in f/2 mediumprepared with 28 PSU seawater. C. cohnii (strain WHd)was kindly provided by Dr Carl Beam (BrooklynCollege, New York, NY) and cultured axenically inMLH liquid medium.9,11,12

Cells of P. piscicida and P. minimum were harvested inexponential growth phase and resuspended in DNAextraction buffer36 or Trizol Reagent (Gibco/BRL). DNAwas extracted following a standard protocol36 and RNAwas isolated according to the Trizol instructions. As acontrol, nucleic acids were also extracted from the foodorganism (Rhodomonas sp.). Nucleic acids were preparedfrom C. cohnii as described.9,11,12

Genomic and cDNA clones from P. piscicida andP. minimum

Total RNA from P. piscicida strain CCMP1831 wasreverse transcribed using a modified oligo(dT) primerand the TimeSavere cDNA Synthesis Kit (AmershamPharmacia Biotech). The resulting cDNA was clonedinto pBluescript II vector (pBS) (Stratagene). A cox1probe was obtained by PCR amplification of theP. piscicida cDNA library, with primer design based on amultiple cox1 alignment, including that of C. cohnii(AF186994). A cob DNA probe was obtained in a similarmanner using a set of degenerate primers based onconserved regions of cob and partial cDNA sequencefrom another dinoflagellate, Lingulodinium polyedrum(formerly Gonyaulax polyedra; GenBank acc. no.AF142472). After sequencing to verify their respectiveidentities, the probes were labeled with fluorescein-11-dUTP using the ECF Random-Prime Labeling andDetection System (Amersham Pharmacia Biotech) andthen used to screen the cDNA library for cox1 and cobcDNA clones. Five cox1 and 20 cob clones were obtainedand sequenced.

First-strand cDNAs for all six P. piscicida strains andfor P. minimum were synthesized from total RNA usingthe SuperScripte Preamplification cDNA Synthesis Kit(Invitrogen) with modified oligo-dT (50-CCGAGGCGGC-CGACATG(T)16-3

0), then used as templates for second-strand synthesis and subsequent PCR amplification.PCR primers (Table 1) were designed based on thesequences of the longest cox1 and cob clones fromthe CCMP1831 cDNA library. PCR reactions comprised35 cycles of 25 seconds at 94 8C, 30 seconds at 55 8Cand 40 seconds at 72 8C, followed by ten minutes at72 8C.

Genomic cox1 and cob sequences from both P. piscicidaand P. minimum were amplified using appropriate gene-specific primers (Table 1). In addition, sequence fromthe 50 flanking region of P. piscicida (CCMP1831) cobwas obtained as described,13 in order to reveal apotential start codon. Approximately 50 ng of total cellu-lar ( ¼ genomic) DNA were used as template, with PCRperformed under the conditions described above.

PCR products were cloned (TOPO TA Cloning Kit,Invitrogen) and ten to 50 clones of each product weresequenced. To assess possible polymorphisms, bothgenomic and cDNA PCR products were also directlysequenced without subcloning.

Mitochondrial mRNA Editing in Dinoflagellates 737

Genomic and cDNA clones from C. cohnii

Clones containing genomic copies of C. cohnii cox1 andcob were obtained by screening an Eco RI library pre-pared from a nucleic acid fraction enriched inmtDNA.9,11,12 Detailed characterization of cox1 clones11,12

revealed two distinct cox1 ORFs differing within theextreme 30-terminal region. A total of 15 cob-containingclones were isolated and characterized, four by completesequencing. The results12 (J.E.N. & M.W.G., unpublishedresults) revealed three distinct cob ORFs differing withinthe extreme 30-terminal region, beyond the last block ofconserved aa.

Clones of C. cohnii cox1 and cob cDNAs were obtainedby RT-PCR using gene-specific primers (Table 1). RTreactions were performed at 45 8C with 1 mg of either oftwo independently isolated, DNase I-treated RNA prep-arations, using the SuperScript II protocol (Gibco/BRL).Aliquots (2 ml) of each RT reaction were taken for PCRwith either of two forward primers together with theparticular reverse primer that had been used initiallyfor RT.

Several independent PCR reactions were carried outfor each individual C. cohnii RT reaction to check for arti-facts introduced by either RT or PCR. PCR products werecloned (TOPO TA Cloning Kit, Invitrogen) andsequenced.

Nucleotide sequence accession numbers

The new sequences reported here have been depositedin GenBank under the following accession numbers:AF463412, AF463413, AF463414, AF463415, AF487783,AF357518, AY030285, AY030286, AF403220, AF403221.

Acknowledgments

We thank Dr JoAnne Burkholder and Bigelow Labora-tory CCMP for kind gifts of cultures. This work was sup-ported by ECOHAB Grant NA86OP0491 (to S.L.) andCanadian Institutes for Health Research Grant MOP-4124 (to M.W.G.). M.W.G. is a Fellow in the Program inEvolutionary Biology of the Canadian Institute forAdvanced Research and Canada Research Chair inGenomics and Genome Evolution at DalhousieUniversity. ECOHAB publication no.47.

References

1. Cavalier-Smith, T. (1998). A revised six-kingdomsystem of life. Biol. Rev. 73, 203–266.

2. Spector, D. L. (1984). Dinoflagellates: an introduction.In Dinoflagellates (Spector, D. L., ed.), pp. 1–15,Academic Press, Orlando.

3. Gajadhar, A. A., Marquardt, W. C., Hall, R.,Gunderson, J., Ariztia-Carmona, E. V. & Sogin, M. L.(1991). Ribosomal RNA sequences of Sarcocystismuris, Theileria annulata and Crypthecodinium cohniireveal evolutionary relationships among apicom-plexans, dinoflagellates, and ciliates. Mol. Biochem.Parasitol. 45, 147–154.

4. Wilson, R. J. M. & Williamson, D. H. (1997). Extra-chromosomal DNA in the Apicomplexa. Microbiol.Mol. Biol. Rev. 61, 1–16.

5. Feagin, J. E. (2000). Mitochondrial genome diversityin parasites. Int. J. Parasitol. 30, 371–390.

6. Cummings, D. J. (1992). Mitochondrial genomes ofthe ciliates. Int. Rev. Cytol. 141, 1–64.

7. Burger, G., Zhu, Y., Littlejohn, T. G., Greenwood, S. J.,Schnare, M. N., Lang, B. F. & Gray, M. W. (2000).Complete sequence of the mitochondrial genome ofTetrahymena pyriformis and comparison withParamecium aurelia mitochondrial DNA. J. Mol. Biol.297, 365–380.

8. Edqvist, J., Burger, G. & Gray, M. W. (2000).Expression of mitochondrial protein-coding genes inTetrahymena pyriformis. J. Mol. Biol. 297, 381–393.

9. Norman, J. E. & Gray, M. W. (1997). The cytochromeoxidase subunit 1 gene (cox1 ) from the dinoflagel-late, Crypthecodinium cohnii. FEBS Letters, 413,333–338.

10. Inagaki, Y., Hayashi-Ishimaru, Y., Ehara, M.,Igarashi, I. & Ohama, T. (1997). Algae or protozoa:phylogenetic position of euglenophytes and dino-flagellates as inferred from mitochondrial sequences.J. Mol. Evol. 45, 295–300.

11. Norman, J. E. & Gray, M. W. (2001). A complexorganization of the gene encoding cytochrome oxi-dase subunit 1 in the mitochondrial genome of thedinoflagellate, Crypthecodinium cohnii: homologousrecombination generates two different cox1 openreading frames. J. Mol. Evol. 53, 351–363.

12. Norman, J. E. (2000). Mitochondrial genome organiz-ation, expression and evolution in the dinoflagellate,Crypthecodinium cohnii. PhD thesis, DalhousieUniversity.

13. Zhang, H. & Lin, S. (2002). Detection and quantifi-cation of Pfiesteria piscicida by using the mito-chondrial cytochrome b gene. Appl. Environ.Microbiol. 68, 989–994.

14. Burkholder, J. M., Noga, E. J., Hobbs, C. H., Glasgow,H. B., Jr & Smith, S. A. (1992). New ‘phantom’ dino-flagellate is the causative agent of major estuarinefish kills. Nature, 358, 407–410. Erratum in: Nature360, 768 (1992).

15. Oldach, D. W., Delwiche, C. F., Jakobsen, K. S., Tengs,T., Brown, E. G., Kempton, J. W. et al. (2000). Hetero-duplex mobility assay-guided sequence discovery:elucidation of the small subunit (18S) rDNAsequences of Pfiesteria piscicida and related dino-flagellates from complex algal culture and environ-mental sample DNA pools. Proc. Natl Acad. Sci.USA, 97, 4303–4308.

16. Bowers, H. A., Tengs, T., Glasgow, H. B., Jr,Burkholder, J. M., Rublee, P. A. & Oldach, D. W.(2000). Development of real-time PCR assays forrapid detection of Pfiesteria piscicida and relateddinoflagellates. Appl. Environ. Microbiol. 66,4641–4648.

17. Gray, M. W. (2001). Speculations on the origin andevolution of editing. In RNA Editing (Frontiers inMolecular Biology) (Bass, B. L., ed.), pp. 160–184,Oxford University Press, Oxford.

18. Stuart, K. D., Panigrahi, A. K. & Salavati, R. (2001).RNA editing in kinetoplastid mitochondria. In RNAEditing (Frontiers in Molecular Biology) (Bass, B. L.,ed.), pp. 1–19, Oxford University Press, Oxford.

19. Gott, J. M. (2001). RNA editing in Physarumpolycephalum. In RNA Editing (Frontiers in MolecularBiology) (Bass, B. L., ed.), pp. 20–37, OxfordUniversity Press, Oxford.

20. Vanfleteren, J. R. & Vierstraete, A. R. (1999). Inser-tional RNA editing in metazoan mitochondria: thecytochrome b gene in the nematode Teratocephaluslirellus. RNA, 5, 622–624.

738 Mitochondrial mRNA Editing in Dinoflagellates

21. Bock, R. (2001). RNA editing in plant mitochondriaand chloroplasts. In RNA Editing (Frontiers inMolecular Biology) (Bass, B. L., ed.), pp. 38–60, OxfordUniversity Press, Oxford.

22. Driscoll, D. M. & Innerarity, T. L. (2001). RNA editingby cytidine deamination in mammals. In RNAEditing (Frontiers in Molecular Biology) (Bass, B. L.,ed.), pp. 61–76, Oxford University Press, Oxford.

23. Emeson, R. B. & Singh, M. (2001). Adenosine-to-inosine RNA editing: substrates and consequences.In RNA Editing (Frontiers in Molecular Biology) (Bass,B. L., ed.), pp. 109–138, Oxford University Press,Oxford.

24. Hough, R. F. & Bass, B. L. (2001). Adenosine deami-nases that act on RNA. In RNA Editing (Frontiers inMolecular Biology) (Bass, B. L., ed.), pp. 77–108,Oxford University Press, Oxford.

25. Adams, K. L., Song, K., Roessler, P. G., Nugent, J. M.,Doyle, J. L., Doyle, J. J. & Palmer, J. D. (1999). Intra-cellular gene transfer in action: dual transcriptionand multiple silencings of nuclear and mitochondrialcox2 genes in legumes. Proc. Natl Acad. Sci. USA, 96,13863–13868.

26. Trumpower, B. L. & Gennis, R. B. (1994). Energytransduction by cytochrome complexes in mito-chondrial and bacterial respiration: the enzymologyof coupling electron transfer reactions to trans-membrane proton translocation. Annu. Rev. Biochem.63, 675–716.

27. Corneille, S., Lutz, K. & Maliga, P. (2000). Conserva-tion of RNA editing between rice and maize plastids:are most editing events dispensable? Mol. Gen. Genet.264, 419–424.

28. Price, D. H. & Gray, M. W. (1998). Editing of tRNA.In Modification and Editing of RNA (Grosjean, H. &

Benne, R., eds), pp. 289–305, ASM Press,Washington.

29. Suplick, K., Morrisey, J. & Vaidya, A. B. (1990). Com-plex transcription from the extrachromosomal DNAencoding mitochondrial functions of Plasmodiumyoelii. Mol. Cell. Biol. 10, 6381–6388.

30. Feagin, J. E. (1992). The 6-kb element of Plasmodiumfalciparum encodes mitochondrial cytochrome genes.Mol. Biochem. Parasitol. 52, 145–148.

31. McIntosh, M. T., Srivastava, R. & Vaidya, A. B.(1998). Divergent evolutionary constraints on mito-chondrial and nuclear genomes of malaria parasites.Mol. Biochem. Parasitol. 95, 69–80.

32. Rehkopf, D. H., Gillespie, D. E., Harrell, M. I. &Feagin, J. E. (2000). Transcriptional mapping andRNA processing of the Plasmodium falciparum mito-chondrial mRNAs. Mol. Biochem. Parasitol. 105,91–103.

33. Orr, A. T., Rabets, J. C., Horton, T. L. & Landweber,L. F. (1997). RNA editing missing in mitochondria.RNA, 3, 335–336.

34. Saldarriaga, J. F., Taylor, F. J. R., Keeling, P. J. &Cavalier-Smith, T. (2001). Dinoflagellate nuclear SSUrRNA phylogeny suggests multiple plastid lossesand replacements. J. Mol. Evol. 53, 204–213.

35. Guillard, R. R. & Ryther, J. H. (1962). Studies ofmarine planktonic diatoms. I. Cyclotella rana Hustedtand Detonula confervacea (Cleve.) Gran. Can.J. Microbiol. 18, 229–239.

36. Sambrook, J., Fritsch, E. F. & Maniatis, T. (1989).Molecular Cloning: A Laboratory Manual, 2nd edit.,Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY.

Edited by D. E. Draper

(Received 20 March 2002; received in revised form 6 May 2002; accepted 7 May 2002)

Mitochondrial mRNA Editing in Dinoflagellates 739