Functional Remodeling of RNA Processing in Replacement Chloroplasts by Pathways Retained from Their Predecessors

Functional remodeling of RNA processing in replacement chloroplasts by pathways retained from their predecessors

Richard G. Dorrell1 and Christopher J. Howe

Department of Biochemistry, University of Cambridge, Cambridge CB2 1QW, United Kingdom

Edited* by J. Woodland Hastings, Harvard University, Cambridge, MA, and approved October 3, 2012 (received for review July 19, 2012) Chloroplasts originate through the endosymbiotic integration of stood that processes originating in the host play important roles a host and a photosynthetic symbiont, with processes established in chloroplast biogenesis and maintenance (2, 20, 21). This raises within the host for the biogenesis and maintenance of the nascent the question whether processes established during an earlier chlo- chloroplast. It is thought that several photosynthetic eukaryotes roplast symbiosis could be retained through serial endosymbiosis have replaced their original chloroplasts with others derived from and contribute to the function of the replacement chloroplast. different source organisms in a process termed “serial endosymbi- Furthermore, if such processes were made use of by replacement osis of chloroplasts.” However, it is not known whether replace- chloroplasts lacking them, the biology of the replacement might ment chloroplasts are affected by the biogenesis and maintenance be dramatically changed as a result. We therefore wanted to de- pathways established to support their predecessors. Here, we in- termine whether the biochemical activities of replacement chlor- vestigate whether pathways established during a previous chloro- oplasts could be functionally altered by pathways retained from plast symbiosis function in the replacement chloroplasts of the ancestral symbioses. dinoflagellate alga Karenia mikimotoi. We show that chloroplast We studied RNA processing in a model example of chloroplast transcripts in K. mikimotoi are subject to 3′ polyuridylylation and replacement, the dinoflagellate Karenia mikimotoi. This globally extensive sequence editing. We confirm that these processes do distributed species, implicated in the formation of fish-killing not occur in free-living relatives of the replacement chloroplast blooms (22), is a particularly well-studied example (6, 23, 24) of EVOLUTION lineage, but are otherwise found only in the ancestral, red algal- dinoflagellates containing the light-harvesting carotenoid pigment fl derived chloroplasts of dino agellates and their closest relatives. fucoxanthin. The majority of photosynthetic dinoflagellates harbor This indicates that these unusual RNA-processing pathways have red algal-derived chloroplasts, which contain the light-harvesting been retained from the original symbiont lineage and made use of carotenoid pigment peridinin and have an unusual, highly reduced by the replacement chloroplast. Our results constitute an addition genome, which is fragmented into a number of small plasmid-like to current theories of chloroplast evolution in which chloroplast elements termed “minicircles” (4, 11, 25–27). Fucoxanthin- biogenesis may be radically remodeled by pathways remaining containing dinoflagellates, by contrast, contain chloroplasts with a from previous symbioses. much less reduced genome that is not encoded on minicircles, although some subgenomic molecules have been reported (15, 28). organelle | fucoxanthin | haptophyte | poly(U) | alveolate The chloroplasts of fucoxanthin dinoflagellates group phylo- genetically with those of another group of algae, the haptophytes, hloroplasts originate through endosymbiosis, in which a free- to the exclusion of other lineages (7, 15, 16, 18). Although the Cliving photosynthetic symbiont is taken up by a eukaryotic phylogenetic relationship between the peridinin-containing host, with processes becoming established within the host to and fucoxanthin-containing chloroplast lineages has historically support the biogenesis and maintenance of the symbiont (1, 2). It proved controversial (6, 24), recent studies have confirmed that has long been understood that many such endosymbiotic events the peridinin-containing chloroplast lineage is the ancestral type have occurred across the eukaryotes, giving rise to a diverse array and was acquired before the radiation of photosynthetic dino- – of extant chloroplast lineages (2 4). Genomic and phylogenetic flagellates (15, 29). Thus, the fucoxanthin-containing chloroplasts evidence has suggested that several major chloroplast-containing were derived from a subsequent independent endosymbiotic event lineages have replaced their original chloroplasts with others that replaced the ancestral red-algal chloroplast in K. mikimotoi derived from a different evolutionary lineage in a process termed fl “ ” and related dino agellate lineages. serial endosymbiosis of chloroplasts (2, 5). The best-supported The ancestral chloroplasts of peridinin-containing dinoflagellates examples of chloroplast replacement are found within the di- fi fl show a number of unusual posttranscriptional RNA modi cations. no agellate algae, within which at least three distinct chloroplast These include the addition of 3′ terminal poly(U) tracts, and (in replacement events are known to have occurred (2, 6, 7). Two some species) extensive transcript editing (3, 29–33). Outside the even more dramatic hypotheses of chloroplast replacement have fl fi peridinin dino agellates, chloroplast RNA polyuridylylation has recently been put forward. The rst proposes that the ancestors so far been reported only in the closely related species Chromera of taxa currently harboring red algal-derived chloroplasts, such as diatoms and apicomplexan parasites, contained green algal symbionts (5, 8–11). In the second hypothesis, even the cya- Author contributions: R.G.D. and C.J.H. designed research; R.G.D. performed research; nobacterial-derived chloroplasts of plants and their closest rel- R.G.D. analyzed data; and R.G.D. and C.J.H. wrote the paper. atives are replacements for organelles derived from an The authors declare no conflict of interest. ancestral symbiosis involving a chlamydiobacterial symbiont (12, 13). Although both these proposed replacement events remain *This Direct Submission article had a prearranged editor. controversial (8, 14), serial endosymbioses may constitute Data deposition: Full-length DNA sequences, nonpolyuridylylated transcripts, and polyuridylylated transcripts for each of Karenia mikimotoi psbA, psbC, psbD, psaA,andrbcL a widespread feature of chloroplast evolution. genes have been deposited in the GenBank database (accession nos. JX899682– Regardless of the number of chloroplast replacement events JX899726). that have occurred, one outstanding question is whether, in any 1To whom correspondence should be addressed. E-mail: [email protected]. given instance, the ancestral chloroplast symbiosis might affect This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. the biology of the incoming chloroplast (15–19). It is well under- 1073/pnas.1212270109/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1212270109 PNAS Early Edition | 1of6 Downloaded by guest on September 25, 2021 velia (29), and chloroplast transcript editing has been reported species (Fig. 1). We found no evidence of polyuridylylated tran- only in land plants (3). Here, we report that these unusual scripts for the psbA or psbD genes of either species (Fig. 1, chloroplast transcript-processing pathways, derived from the an- lanes 11, 12, 15, 16), whereas positive control experiments using cestral peridinin-containing dinoflagellate chloroplast symbiosis, internal gene-specific primers yielded products of the expected function in the replacement chloroplasts of fucoxanthin-containing sizes (Fig. 1, lanes 13, 14, 17, 18). We confirmed the absence of dinoflagellates. This demonstrates that the biology of serial en- transcript polyuridylylation in each species through independent dosymbionts can be dramatically remodeled by host functions repeats of each oligo(dA) RT-PCR, using forward primers against remaining from previous symbioses. the E. huxleyi and P. tricornutum psbA, psbC, psbD, psaA,andrbcL genes (Fig. S1). Results We confirmed the results showing that transcripts of K. miki- 3′ Terminal Polyuridylylation of Chloroplast Transcripts in K. mikimotoi. motoi chloroplast genes were polyuridylylated by cloning and We first tested if transcripts in the serially acquired chloroplasts sequencing RT-PCR products for each reaction (Fig. 2). In each of K. mikimotoi carried 3′ poly(U) tracts. We generated cDNA clone sequenced, we observed poly(U) tracts similar to those pre- from total cellular RNA of K. mikimotoi using an oligo(dA) primer viously reported in peridinin dinoflagellates and C. velia (29, 31, and then carried out PCR reactions with oligo(dA) as a reverse 34). Each sequenced clone was polyuridylylated 8–22 nt down- primer, and forward primers that annealed within specific genes stream of the translation termination codon (Fig. 2 and Fig. S2). (Fig. 1 and Dataset S1). All five K. mikimotoi chloroplast tran- In the case of psbA, a single consensus poly(U) site was observed scripts tested (psbA, psbC, psbD, psaA,andrbcL)gavePCR in every sequenced clone, whereas the precise poly(U) site varied products of the expected sizes, indicating that the corresponding by up to 5 nt in all other genes (Fig. 2). To verify that the poly(U) transcripts were indeed polyuridylylated (Fig. 1, lanes 1–5). We tracts observed were added posttranscriptionally, we amplified the confirmed this by sequencing (see below). As a positive control, 3′ UTR of each gene from genomic DNA templates by thermal a representative chloroplast transcript (psbA) from the peridinin asymmetric interlaced PCR and determined the UTR sequences dinoflagellate Amphidinium carterae also yielded a positive result by direct sequencing. We subsequently confirmed these of the expected size (Fig. 1, lane 19). Representative nuclear sequences by direct PCR using reverse primers located within (PCNA) and mitochondrial sequences (cox1)forK. mikimotoi the 3′ UTR sequence (Dataset S1). In each case we observed that, were also tested (Fig. 1, lanes 6–7). No polyuridylylated transcripts although the 3′ flanking sequences of each gene were A/T rich, the of either of these genes could be amplified, whereas control poly(U) tracts identified did not correspond to poly(T) tracts in RT-PCRs using internal gene-specific cDNA primers that did not genomic DNA (Fig. 2). To confirm the structure and composition depend on a 3′ poly(U) tract generated products of the expected of the poly(U) tracts, we carried out RT-PCR on circularized sizes (Fig. 1, lanes 9–10). transcripts of psbA and psbC (Fig. S2 and Dataset S1) (35, 36). To determine whether transcript polyuridylylation also occurs We found that the poly(U) tracts observed were generally be- in haptophyte chloroplasts (from which group the K. mikimotoi tween 15 and 30 nt in length, homopolymeric, and on the 3′ chloroplasts were derived), or even in other lineages containing end of each transcript (as opposed to an internal insertion). secondary chloroplasts of red algal origin such as diatoms, we Phylogenetic analyses confirmed that the polyuridylylated tran- performed similar oligo(dA) RT-PCRs for representative hapto- scripts were derived from an ancestral haptophyte chloroplast phyte (Emiliania huxleyi) and diatom (Phaeodactylum tricornutum) lineage, as expected for replacement fucoxanthin chloroplasts

Fig. 1. Oligo(dA) and gene-specific RT-PCRs for transcripts from K. mikimotoi, A. carterae, E. huxleyi, and P. tricornutum. The gel photographs display the products from a series of RT-PCRs to detect polyuridylylated transcripts from K. mikimotoi as well as representative haptophyte and diatom species. The size standard is DNA Hyperladder I (Bioline). Lanes 1–5: oligo(dA) RT-PCR of K. mikimotoi psbA, psbC, psbD, psaA, and rbcL. Lanes 6–7: oligo(dA) RT-PCR of K. mikimotoi PCNA and cox1. Lanes 8–10: gene-specific RT-PCR of K. mikimotoi psbA, PCNA, and cox1. Lanes 11–12: oligo(dA) RT-PCR of E. huxleyi psbA and psbD. Lanes 13–14: gene-specific RT-PCR of E. huxleyi psbA and psbD. Lanes 15–16: oligo(dA) RT-PCR of P. tricornutum psbA and psbD. Lanes 17–18: gene- specific RT-PCR of P. tricornutum psbA and psbD. Lane 19: oligo(dA) RT-PCR of A. carterae psbA. Lane 20: reverse transcriptase negative control for oligo(dA) RT-PCR of K. mikimotoi psbA.

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1212270109 Dorrell and Howe Downloaded by guest on September 25, 2021 Fig. 2. Aligned 3′ ends of K. mikimotoi polyuridylylated chloroplast transcripts. This alignment shows the first 50 nt of the 3′ UTR of the K. mikimotoi psbA, psbC, psbD, psaA, and rbcL genes, labeled “gDNA.” The 3′ ends of sequenced, cloned polyuridylylated transcripts are shown below each gDNA sequence. Transcript sequences are shown for each unique poly(U) site observed. Only the first 18 nt of the poly(U) tract are shown for each transcript. Numbers in parentheses correspond to the full length of the poly(U) tract as sequenced in different clones. Specific sequences that are asterisked were obtained directly from crude RT-PCR products. EVOLUTION

(Fig. S3) (6, 15, 29). In addition, the rbcL transcript was, as and genomic DNA sequences for E. huxleyi or P. tricornutum previously reported, of a form ID gene as found in haptophyte psbA and psbD (37, 39), suggesting that editing is absent from and fucoxanthin-containing chloroplast genomes (15, 23, 37), both haptophyte and diatom chloroplast lineages. whereas the rbcL gene used by the ancestral peridinin-containing In peridinin dinoflagellate chloroplasts, some editing events are lineage is located in the nucleus and encodes a form II enzyme found only in polyuridylylated transcripts (34). We investigated if (29, 38). there was a relationship between transcript polyuridylylation and editing in K. mikimotoi by directly sequencing RT-PCR Editing of Chloroplast Transcripts in K. mikimotoi. In addition to products for nonpolyuridylylated transcripts of each gene, using polyuridylylation, transcripts in chloroplasts of peridinin dino- gene-specific cDNA primers that annealed to the 3′ UTR of each flagellates undergo an extensive array of base editing events. gene ∼100 nt downstream of the poly(U) site and the same PCR Outside dinoflagellates, chloroplast RNA editing has been reported forward primers as before. In total, one-fifth of the residues that only in land plants (3, 32, 34). We therefore tested if chloroplast were edited in polyuridylylated transcripts were unedited in transcripts in K. mikimotoi were also edited, by comparing the the nonpolyuridylylated transcripts (Fig. 3A and Dataset S2), sequences of oligo(dA) primed RT-PCR products for each gene whereas we could not identify any editing events specifictonon- with similar sequences obtained from genomic DNA. We observed polyuridylylated transcripts. The degree of difference in the extent extensive editing in oligo(dA) RT-PCR sequences (Fig. 3 and of editing between polyuridylylated and nonpolyuridylylated tran- Dataset S2), with 4.8% of bases differing between oligo(dA) scripts varied among genes. All of the editing events observed RT-PCR sequence and genomic DNA. Although the oligo(dA) in polyuridylylated psbC transcripts were also found in non- RT-PCR sequencing products would be expected to be represen- polyuridylylated transcripts, except for two residues, in each case tative of the entire population of transcripts, and might therefore at the very 3′ end, whereas more substantial differences were ob- contain a mixture of edited and unedited sequences, only a small served for psbA and rbcL (Fig. 3A and Dataset S2). Thus, it appears number of bases in the oligo(dA) RT-PCR sequences were am- that base editing commences before transcript polyuridylylation, biguous. Likewise, individual cloned RT-PCR products showed and only completely edited transcripts are polyuridylylated, as few differences in sequence (Dataset S2). These results indicate with other peridinin dinoflagellates. that, in the population of polyuridylylated transcripts, editing at the majority of individual sites had essentially gone to completion. Discussion Fifty-eight percent of the editing events were predicted to alter Our observations confirm that 3′ terminal poly(U) tracts are added the sequence of the translation product of the transcript in ques- during RNA processing in the chloroplasts of K. mikimotoi,as tion, and there were two instances—in the psaA transcript—of the seen in the ancestral chloroplasts of peridinin dinoflagellates and conversion of premature termination codons to coding sequence C. velia. We additionally show that transcripts from K. mikimotoi by editing (Fig. 3A and Dataset S2). Eight different types of base chloroplasts are subject to extensive base editing, as observed in interconversion were represented within the total transcript pool peridinin chloroplast lineages. As we found no evidence for sim- (Fig. 3B). Although the extent of bias varied, the psbA, psbC, ilar transcript polyuridylylation or editing in either the haptophyte psbD,andpsaA transcripts appeared to contain particularly high E. huxleyi or the diatom P. tricornutum, the most parsimonious frequencies of uracil-to-cytosine conversions (Fig. 3B), in con- explanation is that these transcript-processing pathways arose trast to peridinin dinoflagellates, which predominantly perform once to support the chloroplast lineage in the common ancestor adenine-to-guanine conversions, or plant chloroplasts, which typi- of peridinin dinoflagellates and C. velia (Fig. S4). The presence of cally perform cytosine-to-uracil conversions (3, 20, 31, 32, 34). We poly(U) tracts in K. mikimotoi therefore implies that transcript- could not observe any differences between the RT-PCR product processing pathways were retained from the peridinin chloroplast

Dorrell and Howe PNAS Early Edition | 3of6 Downloaded by guest on September 25, 2021 Fig. 3. Base editing in K. mikimotoi chloroplast transcripts. (A) Tabulated base-editing events observed in K. mikimotoi psbA, psbC, psbD, psaA, and rbcL transcripts. Sequences were obtained using PCR forward primers against the 5′ end of each gene (Dataset S2). (B) The total extent of different types of base editing observed. (C) Poly(U)-associated base editing. The sequences cover positions 890–940 of K. mikimotoi psbA as obtained from (top to bottom): genomic DNA, cDNA generated from nonpolyuridylylated transcripts, and oligo(dA) cDNA. Although three editing events (at positions 897, 914, and 937, labeled with gray arrows) occur before transcript polyuridylylation, two events (at positions 909 and 918, labeled with black arrows) are speciﬁcally associated with polyuridylylated transcripts.

and applied to the replacement fucoxanthin chloroplast lineage, the fast sequence evolution observed in dinoflagellate chloroplasts dramatically altering its RNA metabolism (Fig. S4). The associa- is a feature of the host, the superimposition of ancestral chloro- tion of polyuridylylation with completion of editing demonstrates plast transcript-processing pathways on replacement chloroplasts that the processes remain functionally interconnected in their could be an important step in the replacement, allowing the new new environment. The application of the pathways to the K. lineage to tolerate a host environment that might otherwise prove mikimotoi chloroplast rbcL transcript is particularly striking, as disadvantageous (2, 19). the rbcL gene of peridinin dinoflagellates and C. velia is located in One outstanding question is which RNA-processing factors have the nucleus, and its transcripts do not receive a poly(U) tract (29, specifically been retained through serial endosymbiosis in fuco- 38). Moreover, the fact that the RuBisCo of K. mikimotoi is xanthin chloroplast lineages. Given the limited coding capacity a form ID enzyme (comprising eight large and eight small of the chloroplast genome of peridinin dinoflagellates, poly- subunits), unlike the form II enzyme (comprising two large and uridylylation and transcript editing are believed to depend on two small subunits) found in peridinin dinoflagellates and C. nuclear-encoded proteins (3, 27). Characterizing the factors velia (23, 29, 38), shows that the poly(U) and editing pathways can involved in polyuridylylation and editing will be a major task, be successfully applied to chloroplast transcripts that do not have as the effector proteins involved in chloroplast transcript editing direct homologs in the ancestral, peridinin-containing lineage. Al- and 3′ modification remain poorly characterized even in plants though it has previously been demonstrated that some genes for (20, 40, 41), and poly(U) polymerases involved in nuclear and chloroplast proteins derived from the ancestral peridinin symbi- mitochondrial RNA metabolism of other lineages appear to have osis may be retained in serial dinoflagellates (16–18), none of arisen independently from a range of nucleotide polymerase the examples found thus far have been shown to generate families (41, 42). In addition, it remains to be shown whether products that are functional in the respective replacement chlo- polyuridylylation and editing are functionally interconnected to roplast lineages. In addition, none of the predicted products other features of chloroplast RNA metabolism in K. mikimotoi— would confer a biochemical activity on the replacement chloro- such as transcript end maturation—as has been hypothesized in plast that it was previously lacking. peridinin dinoflagellate lineages (30, 33, 34). The acquisition of transcript polyridylylation and editing may We have demonstrated the application of two previously exist- have had profound long-range evolutionary implications for the ing transcript-processing steps—polyuridylylation and editing—to fucoxanthin chloroplast lineage. The chloroplasts of both peridinin a replacement chloroplast previously lacking these pathways and and fucoxanthin-containing dinoflagellates have long been known newly acquired through serial endosymbiosis. Our observations to be subject to particularly fast sequence evolution (6, 24, 27). It suggest an important addition to conventional models of chloro- has recently been shown that fucoxanthin-containing chloroplast plast evolution: that host lineages can retain biogenesis pathways genomes differ in several major ways from those of haptophytes. from prior symbioses and apply them to replacement chloroplasts, These include the loss of a significant proportion of genes from in which they may confer new functions. This might enhance the chloroplast genome and the accumulation of premature ter- the stability of the replacement chloroplast in the host cell or mination codons in several predicted genes (15, 28, 37). In such adapt the metabolic or regulatory pathways of the chloroplast to a highly divergent chloroplast system, a concerted transcript- the needs of the host (1, 2, 19). In the light of recent data in- editing and polyuridylylation pathway might correct otherwise dicating that serial endosymbiosis has occurred extensively across potentially deleterious mutations in genome sequence, for example, the eukaryotes, we propose that features of the chloroplast bi- by converting premature termination codons to coding sequence, ology of many prominent photosynthetic eukaryotes have been as in the case of the K. mikimotoi psaA transcript (Dataset S2). If optimized by functions derived from previous endosymbioses.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1212270109 Dorrell and Howe Downloaded by guest on September 25, 2021 Experimental Procedures was used in the initial amplification step, and nested forward-directed primers ′ Growth Conditions. K. mikimotoi RCC1513 was grown in modified k/2 me- to the 3 end of each gene were used in the primary and secondary ream- fi fi dium as per http://www.sb-roscoff.fr; A. carterae CCAP1102/6, P. tricornutum pli cations (Dataset S1). Products were con rmed by direct PCR using primers ′ CCAP1052/6, and E. huxleyi CCMP1516 were grown in f/2 medium. Cultures within the 3 UTR of each gene (Dataset S1). For psbA, we performed only the were maintained at 15 °C under 30 μE 12h:12h light/dark cycling. Cells were confirmatory PCR step, as a 3′ UTR sequence had been obtained in a previous harvested in late log phase. study (5).

Nucleic Acid Preparation. Total cellular RNA was extracted from K. mikimotoi Sequencing. PCR products were separated by gel electrophoresis, and cut using TRIzol reagent as previously described (29, 34). RNA was purified with and purified using the MinElute gel extraction kit (Qiagen). Cloned products an additional chloroform phase and precipitated in isopropanol at −20 °C were ligated into pGEM-T vector (Promega), transformed into competent overnight. RNA was treated with DNaseI (Qiagen) as per the manufacturer’s Escherichia coli DH5α, and purified with a GeneJET miniprep kit (Fermentas) instructions and repurified with an RNeasy kit (Qiagen). RNA was confirmed before sequencing. Products were sequenced using an Applied Biosystems to be DNA-free through two rounds of direct PCR. Genomic DNA was 3730xl DNA Analyzer. purified as previously described (35) from 100 mL K. mikimotoi. Nucleic acids − o were stored in diethylpyrocarbonate-treated water at 80 C. Phylogenetic Analysis. A32-× 1,796-aa concatenated PsaA/PsbA/PsbC/PsbD phylogeny was assembled using MAFFT (http://mafft.cbrc.jp/alignment/ RT-PCR. For each reaction, 200 ng total cellular RNA was reverse-transcribed software/) and hand-curated using MacClade (http://macclade.org/macclade. with SuperScript II reverse transcriptase (Invitrogen). For circular RT-PCR, html). Protein sequences for K. mikimotoi were defined by the translation RNA was precircularized using T4 RNA ligase as previously described (33, 35). products of polyuridylylated transcript RT-PCR sequences, as generated using PCRs were performed using GoTaq polymerase (Promega) as previously de- forward primers against the 5′ end of each gene (Fig. 3 and Dataset S1). Sites scribed (33, 35). For the initial diagnostic oligo(dA) RT-PCR, PCR forward that were absent or gapped in more than two taxa were manually removed primers directed against the interior of each gene were used, whereas for from the alignment. PhyML phylogenies were calculated using the MABL whole-gene substitution mapping, PCR forward primers that annealed online server (http://www.phylogeny.fr/version2_cgi/one_task.cgi?task_type= against the 5′ end of each gene were used (Dataset S1). For circular RT-PCR phyml), using the Dayhoff amino acid substitution matrix + Γ correction. of K. mikimotoi psbA and psbC, cDNA was generated using gene-specific primers against the interior of the coding sequence, and PCRs were per- Bootstrap values were calculated using 100 replicate phylogenetic analyses formed using outward-directed reverse and forward primers, respectively, with the same model. against the 5′ and 3′ ends of each gene (Dataset S1 and Fig. S2). ACKNOWLEDGMENTS. The authors thank Dr. Adrian Barbrook, Professor

Alison Smith, Dr. Irmtraud Horst, and Dr. Katherine Helliwell (University of EVOLUTION Thermal Asymmetric Interlaced PCR of 3′ UTR Sequences. Thermal asymmetric Cambridge) for assistance with the establishment of A. carterae, P. tricornu- interlaced PCR was performed for K. mikimotoi psbC, psbD, psaA,andrbcL tum, and E. huxleyi cultures. R.G.D. was supported by a Biotechnology and using 20 ng template genomic DNA per reaction, cycling conditions, and one Biological Sciences Research Council doctoral training grant. Cultures of of four arbitrary degenerate primers adapted from previous studies (Dataset K. mikimotoi were provided by the Roscoff Culture Collection under an ASSEM- S1) (6, 43). The gene-speciﬁc forward-directed primer used in oligo(dA) RT-PCR BLE grant (EFETPIDC).

1. Howe CJ, Barbrook AC, Nisbet RER, Lockhart PJ, Larkum AWD (2008) The origin of 19. Larkum AWD, Lockhart PJ, Howe CJ (2007) Shopping for plastids. Trends Plant Sci 12 plastids. Philos Trans R Soc Lond B Biol Sci 363(1504):2675–2685. (5):189–195. 2. Dorrell RG, Howe CJ (2012) What makes a chloroplast? Reconstructing the estab- 20. Barkan A (2011) Expression of plastid genes: Organelle-specific elaborations on a lishment of photosynthetic symbioses. J Cell Sci 125(Pt 8):1865–1875. prokaryotic scaffold. Plant Physiol 155(4):1520–1532. 3. Green BR (2011) Chloroplast genomes of photosynthetic eukaryotes. Plant J 66(1): 21. Ast M, et al. (2009) Diatom plastids depend on nucleotide import from the cytosol. 34–44. Proc Natl Acad Sci USA 106(9):3621–3626. 4. Yoon HS, Hackett JD, Bhattacharya D (2006) A genomic and phylogenetic perspective 22. Brand LE, Campbell L, Bresnan E (2012) Karenia: The biology and ecology of a toxic on endosymbiosis and algal origin. J Appl Phycol 18(3–5):475–481. genus. Harmful Algae 14:156–178. 5. Moustafa A, et al. (2009) Genomic footprints of a cryptic plastid endosymbiosis in 23. Takishita K, Nakano K, Uchida A (2000) Origin of the plastid in the anomalously diatoms. Science 324(5935):1724–1726. pigmented dinoflagellate Gymnodinium mikimotoi (Gymnodiniales, Dinophyta) as 6. Takishita K, Nakano K, Uchida A (1999) Preliminary phylogenetic analysis of plastid- inferred from phylogenetic analysis based on the gene encoding the large subunit – encoded genes from an anomalously pigmented dinoflagellate Gymnodinium mikimotoi of form I-type RuBisCO. Phycol Res 48(2):85 89. (Gymnodiniales, Dinophyta). Phycol Res 47(4):257–262. 24. Yoon HS, Hackett JD, Bhattacharya D (2002) A single origin of the peridinin- and fl 7. Ishida K, Green BR (2002) Second- and third-hand chloroplasts in dinoflagellates: fucoxanthin-containing plastids in dino agellates through tertiary endosymbiosis. – Phylogeny of oxygen-evolving enhancer 1 (PsbO) protein reveals replacement of a Proc Natl Acad Sci USA 99(18):11724 11729. nuclear-encoded plastid gene by that of a haptophyte tertiary endosymbiont. Proc 25. Howe CJ, Nisbet RER, Barbrook AC (2008) The remarkable chloroplast genome of dinoflagellates. J Exp Bot 59(5):1035–1045. Natl Acad Sci USA 99(14):9294–9299. 26. Iida S, Kobiyama A, Ogata T, Murakami A (2010) Differential DNA rearrangements of 8. Burki F, et al. (2012) Re-evaluating the green versus red signal in eukaryotes with plastid genes, psbA and psbD, in two species of the dinoflagellate Alexandrium. Plant secondary plastid of red algal origin. Genome Biol Evol 4(6):626–635. Cell Physiol 51(11):1869–1877. 9. Cuvelier ML, et al. (2010) Targeted metagenomics and ecology of globally impor- 27. Zhang Z, Green BR, Cavalier-Smith T (1999) Single gene circles in dinoflagellate tant uncultured eukaryotic phytoplankton. Proc Natl Acad Sci USA 107(33):14679– chloroplast genomes. Nature 400(6740):155–159. 14684. 28. Espelund M, et al. (2012) Genome fragmentation is not confined to the peridinin 10. Archibald JM, Rogers MB, Toop M, Ishida K, Keeling PJ (2003) Lateral gene transfer plastid in dinoflagellates. PLoS ONE 7(6):e38809. and the evolution of plastid-targeted proteins in the secondary plastid-containing 29. Janouskovec J, Horák A, Oborník M, Lukes J, Keeling PJ (2010) A common red algal alga Bigelowiella natans. Proc Natl Acad Sci USA 100(13):7678–7683. origin of the apicomplexan, dinoflagellate, and heterokont plastids. Proc Natl Acad 11. Hackett JD, et al. (2004) Migration of the plastid genome to the nucleus in a peridinin Sci USA 107(24):10949–10954. fl – dino agellate. Curr Biol 14(3):213 218. 30. Dang Y, Green BR (2010) Long transcripts from dinoflagellate chloroplast minicircles 12. Price DC, et al. (2012) Cyanophora paradoxa genome elucidates origin of photosyn- suggest “rolling circle” transcription. J Biol Chem 285(8):5196–5203. – thesis in algae and plants. Science 335(6070):843 847. 31. Wang YL, Morse D (2006) Rampant polyuridylylation of plastid gene transcripts in 13. Becker B, Hoef-Emden K, Melkonian M (2008) Chlamydial genes shed light on the the dinoflagellate Lingulodinium. Nucleic Acids Res 34(2):613–619. evolution of photoautotrophic eukaryotes. BMC Evol Biol 8(1):203. 32. Zauner S, Greilinger D, Laatsch T, Kowallik KV, Maier UG (2004) Substitutional editing 14.WoehleC,DaganT,MartinWF,GouldSB(2011) Red and problematic green phylogenetic of transcripts from genes of cyanobacterial origin in the dinoflagellate Ceratium signals among thousands of nuclear genes from the photosynthetic and apicomplexa- horridum. FEBS Lett 577(3):535–538. related Chromera velia. Genome Biol Evol 3:1220–1230. 33. Barbrook AC, et al. (2012) Polyuridylylation and processing of transcripts from mul- 15. Gabrielsen TM, et al. (2011) Genome evolution of a tertiary dinoflagellate plastid. tiple gene minicircles in chloroplasts of the dinoflagellate Amphidinium carterae. PLoS ONE 6(4):e19132. Plant Mol Biol 79(4–5):347–357. 16. Patron NJ, Waller RF, Keeling PJ (2006) A tertiary plastid uses genes from two 34. Dang Y, Green BR (2009) Substitutional editing of Heterocapsa triquetra chloroplast endosymbionts. J Mol Biol 357(5):1373–1382. transcripts and a folding model for its divergent chloroplast 16S rRNA. Gene 442(1–2): 17. Minge MA, et al. (2010) A phylogenetic mosaic plastid proteome and unusual plastid- 73–80. targeting signals in the green-colored dinoflagellate Lepidodinium chlorophorum. 35. Nash EA, et al. (2007) Organization of the mitochondrial genome in the dinoflagellate BMC Evol Biol 10:191. Amphidinium carterae. Mol Biol Evol 24(7):1528–1536. 18. Nosenko T, et al. (2006) Chimeric plastid proteome in the Florida “red tide” dinoflagellate 36. Zghidi-Abouzid O, Merendino L, Buhr F, Malik Ghulam M, Lerbs-Mache S (2011) Charac- Karenia brevis. Mol Biol Evol 23(11):2026–2038. terization of plastid psbT sense and antisense RNAs. Nucleic Acids Res 39(13):5379–5387.

Dorrell and Howe PNAS Early Edition | 5of6 Downloaded by guest on September 25, 2021 37. Sánchez Puerta MV, Bachvaroff TR, Delwiche CF (2005) The complete plastid genome 40. Bentolila S, et al. (2012) RIP1, a member of an Arabidopsis protein family, interacts with the sequence of the haptophyte Emiliania huxleyi: A comparison to other plastid genomes. protein RARE1 and broadly affects RNA editing. Proc Natl Acad Sci USA 109(22):E1453–E1461. DNA Res 12(2):151–156. 41. Lange H, Sement FM, Canaday J, Gagliardi D (2009) Polyadenylation-assisted RNA 38. Morse D, Salois P, Markovic P, Hastings JW (1995) A nuclear-encoded form II RuBisCO degradation processes in plants. Trends Plant Sci 14(9):497–504. in dinoﬂagellates. Science 268(5217):1622–1624. 42. Aphasizhev R (2005) RNA uridylyltransferases. Cell Mol Life Sci 62(19–20):2194–2203. 39. Oudot-Le Secq MP, et al. (2007) Chloroplast genomes of the diatoms Phaeodactylum 43. Liu YG, Mitsukawa N, Oosumi T, Whittier RF (1995) Efﬁcient isolation and mapping tricornutum and Thalassiosira pseudonana: Comparison with other plastid genomes of Arabidopsis thaliana T-DNA insert junctions by thermal asymmetric interlaced PCR. of the red lineage. Mol Genet Genomics 277(4):427–439. Plant J 8(3):457–463.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1212270109 Dorrell and Howe Downloaded by guest on September 25, 2021