Horizontal gene transfer of the algal nuclear genepsbO to the photosynthetic sea slug Elysia chloroticaMary E. Rumphoa,1, Jared M. Worfula, Jungho Leeb, Krishna Kannana, Mary S. Tylerc, Debashish Bhattacharyad,Ahmed Moustafad, and James R. Manharte

aDepartment of Biochemistry, Microbiology, and Molecular Biology, University of Maine, Orono, ME 04469; bGreen Plant Institute, Seoul NationalUniversity, Gwonseon, Suwon, Gyeonggi 441-853, Korea; cSchool of Biology and Ecology, University of Maine, Orono, ME 04469; dDepartment of BiologicalSciences and the Roy J. Carver Center for Comparative Genomics, Interdisciplinary Program in Genetics, University of Iowa, Iowa City, IA 52242-1324;and eDepartment of Biology, Texas A&M University, College Station, TX 77843

Edited by Lynn Margulis, University of Massachusetts, Amherst, MA, and approved September 17, 2008 (received for review June 9, 2008)

The sea slug Elysia chlorotica acquires plastids by ingestion of itsalgal food source Vaucheria litorea. Organelles are sequestered inthe mollusc’s digestive epithelium, where they photosynthesizefor months in the absence of algal nucleocytoplasm. This is per-plexing because plastid metabolism depends on the nuclear ge-nome for >90% of the needed proteins. Two possible explanationsfor the persistence of photosynthesis in the sea slug are (i) theability of V. litorea plastids to retain genetic autonomy and/or (ii)more likely, the mollusc provides the essential plastid proteins.Under the latter scenario, genes supporting photosynthesis havebeen acquired by the animal via horizontal gene transfer and theencoded proteins are retargeted to the plastid. We sequenced theplastid genome and confirmed that it lacks the full complement ofgenes required for photosynthesis. In support of the secondscenario, we demonstrated that a nuclear gene of oxygenic pho-tosynthesis, psbO, is expressed in the sea slug and has integratedinto the germline. The source of psbO in the sea slug is V. litoreabecause this sequence is identical from the predator and preygenomes. Evidence that the transferred gene has integrated intosea slug nuclear DNA comes from the finding of a highly divergedpsbO 3� flanking sequence in the algal and mollusc nuclear homo-logues and gene absence from the mitochondrial genome of E.chlorotica. We demonstrate that foreign organelle retention gen-erates metabolic novelty (‘‘green animals’’) and is explained byanastomosis of distinct branches of the tree of life driven bypredation and horizontal gene transfer.

symbiosis � Vaucheria litorea � evolution � plastid � stramenopile

Symbiotic associations and their related gene transfer eventsare postulated to contribute significantly to evolutionary

innovation and biodiversity. This comes from extensive analysisof organelles such as plastids (e.g., chloroplasts) that originatedvia primary endosymbiosis of a free-living cyanobacterium (1, 2).The cyanobacterial genome was greatly reduced by endosymbi-otic gene transfer (EGT) to the host nucleus and wholesale geneloss, giving rise to the primary lineages of plants and green algae(streptophytes and chlorophytes), red algae (rhodophytes), andglaucophytes (3–6) [see the scheme in supporting information(SI) Fig. S1]. The diverse group of secondary or ‘‘complex’’ algae(e.g., chromalveolates, euglenids), in turn, arose by secondaryendosymbiosis—the uptake of a eukaryotic alga (green or redlineage) by a heterotrophic eukaryotic host. In this case, inaddition to EGT, transfer of genes between the unrelatedorganisms by lateral or horizontal gene transfer (HGT) and lossof genes occurred as a result of the ‘‘merger’’ of the two nuclei(host and endosymbiont) (7). As a result of primary and sec-ondary endosymbiosis, plastid genomes (ptDNAs) encode lessthan 10% of the predicted 1,000 to 5,000 proteins required tosustain the metabolic capacity of the plastid (8, 9).

Examples of HGT between unrelated or nonmating species areabundant among prokaryotes (10, 11) but less so between pro-karyotes and unicellular (12–14) or multicellular eukaryotes (15–

20). Most of these latter examples are associated with parasitism orphagotrophy, including the elegant studies of HGT from the�-proteobacteria Wolbachia to insects and nematodes (16–18), andthe finding of rhizobial-like genes in plant parasitic nematodes (19,20). The exchange of genetic material between two eukaryotes isextremely rare, or at least not well documented to date. Thebest-studied cases include the transfer of mitochondrial DNA fromachlorophyllous or epiphytic plants to the mitochondrial genome(mtDNA) of their closely related photosynthetic hosts (21), theexchange of transposons between two animal (22) or two plant (23)species, and the presence of plant genes in plant parasitic nematodes(in addition to the rhizobial genes discussed previously), which arehypothesized to be ‘‘defense’’ genes whose products protect theparasite from host detection (20).

The sacoglossan mollusc (sea slug) Elysia chlorotica representsa unique model system to study the potential for interdomainHGT between two multicellular eukaryotes—in this case, froma filamentous secondary (heterokont) alga (Vaucheria litorea) toa mollusc. This emerald green sea slug owes its coloring andphotosynthetic ability to plastids acquired during herbivorousfeeding (24–29). The plastids do not undergo division in the seaslug and are sequestered intracellularly in cells lining the finelydivided digestive diverticula. The plastids continue to carry outphotosynthesis, providing the sea slug with energy and carbonduring its approximately 10-month life span (27, 28). Long-termplastid activity continues despite the absence of algal nuclei (27,29), and hence a source of nuclear-encoded plastid-targetedproteins. We hypothesize that the algal nuclear genes encodingessential plastid proteins are present in the sea slug, presumablyas a result of HGT. Here, we present evidence for such inter-domain HGT of psbO, a nuclear gene encoding the plastidmanganese-stabilizing protein (MSP � PsbO). MSP is a subunitof the photosystem II complex associated with photosyntheticoxygen evolution (30, 31), which is, unquestionably, the mostimportant enzyme complex of oxygenic life.

Results and DiscussionPlastid Genetic Autonomy. The plastids in E. chlorotica are nottransmitted vertically; rather, they must be acquired with eachgeneration early in development to ensure maturation to the

Author contributions: M.E.R. and J.R.M. designed research; J.M.W., J.L., and K.K. performedresearch; J.L., M.S.T. contributed new reagents/analytic tools; K.K., D.B., A.M., and J.R.M.analyzed data; and M.E.R. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Data deposition: The sequences reported in this paper have been deposited in the GenBankdatabase [accession nos. EU912438 (V. litorea complete ptDNA), EU599581 (E. chloroticacomplete mtDNA), DQ514337 (V. litorea psbO cDNA), EU621881 (V. litorea psbO DNA), andEU621882 (E. chlorotica psbO DNA).

1To whom correspondence should be addressed. E-mail: mrumpho@umit.maine.edu.

This article contains supporting information online at www.pnas.org/cgi/content/full/0804968105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0804968105 PNAS � November 18, 2008 � vol. 105 � no. 46 � 17867–17871

EVO

LUTI

ON

adult sea slug (32). Laboratory coculturing studies were carriedout to establish that the alga V. litorea, a derived heterokont algathat contains secondary plastids of red algal origin (33) (Fig. S1),was the sole source of plastids in the sea slugs (Fig. 1, Movie S1,Movie S2, and SI Methods). Subsequent sequencing and mappingof the V. litorea ptDNA (only the fifth heterokont ptDNA to bepublished to date) revealed a very compact 115,341-bp double-stranded circular genome encoding 169 genes, including 139protein-encoding genes (14 are conserved ORFs [ORF-designated ycfs] and 2 are unknown ORFs), 27 tRNA genes, and3 rRNA genes (Fig. 2). The genes are densely arranged with anaverage intergenic region of 74bp and 11.1% noncoding DNA.The overall G � C content is 28%, which is low compared withmost plastids, including other related heterokonts (34, 35). Thegenome is separated into two large single-copy regions (62,002bpand 43,469bp) by two smaller inverted repeats (both 4935bp). Allthe plastid rRNA genes are found on both copies of the invertedrepeat in the highly conserved operon rrs-trnI-trnA-rrl-rrf. Unlikethe four published heterokont ptDNAs that lack introns (34, 35),V. litorea contains one intron in the trnL UAA gene—an ancientintron that is also present in cyanobacteria (36). In addition,V. litorea has retained the genes for the light-independentchlorophyll biosynthesis pathway: chlB, chlL, and chlN. How-ever, as expected, the V. litorea ptDNA shares more similaritywith heterokonts and red algae than it does with green plastids(34, 35, 37) (see complete inventory of plastid genes by categoryin Table S1).

Examining the genetic autonomy of V. litorea ptDNA revealedthe absence of the major core protein of the oxygen evolvingcomplex of photosystem II, MSP (encoded by psbO). MSP has

been reported to be critical to the stability of the water-splittingreaction of photosynthesis that generates atmospheric oxygen(30, 31). The evolutionary conservation of this reaction isdemonstrated by the presence of MSP in all oxygenic photosyn-thetic organisms (30). Likewise, animal genomes have neverbeen shown to contain psbO; hence, MSP cannot be made by thesea slug in the absence of HGT. We have previously demon-strated that oxygen evolution is linked to photosynthetic electrontransport in the sea slug for at least 5 months after being removedfrom its algal prey (27), and photosystem II is generally highlysusceptible to photo-oxidative damage requiring de novo syn-thesis and reassembly of its subunits (38, 39). For these reasons,we targeted psbO for HGT from V. litorea to E. chlorotica.

HGT and Expression of psbO. Heterologous degenerate primers(Table S2) were designed based on alignments of published psbOsequences to amplify an internal fragment using reverse tran-scriptase (RT)-PCR. A 452-bp fragment was amplified fromboth algal and sea slug cDNA (5 months after algal feeding) (Fig.3A). The translated product was used to blast the GenBankdatabase, which revealed a relatively high identity (48%–68%)to several secondary red algal–derived MSP amino acid se-quences. The entire V. litorea psbO cDNA sequence was thenobtained using 5�- and 3�-RACE, and this sequence was used todesign homologous primers to amplify a 963-bp internal frag-ment of the V. litorea psbO cDNA (Fig. 3C). These same primersyielded a similar sized PCR product from sea slug cDNA andDNA templates (Fig. 3C), the sequences of which were identicalto the V. litorea psbO cDNA sequence beginning with the startcodon (there are no introns in the algal gene; Fig. S2).

Although it had been several months since the sea slugs hadbeen in contact with any algal prey, the possibility of algalnuclei remaining in the gut of the sea slug and contaminatingthe total genomic DNA preparation was eliminated by carryingout the same PCR on sea slug egg DNA. Because plastids are

B

C

A

D

Fig. 1. Laboratory culturing of E. chlorotica. (A) Free-swimming E. chloroticaveliger larvae. (Scale bar, 100 �m.) Under laboratory conditions, the veligerlarvae develop and emerge from plastid-free sea slug–fertilized eggs withinapproximately 7 days. The green coloring in the digestive gut is attributableto planktonic feeding and not to the acquisition of plastids at this stage.Metamorphosis of the larvae to juvenile sea slugs requires the presence offilaments of V. litorea. (B) Metamorphosed juvenile sea slug feeding for thefirst time on V. litorea. (Scale bar, 500 �m.) The grayish-brown juveniles losetheir shell, and there is an obligate requirement for plastid acquisition forcontinued development. This is fulfilled by the voracious feeding of thejuveniles on filaments of V. litorea. (Also see Movie S1). (C) Young adult seaslug 5 days after first feeding. (Scale bar, 500 �m.) By a mechanism not yetunderstood, the sea slugs selectively retain only the plastids in cells that linetheir highly branched digestive tract. (D) Adult sea slug. (Scale bar, 500 �m.)As the sea slugs further develop and grow in size, the expanding digestivediverticuli spread the plastids throughout the entire body of the mollusc,yielding a uniform green coloring. (Also see Movie S2.) From these controlledrearing studies, we were able to conclude that the only source of plastids inour experimental sea slugs was V. litorea.

Vaucheria litoreachloroplast genome

Fig. 2. Map of the ptDNA of V. litorea. Genes on the outside are transcribedin the clockwise direction, whereas genes on the inside are transcribed in thecounterclockwise direction. Genes are color coded according to their functionas shown. tRNAs are listed by the one-letter amino acid code followed by theanticodon. The only gene with an intron (L-uaa) is indicated by an asterisk.

17868 � www.pnas.org�cgi�doi�10.1073�pnas.0804968105 Rumpho et al.

not inherited in E. chlorotica, eggs provide a source of animalDNA and RNA that is free of algal contamination (27).Amplification of the sea slug egg DNA with the same primersresulted in a 963-bp fragment (Fig. 3C) with a sequenceidentical to the algal and sea slug psbO fragments (Fig. S2). Asfurther PCR controls, primers complementary to the V. litoreainternal transcribed spacer region (ITS1; GenBank EF441743)were used as a positive algal nuclear control (27); no productwas observed from sea slug or sea slug egg DNA templates.Likewise, attempts to amplify psbO from negative controls(pufferfish and Dictyostelium DNA) using the same primersand reagents did not yield a positive PCR product. Finally,identical PCR results have been obtained from sea slugscollected from the same site on multiple occasions over a3-year period (results not shown). Expression of the foreigngene in the sea slug was further supported by Northern blotanalysis. The V. litorea psbO probe cross-reacted with a 1.2-kbtranscript for both V. litorea and E. chlorotica RNA, along witha slightly larger transcript (1.6 kb) in the sea slug (Fig. 3B).

The identical translated MSP amino acid sequences for bothV. litorea and E. chlorotica (Fig. 3D) were analyzed by phyloge-netic methods (40, 41) incorporating 23 published MSP se-

quences. This revealed nesting of the sequences in the red algallineage in a clade containing other heterokonts (Fig. S3). Asexpected for this secondary lineage, the V. litorea MSP prepro-tein contains a tripartite targeting sequence (Fig. 3D; as pre-dicted by refs. 42, 43). This reflects the presence of chloroplastendoplasmic reticulum around the complex plastids, which mustbe traversed by proteins targeted to the plastid (44). What isinteresting is that the MSP preprotein in the sea slug has retainedthe entire tripartite targeting sequence (Fig. 3D), especially inlight of the observation that the chloroplast endoplasmic retic-ulum does not appear to be retained around the engulfedchloroplasts (28).

Recently, it was reported that nuclear genes encoding plastid-localized light harvesting complex proteins ( fcp, lhcv1, andlhcv2) have also been transferred from V. litorea to E. chlorotica(45). Using a similar PCR approach, identical nucleotide se-quences were reported for sea slug and algal fcp and lhcv1, andonly a single base substitution was found between larval lhcv2and adult sea slug or alga lhcv2. Although evidence fromSouthern blotting has not been achieved in the study reportedhere or for the light harvesting complex protein genes (45), wewere able to obtain sequence information using genome walkingfor the 3� untranslated flanking region of the psbO gene fromboth algal DNA and sea slug egg DNA. A nested gene-specificprimer coupled with an adapter-specific primer (Table S2)yielded a 3� f lanking sequence from both organisms that wasidentical for the first 81bp corresponding to the 3� end of thepsbO gene and ending with the stop codon (Fig. S4). Thissequence was followed by a highly diverged sequence corre-sponding to the 3� untranslated region in each genome. Theseresults support the interdomain transfer of an algal gene to amollusc, its expression in the foreign host, and also that the genehas been inserted into the germline, even though the plastids arenot yet transmitted vertically in the sea slug.

Mechanism and Site of Integration of Transferred Genes. Similar tomany other phagocytic or parasitic relations that lead to pre-sumptive HGT events, the E. chlorotica/V. litorea plastid endo-symbiosis involves intimate physical contact between predatorand prey. During the sea slug’s phagocytic feeding, the algalnuclei come into direct physical contact with the sea slugdigestive epithelium. Upon nuclear rupture in the gut, pieces ofalgal chromosomal DNA (and possibly transcripts) may havebeen randomly transferred by ‘‘bulk transfer’’ or viral transmis-sion (46) to the sea slug. Two potential sites for insertion offoreign genes in the sea slug are the nuclear genome and themtDNA. Mitochondrion-to-mitochondrion gene transfer is nowrecognized as a dominant mode of HGT in plants because of thelarger and more plastic mtDNAs in these taxa (21). The smallercompact animal or metazoan mitochondrion genome is generallybelieved to be a poorer target for foreign gene insertion.However, some basal metazoans do exhibit greater variation inmtDNA size and gene content (47). This includes multipleexamples of HGT of group I intron sequences (normally notfound in animals) into the mtDNA of a sponge (48), a seaanemone (49), and a coral (50).

To determine if the mtDNA of E. chlorotica serves as a targetfor any foreign genes, including psbO, we used PCR and primerwalking to obtain the complete sequence and map the 14,132-bpmtDNA from sea slug eggs (Fig. 4). The genome was found toencode the standard 37 genes found in other typical animalmitochondria (see ncbi.nlm.nih.gov/genomes/ORGANELLES/33208). No introns were identified, and only 0.0125% of theDNA was noncoding. By measuring the G � C content overadjacent windows of 500 nt with 200-nt overlaps, the values werefound to be uniformly distributed across the windows, suggestinghomogeneity in GC content of the mtDNA and not supporting

500 bp400 bp

V. l

itore

a

1.2 kb 1.6 kb

Vli Ecl (-) Stds Vli Ecl

C

~960 bp

PCR RT-PCR

egg adult adultV. litorea E. chlorotica

A B

D MKVPSALVALSAFSVKTSAFRPAFAGLKTNAKSSSALTMSVQDDIKTLAVGALTILAGVSILNAPVEAITKDQIESLSYLQVKGTGLANRCPEVFGTGSIDVNGKTKIVDMCIEPKTFQVLEETSSKRGEAKKEYVNTKLMTRQTYTLYGIDGSFAPENGKITFREKDGIDYAATTIQLPGGERVPFLFTVKELVAQATTPGNSVTPGLQFGGPFSTPSYRTGLFLDPKGRGGSTGYDMAVALPGHQSGIEGDAELFGENNKTFDVTKGNIEFEVNRVDPSNGEIGGVFVSKQKGDTDMGSKVPKDLLIKGIFYGRLESE

Fig. 3. Expression of psbO in V. litorea and E. chlorotica. (A) RT-PCR usingheterologous primers to psbO amplified a 452-bp fragment from both algaland adult sea slug cDNA. Water served as the negative control. Standardswere a 1-kb Plus DNA ladder (Invitrogen). Vli, Vaucheria litorea, Ecl, Elysiachlorotica. (B) Northern blot analysis employing the cloned V. litorea 452bp psbO product as probe hybridized with a 1.2-kb transcript for V. litoreaand E. chlorotica as well as a 1.6-kb transcript in the sea slug. RNAMillennium Size Markers (Ambion) were run to estimate transcript size. (C)Homologous primers were designed from the RACE-amplified sequence ofthe V. litorea psbO fragment in A. By PCR, these primers amplified a 963-bpproduct from genomic DNA of V. litorea and E. chlorotica eggs and adulttissue as well as E. chlorotica adult cDNA by RT-PCR. (D) Translation of thepsbO sequences obtained from the 963-bp products in B, for both V. litoreaand E. chlorotica, yielded an identical amino acid sequence with a putativetripartite targeting signal for MSP. The signal sequence is in red, the transitpeptide is in blue, and the thylakoid targeting domain is in green. Note thehighly conserved phenylalanine residue at the cleavage site of the signalsequence.

Rumpho et al. PNAS � November 18, 2008 � vol. 105 � no. 46 � 17869

EVO

LUTI

ON

the existence of a chimeric region (Fig. S5). To assess further thepossibility of HGT in E. chlorotica mitochondria, we did phylo-genetic analyses with nucleotide data generated using a slidingwindow approach with the genome data (i.e., DNA sequencesthat are independent of gene structure) and using the completetranslated ORFs. The maximum likelihood phylogenetic treesinferred with these alignments showed that the E. chloroticasequences are monophyletic with molluscs, consistent with avertical evolutionary history for E. chlorotica mtDNA (e.g., seethe phylogenetic tree of cytochrome b in Fig. S6). These analysespoint to an intact and ‘‘typical’’ animal mtDNA in E. chlorotica.We, however, do not argue absolutely against the possibility ofa partial DNA insertion from an algal or other source in thisgenome; rather, that if such an insertion exists, it is not detect-able using the approaches described here. In any case, it isundoubtedly more likely that large-scale gene insertion would bemore readily accommodated in sea slug nuclear DNA than inmtDNA, and high-throughput genome sequencing will be nec-essary to prove this idea.

ConclusionsMolecular evidence is presented supporting eukaryotic multi-cellular interdomain HGT (including into the germline) using amollusc model and expression of an essential algal nuclear generequired for photosynthesis. Many questions remain to be an-swered, however; for example, the chromosomal location andadditional f lanking sequences of the psbO gene in the sea slug.Key will be to establish how this gene was activated in the molluscand to identify the mechanism of plastid protein targeting. It isalso very likely that HGT contributes to the long-term survivaland functioning of V. litorea plastids in E. chlorotica and thatmany more algal nuclear genes have been transferred in the seaslug. In light of these findings, the prospect of natural HGTtaking place between distantly related organisms, especially with

any physical contact, must be considered formally possible. Thisis especially true in the context of genetically modified organ-isms. The implications for evolution and speciation throughacquisition of foreign parts and selected genes to produce newlineages, as proposed by Margulis (2), are heightened by thisunusual photosynthetic mollusc.

MethodsExperimental Materials. E. chlorotica was collected from Martha’s VineyardIsland in Massachusetts and maintained without algae in aquaria containingaerated artificial seawater (925 mosmol; Instant Ocean, Aquarium Systems) at12 °C during a 14-h photoperiod (27). After 3 months, eggs produced by E.chlorotica were used to initiate culturing experiments as described in SIMethods. V. litorea CCMP2940 filaments were maintained in a modified f/2medium (27).

Nucleic Acid Preparation. DNA and RNA were isolated from sea slugs (5 monthsafter feeding or collection), sea slug eggs, and algal filaments using DNAzol orDNAzol extra strength (Molecular Research Center, Inc.) and the RNeasy minikit (Qiagen), respectively, unless noted differently. RNase and DNase wereadded during the extraction process for DNA and RNA, respectively, andnegative controls were run on each. First-strand cDNA was synthesized usingSuperScript II ribonuclease H� RT (Invitrogen) and oligo d(T) priming onDNase-treated RNA.

PCR Amplification and Northern Blotting. Degenerate primers (psbO R andpsbO L2; see Table S2) were designed to amplify an internal fragment of V.litorea psbO based on the conserved regions of several heterokont and redalga psbO sequences (for list, see SI Methods). This 452-bp psbO fragment wasthen used as a probe for Northern blot analysis with the Northern Maxkit (Ambion) and RediprimeII random prime labeling system (AmershamBiosciences).

RACE and Phylogenetic Analysis of psbO. The complete V. litorea psbO genewas obtained by rapid amplification of cDNA ends (RACE) using the Gene-Racer Kit (Invitrogen) and primers listed in Table S2. Homologous primers(psbO L5 and psbO R8) were then designed to amplify a larger (963-bp)internal fragment of the V. litorea psbO cDNA. Phylogenetic analysis ofpsbO (MSP) was based on amino acid sequences of 25 mature proteins (forlist, see SI Methods) and carried out using maximum parsimony in PAUP4.0b10 (41).

ptDNA Sequencing. ptDNA was purified from V. litorea filaments as de-scribed (51). ptDNA was digested with the restriction endonucleases PstI,HindIII, and EcoRI. The PstI and HindIII fragments were cloned, and arestriction site map using all three enzymes was produced as described byLehman and Manhart (52). A total of 104 kb was obtained from clonedrestriction fragments. The remainder of the genome (11 kb) was obtainedby PCR amplification (53). Thirty-eight oligonucleotide primers were usedto fill gaps between cloned restriction fragments and to check fragmentconnections, using all possible combinations of these primers. Fragmentswere sequenced by primer walking (53).

Genome Walking. Clontech’s Genome Walking Kit was used with gene-specificand adapter primers (sequences presented in Table S2) to amplify the 3� endof the psbO gene and the flanking untranslated region using algal DNA andsea slug egg DNA (see SI Methods for additional details).

mtDNA Sequencing. Universal primers (ref. 54; Table S3) were used to amplifyfragments of the mitochondrial rrnL, cob, and cox1 regions from sea slug eggDNA and then in various combinations to amplify the entire mtDNA. Themitochondrial sequence was annotated using Dual Organellar GenoMe An-notator (DOGMA) (55), and the map was drawn using OrganellarGenome-DRAW (OGDRAW) (56). Additional information, including analysis for HGT, isprovided in SI Methods.

ACKNOWLEDGMENTS. This research was supported by National Science Foun-dation grants IBN-9808904 (M.R. and J.M.) and IOS-0726178 (M.R. and M.T.);the American Society of Plant Biologists’ Education Foundation (M.R. andM.T.); the Maine Technology Institute (M.R.); Ministry for Food, Agriculture,Forestry, and Fisheries, Korean Government, Korea Research Foundation (J.L.);the National Institutes of Health (grant R01ES013679 to D.B.), and the UniversityofMaine (M.R.). This ismanuscriptno.3024of theMaineAgricultureandForestryExperiment Station Hatch Project no. ME08361-08MRF (NC 1168).

+

-

-

+

Fig. 4. Map of the mtDNA of the sacoglossan mollusc E. chlorotica. Genestranscribed clockwise are shown on the outside of the circle, whereas thosetranscribed counterclockwise are shown to the inside of the circle. Namesof tRNA genes are indicated by the three-letter amino acid code with thetwo leucine and two serine tRNAs differentiated by � and � signs, recog-nizing codons UAG and UAA for leucine and AGN and UCN for serine,respectively.

17870 � www.pnas.org�cgi�doi�10.1073�pnas.0804968105 Rumpho et al.

1. Martin W, Kowallik KV (1999) Annotated English translation of Mereschkowsky’s 1905paper Uber Natur und Ursprung der Chromatophoren im Pflanzenreiche. Eur J Phycol34:287–295.

2. Margulis L, Sagan D (2003) Acquiring genomes: A theory of the origins of species. (BasicBooks, New York, NY).

3. Martin W (2003) Gene transfer from organelles to the nucleus: Frequent and in bigchunks. Proc Natl Acad Sci USA 100:8612–8614.

4. Reyes-Prieto A, Weber AP, Bhattacharya D (2007) The origin and establishment of theplastid in algae and plants. Annu Rev Genet 41:147–168.

5. Margulis L (1970) Origin of Eukaryotic Cells. (Yale Univ Press, New Haven, CT).6. Mulkidjanian AY, et al. (2006) The cyanobacterial genome core and the origin of

photosynthesis. Proc Natl Acad Sci USA 103:13126–13131.7. Lane CE, Archibald JM (2008) The eukaryotic tree of life: Endosymbiosis takes its TOL.

Trends Ecol Evol 23:268–275.8. Martin W, et al. (2002) Evolutionary analysis of Arabidopsis, cyanobacterial, and

chloroplast genomes reveals plastid phylogeny and thousands of cyanobacterial genesin the nucleus. Proc Natl Acad Sci USA 99:12246–12251.

9. Richly E, Leister D (2004) An improved prediction of chloroplast proteins revealsdiversities and commonalities in the chloroplast proteomes of Arabidopsis and rice.Gene 329:11–16.

10. Koonin EV, Makarova KS, Aravind L (2001) Horizontal gene transfer in prokaryotes:Quantification and classification. Annu Rev Microbiol 55:709–742.

11. Dagan T, Martin W (2007) Ancestral genome sizes specify the minimum rate of lateralgene transfer during prokaryote evolution. Proc Natl Acad Sci USA 104:870–875.

12. Andersson JO (2005) Lateral gene transfer in eukaryotes. Cell Mol Life Sci 62:1182–1197.

13. Loftus B, et al. (2005) The genome of the protist parasite Entamoeba histolytica. Nature33:865–868.

14. Nosenko T, Bhattacharya D (2007) Horizontal gene transfer in chromalveolates. BMCEvol Biol 7:173.

15. Starcevic A, et al. (2008) Enzymes of the shikimic acid pathway encoded in the genomeof a basal metazoan, Nematostella vectensis, have microbial origins Proc Natl Acad SciUSA 105:2533–2537.

16. Nikoh N, et al. (2008) Wolbachia genome integrated in an insect chromosome: evolu-tion and fate of laterally transferred endosymbiont genes. Genome Res 18:272–280.

17. Fenn K, et al. (2006) Phylogenetic relationships of the Wolbachia of nematodes andarthropods. PLoS Pathog 2:e94.

18. Dunning-Hotopp JC, et al. (2007) Widespread lateral gene transfer from intracellularbacteria to multicellular eukaryotes. Science 317:1753–1756.

19. Scholl EH, Thorne JL, McCarter JP, Bird DMcK (2003) Horizontally transferred genes inplant-parasitic nematodes: A high-throughput genomic approach. Genome Biol 4:R39.

20. Wasmuth J, Schmid R, Hedley A, Blaxter M (2008) On the extent and origins of genicnovelty in the Phylum Nematoda. PLoS Negl Trop Dis 2:e258.

21. Richardson AO, Palmer JD (2007) Horizontal gene transfer in plants. J Exp Bot 58:1–9.22. Lampe DJ, Witherspoon DJ, Soto-Adames FN, Robertson HM (2003) Recent horizontal

transfer of Mellifera subfamily mariner transposons into insect lineages representingfour different orders shows that selection acts only during horizontal transfer. Mol BiolEvol 20:554–562.

23. Roulin A, Piegu B, Wing RA, Panaud O (2008) Evidence of multiple horizontal transfersof the long terminal repeat retrotransposon RIRE1 within the genus Oryza. Plant J53:950–959.

24. Trench RK (1975) Of ‘leaves that crawl’: Functional chloroplasts in animal cells. Sym-posia of the Society for Experimental Biology, ed Jennings DH (Cambridge Univ Press,London), pp 229–265.

25. Rumpho ME, Summer EJ, Manhart JR (2000) Solar-powered sea slugs. Mollusc/algalchloroplast symbiosis. Plant Physiol 123:29–38.

26. Rumpho ME, Dastoor FP, Manhart JR, Lee J (2006) in Advances in Photosynthesis andRespiration—The Structure and Function of Plastids, eds Wise RR, Hoober JK (Springer,Berlin), pp 451–473.

27. Green BJ, et al. (2000) Mollusc-algal chloroplast endosymbiosis: Photosynthesis, thy-lakoid protein maintenance, and chloroplast gene expression continue for manymonths in the absence of the algal nucleus. Plant Physiol 124:331–342.

28. Rumpho ME, Summer EJ, Green BJ, Fox TC, Manhart JR (2001) Mollusc/algal chloroplastsymbiosis: How can isolated chloroplasts continue to function for months in the cytosolof a sea slug in the absence of an algal nucleus? Zoology 104:303–312.

29. Mujer CV, Andrews DL, Manhart JR, Pierce SK, Rumpho ME (1996) Chloroplast genesare expressed during intracellular symbiotic association of Vaucheria litorea plastidswith the sea slug Elysia chlorotica. Proc Natl Acad Sci USA 93:12333–12338.

30. De Las Rivas J, Barber J (2004) Analysis of the structure of the PsbO protein and itsimplications. Photosynth Res 81:329–343.

31. Roose JL, Wegener KM, Pakrasi HB (2007) The extrinsic proteins of photosystem II.Photosynth Res 92:369–387.

32. West HH, Harrigan J, Pierce SK (1984) Hybridization of two populations of a marineopistobranch with different development patterns. Veliger 26:199–206.

33. Lee RE (1989), in Phycology, ed Lee RE (Cambridge Univ Press, New York), pp 507–522.34. Oudot-Le Secq MP, et al. (2007) Chloroplast genomes of the diatoms Phaeodactylum

tricornutum and Thalassiosira pseudonana: Comparison with other plastid genomes ofthe red lineage. Mol Genet Gen 277:427–439.

35. Cattolico RA, et al. (2008) Chloroplast genome sequencing analysis of Heterosigmaakashiwo CCMP452 (West Atlantic) and NIES293 (West Pacific) strains. BMC Genomics9:211.

36. Paquin B, Kathe SD, Nierzwicki-Bauer SA, Shub DA (1997) Origin and evolution ofgroup I introns in cyanobacterial tRNA genes. J Bacteriol 179:6798–6806.

37. Reith M, Munholland JA (1993) High-resolution gene map of the chloroplast genomeof the red alga Porphyra purpurea. Plant Cell 5:464–475.

38. Roose JL, Kashino Y, Pakrasi HB (2007) The PsbQ protein defines cyanobacterialphotosystem II complexes with highest activity and stability. Proc Natl Acad Sci USA104:2548–2553.

39. Andersson B, Aro E-M (2001) Photodamage and D1 protein turnover in photosystem II.Regulation of Photosynthesis, eds Andersson B, Aro E-M (Kluwer, Dordrecht, TheNetherlands), pp 377–393.

40. Felsenstein J (1985) Confidence limits on phylogenies: An approach using the boot-strap. Evolution 39:783–791.

41. Swofford DL, et al. (2001) Bias in phylogenetic estimation and its relevance to thechoice between parsimony and likelihood methods. Syst Biol 50:525–539.

42. Westerlund I, von Heijne G, Emanuelsson O (2003) LumenP—A neural network pre-dictor for protein localization in the thylakoid lumen. Protein Sci 12:2360–2366.

43. Emanuelsson O, Brunak S, von Heijne G, Nielsen H (2007) Locating proteins in the cellusing TargetP, SignalP, and related tools. Nat Protoc 2:953–971.

44. Chaal BK, Green BR (2005) Protein import pathways in ‘complex’ chloroplasts derivedfrom secondary endosymbiosis involving a red algal ancestor. Plant Mol Biol 57:333–342.

45. Pierce SK, Curtis NE, Hanten JJ, Boerner SL, Schwartz JA (2007) Transfer, integrationand expression of functional nuclear genes between multicellular species. Symbiosis43:57–64.

46. Pierce SK, Maugel TK, Rumpho ME, Hanten JJ, Mondy WL (1999) Annual viral expres-sion in a sea slug population: Life cycle control and symbiotic chloroplast maintenance.Biol Bull 197:1–6.

47. Lavrov DV (2007) Key transitions in animal evolution: A mitochondrial DNA perspec-tive. Integr Comp Biol 47:734–743.

48. Rot C, Goldfarb I, Ilan M, Huchon D (2006) Putative cross-kingdom horizontal genetransfer in sponge (Porifera) mitochondria. BMC Evol Biol 6:71.

49. Beagley CT, Okada NA, Wolstenholme DR (1996) Two mitochondrial group I introns ina metazoan, the sea anemone Metridium senile: One intron contains genes forsubunits 1 and 3 of NADH dehydrogenase. Proc Natl Acad Sci USA 93:5619–5623.

50. van Oppen MJH, et al. (2002) The mitochondrial genome of Acropora tenuis (Cnidaria:Scleractinia) contains a large group I intron and a candidate control region. J Mol Evol55:1–13.

51. Palmer JD (1991) Plastid chromosomes: Structure and evolution. The Molecular Biologyof Plastids, eds Bogorad L, Vasil IK (Academic, San Diego), pp 5–53.

52. Lehman RL, Manhart JR (1997) Preliminary comparison of restriction fragment patternsin the genus Caulerpa (Chlorophyta) and the unique structure of the chloroplastgenome of Caulera sertularioides. J Phycol 33:1055–1062.

53. Lee J, Manhart JR (2002) Four embryophyte introns and psbB operon indicate Chlo-rokybus as a basal streptophyte lineage. Algae 17:53–58.

54. Folmer O, Black M, Hoeh R, Lutz R, Vrijenhoek R (1994) DNA primers for amplificationof mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates.Mol Mar Biol Biotechnol 3:294–299.

55. Wyman SK, Jansen RK, Boore JL (2004) Automatic annotation of organellar genomeswith DOGMA. Bioinformatics 20:3252–3255.

56. Lohse M, Drechsel O, Bock R (2007) OrganellarGenomeDRAW (OGDRAW)—A tool forthe easy generation of high-quality custom graphical maps of plastid and mitochon-drial genomes. Curr Genet 52:267–274.

Rumpho et al. PNAS � November 18, 2008 � vol. 105 � no. 46 � 17871

EVO

LUTI

ON

Supporting Information

Rumpho et al. 10.1073/pnas.0804968105SI MethodsSea Slug and Algal Culturing. Sea slug egg masses were cultured inPetri dishes containing sterile autoclaved artificial sea water(Instant Ocean) and incubated under room conditions of tem-perature and lighting. Veliger larvae emerged from egg massesafter 5–7 days and were fed aliquots of Rhodomonas salina orIsochrysis galbana. Within 16–22 days, the veligers started todevelop pigmented striations on their shells; at this point,filaments of V. litorea were provided to induce metamorphosis.Cultures were supplemented with additional V. litorea on a dailybasis to promote growth and establishment of the symbioticassociation.

ptDNA Sequencing. ABI 3100 automated sequencers (AppliedBiosystems) at the University of Maine Sequencing Facility andthe Institute for Plant Genomics and Biotechnology, TexasA&M University, were used to sequence the ptDNA fragments.Sequences were assembled with Sequencher version 4.0, andgenes were identified using DNA and protein sequences sub-mitted to BLAST and Genetic Computer Group (www.accelry-s.com/products/gcg/). The map was constructed using Canvasversion 6.0.

Primer Design and PCR Amplification of psbO. Conserved regions ofseveral heterokont and red alga psbO sequences were alignedusing Clustal X (1) to design primers, including Karenia brevis(GenBank AY116667), Isochrysis galbana (GenBankAY116669), Heterocapsa triquetra (GenBank AY116668), Het-erosigma akashiwo (GenBank AY191862), Phaeodactylum tricor-nutum (GenBank AY191862), and Cyanidium caldarium (2).The resulting primers, psbO R and psbO L2, amplified an internalregion of the V. litorea psbO gene. A list of all primers used canbe found in Table S2. PCR conditions included 1 � enzymereaction buffer, 1.5 mM MgCl2, 0.2 mM dNTP mix (Invitrogen),primers (0.5 �M each), 1 to 10 ng of DNA, and 2.5 U of platinumTaq polymerase (Invitrogen). Cycle conditions were 94°C for 2min, 25 cycles of 94°C for 30 sec, 58.7°C for 30 sec, and 72°C for1 min, with a final elongation step at 72°C for 10 min. Productswere separated using a 1% agarose gel in 1 � Tris borate EDTAbuffer (TBE) and visualized by staining with ethidium bromide.

5�- and 3�-RACE. The complete V. litorea psbO transcript wasobtained by RACE using the GeneRacer Kit (Invitrogen) andoligo d(T) and random hexamer priming. Amplification of the 5�end was performed with the following primary primer sets: psbORev1 and GeneRacer 5� primer and psbO For1 and GeneRacer3� primer (Invitrogen). The full-length V. litorea psbO sequencewas used to generate homologous primers (psbO L5 and R8),which amplified a larger (�963-bp) psbO fragment. A list of allprimers used can be found in Table S2.

Northern Blotting. The procedures for Northern Max (formalde-hyde-based system for Northern blots; Ambion) were followedfor Northern blot analysis of the psbO transcripts using 4 �g ofDNAse-treated total RNA for each sample. RNA MillenniumSize Markers (Ambion) were run in an adjacent gel lane toestimate transcript size. RNA was transferred to Hybond-n �nylon membranes (Amersham Biosciences) using the NytranSuPerCharge Turboblotter (Schleicher & Schuell) and theNorthern Max 20 � SSC transfer buffer, and was UV cross-linked. The psbO probe was prepared using the Rediprime II

random prime labeling system (Amersham Biosciences) and[32P] dCTP (50 �Ci). Prehybridization was performed withULTRAhyb (Ambion) preheated to 42°C for 1 to 2 h beforehybridizing overnight at 42°C. Following hybridization, the mem-brane was washed with 2 � SSC/0.1% SDS for 5 min at roomtemperature, followed by two washes in 0.1 � SSC, 0.1% SDS for15 min at 42°C. The blot was exposed to x-ray film (Image Plus;Diagnostic Imaging, Inc.) at �80°C for 5 days before manualdevelopment.

Genome Walking. Algal and sea slug egg genomic DNA wasisolated using DNAzol extra strength (Molecular ResearchCenter, Inc.). The egg DNA was further purified on a CsClgradient to remove inhibiting mucopolysaccharides (3). AllDNA (2.5 �g) was restriction digested with 80 U of MscI, PvuII,or SspI overnight at 37°C; precipitated; and concentrated to 50ng/�l in deionized H2O. Genome walking was performed fol-lowing the method detailed in Clontech’s Genome Walking Kit.Briefly, 0.2 �g of restriction-digested DNA, 47.5 pmol modifiedadapter (T � N mixed 1:1), and 3 U of T4 DNA ligase weremixed and incubated at 16°C overnight. Adapter-ligated DNAwas precipitated and used in sequential inverse PCR reactions.Gene-specific primers and adapter primers (Table S2) were usedin four sequential nested amplifications starting with PsbO GW3�-3. Amplified products were band isolated using the QiagenGel Extraction Kit and cloned into pGEM-TEZ vector (Pro-mega), plated on LB agar plates containing 50 �g ml�1 Amp, andgrown overnight at 37°C. Inserts were amplified with Sp6 and T7primers and analyzed by gel electrophoresis. Amplified product(5 �l) was mixed with 2 �l of ExoSap-IT (USB Corp.) to removeresidual dNTP and primers and was incubated for 15 min at 37°Cbefore inactivating at 80°C for 15 min. Samples were sequencedby the University of Maine DNA Sequencing Facility.

Cloning and Sequence Analysis. Unless stated otherwise, the TOPOTA Cloning Kit for Sequencing (Invitrogen) was used to cloneDNA fragments excised from 1% agarose gel slices by S.N.A.P.(Invitrogen) column centrifugation. Plasmids were isolated usingthe Qiaprep Miniprep Spin Kit (Qiagen), and inserts weresequenced in both directions by the University of Maine DNASequencing Facility. A minimum of two totally independentPCR reactions and cloning events and three to six differentplasmids were sequenced for all PCR products reported here.Consensus nucleotide sequences [manually identified after align-ing with ClustalX (1)] and translated amino acid sequences wereused to search the GenBank database (www.ncbi.nlm.nih.gov/BLAST/).

Phylogenetic Analysis of psbO (MSP). The psbO phylogeny wasconstructed using maximum parsimony in PAUP 4.0b10 (4). Max-imum parsimony heuristic searches included 1,000 replications ofRANDOM addition and tree bisection-reconnection. Sets ofequally parsimonious trees were summarized using strict consensus.Bootstrapping (5) was implemented in PAUP using 1,000 replicatesof heuristic searches with SIMPLE addition sequence and treebisection-reconnection. Sources for the sequences used are asfollows: green lineage: Spinacia oleracea (S00415), Oryza sativa(NP�001043134), Chlamydomonas reinhardtii (CAA32053), Volvoxcarteri (AAD55562), Bigelowiella natans (AAP79149), and Euglenagracilis (BAA03529); red lineage: Karenia brevis (AAM77464),Isochrysis galbana (AAM77466), Heterocapsa triquerta

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 1 of 13

(AAM77465), Guillardia theta (ABD51936), Porphyra yezoensis(AAW33888), Cyanidioschyzon merolae (BAD36767), Phaeodacty-lum tricornutum (AAO43192), Thalassiosira pseudonana (thaps1/scaffold�39:162585–163984 from genome.jgi-psf.org/), Heterosigmaakashiwo (AAN11311), and Vaucheria litorea (DQ514337); mol-lusc: Elysia chlorotica (EU621882); glaucophyta: Cyanophora para-doxa (CAH04962); and prokaryotes: Nostoc punctiforme PCC73102 (ZP�00111456), Trichodesmium erythraeum (YP�723422),Thermosynechococcus elongatus BP-1 (BAC07996), Crocosphaerawatsonii WH 8501 (ZP00515383), Cyanothece sp. ATCC 51142(AAF13997), Synechococcus sp. WH 8102 (CAE06818), andGloeobacter violaceus PCC 7421 (BAC91632).

mtDNA Sequencing. After 3 months in culture in the absence ofany algae, the sea slugs produced eggs and these were used forDNA extraction. Universal primers (6, 7) were used to amplifyfragments of the mitochondrial rrnL and cox1 regions, and cobprimers were designed based on published mollusc sequences(ref. 8; Table S3). Standard PCR mixtures contained 10 mMTris-HCl, 50 mM KCl, 1.5 mM MgCl2, 0.5 �M each primer,template DNA (30–40 ng), and Taq Polymerase (1 U; NewEngland Biolabs) in a final volume of 50 �l. They were subjectedto an initial denaturation cycle at 94°C for 2 min; 25 cycles at94°C for 30 sec; annealing at 42°C for rrnL, 51°C for cob, and48°C for cox1, each for 30 sec; and then extension at 72°C for 1min. All PCR products were separated in a 1% agarose gel, andfragments were isolated using the QIAquick Gel Extraction Kit(Qiagen). Fragments were ligated using the pGEM-T EasyVector system (Promega) and transformed into NEB (NewEngland Biolabs) 5� competent cells. Three to five colonies ofeach cloned PCR product were selected, and the cloned frag-ments were sequenced using T7 and SP6 universal primers.Following identification of primers that yielded known mito-chondrial products from sea slug egg DNA, combinations ofthese primers (Table S3) were used to generate longer productsto amplify the entire circular mtDNA. The mitochondrial geneorder of the mollusc Roboastra europaea was used as a guide (10).Two cox1 primers were used with two cob primers in fourdifferent combinations of separate PCR reactions. These reac-

tions yielded two products of approximately 7,000bp each. Thelong PCR products were directly sequenced by primer walking.The two cob primers were also used in combination with the tworrnL primers, and the resulting fragments were directly se-quenced following gel extraction. The remainder of the genomewas obtained in the same manner using a combination of tworrnL and cox1 primers (see Table S3). For long PCR reactions,the Phusion High-Fidelity PCR Kit (New England Biolabs) wasused and PCR mixtures were subjected to an initial denaturationat 98°C for 30 sec, 35 cycles of denaturation at 98°C for 20 sec,annealing for 20 sec, and extension at 72°C for 5 min. Theannealing temperature was adjusted in accordance with themelting temperature (Tm) of the primer pairs. The resultingfragments were extracted from an agarose gel as describedpreviously and sequenced by primer walking on both strands.

The gene for ATPase8 was annotated by alignment with otheropisthobranch ATPase8 genes, including Aplysia californica(NC�005827), Roboastra europaea (NC�004321), and Pupa stri-gosa (AB028237), using CLUSTAL X (1). Two trnS genes thatwere not identified by tRNA-SE Scan or DOGMA programswere identified by their ability to form characteristic secondarystructures. Both nucleotide and amino acid sequence alignmentswere used to define the start and stop codons for each gene.

Analysis of the mtDNA for HGT. The E. chlorotica mitochondrialnucleotide windows (500 nt each) were analyzed by compiling adatabase of 1,429 RefSeq (11) complete mtDNAs (animals,heterokonts, green plants and algae, red algae, alveolates, andfungi) for homologous segments to each of the windows usingWashington University (WU)-BLAST. Each window was thenaligned along with its homologues using MUSCLE (12); inferredmaximum likelihood topologies were then determined usingPhyML (13). In the case of translated ORFs, a database wasassembled of 24,722 RefSeq (11) proteins from the mitochondriaof 24 species spanning animals (including molluscs), het-erokonts, green plants and algae, fungi, alveolates and slimemolds, and six bacteria. Then, each ORF was analyzed bysearching for homologues, aligning, and inferring trees (using thesame set of tools as for the nucleotide windows, adjusting theparameters appropriately for amino acid sequences).

1. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG (1997) The ClustalXwindows interface: Flexible strategies for multiple sequence alignment aided byquality analysis tools. Nucleic Acids Res 24:4876–4882.

2. Tohri A, et al. (2002) Comparison of the structure of the extrinsic 33 kDa protein fromdifferent organisms. Plant Cell Physiol 43:429–439.

3. Rumpho ME, Mujer CV, Andrews DL, Manhart JR, Pierce SK (1994) Extraction of DNAfrom mucilaginous tissues of a sea slug (Elysia chlorotica). BioTechniques 17:1097–1101.

4. Swofford DL, et al. (2001) Bias in phylogenetic estimation and its relevance to thechoice between parsimony and likelihood methods. Syst Biol 50:525–539.

5. Felsenstein J (1985) Confidence limits on phylogenies: an approach using the boot-strap. Evolution 39:783–791.

6. Palumbi S, et al. (1991) The Simple Fool’s Guide to PCR (Department of Zoology,University of Hawaii, Honolulu).

7. Folmer O, Black M, Hoeh R, Lutz R, Vrijenhoek R (1994) DNA primers for amplificationof mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates.Mol Mar Biol Biotechol 3:294–299.

8. Merritt TJS, et al. (1998) Universal cytochrome b primers facilitate intraspecific studiesin molluscan taxa. Mol Mar Biol Biotechnol 3:7–11.

10. Grande C, Templado J, Cervera L, Zardoya R (2002) The complete mitochondrialgenome of the nudibranch Roboastra europaea (Mollusca: Gastropoda) supports themonophyly of opisthobranchs. Mol Biol Evol 19:1672–1685.

11. Pruitt KD, Tatusova T, Maglott DR (2007) NCBI reference sequences (RefSeq): A curatednon-redundant sequence database of genomes, transcripts and proteins. Nucleic AcidsRes 35:D61–D65.

12. Edgar RC (2004) MUSCLE: A multiple sequence alignment method with reduced timeand space complexity. BMC Bioinformatics 5:113.

13. Guindon S, Gascuel O (2003) A simple, fast, and accurate algorithm to estimate largephylogenies by maximum likelihood. Syst Biol 52:696–704.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 2 of 13

Fig. S1. Schematic illustrating the evolutionary origin of secondary plastids in V. litorea and tertiary plastids in E. chlorotica. The drawing highlights the redalgal secondary endosymbiotic origin of V. litorea. Four membranes surround the algal plastids as a result of the two endosymbiotic events, with the outermostmembrane being continuous with the nuclear envelope. The bottom panel with two sea slug digestive epithelial cells illustrates that only two membranes, thetypical plastid double envelope, are typically seen around the plastids in the sea slug. The plastids are colored red in the drawing to reflect the red algal origin.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 3 of 13

Fig. S2. Nucleic acid sequence comparison of psbO from V. litorea cDNA and genomic DNA, E. chlorotica DNA, and E. chlorotica egg DNA. An initial 452-bpfragment was obtained from V. litorea by RT-PCR, and the complete cDNA for psbO was obtained by 5� and 3� RACE. This sequence was used to design primersthat amplified a �963-bp fragment from DNA or cDNA of both organisms. The psbO gene does not contain an intron. The black dots identify identical base pairsin all four sequences beginning with the start codon, and the dashes indicate that the sequence is not available for comparison. Base pair numbering is accordingto the V. litorea cDNA sequence.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 4 of 13

Fig. S3. Maximum parsimony tree based on amino acid sequences of 25 psbO-encoded MSPs. Strict consensus of nine trees (1,530 steps, consistency index �0.650, retention index � 0.576) using maximum parsimony in PAUP 4.0b10 (1, 2). Numbers are shown above branches for all boot strap values �90. Weightedbranches are indicative of branches with boot strap support values �70. For a complete list of sources, see SI Methods.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 5 of 13

Fig. S4. Nucleic acid sequence comparison of the psbO 3� flanking region of V. litorea and E. chlorotica DNA. Genome walking using a nested gene-specificprimer with an adapter-specific primer yielded 3� psbO flanking sequence data from both V. litorea and E. chlorotica. The sequences were identical for the first81bp corresponding to the 3� end of the psbO gene and ending with the stop codon (bold text). This sequence was followed by the highly diverged sequencecorresponding to the 3� untranslated flanking region in each organism.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 6 of 13

Fig. S5. Distribution of E. chlorotica G � C content over the sliding window of 500 nucleotides with overlaps of 200 nucleotides. The values were found to beuniformly distributed across the windows with a linear trend described by the model y � 0.0026x � 36.25. The average G � C content of the mtDNA of E. chloroticawas 36.19%.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 7 of 13

Fig. S6. Maximum likelihood phylogenetic tree of cytochrome b. The tree was inferred using PhyML (13) with 100 bootstrap replicates.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 8 of 13

Movie S1 (MOV)

Movie S1. Young sea slug sucking plastids out of Vaucheria litorea filaments. A juvenile sea slug is observed feeding on filaments of the heterokont algaV. litorea. There is an obligate requirement at this stage for plastid acquisition for continued development to the adult stage. This is fulfilled by juvenilespuncturing the siphonaceous filaments and sucking out the cellular contents. Only the plastids are retained by the sea slug in cells lining the digestive epithelium.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 9 of 13

Movie S2 (MOV)

Movie S2. Mature “solar-powered” sea slug Elysia chlorotica. An adult sea slug is observed feeding on the heterokont alga V. litorea. Algal chloroplasts areretained in the digestive epithelium in an endosymbiotic association yielding an emerald green “solar-powered” sea slug. The sea slug can sustain itself for itsentire life-span of about 10 months photoautotrophically requiring only light and air as a source of carbon dioxide. Adult animals range from about 1.5 to amaximum of 6 cm in length.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 10 of 13

Table S1. Complete listing of Vaucheria litorea chloroplast genes by category

Protein-encoding genes 139Photosynthesis

Photosystem I 10 psaA, psaB, psaC, psaD*, psaE, psaF, psaI, psaJ, psaL,psaM

Photosystem II 18 psbA, psbB, psbC, psbD, psbE, psbF, psbH, psbI, psbJ,psbK, psbL, psbN, psbT, psbV, psbX, psbY, psbZ, psb28

Chlorophyll biosynthesis 4 chlB‡, chlI, chlL‡, chlN‡

Cytochrome 9 petA, petB, petD, petF, petG, petJ§, petL, petM, petNATP synthase 8 atpA, atpB, atpD, atpE, atpF, atpG, atpH, atpIRubisco 3 rbcL, rbcS, cfxQ

Transcription/translation/replication

RNA polymerase 4 rpoA, rpoB, rpoC1, rpoC2Translation factors 2 tufA, tsf¶

Replication helicase 1 dnaBRibosomal proteins

Small subunits 18 rps1§, rps2, rps3, rps4, rps5, rps7, rps8, rps9, rps10, rps11,rps12, rps13, rps14, rps16, rps17, rps18, rps19, rps20

Large subunits 27 rpl1, rpl2, rpl3, rpl4, rpl5, rpl6, rpl9‡, rpl11, rpl12, rpl13,rpl14, rpl16, rpl18, rpl19, rpl20, rpl21†, rpl22, rpl23,rpl24, rpl27, rpl29, rpl31, rpl32, rpl33, rpl34, rpl35,rpl36

Miscellaneous proteinsMaintenance 4 clpC, dnaK, ftsH, groELTransport 5 secA, secY, sufB, sufC, tatCAmino acid biosynthesis 2 ilvB§, ilvH§

Other proteins 8 acpP�, acsF§, ccsA, ccs1, ftrC§, ycf17 (hlip§), thiG, thiSConserved ORFs 14 ycf3, ycf4, ycf12, ycf19‡, ycf33, ycf36§, ycf37‡, ycf39,

ycf41, ycf42, ycf54§, ycf60‡, ycf65†§, ycf66†

Unidentified ORFs 2RNA-encoding genes 35

Ribosomal RNAs 6 (3�2) rrl, rrs, rrfTransfer RNAs 29 trnA(ugc)X2, trnC(gca), trnD(guc), trnE(uuc), trnF(gaa),

trnfM(cau), trnG(gcc), trnG(ucc), trnH(gug), trnI(cau),trnI(gau)X2, trnK(uuu), trnL(uaa)**, trnL(uag),trnM(cau), trnN(guu), trnP(ugg), trnQ(uug), trnR(acg),trnR(ccg), trnR(ucu), trnS(gcu), trnS(uga), trnT(ugu),trnV(uac), trnW(cca), trnY(gua)

*Genes not found in Streptophyta or Chlorophyta chloroplast genomes are bold; those found in Chlorophyta but not embryophytes (land plants) aresingle-underlined.

†Gene found in streptophytes but not chlorophytes.‡Gene not found in other published heterokont chloroplast genomes (including Odontella sinensis, Phaeodactylum tricornutum, Thalassiosira pseudonana, andHeterosigma akashiwo).

§Gene found only in Heterosigma akashiwo.¶Gene found only in Phaeodactylum tricornutum.�Gene found in Odontella sinensis, Phaeodactylum tricornutum, and Heterosigma akashiwo but not in Thalassiosira pseudonana.**Gene-containing intron.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 11 of 13

Table S2. Primers for psbO*

Probe Product size Sequence (5� to 3�)

PCR and RT-PCRpsbO R 452bp RCC DCG KCC YTT SGG RTC MAG GAApsbO L2 ARG GGH WSH GGY YTB GCV AACpsbO L5 963bp GAA GGT CCC ATC TGC TTT GGT CpsbO R8 ATT CGC TCT CAA GCC TTC CAT AG

5� RACEpsbO Rev1 CCA CGT CCT TTG GGG TCA AGGGeneRacer RNA oligo (Invitrogen)

3� RACEpsbO For1 AAG GGA AGN GGT TTG GCC AAC AGOligo d(T) (Invitrogen)

Genome WalkingPsbO GW 3�-1 GGGGAGATTGGAGGAGTCTTTGTTTCGPsbO GW 3�-2 GGAGACACTGATATGGGCTCTAAAGTCCPsbO GW 3�-3 CACCTTGTACGGGATTGACGGCTCTTTCGPsbO GW 3�-4 GAAAGACGGGATTGATTATGCTGCCACTACAdapter Primer 1 GTAATACGACTCACTATAGGGCAdapter Primer 2 ACTATAGGGCACGCGTGGTGenome Walker AdapterT GTAATACGACTCACTATAGGGCACGCGTGGTC

GACGGCCCGGGCTGGTGenome Walker AdapterN (P)ACCAGCCC(L)

*All primers were synthesized by Integrated DNA Technologies, Inc. unless stated otherwise. L, forward primer, R, reverse primer.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 12 of 13

Table S3. Primer pairs for Elysia chlorotica mitochondrial genome*

Primer Primer sequence Gene amplified

A. Initial primersRrnL (6)† F: GAAAAAAGACGAGAAGACCC rrnL

R: GGGTCTTCTCGTCTTTTTTCCox1 (7) F: GGTCAACAAATCATAAAGATATTGG cox1

R: TAAACTTCAGGGTGACCAAAAAATCACob (8) F: TGTGGRGCNACYGTWATYACTAA cob

R: AANAGGAARTAYCAYTCNGGYTGB. Final primer

combinationsPrimer Primer sequence Regions amplifiedCobp3 TGTGGRGCNACYGTWATYACTAA Part of cob, coxII, ATP8, ATP6, rrnS, nad3, nad4, coxIII, nad2, part of coxICox1f1 GGTCAACAAATCATAAAGATATTGGCobp4 AANAGGAARTAYCAYTCNGGYTG Part of cob, nad4L, nad1, nad5, nad6, part of rrnL16SarL CGCCTGTTTAACAAAAACAT16SbrH CCGGTCTGAACTCAGATCACGT Part of rrnL, part of cox1Cox1f2 TAAACTTCAGGGTGACCAAAAAATCA

*All primers were synthesized by Integrated DNA Technologies, Inc.†Number in parentheses refers to supplementary reference number. F, forward primer, R, reverse primer.

Rumpho et al. www.pnas.org/cgi/content/short/0804968105 13 of 13