Keywords

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

I. Plastid Origin

A. Plastids Acquired via Eukaryote–Eukaryote Endosymbiosis

Algae is a widely used informal name that refers to a diverse group of photosynthetic eukaryotes such as euglenids and diatoms that have a polyphyletic origin (Reyes-Prieto et al. 2007). The Plantae and the Chromalveolata comprise the two largest eukaryote supergroups of presumed photosynthetic ancestry. Current data clearly show that the primary plastid traces its origin to primary (i.e., eukaryote–prokaryote) endosymbiosis, in which a unicellular protist (the “host”) engulfed and retained a photosynthetic cyanobacterium (the endosymbiont) (Chap. 1). The resulting photosynthetic eukaryote is the putative common ancestor of the Plantae (Moreira et al. 2000; Rodriguez-Ezpeleta et al. 2005; Hackett et al. 2007) that is comprised of red, green (including plants), and glaucophyte algae (Cavalier-Smith 1981; Chan et al. 2011). Upon establishment, the primary plastid was apparently maintained in all extant Plantae. There are many colorless (non-photosynthetic) Plantae known, for example in parasitic plants and pathogenic algae (e.g., Epifagus virginiana, Cuscuta spp., Aneura mirabilis, Polytomella sp., Helicosporidium spp; Wolfe et al. 1992; Bungard 2004; McNeal et al. 2007; Wickett et al. 2008; Tartar et al. 2002), however, each of these taxa retains a vestigial plastid to perform other organelle functions, such as carbohydrate storage and heme biosynthesis (Atteia et al. 2005). In contrast, the chromalveolate ancestor gained its plastid through secondary (i.e., eukaryote–eukaryote) endosymbiosis (see Fig. 2.1a), whereby under the original hypothesis (Cavalier-Smith 1992), a red alga was engulfed and reduced to a secondary plastid. Phylogenomic data have shown, however, that a number of genes shared by chromalveolates are of green algal origin (e.g., Li et al. 2006; Nosenko and Bhattacharya 2007). Of particular significance is the finding of five enzymes of green algal origin involved in carotenoid biosynthesis in “chromist” (stramenopile, cryptophyte, and haptophyte) algae (Frommolt et al. 2008). Three of these genes branch deeply within prasinophyte green algae in phylogenetic analyses. This suggests they may have originated via endosymbiotic gene transfers (EGTs) or extensive horizontal gene transfer (HGT) from an ancient green algal endosymbiont (i.e., prasinophytes form a basal split among green algae; e.g., Steinkoetter et al. 1994; Fawley et al. 2000) that predates the widespread red algal plastid in chromalveolates (Fig. 2.1b). This hypothesis gained further support when Moustafa et al. (2009) found evi­dence of hundreds of genes of green algal origin in the diatoms (stramenopiles, Bacillariophyta) Thalassiosira pseudonana and Phaeodactylum tricornutum. Similarly, phylogenomic analysis of the brown alga (stramenopiles, Phaeophyta) Ectocarpus siliculosus turned up ca. 2,600 genes of putative green algal origin, that contrast with only 611 genes of red algal provenance in this species (Cock et al. 2010). Whether these “green genes” trace their origin to a single cryptic endosymbiosis, to repeated HGTs, or more likely a combination of the two, these data highlight the complex nature of chromalveolate genome evolution that is only now being fully appreciated (see Fig. 2.1b and Baurain et al. 2010). The working hypothesis favored by Moustafa et al. (2009) is that the presence of a red algal-derived plastid in many chromalveolates conceals a past endosymbiosis with the “green” nuclear encoded genes acting as footprints of ancient E/HGT (see also Elias and Archibald 2009; Dagan and Martin 2009).

Fig. 2.1.
figure 1

Secondary endosymbiosis in eukaryote evolution. (a) Schematic representation of major events that presumably occurred during plastid evolution in chromalveolate lineages. Red algal secondary plastids are present in most photosynthetic chromalveolates (yellow cells). Secondary endosymbiotic gene transfer (EGT) into the host (host ancestor) genome (N2) from both nucleus (N1; red dotted line) and plastid (green dotted line) of the captured red alga are shown. There are several cases (black dotted lines) of photosynthesis and plastid loss in different chormalveolate lineages. Representative genera (or groups) are indicated close to each known non-photosynthetic variant (grey cells). (b) Mock phylogeny showing what is currently known about the origin of plastids and the interrelationships of the ‘Hacrobia’ and ‘SAR’ clades. The common photosynthetic origin of the chromalveolate red-algal plastid (e.g., the chromalveolate hypothesis; secondary endosymbiosis) is a contentious scenario, and recent genome analyses suggest an ancestral green algal endosymbiosis (green line at base) that predates the red algal capture. Photosynthetic lineages (red lines) are intermingled with non-photosynthetic (blue lines). Some of the non-photosynthetic groups, such as the ciliates and oomycetes, presumably evolved after independent losses of the red algal plastid. The Rhizaria regained two different types of plastids (also presumably after loss of the ancestral red algal plastid), the green algal plastid in chlorarachniophytes and the cyanobacterium-derived plastid in Paulinella species (Yoon et al. 2006; Reyes-Prieto et al. 2010) are indicated in green and purple, respectively.

Full size image

Two other eukaryote supergroups, the Rhizaria and Excavata also contain photosynthetic members (Chlorarachniophyta and Euglenozoa, respectively) but these algae are derived branches of what are believed to be anciently plastid-lacking lineages (hereafter, plastid [−]; in contrast to plastid [+]). Recent molecular studies have unveiled phylogenetic ties (see Fig. 2.1) between Rhizaria, some groups of chromalveolates, and non-photosynthetic lineages such as telonemids and katablepharids (Shalchian-Tabrizi et al. 2006b, Okamoto and Inouye 2005; Hackett et al. 2007; Burki et al. 2007, 2009; Reeb et al. 2009). If these hypotheses are substantiated with genome data, then it will be important to trace the evolution of plastids via secondary endosymbioses across trees with intermingled plastid [−] and plastid [+] lineages whose origins extend back hundreds of millions of years to the time of eukaryote origin (see below and Moustafa et al. 2009; Cavalier-Smith 2010).

Understanding the convoluted history of secondary plastid evolution is aided significantly by a well-resolved nuclear host phylogeny. The chromalveolates however pose a great challenge in this respect. Plastid data usually support the monophyly of photosynthetic chromalveolates (e.g., Yoon et al. 2002; Khan et al. 2007), however nuclear gene trees disagree with the chromalveolate hypothesis in two key respects. First, recent multi-gene analyses support the inclusion of the Rhizaria within chromalveolates (Hackett et al. 2007; Burki et al. 2007 2009; the ‘SAR’ clade in Fig. 2.1b), and second, chromists as originally proposed by Cavalier-Smith (1992) are polyphyletic with cryptophytes often found sister to haptophytes in a major lineage that also may include telonemids (Shalchian-Tabrizi et al. 2006b), katablepharids, centrohelids (Okamoto and Inouye 2005; Okamoto et al. 2009), picobiliphytes (Not et al. 2007), and the recently described photosynthetic rappemonads (Kim et al. 2011). This assemblage is sometimes referred to as the ‘Hacrobia’ (Okamoto et al. 2009) and is putatively sister to the SAR clade (Hackett et al. 2007; Burki et al. 2007; Patron et al. 2007; Burki et al. 2008; Okamoto et al. 2009). Clearly much more work has to be done to resolve the origins of chromalveolate-affiliated taxa.

B. How Is the Nuclear Genome Affected by Plastid Origin and Loss?

A key aspect of organelle evolution is the transfer of endosymbiont genes to the host nucleus, followed by import of the gene products into the organelle (Herrmann 1997; Martin et al. 1998; Martin and Herrmann 1998). Over time, EGT enriches the host genome with hundreds of transferred genes (Martin and Herrmann 1998; Moustafa et al. 2008a, b, 2009). The magnitude of EGT in Plantae genomes was estimated for Arabidopsis thaliana (Martin et al. 2002; Sato et al. 2005), and the unicellular algae Chlamydomonas reinhardtii (Moustafa and Bhattacharya 2008), Cyanophora paradoxa (Reyes-Prieto et al. 2006) and Cyanidioschyzon merolae (Sato et al. 2005). The results of these studies suggest that unicellular algae contain ca. 600–900 genes of cyanobacterial origin in their nucleus, and the vast majority encode proteins with plastid functions (Reyes-Prieto et al. 2006).

Secondary Endosymbiotic Gene Transfer

In contrast to primary plastid evolution, the acquisition of secondary plastids involves a more complex scenario for EGT. In these cases, genes are transferred both from the nucleus of the eukaryotic endosymbiont (Robertson and Tartar 2006; Li et al. 2006) into the host genome as well as directly from the plastid (Sanchez-Puerta et al. 2005; Oudot-Le Secq et al. 2007; see Fig. 2.1a). Nucleus–nucleus EGT facilitates the transfer of eukaryotic genes required for plastid function and maintenance (e.g., Archibald et al. 2003; Li et al. 2006), as well as other genes to provide redundancy and/or perform novel non-plastid functions (e.g., Stibitz et al. 2000), or replace the existing host gene copies (e.g., Hackett et al. 2007). Eukaryote–eukaryote endosymbiosis provides therefore the potential for extensive nuclear gene transfers that can significantly alter the host gene pool. Given the clear evidence for green algal genes in photosynthetic chromalveolates (e.g., Frommolt et al. 2008; Moustafa et al. 2009; Cock et al. 2010), it is likely that secondary EGT and/or HGT in these lineages encompasses both red and green algal (as well as potentially other) sources. This leads to an important point to consider with regard to anciently phagotrophic lineages such as chromalveolates: the evolutionary history of the plastid and nuclear genome may be uncoupled in these taxa. The red-algal plastid most likely represents the most recent organelle capture and associated EGT to the nucleus, whereas the nucleus contains not only this information but potentially evidence of all endosymbioses/EGTs and HGTs that have occurred in the history of the lineage (e.g., dinoflagellates; see Li et al. 2006; Patron et al. 2006; Nosenko et al. 2006). Nuclear genome data (in spite of its convoluted history) offers therefore a more accurate view of host evolution than does the plastid. Discriminating between competing evolutionary scenarios will however require extensive taxon sampling (that is not yet available) to understand better the impact of E/HGT on chromalveolate nuclear genome evolution (e.g., ortholog gene replacement, gene losses, impacts of homologous recombination, heterogeneous evolutionary rates; Harper and Keeling 2004; Richards et al. 2006), to ameliorate their misleading effects (e.g., phylogenetic artifacts) on multiprotein phylogenies.

Alveolate Plastids

Genome analysis of different apicomplexans (e.g., the parasitic protists Plasmodium falciparum, Theileria parva, and Toxoplasma gondii; Huang et al. 2004a) has turned up dozens of genes of algal (endosymbiotic) origin, with some of them encoding apicoplast-targeted proteins (Gardner et al. 2002). Even in the apicoplast-lacking apicomplexan Cryptosporidium parvum, several dozen genes of putative endosymbiotic origin have been identified (Huang et al. 2004b). The recent discovery of the photosynthetic relatives of apicomplexans, the coral-endosymbiont Chromera velia (Moore et al. 2008) and its relative Chromerida sp. CCMP3155 supports an algal ancestry for this group and likely as well for the common ancestor of dinoflagellates and apicomplexans (Moore et al. 2008; Janouskovec et al. 2010). Numerous examples of mixotrophic (Stoecker 1990) or non-photosynthetic (plastid-lacking or with a relic plastid) dinoflagellates (or closely related taxa) have been described in the past (e.g., Noctiluca, Pfiesteria, Gymnodinium, Protoperidinium and some Dinophysis species; Gaines and Elbrächter 1987; Jeong 1999), but until recently the genomic footprint of a past plastid (i.e., endosymbiont) has been described only in the ­heterotroph Crypthecodinium cohnii (Sanchez-Puerta et al. 2007), the early branching (plastid-lacking) dinoflagellate Oxyrrhis marina (Slamovits and Keeling 2008), and the bivalve parasite Perkinsus marinus (Matsuzaki et al. 2008).

The evolution of dinoflagellate plastids is marked by multiple independent examples of tertiary endosymbiosis involving the capture of algae harboring secondary plastids (e.g., photosynthetic stramenopiles, haptophytes, cryptophytes). The phylogeny of these taxa suggests that each of these events has involved independent replacements of the ancestral red algal plastid (Saldarriaga et al. 2001; Shalchian-Tabrizi et al. 2006a). Some gymnodiniacean dinoflagellates, such as Karenia brevis and Karlodinium micrum, harbor multi-membrane-bound plastids containing the typical haptophyte photopigment 19′ hexanoyl oxy-fucoxanthin. These fucoxanthin-containing plastids of tertiary origin presumably replaced the original secondary organelle of the Karenia–Karlodinium ancestor after an endosymbiosis involving a haptophyte alga. Phylogenies of nuclear-encoded plastid targeted proteins in K. brevis (Nosenko et al. 2006) and K. micrum (Patron et al. 2006) indicate that the proteome of these tertiary plastids accumulates a complex collection of proteins of bacterial, haptophyte, red and even green algal origin (Patron et al. 2006; Nosenko et al. 2006) as a consequence of multiple E/HGT events. The “recycling” of a fraction of the ancestral secondary-plastid proteome for the tertiary plastid suggests that both algal-derived organelles likely co-existed in the Karenia–Karlodinium ancestor during consolidation of the tertiary endosymbiosis (Patron et al. 2006).

Dinoflagellates have also recruited diatom endosymbionts on multiple occasions (Dodge 1971; Inagaki et al. 2000; Imanian and Keeling 2007). Relevant cases are Durinskia baltica and Kryptoperidinium foliaceum that harbor permanent tertiary plastids derived from a common pennate diatom ancestor (Inagaki et al. 2000; Imanian and Keeling 2007). The tertiary plastids of D. baltica and K. foliaceum maintain the endosymbiont nuclear membrane, endoplasmic reticulum surrounding the plastids, mitochondria, and ribosomes (Eschbach et al. 1990). Moreover, other dinoflagellate species have acquired independently plastids from centric diatoms (Imanian et al. 2010), and a putative case of successive replacement of diatom-derived plastids was reported recently (Takano et al. 2008). Some species of the genus Dinophysis harbor plastids (possibly kleptoplastids) of cryptophyte origin (Schnepf and Elbrachter 1988; Takishita et al. 2002). In this case, the two-membrane-bound plastids of tertiary origin contain phycobilins instead of peridinin as the main accessory photosynthetic pigment. Typical features of cryptophyte photosynthetic organelles, such as four bounding plastid membranes and the nucleomorph, are no longer present in Dinophysis plastids. It is plausible that the common ancestor of Dinophysis species lost the peridinin-containing plastid and some of these lineages are able to retain temporarily plastids from cryptophyte prey (Minnhagen and Janson 2006; see also Elysia section below). These results provide strong evidence that the ancestral red algal plastid has been lost repeatedly during dinoflagellate evolution.

Ciliates constitute the third branch of alveolates and form a non-photosynthetic sister to apicomplexans and dinoflagellates. In contrast to other alveolates, ciliates have no apparent physical remnant of a plastid, begging the question of a potential photosynthetic past for this group and for the alveolates as a whole. Until now, no Chromera-like taxon has turned up as an early diverging “ciliate” but a recent analysis done by our lab identified multiple examples of algal genes in this group (Reyes-Prieto et al. 2008; see also Archibald 2008).

Were Ciliates Once Algae?

Phylogenomic analysis can generate thousands of single-gene trees that are placed into categories reflecting phylogenetic origin (e.g., vertically inherited in eukaryotes or candidates for E/HGT). Single gene trees that address billion year-old splits are however particularly prone to stochastic behavior due to a paucity of signal in the limited data (e.g., Martin et al. 2002). Therefore, single trees need to be interpreted with great ­caution. As an example of the type of result produced by phylogenomics, Fig. 2.2 shows two trees generated by past analyses done in our lab (see Moustafa and Bhattacharya 2008; Moustafa et al. 2008b). The first (Fig. 2.2a) was inferred using RuvB-like 2 (DNA helicase-like) protein, a putative vertically inherited gene that provides a reasonable single-protein estimate of the eukaryote tree of life. The second example (Fig. 2.2b) is a tree inferred from an aspartyl protease family protein that is clearly of algal origin (i.e., the tree is arbitrarily rooted on the branch leading to red algae) and has been maintained in a broad diversity of plastid [+] and [−] chromalveolates. Note however that both single-protein trees fail to provide bootstrap support for deeper splits (e.g., for supergroups), but do substantiate the monophyly of most phyla (e.g., stramenopiles, cryptophytes, green algae).

Fig. 2.2.
figure 2

RAxML phylogenetic trees of (a) RuvB-like 2 DNA helicase-like and (b) aspartyl protease family proteins. These trees represent typical output from phylogenomic analysis with the RuvB-like tree showing a putatively vertically inherited gene and the aspartyl protease tree showing a case of EGT from a red or green algal source into the putative common ancestor of chromalveolates. Bootstrap values (when ≥50%) from a RAxML analysis are indicated above the branches and PhyML bootstrap values are shown in Italic text below the branches. The branch lengths are proportional to the number of substitutions per site (see scales in the figure). Green algae and plants are shown in green text, red algae in red text, chromalveolates in brown text and all others in black. Non-photosynthetic chromalveolates are in boldface text and taxa that have lost the plastid are indicated with the filled brown circles. The tree shown in (a) was rooted on the branch leading to the Amoebozoa, whereas the tree shown in (b) was rooted arbitrarily on the branch leading to red algae.

Full size image

Using this type of approach, we reported 16 genes of putative algal origin in the genome of the ciliate Tetrahymena thermophila that are shared with another distantly related ciliate, Paramecium tetraurelia (Reyes-Prieto et al. 2008). It should be noted that for some ciliate proteins, single homologous sequences are also returned from other taxa such as Amoebozoa, excavates or opisthokonts (the clade comprising fungi, animals and their immediate protist ancestors). We suggest these genes originated via independent HGTs in non-chromalveolate taxa. The complex topology of most of the phylogenies presented in Reyes-Prieto et al. (2008) and potential for recurrent HGTs from eukaryotic sources, however, renders it difficult to unambiguously distinguish between rare, ancient EGT and recurrent, independent HGTs as explanation for gene origin in ciliates. The wide phylogenetic distribution of some genes in both plastid [+] and plastid [−] chromalveolates does not prove, but is consistent with, ancient origin via EGT, under the dual endosymbiosis hypothesis described above (Nosenko et al. 2006; Frommolt et al. 2008; Moustafa et al. 2009). Most of the ciliate algal-derived genes are shared with at least one other chromalveolate group, and several are present in at least two other lineages. The presence of algal-derived genes in non-photosynthetic and photosynthetic chromalveolates strongly suggests therefore a common and ancient origin of these sequences (e.g., Fig. 2.2b). Although some of the plant (e.g., Arabidopsis thaliana) homologs are potentially plastid-targeted, most of the putative algal proteins in ciliates are not derived from plastid-targeted sequences in photosynthetic eukaryotes (Reyes-Prieto et al. 2008). Therefore genome analysis of ciliates provides tantalizing evidence of the footprints of endosymbiosis that have persisted over hundreds of millions of years in spite of presumptive plastid loss. It should be noted that the set of algal genes identified in ciliates (even when likely to grow in number with more sophisticated genome analysis) is by definition an underestimate of the true value given that over time, sequence divergence blurs the evolutionary history of some genes, making it impossible to determine their origin using standard molecular phylogenetic methods (e.g., Martin et al. 2002; Dagan and Martin 2006; Reyes-Prieto et al. 2008).

Stramenopile Plastids

The extraordinarily diverse stramenopiles comprise many ecologically relevant photosynthetic groups such as diatoms (Bacilariophyta), phaeophytes, and chrysophytes, but includes as well members with vestigial, non-photosynthetic plastids (e.g., the chrysophytes Spumella spp. and Antophysa vegetans and the dictyochophytes (axodines) Pteridomonas danica and Ciliophrys infusionum; Sekiguchi et al. 2002), and plastid [−] lineages or lineages with plastid-derived vestigial structures, such as oomycetes, bicosoecids, labyrinthulids and opalinids. Some studies suggest that outright plastid loss has occurred only twice, early in stramenopile evolution; i.e., in the ancestors of oomycetes and a putative monophyletic group formed by opalinids, labyrinthulids and bicosoecids (Cavalier-Smith and Chao 2006). The placement of the many heterotrophic environmental MAST (Marine Stramenopile) picoeukaryotes within the tree (e.g., the MAST-1 clade in Not et al. 2007) may however inflate the number of putative plastid losses in this group if they are intermingled with photosynthetic groups. Nevertheless, consistent with the idea of an algal past for plastid [−] stramenopiles, analysis of the complete nuclear genome sequence from the oomycetes Phytophthora ramorum and P. sojae revealed at least 30 (with up to several hundred) genes of putative cyanobacterial or algal (i.e., endosymbiotic) origin, including 12 Phytophthora genes with plant/algal homologs that encode plastid-targeted proteins (Tyler et al. 2006; see Stiller et al. 2009 for an alternative explanation for algal/cyanobacterial genes in oomycetes).

‘Hacrobia’: Cryptophyte and Haptophyte Plastids

Most members of this clade contain a red algal-derived secondary plastid that was likely acquired by their putative common ancestor (Rice and Palmer 2006). However examples of non-­photosynthetic haptophytes (Andersen 2004) and cryptophytes (Clay et al. 1999) are known. Cryptophytes are a notable case to highlight the history of secondary plastid evolution in chromalveolates given the persistence in most of these taxa of a reduced nucleus (nucleomorph) that can be traced back to the red algal endosymbiont (Douglas et al. 2001). Two fully sequenced cryptophyte nucleomorph genomes show the conservation of hundreds of protein-coding genes (ca. 470, mostly housekeeping), including some essential players in photosynthesis (Douglas et al. 2001; Lane et al. 2007; Kim et al. 2008). The cryptophytes comprise a number of non-photosynthetic lineages, such as members of the genus Cryptomonas (Hoef-Emden 2005) with a relic plastid and a nucleomorph and members of the genus Goniomonas that contain neither a plastid nor a nucleomorph (McFadden et al. 1994). Phylogenetic analyses show that Goniomonas diverges earliest from the branch leading to photosynthetic cryptophytes, suggesting to some that a red algal endosymbiosis may have occurred independently in cryptophytes after the Goniomonas split (McFadden et al. 1994; Deane et al. 2002), whereas an alternative interpretation is that the cryptophyte ancestor was photosynthetic and Goniomonas lost outright the organelle (Cavalier-Smith et al. 1996). This last scenario is the most parsimonious considering the likely common ancestry of cryptophytes and haptophytes (Burki et al. 2007, 2009; Hackett et al. 2007; Rice and Palmer 2006; Okamoto et al. 2009), strongly supported by a unique plastid gene (rpl36) replacement shared by these two lineages (Rice and Palmer 2006). Finally, the phylogenetic affiliation between cryptophytes and katablepharids, and the intermingling of haptophytes with other lineages, such as telonemids, picobiliphytes and centrohelid heliozoa (Burki et al. 2009; Okamoto et al. 2009), within the proposed Hacrobia, suggest the putative ancestor of this assembly was photosynthetic. Additional genome data are needed to confirm the existence of footprints of ancient endosymbioses in the Hacrobia that putatively constitutes a major eukaryotic lineage of photosynthetic ancestry.

C. Future Directions

A key question is raised as genome data accumulate for members of the SAR, Hacrobia and other under-studied microbial eukaryotes: what role did cryptic secondary endosymbiosis of red and/or green algae versus recurrent HGT from these sources play in the evolutionary history of these taxa? Here we stress that the question remains largely unanswered but some key insights can already be made. First, E/HGT is substantial in the genomes of taxa originally included in the Chromalveolata and presumably in recently erected groups (see Okamoto et al. 2009; Cavalier-Smith 2010). The magnitude of E/HGT remains a challenging issue for projects that aim to reconstruct organism and plastid history using multi-gene data. The inability to convincingly settle the issue of chromalveolate phylogenetic history, the relationship among major photosynthetic lineages (e.g., Plantae, SAR, Hacrobia) and the phylogenetic positions of novel taxa (e.g., the heterotophic biflagellate Palpitomonas; Yabuki et al. 2010), even when using large data sets is worrisome (e.g., Nozaki et al. 2007; Patron et al. 2007; Burki et al. 2008; Kim and Graham 2008; Yoon et al. 2008; Archibald 2009). What has however not yet happened is the combination of broad taxon sampling and a large data set (dozens of proteins) derived from analysis of complete genome data (e.g., Rodriguez-Ezpeleta et al. 2005; Hackett et al. 2007; Burki et al. 2008). It is this missing piece of the tree of life puzzle that needs to be filled to accurately infer the number of plastid endosymbioses that have occurred during eukaryote evolution.

II. The Evolution of Plastid Protein Topogenesis in Chromalveolates

Under the chromalveolate hypothesis, Cavalier-Smith emphasized that any event of organellogenesis leading to the transformation of a red algal endosymbiont into a plastid must include the evolution of an organized molecular system (e.g., protein translocons) to catalyze the transport of nuclear-encoded proteins into the endosymbiont subcompartments (Cavalier-Smith 1999). An additional “difficulty” was assumed to be that each gene transferred from the red algal chromosome to the host nucleus needed to acquire specific sorting signals to “inform” the final destination of the encoded product into the organelle. Cavalier-Smith argued that the de novo emergence of such protein translocons and topogenic signals is a complex, if not highly improbable, evolutionary leap. One should therefore favor a parsimonious scenario in which only a single endosymbiotic event and unique evolution of the organellar protein targeting explains the chromalveolate plastid origin (Cavalier-Smith 1999). Cell biologists are now constructing a picture of how protein topogenesis operates into secondary plastids. The new data open up the opportunity to revisit the original premises of the chromalveolate hypothesis and to approach the fundamental question of how a complex plastid can derive from a secondary endosymbiont.

A. Protein Targeting to Secondary Plastids

Organellogenesis involving a secondary red algal endosymbiont resulted in a four membrane-bound plastid in most members of the chromalveolate clade (one exemption is the plastid of peridinin-containing dinoflagellates which is surrounded by three membranes; Bolte et al. 2009; Keeling 2009). This defines new compartments not found in primary plastids; i.e., the periplastid compartment (PPC), corresponding to the remnant of the red algal cytosol, and the periplastid membrane (PPM), derived from the endosymbiont plasma membrane. The outermost membrane of cryptophyte, stramenopile and haptophyte plastids is contiguous with that of the endoplasmic reticulum (ER), indicating these complex organelles are embedded within the endomembrane system. This fact is compatible with the idea that the original red algal endosymbiont was acquired via phagocytosis (Patron and Waller 2007; Bolte et al. 2009). Accordingly, proteins directed to the complex chromalveolate plastids are first routed to the ER via the Sec61 translocon. Typically, proteins imported into secondary plastids have bipartite topogenic signals (BTS) that comprise an N-terminal signal peptide (SP) to cotranslationally direct the imported protein into the ER lumen, followed by a canonical transit peptide (TP) for plastid targeting (Waller et al. 2000; Apt et al. 2002; Patron et al. 2005; Gould et al. 2006; Patron and Waller 2007; Bolte et al. 2009). The SP is cleaved off upon substrate entrance into the ER, thereby exposing the TP-like leader sequence. In cryptophytes, stramenopiles, and haptophytes this TP further directs the pre-proteins across the PPM (Apt et al. 2002; Gould et al. 2006; Gruber et al. 2007; Bolte et al. 2009). In apicomplexans and peridinin-containing dinoflagellates, the imported proteins are presumably routed via vesicular transport from the ER to the outermost membranes of their respective complex plastids (Patron et al. 2005; Agrawal and Striepen 2010).

Recent data suggest that in chromalveolates, a molecular system originally derived from the endosymbiont ERAD (Endoplasmic Reticulum Associated Degradation) is responsible for protein translocation across the PPM. In eukaryotes, ERAD components are involved in an energy-depended retro-translocation of misfolded proteins from the ER lumen into the cytosol, where they are tagged with poly-ubiquitins to be routed to the degradosome (Xie and Ng 2010). The homologs of the ERAD subunits Der1-1, Der1-2, Hrd1, and Udf1 are still encoded in the nucleomorph genome of the cryptophyte Guillardia theta (Sommer et al. 2007). The nucleomorph Der gene (ORF201) can complement a yeast strain with a defective homolog allele, indicating a conserved function despite of the fact that the secondary red algal plastid is apparently devoid of a remnant ER (Sommer et al. 2007). In diverse chromalveolates, two sets of ERAD components are encoded in the nucleus. One group of homologs corresponds to the canonical ER retro-translocon of host origin. The other set of encoded ERAD homologs is phylogenetically distinct and contains standard BTSs capable of targeting reporter markers to the complex plastids in experiments of subcellular localization (Sommer et al. 2007; Hempel et al. 2009; Spork et al. 2009; Felsner et al. 2010b). A detailed understanding of the cell biology of the ERAD system in the ER is still lacking (Xie and Ng 2010). One model posits that oligomers of polytopic Der subunits may form the protein-conducting pore of the retro-translocon at the ER. Such an idea that Der components compose a protein channel may be applicable to translocation into the secondary plastid because the Phaeodactylum tricornutum Der1-1 and Der1-2 homologs form homo- and hetero-oligomers that are associated with the PPM (Hempel et al. 2009). In addition, Der complexes interact with TPs of imported intermediates directed to the PPC (but not with TPs of stromal-targeted proteins; Hempel et al. 2009). Genetic evidence that chromalveolate symbiont-derived Der components function in protein import to the complex plastids comes from an engineered conditional Der1 mutant of Toxoplasma gondii in which the decrease in protein translocation into the apicoplast is directly proportional to the ablation of the conditionally expressed Der1 protein (Agrawal et al. 2009). Empirical data in P. tricornutum also support the idea that the ubiquitin ligase (ptE3P) and the de-ubiquitinase (ptDUP) homologs have conserved enzymatic functions and are located in the PPM and PPC of the secondary plastid, respectively (Hempel et al. 2010). Although the hypothesis of ubiquitination-dependent protein translocation into the complex plastids still needs to be verified, current evidence favors the idea that the ERAD system was co-opted from the red algal endosymbiont to mediate protein import across the PPM.

The ERAD system likely represented an “immediate” evolutionary solution for protein import into the new organelle. Secondary plastids seem to have retained components of the TOC and TIC machineries (translocon at the outer/inner envelope of chloroplasts), which are responsible for protein import across the outer and inner envelope membranes (OEM and IEM, respectively) of the primary red algal plastid (Gross and Bhattacharya 2009a). Toc75 has been identified by bioinformatic analysis in the diatoms P. tricornutum, Thalassiosira pseudonana, the haptophyte Emiliania huxleyi and the apicomplexan parasites Plasmodium falciparum and T. gondii (Bullmann et al. 2010). The P. tricornutum Toc75 forms a channel with electrophysiological properties similar to cyanobacterial and land plant homologs and is targeted to the second innermost plastid membrane by a pathway that may involve its transient accumulation in the intermembrane space (Bullmann et al. 2010). This route of protein sorting is analogous to that observed for Pea Toc75 (Baldwin and Inoue 2006), indicating remarkable conservation of protein topogenesis in plastids across widely separated taxa and after remodeling of the organelle by secondary endosymbiosis. The TIC translocon also seems to be conserved during the evolution of secondary plastids. Tic110 and Tic22 are encoded in the nucleomorph of cryptophyte algae and together with Tic20 are also found in the in the nuclear genome of diatoms and E. huxleyi (McFadden and van Dooren 2004; Gross and Bhattacharya, personal observations). Tic20 and Tic22 are also encoded in the nucleus of apicomplexan parasites and are targeted to the plastid-derived compartment (van Dooren et al. 2008; Kalanon et al. 2009; Agrawal and Striepen 2010). A conditional null mutant of T. gondii Tic20 showed that ablation of Tic20 expression impacts protein import into the apicoplast and is lethal to the parasite. However protein import into the organelle is only extinguished after 2 days of the complete absence of immunological detection of Tic20 (van Dooren et al. 2008). This situation contrasts with an immediate pronounced drop in protein import rates into the apicoplast once the T. gondii Der homolog is depleted (Agrawal et al. 2009). The lag time between Tic20 knockout and apicoplast protein import decline may suggest that Tic20 does not represent a central ­protein-conducting pore (van Dooren et al. 2008), but rather is an accessory component, or a factor for biogenesis of the IEM translocon. Despite many remaining uncertainties, increasing evidence for the involvement of TOC and TIC components in protein import across the second and first innermost membranes of chromalveolate plastids, respectively, provides a novel perspective on the evolution of secondary endosymbiosis.

B. A Bottleneck to Evolve a Secondary Plastid?

What can recent experimental data tell us about the initial premises of the chromalveolate hypothesis? Phylogenetic trees of ERAD components Cdc48 and Uba1, and Toc75 tend to place ­plastid-targeted chromalveolate proteins in the same branch forming a sister group to red algal homologs (Agrawal et al. 2009; Bullmann et al. 2010; Felsner et al. 2010b). These important data lend strong support to key ideas of the chromalveolate hypothesis; i.e., the red algal secondary plastid had a single origin and was made possible by the unique emergence of an organelle protein sorting system (Cavalier-Smith 1999). Recent observations indicate that the chromalveolate plastids may conserve the TOC and TIC pathways (Agrawal and Striepen 2010; Bullmann et al. 2010). This raises an intriguing perspective in which the bottleneck to evolve protein targeting to the red algal endosymbiont captured within the endomembrane system was the rerouting of proteins from the ER lumen across the endosymbiont plasma membrane. The initial steps of organellogenesis probably included transfer to the nucleus of genes encoding TP-contained products that eventually were mistargeted to the host ER. The redirection of ERAD components (e.g., the Der subunits) from the red algal ER to its plasma membrane may have been an adaptation to retro-translocate ER dispersed TP-containing proteins into the endosymbiont cytosol. Once entering that compartment, proteins equipped with a TP would by default have been routed to the red algal primarily plastid via TOC and TIC translocons. This scenario is supported by the observation that ERAD components once relocated to the PPM would retain the same topology as in the ER membrane.

C. Co-option of Pre-existing Topogenic Signals

Another important assumption of the chromalveolate hypothesis is the “difficulty” to establish a system of topogenic signals for proteins directed to the new organelle (Cavalier-Smith 1999). Chromalveolate plastid-directed proteins are first targeted into the ER via a standard N-terminal SP contained in the BTS, and then further directed across the PPM by a canonical TP for targeting into the primary plastids (Bolte et al. 2009). The TP-like sequences of the chromalveolates tend to preserve features found in TPs of Plantae (Tonkin et al. 2006; Patron and Waller 2007; Felsner et al. 2010a); i.e., the bias for the hydroxylated amino acids serine and threonine and under-representation of acidic residues, conferring a net positive charge to the TP. More important is the tendency for conservation of a phenylalanine near the first amino acid of the TP, that is a hallmark of the Rhodophyta (Patron and Waller 2007). This observation points to a “less-difficult” scenario, whereby topogenic signals were not created de novo during the evolution of secondary plastids, but instead recycled from pre-existing systems, probably via exon shuffling of SPs and TPs to newly established protein-coding genes with a function in the organelle (Kilian and Kroth 2004). It is noteworthy that precursors of proteins with a final destination in the PPC are also equipped with TP-like sequences. Empirical and bioinformatic evidence in cryptophytes and stramenopiles indicates that absence of the critical N-terminal phenylalanine at the TP seems to be the topogenic determinant to retain import substrates in the PPC once the import intermediate crosses the PPM (Gould et al. 2006; Patron and Waller 2007; Felsner et al. 2010a). Such a feature suggests that a standard sorting system in the new organelle initially evolved to target precursors to the primary plastid, and a mechanism to halt proteins in the PPC was superposed on to this feature. It is likely that the translocon at the PPM was initially under selective pressure to evolve affinity for TP-containing substrates directed to the primary plastid, indicating that the functions of the red algal primary plastid (e.g., photosynthesis) were the target of selection. That TPs emerged as a canonical signal to cross the red algal former plasma membrane is in accordance with our previous hypothesis that topogenic signals tend to emerge from physical properties already present in the import substrate (see Gross and Bhattacharya 2009b).

D. Evolution of Secondary Plastids, an Insiders’ Perspective?

In light of the ideas discussed above, can we draw comparisons between the evolution of protein targeting to the primary plastid in Plantae and to the secondary plastid in chromalveolates? We previously postulated that the evolution of the primary plastid and the mitochondrion was constrained by topological factors (Gross and Bhattacharya 2009b). Initially, host-encoded proteins that were synthesized in the cytosol could not easily cross the two membranes of the Gram-negative endosymbiont progenitors of mitochondria and plastids to directly gain access to their interior. Therefore, organellogenesis of plastid and mitochondria was hypothetically initiated by limited targeting of host-proteins constrained to the OM of the captive endosymbionts. Such a view, referred to as an “outsiders’ perspective”, suggests that organelle evolution then progressed by gradually establishing an inward organized topological system to finally direct proteins into the organelle lumen (Gross and Bhattacharya 2009b). However, organellogenesis of secondary plastids may be conceptually different from that of primary plastids and mitochondria. The fact that TPs seem to be the standard topological signal to move imported proteins across the PPM indicates that organelle protein sorting was initially selected to import substrates directly to the primary plastid located inside the endosymbiont (i.e., an “insiders’ perspective”). The Toc and Tic translocons and pathways to further route proteins to the thylakoid membranes of Plantae plastids seem to be conserved in chromalveolates (Broughton et al. 2006; Gould et al. 2007; van Dooren et al. 2008; Bullmann et al. 2010). If the red algal endosymbiont was trapped within the host endomembrane system, it is then conceivable that the only physical obstacle for host proteins to reach the red algal innermost compartments was to cross the endosymbiont plasma membrane. The recruitment of the endosymbiont ERAD translocon to the PPM likely represented a one-step solution to overcome this topological constraint. Interestingly, ERAD homolgs are still encoded in the nucleomorph genome of G. theta indicating that these molecular components were directly co-opted from the endosymbiont genome (Sommer et al. 2007). This case represents a deviation from the pattern that most molecular components supporting the evolution of primary plastids and mitochondria arguably evolved in the host genome (Gross and Bhattacharya 2009b, 2011). Conceivably, direct recruitment of the ERAD components from the endosymbiont chromosome only involved minor modifications. Nonetheless, the overall tendency for shrinkage and disappearance of the former red algal nuclear genome and occurrence of EGT requires that the secondary plastid evolution should be interpreted as a result of events predominantly selected in the host nuclear genome (Gross and Bhattacharya 2009b. 2011).

E. Convergent Evolution of Secondary Plastids

Taxa belonging to Chlorarachniophyta and Euglenophyta also have a secondary plastid surrounded by four and three membranes, respectively (Bolte et al. 2009). These organelles are derived from green algal plastids via two independent secondary endosymbiotic events. Curiously, many overlapping features are observed between protein topogenesis of these green algal secondary plastids and chromalveolates. For example, proteins targeted to the chlorarachniophyte and euglenophyte plastids also have BTSs analogous to that of chromalveolates (Durnford and Gray 2006; Bolte et al. 2009; Hirakawa et al. 2010). The N-terminus of these BTSs is a SP that specifies routing through the ER. It is followed by a canonical TP that shares features with green plastid TPs, such as an overall positive charge and enrichment of hydroxylated amino acids. In accordance, components of the TOC and TIC translocons may also be conserved in green algal-derived secondary plastids. The Toc75 and Tic20 components are encoded in the nucleomorph genome of the chlorarachniophyte Bigelowiella natans (Gilson et al. 2006). In addition, as in chromalveolates, the routing of proteins destined to the PPC of chlorarachniophytes seems to rely on topogenic signals that retain TP-containing proteins in that compartment (Hirakawa et al. 2010). Finally, protein targeting into the complex plastids of euglenophytes most likely proceeds via vesicular transport and is determined by a BTS that contains an additional hydrophobic stop-transfer signal immediately downstream to the TP (Durnford and Gray 2006). This may serve to anchor import substrates into the vesicular membrane. Despite having a different origin, an analogous stop-transfer signal following the TP is also observed in proteins targeted to the plastid of peridinin-containing dinoflagellates (Patron et al. 2005). In light of all these examples implying convergent evolution between complex plastids of different origins in chromalveolates, chlorarachniophytes and euglenophytes it is tempting to speculate that there is a defined trajectory for the evolution of a secondary plastid. This may be the result of strong selection for increasing host control over the photosynthetic organelle. The establishment of a translocation system in the endosymbiont plasma membrane (e.g., the ERAD machinery) and conservation of pre-existing endosymbiont protein-sorting components (e.g., the TOC and TIC translocons) and topologic signals (e.g., SPs and TPs) may be recurrent evolutionary solutions to this problem. Convergent evolution of protein targeting principles in unrelated secondary plastids surrounded by multiple membranes also corroborates the notion that topological constrains are critical barriers that impose a trajectory to evolutionary processes, such as organellogenesis, as recognized by the outsiders’ hypothesis (Gross and Bhattacharya 2009b). Finally, the idea that minimal topological innovations are required for establishment of a secondary plastid may reinforce the feasibility of a past existence of a cryptic green algal secondary organelle/endosymbiont that presumably left a phylogenetic footprint of EGT in the genome of the chromalveolate ancestor (Moustafa et al. 2009).

III. Kleptoplasty of a Secondary Endosymbiont in a Metazoan System

A. Introduction

Sacoglossan molluscs are marine invertebrates generally referred to as sea slugs. Many sacoglossans have evolved close evolutionary relationships with their algal food source (though not all are herbivorous) and, for some, the feeding mechanism and apparatus are highly specialized for suctorial feeding on their algal prey (Jensen 1997). These molluscs break down the algal components, except for the plastids which are retained intact in the animal digestive tissue for anywhere from 24 h to 10 months (Händeler et al. 2009; Yamomoto et al. 2009; Rumpho et al. 2011). This unique relationship between animal host and algal plastids is referred to as a symbiosis (more correctly, an endosymbiosis), because the host retains the organelle intracellularly and is endowed with a novel metabolic trait – photosynthesis. Photosynthate from the plastid has been traced to the host and the animals can be maintained in the laboratory for months with light and CO2 alone; no additional energy or food sources are required (reviewed by Rumpho et al. 2011).

The vast majority of these specialized ­sacoglossans feed on chlorophyte algae possessing plastids of primary endosymbiotic origin (see Fig. 2.3; Jensen 1997; Händeler et al. 2009; Händeler et al. 2010; Wägele et al. 2010); thus, these host animals are models of kleptoplasty of a secondary endosymbiont. In contrast, E. chlorotica is unique in that its algal prey is the stramenopile alga, Vaucheria litorea (Xanthophytes, yellow–green algae). The plastids in Vaucheria are products of a secondary endosymbiosis involving the uptake of a red alga (Fig. 2.3). Stramenopile plastids, such as in Vaucheria sp., are typically surrounded by multiple bounding membranes (three or four) including the two original primary plastid membranes from the uptake of the cyanobacterial ancestor, a third membrane derived from the cellular membrane of the symbionts (red algal cell in the case of Vaucheria), and the fourth membrane which presumably originates from the plasma membrane of the phagocytic host and is continuous with the host’s outer nuclear membrane. After E. chlorotica feeds on Vaucheria, the plastids within the animal cells definitively retain the two original plastid membranes. The presence of a third membrane surrounding the plastids is variable (but see Pierce et al. 2009); the fourth membrane appears to be lost through the mechanics of feeding and digestion. In cases where a third membrane has been visualized via electron microscopy, the source of the membrane (algal plastid membrane or host vacuolar/phagocytic membrane) remains to be determined.

Fig. 2.3.
figure 3

Putative mechanisms that may work synergistically to support long-term plastid function in the sea slug Elysia chlorotica. (a) A limited amount of horizontal gene transfer (HGT) from the algal nucleus to the animal nucleus. (b) In the environment, the animal can replenish algal plastids, proteins and transcripts through feeding. (c) Multiple mechanisms are in place in the animal to provide protection of the plastids from photo-oxidation and free radicals, including the parapodial extensions covering the body when folded, protective muco-polysaccharides, and intracellular protective compounds. (d) Proteins encoded in the animal nuclei and targeted to the mitochondria may be co-opted and used in plastid function. (e) Cellular uptake and transient use of materials encountered in the environment or during feeding, such as nucleic acids and proteins (algal, bacterial or viral), may occur in addition to the integration of plastids into the digestive cells of E. chlorotica.

Full size image

Because the plastids are secondarily derived in the algal prey for E. chlorotica, this sea slug kleptoplasty is most analogous to tertiary endosymbiosis. The most common models of tertiary symbiosis are those observed in dinoflagellates (Yoon et al. 2005; Hackett et al. 2004; Palmer 2003), but the E. chlorotica-Vaucheria system has received much attention regarding the temporary establishment, sustainment and ultimately the evolution of photosynthesis in a multicellular heterotrophic host. Thus, both the source and stability of the plastids in this model render E. chlorotica unique from its sacoglossan relatives that exhibit similar abilities, but on much shorter time scales with plastids of primary endosymbiotic origin.

B. The Stability Dilemma

Although this unique example of functional photosynthesis in a metazoan has been researched for decades (reviewed by Trench 1975; Rumpho et al. 2006; Rumpho et al. 2011; Pelletreau et al. 2011), questions surrounding the mechanisms involved in obtaining and maintaining the foreign organelle, remain unanswered. Research into the intracellular sequestration of the plastids, recognition of the plastids by the digestive cells, and the host immune response (or lack thereof), are just beginning to receive attention. Perhaps the most intriguing aspect of the E. chlorotica-Vaucheria symbiosis is the lack of algal nuclei in the animal tissue. Repeated work using a variety of techniques (from microscopy to molecular tools) has failed to provide evidence for algal nuclei or algal housekeeping genes in the animal tissue. However, like in all other photosynthetic organisms, the V. litorea plastid genome has retained only a small fraction of the genes needed for the synthesis, maintenance and turnover of photosynthesis proteins, along with sigma factors and other transcriptional and translational modifiers. The V. litorea plastid genome shows no exceptional coding capacity that would facilitate the symbiosis (Rumpho et al. 2008). In the absence of an algal nucleus encoding for these presumably requisite proteins, how then does the plastid remain stable and functional in the animal?

The inherent energy-intensive processes involved in photosynthesis typically result in rapid turnover of the protein pool to ensure efficient energy capture. Yet, in E. chlorotica, plastids continue to function, synthesize photosynthate and transfer reduced carbon to the host without a known source for the replenishment of nuclear-encoded plastid proteins (reviewed by Rumpho et al. 2006; Rumpho et al. 2011). With the apparent widespread distribution of horizontal gene transfer (HGT; e.g., Olendzenski and Gogarten 2009; Bock 2010; Boto 2010; Moran and Jarvik 2010), this mechanism has been implicated in helping to sustain plastid stability and function in E. chlorotica. The sea slug shares a physically close relationship with filaments of Vaucheria during development. In fact, this association is essential for the sea slug to undergo metamorphosis from the veliger larva stage to juvenile sea slug, and for the juveniles to mature into adult sea slugs. This requisite physical association further supports the plausibility of transient HGT from the algal nucleus to the sea slug. PCR-based results support the presence of certain genes in aposymbiotic host tissue, i.e., the egg and larval stages that are not exposed to algal prey and do not sequester plastids (Pierce et al. 1996, 2003, 2007; Rumpho et al. 2001, 2008, 2009; Schwartz et al. 2010). However, partial sequencing of the sea slug genome and transcriptome of actively photosynthesizing adult E. chlorotica, has not revealed any photosynthesis-related genes, or genes specifically originating from Vaucheria (Pelletreau et al. 2011 [but see Pierce et al. 2011]). Similar results were also observed for two other sacoglossan species, E. timida, which harbors chlorophyte plastids for several weeks at a time, and Plakobranchus ocellatus, which harbors mixed plastids for several months (Evertsen et al. 2007; Händeler et al. 2009; Wägele et al. 2010). No photosynthetic genes were identified following 454 pyrosequencing of the transcriptomes from both organisms (Wägele et al. 2010). All of these results suggest that extensive HGT has not occurred between host and symbiont, despite the close physical relationship with the symbiont’s nuclear genome during feeding, and/or HGT has not occurred uniformly among populations of sea slugs. Thus, at present, additional mechanisms must be proposed and explored to explain the sustained viability and stability of the plastids observed in the host tissue.

C. Alternate Mechanisms to Explain Plastid Stability

It is apparent that the mechanistic complexity underlying plastid stability in E. chlorotica is much greater than once presumed. Neither the algal plastid genome (Rumpho et al. 2008), nor our current understanding of the nuclear genome and transcriptome of the sea slug, provide a compelling explanation for plastid function. Although there are numerous potential explanations; here, four mechanisms will be discussed which may work synergistically with limited HGT to contribute to plastid function and stability: (1) Plastid replenishment; (2) Plastid durability and protection; (3) Transient transcript expression and protein function; and (4) Dual targeting of animal proteins (Fig. 2.1). The relative contribution of each of these mechanisms remains unknown and all avenues warrant further investigation in order to fully comprehend the processes involved in animal photosynthesis.

Limited HGT

Although emerging sequencing studies do not, at present, support the presence of expressed transferred genes, numerous studies employing a variety of methods have provided evidence for the presence of genes for nuclear-encoded plastid-targeted photosynthesis proteins in E. chlorotica (reviewed by Rumpho et al. 2006, 2011; Schwartz et al. 2010). The majority of these studies have employed single gene investigation of genomic and complementary DNA from E. chlorotica during both the aposymbiotic and symbiotic phases of the animal’s life history (veliger larvae and eggs vs. ‘green’ adults). To date, the available evidence supports the transfer of six algal nuclear genes related to energy capture and photosynthetic electron transport. These genes encode the following proteins: light harvesting complex proteins (Pierce et al. 2007), the manganese stabilizing protein of photosystem II (Rumpho et al. 2008), the Calvin-Benson cycle enzyme phosphoribulokinase (PRK; Rumpho et al. 2009; Schwartz et al. 2010; Soule 2010), and three proteins involved in chlorophyll synthesis (Pierce et al. 2009; Schwartz et al. 2010 [see also Pierce et al. 2011]). In addition, evidence for the synthesis of chlorophyll in E. chlorotica has been reported using 14C radiolabeling, suggesting the presence of additional nuclear encoded algal genes functioning in the animal (Pierce et al. 2009).

Plastid Replenishment

In its natural environment, E. chlorotica encounters Vaucheria specimens sporadically throughout the year. Observations of individuals fed algae in the lab suggest that, when available, plastids from Vaucheria are continually incorporated into the digestive cells of the animals. It is therefore reasonable to assume that proteins, transcripts and other materials that can contribute to plastid stability would likewise be available to the animal when feeding in nature or when provided algal prey in the laboratory. In the natural environment, food availability would presumably be the limiting factor for plastid sustainability for any herbivorous sacoglossan. In the marsh habitat for E. chlorotica, growth of Vaucheria is limited during the winter seasons and during this time the animals would need to support plastid functions without the introduction of new algal materials through feeding. Of interest, the animals are not observed in the field through the winter, and some speculation exists of “hibernation” in sediment or deep water. If this is the case, photosynthesis would be minimal during this time. This behavioral explanation works to explain how in nature the animals could sustain plastid function; however, in the laboratory, animals are maintained in well lit and aerated aquaria for 9–10 months after collection from the field without any additional Vaucheria to feed on (“starved” conditions). In this scenario, it is apparent that replenishment of materials via feeding is not required for long-term plastid function. As a result, one must still seek explanations to explain how plastid proteins are maintained for an extended period of time in the absence of the algal nuclei.

Plastid Durability and Protection

Many of the sacoglossan molluscs that are able to exploit plastid function feed on coenocytic (lacking cross walls) algae. Trench et al. (1973) first suggested that the inherent nature of the plastids and the morphology of coenocytic algae may play a role in the evolution of the sacoglossan-plastid symbiosis. Vaucheria plastids exhibit a unique “robustness” when isolated from the algal filament or from E. chlorotica. In comparison to spinach plastids, Vaucheria litorea plastids remain structurally intact and able to fix CO2 for 3 days after isolation from the alga when simply suspended in a buffered iso-osmotic medium. Conversely, spinach plastids showed a rapid deterioration of shape and function within 24 h of isolation (Green et al. 2005). The V. litorea plastids are also resistant to varying osmotic concentrations and able to translate proteins throughout the 3 day isolation period. When isolated, plastids of land plants and other algal species typically exhibit a precipitous drop in photosynthetic activity and protein translation, ceasing within hours of isolation (Kirk and Tilney-Bassett 1967; Morgenthaler and Morgenthaler 1976; Mayfield et al. 1995). Therefore, the physical properties of V. litorea plastids may facilitate the successful establishment of the symbiosis. It is important to remember that the vast majority of sacoglossan species feed on chlorophyte algae derived from the green algal lineage, which do not have additional envelope membranes, and there is very little known about the physical characteristics of plastids from these other algal species or if the additional membranes “protect” secondary plastids.

Photoprotection of the plastids within the animal, in addition to the robust nature of the plastids, would synergistically aid in long-term photosynthesis. Photoprotection can result from physical shading, sunscreens, antioxidants, and enzymes that counteract the effects of free radicals. All of these mechanisms may be involved in this symbiosis. Elysia species are members of the family Placobranchiodea, a distinguishing feature of which is the presence of parapodia. These wing-like extensions on the animal open and close in response to light, movement, and other environmental cues. The parapodia are thought to have contributed to the evolution of the symbiosis and the longevity of plastid function within these animals (Trench 1975; Rahat and Monselise 1979; Händeler et al. 2009; Wägele et al. 2010). Additionally, sacoglossans produce copious amounts of mucus and polysaccharides that, in other marine invertebrates, contain many UV absorbing compounds and sunscreens such as microsporine-like amino acids (MAAs; reviewed by Karentz 2001). The presence of MAAs in E. chlorotica has not been investigated. These physical characteristics of the host would presumably generate a protective environment for the plastids that could ameliorate potentially damaging effects of light absorption.

Of equal importance are mechanisms involved in intracellular photoprotection of the organelle after it has been sequestered into the digestive cells of the animal. Vaucheria litorea contains relatively high concentrations (∼2.7% dry mass) of mannitol within its tissues. Mannitol is important to marine organisms in variable saline environments, where it functions as a compatible solute (Munda 1964; Reed et al. 1985; Iwamoto and Shiraiwa 2005). In algae, the synthesis of mannitol from glycolytic intermediates only requires two additional enzymes, mannitol-1-P dehydrogenase and mannitol-1-P specific phosphatase (Iwamoto and Shiraiwa 2005; Rousvoal et al. 2011). More recently, mannitol has shown importance as an antioxidant. Plastids from tobacco plants were genetically modified to carry mannitol-1-P dehydrogenase, which synthesizes mannitol-1-P from fructose-6-P, and the transgenic progeny exhibited a greater ability to scavenge free radicals (Shen et al. 1997a). Furthermore, the Calvin-Benson cycle enzyme PRK was protected from inactivation normally caused by free radicals (Shen et al. 1997b). Preliminary investigation of mannitol concentrations in E. chlorotica showed the presence of mannitol in the animal 2 months after removal from its algal prey, V. litorea (the presumed source of mannitol; Rumpho ME and W Loescher, unpublished data). Thus, it is possible that upon feeding on V. litorea and during the uptake of plastids, the host animal also sequesters mannitol (or synthesizes it itself) and the presence of mannitol may play a role in stabilizing and protecting the intracellular photosynthetic machinery.

Preliminary investigation into the transcriptome of actively photosynthesizing E. chlorotica is revealing an abundance of genes which play a role in anti-oxidant function, including catalase, peroxisomal biogenesis factor 16, caspase 7, cytosolic phospholipase A2 beta, Cu/Zn superoxide dismutase, ferritin, manganese superoxide dismutase, peroxiredoxin 6, Ser/Thr-protein phosphatase 2A catalytic subunit beta isoform, selenium-dependent glutathione peroxidase and thioredoxin peroxidase (Pelletreau et al. 2011). Further investigation into expression levels and timing of expression of these various anti-oxidants will clarify their role in plastid protection and stability.

Transient Transcript Expression and Protein Function

As discussed earlier, transcriptome data do not support HGT as a sole or major explanation for plastid function and stability; yet, several studies have provided evidence for the presence of algal nuclear genes in varied phases of the animal life history. This discrepancy may be reconciled via transient processes (rather than permanent integration into the host genome). Several mechanisms may allow host cells to take up and use “foreign” proteins, nucleic acids or other molecules. One involves microvesicles or smaller exosomes as vehicles for transferring proteins and other molecules between cells in a variety of organisms from human and mouse to fungi (see reviews by Valadi et al. 2007; Casadevall et al. 2009; Mansfield and Keene 2009; Feng et al. 2010). Valadi et al. (2007) first demonstrated that exosomes from human and mouse mast cell lines also contain functional mRNAs and regulatory microRNAs, and these RNAs can be transferred in vitro to other cells, translated, and new proteins are observed in the recipient cells (but also see Smalheiser 2007). More recently, reports of similar biologically active vesicles akin to exosomes have been characterized in several fungi facilitating transport across the cell wall and stimulating macrophage activity in animal hosts (Casadevall et al. 2009; Regente et al. 2009; Oliveira et al. 2010).

A second mechanism facilitating transfer and expression of foreign DNA or RNA involves RNA and reverse transcriptase (RT)-mediated inheritance of novel traits. This is well documented in zygotes and spermatozoa of mice (Sciamanna et al. 2003; Rassoulza­degan et al. 2006; Cuzin, et al. 2008; Spadafora 2008; Sciamanna et al. 2009; Garcia-Olmo et al. 2010), cow (Canovas, et al. 2010; Feitosa et al. 2010), pig (Garcia-Vazquez et al. 2010) and fish (Collares et al. 2010), and is now a common mechanism employed in generating transgenic animals (Sciamanna et al. 2009). In these systems, RNA injected directly into zygotes, or sperm containing foreign DNA and RNA and incubated with cells, or cells bathed in free nucleic acids all took up foreign nucleic acids with subsequent expression of the encoded traits in the embryos. In these cases, expression of the DNA or RNA was mosaic in nature; i.e., differential expression was observed between individuals and/or among cells of one individual, and transient over time (Sciamanna et al. 2003, 2009; Rassoulzadegan et al. 2006). Genes transferred via RT/RNA mediation in E. chlorotica would presumably be expressed in a mosaic fashion, which may explain the often confounding results obtained using PCR to amplify gene products, whereby the reproducibility of the PCR reaction among animals or samples can at times be less than 25% (Pierce et al. 2007; Schwartz et al. 2010).

Development and establishment of irreversible plastid endosymbiosis (or kleptoplasty) in E. chlorotica takes place during the development of the animal gut tissue and other advanced features. For this transition to occur, the animal must feed on V. litorea resulting in its digestive tract being bathed in algal-derived nucleic acids and protein, establishing a prime environment for RT-mediated inheritance. Mechanistically, direct injection of RNA requires an RNA-dependent RNA polymerase (RDRP) for amplification, whereas cells bathed in free nucleic acids rely on RT and retrovirus activity (Alleman et al. 2006; Sciamanna et al. 2009). It is interesting to note that viral particles have been observed to increase in density and measurable RT activity increases in older senescing specimens of E. chlorotica (Pierce et al. 1999; Mondy and Pierce 2003). Additionally, the partial transcriptome library obtained to date is replete with top hits to transposases, reverse transcriptases, RDRP and retroviral Gag-Pol (polyprotein-reverse transcriptase) sequences; perhaps even more intriguing are several foreign viral signatures for RDRP (Pelletreau et al. 2011). The possibility exists for RT activity in E. chlorotica that is analogous to that observed in spermatozoa, enabling transient expression of non-animal RNA-derived cDNA in affected cells.

Dual Targeting of Cytosolic Host Proteins

Transient gene expression and DNA encountered upon feeding provide plausible mechanisms for the provision of requisite proteins in the absence of algal nuclei and massive HGT. However, this model fails to account for animals which are maintained without exposure to food for months in the laboratory, yet retain photosynthetic ability. Replenishment of these essential components could be provided through co-option of native cytosolic and mitochondrial proteins to provide analogous functions in the plastid. Several processes are shared by mitochondria and plastids including DNA replication and repair, gene expression, protein processing and proteolysis, and generation of ATP (Mackenzie 2005), and several metabolic enzymes are “shared” between pathways in the cytosol and plastid. For example, in V. litorea all of the enzymes of the Calvin-Benson photosynthetic carbon reduction cycle are nuclear encoded except for ribulose-1,5-bisphosphate carboxylase/oxygenase, Rubisco (Rumpho et al. 2008). However, all but two of these enzymes (sedoheptulose-1,7-bisphosphatase and PRK) have cytosolic counterparts in E. chlorotica. There is increasing evidence supporting dual targeting of some proteins to mitochondria and plastids (reviewed by Carrie et al. 2009) and this possibility should be considered as a contributor to long-term plastid functioning in E. chlorotica.

D. Future Directions

The mechanisms supporting this unique model of long-lasting kleptoplasty of a secondary endosymbiont in a metazoan remain elusive and the data are not reconciled. The theory of massive HGT enabling plastid function is questionable, and investigation into other explanations and mechanisms of gene expression are required, if we are to fully understand the evolution of such a novel and highly complex metabolic trait in an animal. It is likely that a combination of factors, such as those outlined here, enable the plastids to remain functional in the animal cells for long periods of time. Greater understanding of how these varied mechanisms work and their relative contributions to plastid stability and function will enhance the overall understanding of the evolution of photosynthesis and the novel acquisition of such an important metabolic function among Metazoa.