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.

Introduction

Independent primary endosymbioses involving Gram-negative bacteria led to the origin of the bioenergetic organelles mitochondria and plastids (e.g., Margulis 1970; Cavalier-Smith and Lee 1985; Gray 1992; Gross and Bhattacharya 2009). These events had lasting impacts on our planet with plastid endosymbiosis giving rise to algae and plants that became a driving force behind Earth’s climate, geochemistry, and ecology (Falkowski et al. 2004). Primary plastids are shared by three extant lineages that are referred to as the Plantae (Cavalier-Smith 1981) or Archaeplastida (Adl et al. 2005): the Glaucophyta (glaucophyte algae), the Rhodophyta (red algae), and the Viridiplantae (green algae and plants). Determining the number of primary endosymbioses (i.e., single or multiple) that gave rise to the plastid in these three phyla has long been an open question in algal evolution and directly impacts inference of the tree of life. If it happened as many as three times in the different Plantae lineages that would suggest that these taxa are polyphyletic and establishing the combination of a eukaryotic and a prokaryotic cell is relatively “easy,” in evolutionary terms. Most importantly, the resulting chimera converged on similar plastid features for each separate occurrence. If it occurred only once, then the plastid in all major autotrophic eukaryotic lineages traces its origin to this remarkable event in evolution.

A large body of data stemming from phylogenetic and comparative analyses of plastids suggest that primary endosymbiosis occurred a single time in the Plantae ancestor and all plastids [except in the Paulinella lineage (e.g., Yoon et al. 2006; Nowack et al. 2011)] trace their ancestry to this singular event (e.g., Bhattacharya et al. 2004; Delwiche 1999; Palmer 2003; Rodriguez-Ezpeleta et al. 2005; Chan et al. 2011). Nevertheless, many recent nuclear multigene trees provide little (Burki et al. 2007; Patron et al. 2007) or no (Nozaki et al. 2009; Baurain et al. 2010; Parfrey et al. 2010) support for Plantae monophyly and often provide conflicting results. The inability to conclusively support or reject Plantae monophyly (and thereby resolve the number of plastid primary endosymbioses in the eukaryote tree of life) is largely explained by the lack of complete genome data from glaucophytes. This hurdle was recently crossed with the completion and analysis of a draft genome assembly from the glaucophyte Cyanophora paradoxa (Price et al. 2012). Intriguingly, the C. paradoxa plastid (often referred to as the “muroplast”) maintains the ancestral cyanobacterial trait of a peptidoglycan wall (Pfanzagl et al. 1996). This and other traits such as an unconventional carbon-concentrating mechanism (CCM) have made this glaucophyte a model for photosynthesis and endosymbiosis research. Here, we review some of the major features of the C. paradoxa genome and the insights it provides into Plantae evolution.

Genome Data

To generate an initial genome draft, a total of 8.3 billion base pairs (Gbp) of Roche 454 and Illumina GAIIx sequence data from C. paradoxa CCMP329 (Pringsheim strain) were coassembled with 279 Mbp of random-shear Sanger sequence from this taxon. The resulting assembly comprised 60,119 contigs totaling 70.2 Mbp in size with an N50 of 2.7 kbp (minimum 100 bp and maximum 66 kbp). This highly fragmented assembly is currently being improved by the addition of significant Illumina mate-pair library sequence data. Pulsed field gel electrophoresis analysis of C. paradoxa shows the presence of at least seven chromosomes with the smallest being less than 3 Mbp in size (Price et al. 2012). A previous fluorescence-activated cell sorting (FACS) study suggested the haploid genome size in C. paradoxa was 140 Mbp (Löffelhardt et al. 1997). Given that the draft genome assembly converged on ca. 70 Mbp, it is likely that the original FACs sorting was done with diploid cells and the haploid genome size of C. paradoxa is closer to 70 Mbp.

Analysis of the C. paradoxa nuclear genome data (Price et al. 2012) demonstrates a highly enriched G + C-content (i.e., 83.8 % at third codon positions), resulting in difficulties in sequence generation and assembly. To assess whether the assembly was deficient in coding regions due to this issue, we used BLASTN with 3,900 Sanger-derived EST unigenes from C. paradoxa to query the draft assembly. This analysis showed that 99.1 % of the ESTs (i.e., putative protein coding regions) had hits (at e ≤ 10−10). This suggests the majority of expressed genes are represented in the genome data. Thereafter, a total of 15 Gbp of Illumina mRNA-seq data was used to train ab initio gene predictors, resulting in 27,921 weighted consensus gene structures. The organelle genomes of C. paradoxa were also analyzed, including novel data from its sister glaucophyte, Glaucocystis nostochinearum, but will not be discussed here (for details, see Price et al. 2012).

Testing Plantae Monophyly

Several approaches were taken to ascertain support for the monophyly of Plantae. These included phylogenetic analysis of single proteins, elucidating the extent and type of endosymbiotic or horizontal gene transfer (E/HGT), comparative analysis of groups of proteins such as components of the plastid protein translocons and fermentation pathways, and analysis of plastid solute transporters. All of these data strongly support a single origin of Plantae and therefore a single primary plastid endosymbiosis in their common ancestor (Price et al. 2012). Here, we present the results of the single protein and plastid translocon analyses.

For the single proteins, we first used BLASTP to analyze the evolutionary affiliations of the 27,921 predicted protein models in C. paradoxa. A total of 4,628 proteins had significant BLASTP hits (e ≤ 10−10) to sequences in a comprehensive local database that we use for comparative analysis (e.g., Moustafa et al. 2009; Chan et al. 2011). A simplified reciprocal BLAST best-hits approach (Chan et al. 2011) identified a total of 606 proteins that had hits only to one other phylum (i.e., exclusive gene sharing). With the requirement of an increasing number of hits per query (x) from C. paradoxa and the second phylum, as x ≥ 2 (606 proteins), x ≥ 10 (125 proteins), and x ≥ 20 (23 proteins), we found that C. paradoxa and Viridiplantae shared the largest number of exclusive genes (Fig. 1), indicating a close evolutionary relationship between these lineages. Next, using an automated approach (Chan et al. 2011), we generated 4,445 maximum likelihood (ML) trees for C. paradoxa proteins that had significant database hits. To minimize the impact of taxon sampling on this analysis, we considered trees that contained ≥3 phyla and a minimum number of terminal taxa (N) that ranged from 4 to 40 (Fig. 2a). Using this approach, we found that >60 % of all trees support (at bootstrap ≥90 %) a sister group relationship between glaucophytes and red and/or green algae. The glaucophytes were most often positioned as sister to Viridiplantae (105, 83, 48, 19, and 10 trees at N = 4, 10, 20, 30, and 40), consistent with the analysis of exclusive gene sharing with only a small number of trees favoring the monophyly of glaucophytes and red algae. This result was found even though a significant number of trees favored glaucophyte–red–green (Plantae) monophyly (44, 40, 32, 18, and 16 trees at N = 4, 10, 20, 30, and 40) and we had substantial red algal genome data in our database (361,625 sequences). Interestingly, many of the trees showed C. paradoxa to be monophyletic with other Plantae in a clade (“shared”) that also included non-Plantae phyla (GlR/GlGr/GlRGr in Fig. 2a). When we sorted the phylogenomic output using the red or green algae as the query to test Plantae monophyly, these results also identified Plantae as the most frequently recovered clade (Fig. 2b, c). However, both red and green algae show far more gene sharing than glaucophytes because they, unlike glaucophytes, are implicated in secondary endosymbioses that have resulted in their genes being spread throughout the tree of life to groups such as “chromalveolates” and euglenids (Harper and Keeling 2003; Moustafa et al. 2009; Baurain et al. 2010; Chan et al. 2011). Given that single protein trees firmly establish glaucophytes as members of the Plantae we analyzed a landmark trait of Plantae, the plastid protein translocons.

Fig. 1
figure 1

Exclusive gene sharing between Glaucophyta and one other taxon, when the total number of hits (x) was ≥2 (a), ≥10 (b), and ≥20 (c). The green slices get larger as x increases. The vira matches include prasinophyte (Bathycoccus, Micromonas, and Ostreococcus) and Chlorella viruses

Full size image
Fig. 2
figure 2

Testing Plantae monophyly. (a) Percentage of maximum likelihood single protein trees that support the monophyly of Glaucophyta (bootstrap ≥90 %) with other members of the Plantae, or in combination with non-Plantae taxa that interrupt this clade. These latter groups of trees are primarily explained by red or green algal endosymbiotic gene transfer (EGT) into the nuclear genome of “chromalveolates” and euglenids. For each of these algal lineages, the set of trees with different numbers of terminal taxa (N) ≥4, ≥10, ≥20, ≥30, and ≥40 and distinct phyla ≥3 in a tree are shown. Similar analyses that were done using red algae and Viridiplantae as the query to test for Plantae monophyly are shown in panels (b) and (c), respectively

Full size image

A key innovation required for the cyanobacterium-to-plastid evolutionary transition in primary endosymbiosis was the establishment of protein translocons for protein targeting into the emergent organelle (e.g., Reumann et al. 2005; Gross and Bhattacharya 2008, 2009). Components of the Translocons at the outer and inner envelope membranes of chloroplasts (Toc and Tic, respectively) have been described in higher plants, and algae of the green, red, and “chromalveolate” lineages (McFadden and van Dooren 2004). The existence of an analogous protein import system in C. paradoxa is suggested by immunological detection of epitopes in this alga using plant Toc75 and Tic110 antibodies, and heterologous protein import assays (Steiner et al. 2005; Yusa et al. 2008). These data suggest that all Plantae share a key invention that laid the foundation for plastid integration within the host cell. Our analysis of the C. paradoxa genome identified homologs of Toc75 and Tic110 that are OEM (outer envelope membrane) and IEM (inner envelope membrane) protein conducting channels, respectively, two Toc34-like receptors, as well as homologs of the plastid Hsp70 and Hsp93 chaperones, and stromal processing peptidase (Price et al. 2012). Such a minimal set of components is likely to have formed the primitive protein translocation system in the Plantae ancestor (Gross and Bhattacharya 2008, 2009). Candidates for additional translocon subunits were also detected in C. paradoxa. Furthermore, the Tic20 and Toc22 ML phylogenies provide unambiguous evidence for a cyanobacterial provenance of these genes in Plantae and a monophyletic relationship of C. paradoxa with plants and other algae (see Price et al. 2012). In summary, the evolution of protein translocons to the nascent plastid has long been held as a formative event in the emergence of the Plantae ancestor. Analysis of the C. paradoxa genome revealed the presence of the conserved core of translocon subunits derived from the cyanobacterial endosymbiont (i.e., Toc75, Tic20, Tic22) as well as novel genes that apparently evolved de novo in the host (i.e., Toc34 and Tic110). These data provide further unambiguous evidence that the primary plastid was established in a single common ancestor of the Plantae.

Enzymes of Peptidoglycan Biosynthesis

The muroplast wall of glaucophyte algae is the sole documented example of peptidoglycan (PG) in Plantae and its origin from the plastid endosymbiont is noncontroversial. The PG consists of one giant molecule (“sacculus”) and belongs to the A1gamma type, like the cell walls of Escherichia coli and cyanobacteria, but is thicker and more cross-linked than in the former and more reduced than in the latter. A unique feature of the PG is its modification with N-acetylputrescine (Löffelhardt and Bohnert 2001). The space between the inner and outer envelope membranes of muroplasts, the “periplasmic space”, harbors the peptidoglycan layer and enzymes for its synthesis, modification, and degradation.

PG biosynthesis has been studied in great detail in E. coli and can be divided into cytoplasmic, membrane-bound, and periplasmic steps. This three-step compartmentalization process is also present in C. paradoxa: (1) biosynthesis of the disaccharide–pentapeptide precursor occurs in the muroplast stroma (activities of MurA and MurF have been shown), (2) its transfer to the lipid carrier at the inner envelope membrane, and (3) its insertion into growing PG chains in the periplasmic space (Löffelhardt and Bohnert 2001). The latter step is catalyzed by penicillin binding proteins (PBPs) that have transglycosylase and/or transpeptidase activities. Seven PBPs that range in size from 35 to 110 kDa were identified in the muroplast envelope by labeling with a radioactive derivative of ampicillin. In addition, enzymatic activities of dd- and ld-carboxypeptidases and dd-endopeptidase that hydrolyze defined bonds in PG have been demonstrated in muroplasts (Löffelhardt and Bohnert 2001).

Here, three different approaches were used for PBP gene identification (1) domain search; (2) BLAST search against the eight PBP genes of Synechocystis sp. PCC6803 (Marbouty et al. 2009) and the Anabaena sp. PCC7120 homologs; and (3) BLAST search against Physcomitrella patens PBP-like genes. In most cases, the results converged leading to at least 11 genes or gene fragments being identified in C. paradoxa (Table 1). In general, sequence similarity was higher to homologs in cyanobacteria than those in P. patens. No C. paradoxa homologs to the small PBPs 6 and 7 were identified. The PBP numbering scheme applied here is from E. coli. However, sequence similarity (especially among the large PBPs) is significant which is reflected in their redundant function.

Table 1 Nuclear genes involved in the biosynthesis of plastid peptidoglycan in C. paradoxa
Full size table

In some cases of periplasmic proteins, bipartite presequences consisting of a transit peptide and a signal peptide can be envisaged. This suggests import to the muroplast stroma, followed by export to the periplasmic space. This special variant of “conservative sorting” would necessitate a dual location of Sec (already documented) and Tat (seems possible as another parallel to cyanobacteria) translocases on thylakoid and inner envelope membranes of muroplasts. In a Gram-negative background, the low molecular weight (MW) peptidases VanX and VanY are not linked to vancomycin resistance but rather to d-alanine recycling and to an additional endolysin, respectively. Peptidoglycan biosynthesis requires cleavage of existing glycan chains to allow for insertion of new material. This is performed by soluble and membrane-bound lytic transglycosylases: one gene of this kind could also be identified in C. paradoxa. A lysozyme family protein with significant similarity to protist lysozymes displays a signal peptide indicating a vacuolar (lysosomal) location that is likely involved in the autophagosomal digestion of damaged muroplasts. Genes for stromal proteins that are involved in the synthesis of the soluble precursor are denoted as glm. The N-terminal transit peptide identifies one protein in C. paradoxa (glmS) as a member of the muroplast-resident PG biosynthesis pathway, whereas a cytosolic counterpart would be expected to participate in protein glycosylation. The complete list of enzymes in the alga that are involved in UDP-N-acetylmuramate biosynthesis as well as the peptide side-chain adding enzymes, and the alanine (Alr) and glutamate (MurI) racemases are listed in Table 1. The membrane-bound or associated MraY and MurG proteins complete this compilation.

Genes for enzymes of PG biosynthesis were transferred twice into Plantae in the course of evolution—from the mitochondrial ancestor and from the cyanobacterial ancestor of plastids. These remain recognizable in sequence from Arabidopsis thaliana (few genes) to the moss P. patens (almost complete set), but their functions are likely to have changed. As long as chemical and structural proof is lacking (pleiotropic), effects of antibiotics or gene knock-outs of plastid division do not provide sufficient evidence to claim the presence and biosynthesis of PG in the plastids of bryophytes (Takano and Takechi 2010). Glaucophyte PG is unique in Plantae. In Paulinella, the situation is different: there is also PG in this eukaryote, but all genes necessary for its biosynthesis (Marin et al. 2007) are encoded on the endosymbiont (i.e., “chromatophore”, photosynthetic organelle) genome, which exceeds the size of plastid genomes by a factor of 5–10. Unlike their counterparts in C. paradoxa, these genes retain their prokaryotic character, i.e., they were not transferred to the nuclear genome and thus no import of precursor proteins is required for biosynthesis of the sacculus in photosynthetic Paulinella species.

The correlation of more than one gene to a given function is not uncommon among cyanobacteria. A second gene with high sequence similarity to murG is more closely related to MGDG synthases, the likely function of “MurG” in plants. In an analogous fashion, murD-like genes might instead play a role in folate biosynthesis. Until the presence of PG in P. patens is unequivocally proven, one should expect modified functions for “mur-like” genes. The fact that the cyanobacterial counterparts are often, but not always the top hits suggests a mosaic structure of the gene complement for PG biosynthesis in C. paradoxa. HGT from bacteria (e.g., Firmicutes and Verrucomicrobia) is likely to be prominent when the transferred genes provide a required function, i.e., PG biosynthesis in the case of glaucophytes. In addition, gene replacement might have occurred in some cases.

The Rubisco-Containing Microcompartment of Muroplasts: Carboxysome Versus Pyrenoid

The conspicuous, electron-dense central body of C. paradoxa muroplasts described in most publications was named a carboxysome (Raven 2003; Fathinejad et al. 2008). This coinage did not take into account the fact that eukaryotes contain pyrenoids to fulfill the function of a carbon-concentrating mechanism (CCM) and emphasized the often-postulated transitional position of glaucophytes between plastids and cyanobacteria. However, all of our attempts to identify carboxysomal shell proteins in the C. paradoxa genome failed, either with domain searches (BMC = Pfam 00936 or Pfam 03319) or with a concatenated dataset of cyanobacterial CcmKLMNO sequences. Indeed, it might be problematic to harbor shell protein genes in the nucleus, because they have high affinities to each other and likely self-assemble as carboxysomal prestructures (Kinney et al. 2011), thereby interfering with protein import into muroplasts. Thus far, Paulinella constitutes the only example of “eukaryotic carboxysomes”. Again, the necessary genes remain on the plastid genome, interestingly derived via HGT (Marin et al. 2007). In any case, the hypothesis of peptidoglycan retention in C. paradoxa (Raven 2003) to stabilize the plastid against the osmotic pressure of bicarbonate that is enriched more than 1,000-fold in the stroma through the action of the carboxysomal CCM could not be verified. In contrast, evidence was obtained (Table 2) for a number of proteins (LciB, C, and D) with functions in the pyrenoidal CCM of Chlamydomonas reinhardtii (Yamano et al. 2010). LciB and LciC were shown to form a hexameric complex (ca. 360 kDa) under active operation of the CCM: light and low concentration of CO2. This complex localizes close to the pyrenoid but is relocalized from the pyrenoid to the stroma upon high CO2 concentration or darkness. There seems to be no connection to pyrenoid development and/or starch sheath formation. A role is assumed in trapping of CO2 that has escaped from the pyrenoid via interaction with the carbonic anhydrase Cah6 and, eventually, also in accumulating CO2 reaching the stroma from the cytosol, i.e., in the active uptake of CO2 in C. reinhardtii (Wang et al. 2011). Alternatively, physical blockage of CO2 from escaping the pyrenoid by the complex has been postulated (Yamano et al. 2010). The complex is not required under high levels of CO2. In this case, a function similar to the cyanobacterial shell proteins CcmK and CcmL (which, however, are present under all conditions) can be envisaged. Some putative cyanobacterial plastid ancestors contain LciB and LciC, given their filamentous nature (Lyngbya) or capability of producing a starch-like reserve carbohydrate (Cyanothece). These bacteria might use mechanisms of the type discussed above that are superimposed on their carboxysomal CCM. If carboxysomes were transferred to early plastids via endosymbiosis, the separation between carboxysomal and pyrenoidal CCM could have occurred within the phylum Glaucophyta, i.e., C. paradoxa and G. nostochinearum already progressed towards a pyrenoidal CCM, whereas Gloeochaete wittrockiana and Cyanoptyche gloeocystis, with their polyhedral microcompartments confined by an electron-dense shell-like layer, might have retained the carboxysomal CCM (Fathinejad et al. 2008). Under such a scenario, the ccmKLMNO genes would be expected to reside on the muroplast genomes of G. wittrockiana and C. gloeocystis. The PG wall, though no longer necessary, was retained for unknown reasons in the plastids of C. paradoxa and G. nostochinearum. Table 2 includes two genes encoding the putative bicarbonate transporter LciA and several genes with strong sequence similarity to genes for LciB, LciC, and LciD from C. reinhardtii. Because these are closely related, an exact assignment is difficult. However, whenever the N-termini are intact, unequivocal muroplast presequences were found for these enzymes.

Table 2 Genes for proteins involved in the CCM of Cyanophora paradoxa
Full size table

A key enzyme of the CCM is carbonic anhydrase, either copackaged with Rubisco in cyanobacterial carboxysomes or located in the lumen of thylakoids traversing the pyrenoid of C. reinhardtii. The number of CAs can vary among algae, e.g., from 9 in C. reinhardtii to 13 in some diatoms (Tachibana et al. 2011). Five CAs from C. paradoxa are shown in Table 2. Two of these belong to the gamma-CA family with high sequence similarity to homologs in plants. The other three contain the conserved Zn-binding site (VCGHSHCGAMKG) of (cyano)bacterial beta-CAs. In the case of the putative mitochondrial CAs, high sequence similarity to C. reinhardtii Ca1 and Ca2 is observed. A bona fide muroplast CA (e.g., the stromal Cah6 or the lumenal Cah3 of C. reinhardtii) is missing from this compilation. If we assume a pyrenoidal CCM in C. paradoxa, the organism must utilize a mechanism different from that in C. reinhardtii. There is no evidence of a thylakoid-lumenal CA or a muroplast microcompartment traversed by thylakoid membranes. In the diatom Phaeodactylum tricornutum, the carbonic anhydrase CA-1 (CO2 responsive) is copackaged with pyrenoidal Rubisco and does not reside in the lumen of the traversing thylakoid (Tachibana et al. 2011). Mass spectrometric analysis of central body proteins from C. paradoxa did not reveal a CA-like protein. The only outcome of these studies (in addition to Rubisco LSU and SSU) was Rubisco activase that was also corroborated by Western blotting and assembly studies after in vitro import into isolated muroplasts (Fathinejad et al. 2008). C. paradoxa activase, while showing high sequence similarity to both cyanobacterial and plant homologs, lacks the C-terminal extension typical for filamentous cyanobacteria. This protein contains a domain that shares high sequence similarity with repetitive regions found in the largest carboxysome shell protein CcmM. An N-terminal extension present in plant homologs is present in the C. paradoxa protein. Taken together, the domain structure of Rubisco activase from C. paradoxa does not support the carboxysome concept. Several genes listed in Table 2 were shown to be CO2 responsive in the closely related C. paradoxa SAG 45.84 (Kies strain) underlining their postulated role in the CCM (Burey et al. 2007).

Conclusions

Rather than being a relict lineage, the analyses presented here and in Price et al. (2012) paint a picture of the “living fossil” C. paradoxa as a gene- (and function)-rich species that provides many clues to early events in plastid endosymbiosis and Plantae evolution. These data unambiguously support Plantae monophyly, thereby answering a fundamental question about the eukaryote tree of life. The components of the peptidoglycan biosynthetic pathway in C. paradoxa were identified and indicated a cyanobacterial provenance of many key enzymes with likely instances of recruitment of additional genes via HGT from other prokaryote sources. Finally, evidence was found that strongly argues against the existence of a proposed eukaryotic carboxysome in C. paradoxa. The available data are more consistent with a pyrenoidal CCM in this species and in its sister G. nostochinearum. However, the mechanism of CCM function is likely to be different from that found in C. reinhardtii.