The Spheroid Body Genome Is Reduced in Size and Gene Content.

The spheroid bodies in a clonal culture of E. turgida were isolated by discontinuous density Percoll gradient centrifugation. DNA extracted from the spheroid body-rich fraction was subjected to whole-genome amplification followed by 454 and Illumina sequencing. Sequence assembly was carried out by a combination of Velvet (19) and GS de novo assembler software, followed by PCR-based contig gap filling (Sanger sequencing). We obtained a single circular chromosome of 2.79 Mbp with a G+C content of 33.4% (Fig. 1). The genome has 39 tRNA genes (which can translate all 61 aa codons), two identical rRNA gene operons, and 1,720 protein-coding genes (Table 1). We also detected 225 pseudogenes, and no transposable elements were identified. The E. turgida spheroid body (EtSB) genome shares two major syntenic gene clusters with other cyanobacterial genomes—one for nitrogen fixation genes (nif genes) and the other for genes encoding ribosomal proteins. In addition to the EtSB genome, we assembled a 5.7-Kbp fragment with a low G+C content (Dataset S1). Most of the protein-coding genes on this fragment show no specific affinity to homologs of Cyanothece spp., which bear a close affinity to the spheroid bodies in cyanobacterial phylogeny (as detailed later). It is at present unclear whether this 5.7-Kbp fragment is an extrachromosomal element (e.g., plasmid) of EtSB origin; we did not analyze it further.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Map of the circular chromosome of the spheroid body of the diatom E. turgida. Outer two circles (dark and light blue) show positions of protein-coding genes on plus and minus strands. The red bars on the third circle indicate pseudogenes. tRNA genes (yellow bars) and rRNA genes (green bars) are displayed on the fourth circle. The innermost circle shows G+C content (window size: 5,000 bp).

The spheroid bodies of rhopalodiacean diatoms have been shown to be specifically related to members of the cyanobacterial genus Cyanothece (9) (Fig. S1). Among cyanobacteria most closely related to spheroid bodies, complete genome data are available for three free-living Cyanothece species, strains PCC 8801 (National Center for Biotechnology Information BioProject ID PRJNA59027), PCC 8802 (PRJNA59143), and ATCC 51142 (PRJNA20319) (20, 21), as well as an uncultured oceanic species living symbiotically with a haptophyte alga (UCYN-A) (22, 23). The EtSB genome was found to be smaller in size and lower in G+C content than the free-living Cyanothece strains, which possess genomes of 4.68–5.36 Mbp in size and 37.9–39.8% G+C content (Table 1). Consistent with the observed differences in genome size, we found a marked difference in the number of ORFs between the EtSB genome and its free-living relatives. Whereas only 1,720 ORFs were identified in the EtSB genome, the genomes of the three free-living Cyanothece strains possess 4,367–5,304 ORFs. Moreover, 54.5% of the ORFs on the EtSB genome could be assigned to functional categories in Kyoto Encyclopedia of Genes and Genomes (KEGG) orthology (KO), whereas only 32.6–38.0% of the ORFs in the free-living Cyanothece strains could be given a functional designation. Overall, these data suggest that the EtSB genome has undergone significant reductive evolution, as seen in other obligate bacterial symbionts (15, 24⇓–26). Furthermore, the presence of numerous pseudogenes implies that genome reduction is still ongoing in the genome.

The partial nature of the R. gibba spheroid body genomic data obtained by Kneip et al. (12) precluded whole genome-scale comparisons between the spheroid body and free-living cyanobacteria, or between the spheroid body and other symbiotic cyanobacteria. Our KO-based analysis of the ORFs in the complete EtSB genome revealed that the spheroid body has already lost a large part of the metabolic functions presumed to have been present in its free-living ancestors. The numbers of KO IDs assigned to each predicted protein set for Cyanothece spp. ATCC 51142, PCC 8801, and PCC 8802 are 1,309, 1,325, and 1,318, respectively. In sharp contrast, only 849 KO IDs could be assigned to EtSB proteins, and most of these (842 KO IDs) were shared with at least one of the Cyanothece reference genomes (Fig. 2A). Large differences in the number of KO IDs between the EtSB and its free-living relatives were observed in several KEGG functional categories, most notably “energy metabolism,” “metabolism of cofactors and vitamins,” “carbohydrate metabolism,” “membrane transport,” and “signal transduction” (Fig. 2B). For instance, the EtSB genome possesses only half of the IDs present in sequenced Cyanothece genomes in the category of energy metabolism, mostly the result of a reduction in energy production abilities (e.g., photosynthesis, as detailed later). To the extent that the KO ID repertoire represents the metabolism of an organism, the EtSB genome has significantly reduced its metabolic capacity from its cyanobacterial progenitor, which is likely to have had a similar set of genes to strain PCC 8801, PCC 8802, or ATCC 51142.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

KO-based comparison between free-living Cyanothece strains and cyanobacterial symbionts. (A) Venn diagram of KO ID repertoires of the spheroid body of Epithemia turgida and Cyanothece spp. strain PCC 8801, PCC 8802, and ATCC 51142. The numbers in parentheses indicate the total number of unique KO IDs in each genome. (B) Comparison of KO ID numbers between the EtSB, PCC 8801, PCC 8802, and ATCC 51142. Each bar indicates numbers of unique KO IDs in each functional category. (C) Venn diagram of KO ID repertoires of the EtSB, UCYN-A, R. intracellularis and ‘N. azollae.’ (D) Comparison of KO ID numbers between the EtSB, UCYN-A, R. intracellularis, and ‘N. azollae' in functional categories.

In addition to the EtSB, two complete genome sequences are available for cyanobacterial nitrogen-fixing symbionts: ‘Nostoc azollae,’ which is a vertically inherited but extracellular symbiont of the small heterosporous water fern Azolla filiculoides (27), and the uncultured cyanobacterium UCYN-A, which is thought to have an extra- and/or intracellular symbiotic relationship with haptophyte algae (22, 23, 28). Additionally, there is a near-complete genome sequence for another cyanobacterial symbiont, Richelia intracellularis, which has a mutually dependent relationship with a diatom distantly related to rhopalodiacean species (29). Whereas the genomes of ‘N. azollae’ and R. intracellularis are larger than the EtSB genome (5.49 Mbp and 3.24 Mbp, respectively; Table S1), UCYN-A has a severely reduced genome of 1.44 Mbp, nearly half the size of the EtSB genome. As in EtSB, the UCYN-A, R. intracellularis, and ‘N. azollae’ genomes have smaller KO ID repertoires (683, 965, and 1,084, respectively; Fig. 2C and Table S1) than the free-living Cyanothece spp., which have >1,300 KO IDs (Fig. 2A). In the functional categories of energy metabolism and metabolism of cofactors and vitamins, UCYN-A and EtSB exhibit similar reductive trends compared with ‘N. azollae’ and R. intracellularis (Fig. 2D). Nevertheless, UCYN-A has fewer KO IDs than does EtSB in other functional categories, such as “carbohydrate metabolism,” “amino acid metabolism,” and “nucleotide metabolism,” suggesting that the genome of this uncultured cyanobacterial symbiont has experienced more severe reduction in terms of genome size and metabolic capacity than that of EtSB, R. intracellularis, and ‘N. azollae’ (Fig. 2 C and D and Table S1).

Spheroid Bodies Are Nonphotosynthetic Cyanobacterial Obligate Endosymbionts.

To our knowledge, the complete EtSB genome provides us with the first opportunity to predict the entire set of metabolic pathways carried out in the spheroid body (see Fig. 4). Pioneering studies detected no chlorophyll autofluorescence (13), and inactivated photosynthetic genes were found in a partially sequenced spheroid body genome (12). However, given the prominent thylakoid membranes of spheroid bodies (7), it was difficult to exclude the existence of facultative photosynthetic capacity of the endosymbiont based solely on the partial genome sequence of RgSB or microscopic observations. Analysis of the complete genome sequence of EtSB presented herein allows us to conclude that the cyanobacterial endosymbionts of rhopalodiacean diatoms truly lack photosynthetic ability. In the EtSB genome, genes for photosynthetic components such as photosystems I and II and phycobilisomes (the main peripheral antennae of cyanobacteria) were absent or clearly nonfunctional (Fig. 3). We also found that genes for the entire chlorophyll synthesis pathway, which converts protoporphyrin IX to chlorophyll a, appear to be inactivated by pseudogenization (Fig. S2). All things considered, the complete spheroid body genome sequence provides concrete and comprehensive evidence for the loss of photosynthesis. To our knowledge, the EtSB is the only known example of a nonphotosynthetic cyanobacterium, free-living or symbiotic.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Schematic representation of the predicted electron flow in the thylakoid membrane of the spheroid body. Dashed red arrows indicate deduced electron flow. Complexes predicted to be inactivated by gene loss and/or pseudogenization are shown by dashed lines. Yellow- and blue-colored complexes are specific to respiration and photosynthesis, respectively, whereas those in green are involved in respiration and photosynthesis. COX, terminal cytochrome c oxidase; cyt b₆f, cytochrome b₆f complex; FNR, ferredoxin-NADP⁺ reductase; Fd, ferredoxin; Hup, uptake hydrogenase; NDH, type 1 NADPH dehydrogenase; PC, plastocyanin; PBS, phycobilisome; PQ, plastoquinone; PS I, photosystem I; PS II, photosystem II; SDH, succinate dehydrogenase.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Overview of the deduced metabolism of the spheroid body in E. turgida. Single arrows indicate metabolic reactions and pathways involving multiple steps are represented by double arrows. Amino acids and nucleotides are highlighted by orange. Substances of particular interest are in bold. Question marks highlight pathways that are uncertain even among related Cyanothece strains. Transporter types: single oval, P-type ATPase transporter; oval coupled with small oval, ABC-transporters; single rounded rectangle, channels or secondary transporters. Ac-[acp], acetyl-[acyl-carrier protein]; Ac-CoA, acetyl-CoA; 1,3BPG, 1,3-bisphosphoglycerate; AICAR, 5-aminoimidazole-4-carboxamide ribotide; AIR, 5-aminoimidazole ribotide; APS, adenosine 5′-phosphosulfate; COX, terminal cytochrome c oxygenase; cyt b₆f, cytochrome b₆f complex; DHAP, dihydroxyacetone phosphate; FAD, flavin adenine dinucleotide; FMN, flavin mononucleotide; Fruc, fructose; GAP, glyceraldehyde-3-phosphate; Gluc, glucose; Mal-[acp], malonyl-[acyl-carrier protein]; Mal-CoA, malonyl CoA; NDH, type 1 NADPH dehydrogenase; PC, plastocyanin; PEP, phosphoenolpyruvate; 2PG, 2-phosphoglycerate; 3PG, 3-phosphoglycerate; PRPP, 5-phosphoribosyl 1-pyrophosphate.

Despite the absence of a photosynthetic apparatus, a cytochrome b₆f complex, plastocyanin, and ATP synthase are most likely functional on the spheroid body thylakoid membranes, as intact ORFs for such proteins reside in the EtSB genome (Fig. 3). In cyanobacteria, the aforementioned set of proteins is involved in respiration together with NADH dehydrogenase and cytochrome c oxidase, in addition to photosynthesis (30). Thus, we predict that the EtSB still synthesizes ATP by an electrochemical proton gradient. The uptake hydrogenase (Hup, encoded by hupSL) (31), which can recycle dihydrogens produced through nitrogen fixation as electron donors, may be involved in the process described earlier as an additional supplier of electrons (Fig. 3). Therefore, although the spheroid body lacks photosynthetic capacity, its thylakoid membranes may be retained as the site of an intact respiratory chain. Furthermore, it is possible that the activity of cytochrome c oxidase in the spheroid body serves to protect extremely oxygen-sensitive nitrogenases by reducing oxygen molecules to water.

To gain insight into the energy-metabolic dependence of the spheroid body on its diatom host, we searched the EtSB genome for genes encoding enzymes involved in the Calvin and tricarboxylic acid (TCA) cycles. The Calvin cycle appears to be incomplete (Fig. S3). In particular, ORFs for the three subunits of ribulose-1,5-bisphosphate carboxylase/oxygenase (rbcL, rbcS, and rbcX) were found to be disrupted by stop codons, suggesting that the endosymbiont cannot fix carbon dioxide through the Calvin cycle. As the EtSB still possesses apparently functional genes for enzymes to catabolize carbohydrates (e.g., enzymes involved in the pentose phosphate pathway and glycolysis, except phosphofructokinase), the endosymbiont likely energetically depends on sugars that are “extracellularly” supplied from the host. This idea is also supported by the presence of a gene encoding an ABC-transporter for sugars in the EtSB genome (Fig. 4), despite the fact that the total repertoire of transporter genes is highly reduced compared with its free-living relatives (Fig. 2B, “membrane transport”). Additionally, the TCA cycle was found to be incomplete in the EtSB (Fig. 4 and Fig. S4); only intact ORFs for citrate synthase, aconitase, isocitrate dehydrogenase, and fumarase were identified. The TCA cycle in the spheroid body is thus unlikely to function in ATP synthesis, but the products from the incomplete cycle are likely used as substrates for the biosynthesis of certain amino acids (oxaloacetate and 2-oxoglutarate are required for biosynthesis of five and four amino acids, respectively; Fig. 4). The incomplete nature of the TCA cycle and the Calvin cycle deduced from the genome data strongly support the obligate endosymbiosis between spheroid bodies and rhopalodiacean diatoms. It is conceivable that some or all of the missing enzymes in the EtSB are imported from the host cell. Regardless, the obligate nature of the endosymbiosis would not be changed, as the host cell would still be required to complete the two pathways in the endosymbiont.

Amino acid biosynthetic pathways are often partially or entirely discarded in bacterial endosymbionts (15, 22, 25, 26, 32). In sharp contrast, the EtSB genome retains seemingly functional genes for the biosynthesis of all protein amino acids (Fig. 4). Likewise, the biosynthesis of purines, pyrimidines, and most cofactors [e.g., FMN, FAD, NAD(P)⁺, CoA, heme, and thiamine] is likely to be operating in the EtSB.

The key characteristics of the EtSB described earlier—(i) absence of photosynthesis, (ii) incomplete TCA cycle, and (iii) intact biosynthetic pathways for protein amino acids, nucleotides, and cofactors—places the spheroid bodies of rhopalodiacean diatoms on a distinct evolutionary trajectory from that of UCYN-A, which discarded the entire TCA cycle, and several biosynthetic pathways for amino acids and purine nucleotides, and indeed any endosymbiotic bacteria known to date.

Is the Spheroid Body Becoming a Nitrogen Fixing Organelle?

Nitrogenase activity has been experimentally confirmed in R. gibba cells (8, 11). To validate the nitrogen-fixing capacity of E. turgida experimentally, E. turgida cells were cultured in the presence of ¹⁵N₂ gas, followed by analysis of the isotopic composition of chlorophyll a. Incorporation of isotopic nitrogen in chlorophyll a was detected (Table S2), indicating that this diatom species can indeed use gaseous nitrogen. A nearly complete set of nif genes for nitrogen fixation was identified in the EtSB genome, which is homologous to the gene set found in a 51-Kbp genome fragment of the RgSB (12). The composition and order of the nif genes were almost identical between the spheroid body genomes of E. turgida and R. gibba, with the exception of the nifU gene encoding a scaffold protein for [Fe₂S₂] cluster assembly. The partial genome of the RgSB harbors an ORF for NifU, which is highly truncated compared with other homologs (12), whereas the orthologous ORF in the EtSB genome is disrupted by multiple in-frame stop codons. We predict that the EtSB uses a NifU-like protein (ETSB_1030) located in a region distinct from the nif gene cluster, instead of the authentic NifU. It is presently not known how the products of nitrogen fixation are transported into the cytosol of the diatom host. Although a gene encoding a putative ammonium transporter persists in the EtSB genome (Fig. 4), its precise function is unclear.

It is also currently unclear how the spheroid body, which lacks a complete TCA cycle and photosynthetic ability (Fig. 3 and Fig. S4), generates sufficient ATP to support an energetically expensive nitrogenase activity (33). If glycolysis and/or the pentose phosphate pathway are the principal electron donors in the spheroid body, sugars or sugar phosphates need to be effectively imported from the host to the endosymbiont. It is also possible that the host supplies ATP to the spheroid bodies, an interesting hypothesis but one that cannot be confirmed or refuted based on the data presently in hand. Characterization of the complete set of spheroid body transporters will be necessary to elucidate the trafficking of metabolites between host and endosymbiont.

Reductive genome evolution has been demonstrated in EtSB, UCYN-A, ‘N. azollae,’ and R. intracellularis (22, 27, 29), but only the EtSB has discarded photosynthetic ability; photosynthesis genes (e.g., those for photosynthetic complexes) persist in the genomes of UCYN-A, ‘N. azollae,’ and R. intracellularis (22, 27, 29). For instance, UCYN-A retains complete gene sets for photosystem I and chlorophyll a synthesis, despite its small genome size (22). Altogether, with its lack of photosynthesis, the spheroid bodies of rhopalodiacean diatoms are highly integrated into their diatom host cells, and indeed distinct from all other cyanobacterial symbionts examined to date. However, based on the EtSB genome data currently in hand, it is not possible to determine whether the spheroid body is a bona fide organelle in the strictest sense. The evolution of protein import machinery, which enables proteins to move from host to endosymbiont, has been proposed as the critical event in the conversion of endosymbionts into organelles (34, 35). To address this issue, it will be important to survey the host diatom nuclear genome for genes possibly encoding proteins transported to, and functioning in, the spheroid body.