Report

Green Evolution and Dynamic Adaptations Revealed by Genomes of the Marine Picoeukaryotes Micromonas

See allHide authors and affiliations

Science  10 Apr 2009:
Vol. 324, Issue 5924, pp. 268-272
DOI: 10.1126/science.1167222

This article has a correction. Please see:

Abstract

Picoeukaryotes are a taxonomically diverse group of organisms less than 2 micrometers in diameter. Photosynthetic marine picoeukaryotes in the genus Micromonas thrive in ecosystems ranging from tropical to polar and could serve as sentinel organisms for biogeochemical fluxes of modern oceans during climate change. These broadly distributed primary producers belong to an anciently diverged sister clade to land plants. Although Micromonas isolates have high 18S ribosomal RNA gene identity, we found that genomes from two isolates shared only 90% of their predicted genes. Their independent evolutionary paths were emphasized by distinct riboswitch arrangements as well as the discovery of intronic repeat elements in one isolate, and in metagenomic data, but not in other genomes. Divergence appears to have been facilitated by selection and acquisition processes that actively shape the repertoire of genes that are mutually exclusive between the two isolates differently than the core genes. Analyses of the Micromonas genomes offer valuable insights into ecological differentiation and the dynamic nature of early plant evolution.

Ancestral green algae were of fundamental importance to the eukaryotic greening that shaped the geochemistry of our planet. This process began over a billion years ago when a cyanobacterium was captured by a heterotrophic protist and incorporated as an endosymbiont, giving rise to the first eukaryotic alga (1). The extant Prasinophytae retain characteristics that are believed to have been present in the last common ancestor of green algae (chlorophytes) and land plants (streptophytes, including charophyte algae) (2). Most prasinophytes within the monophyletic marine order Mamiellales (Fig. 1A and fig. S1), such as Micromonas, are tiny (≤2 μm in diameter) and known as picoeukaryotes. Micromonas is a motile unicell, with a single chloroplast and mitochondrion (Fig. 1A, inset), first reported as a dominant phytoplankter in the 1950s (3) and now recognized as having a global distribution (Fig. 1B) (4).

Fig. 1.

Micromonas phylogeny and distribution. (A) A consensus neighbor-joining (NJ) distance 18S rRNA gene tree illustrating the distinct Micromonas clades (12). Bootstrap values represent a percent of 1000 replicates (NJ), and where provided the second value represents the maximum-likelihood bootstrap percentages. The genome isolates sequenced in this work are highlighted (yellow). The previously sequenced Ostreococcus tauri and O. lucimarinus neighbor each other in clade O_I. The relationship to plants and other photosynthetic lineages is shown in fig S1. (Inset) Micromonas thin section showing the nucleus (n), chloroplast (c), flagellum (f), and mucronate extension (the thin tip at the end of the flagellum, indicated by the arrow). (B) Mean sea surface temperature (SST) for 2006 measured with global high-resolution SST (GHRSST) blended infrared and microwave SSTs, and locations where Micromonas (solid lines and circles around the isolates used in this work) and Ostreococcus (dashed lines) 18S rDNA sequences have been recovered. Micromonas appeared in all temperature regimes.

Today's oceans contain a polyphyletic diversity of algae, some with plastids that share ancestry with land plants (green algae) and others (chromalveolates) that are derived from red algae through secondary or tertiary (eukaryotic-eukaryotic) endosymbioses (5, 6). Unlike most episodic chromalveolate bloomers and the freshwater green alga Chlamydomonas (7), the Mamiellales have reduced genomes, as first shown in Ostreococcus (8, 9). Ostreococcus has a narrower environmental distribution than Micromonas (Fig. 1B) and a smaller genome (12 to 13 Mb containing only ∼8000 genes). Open-ocean bacteria, including SAR11 and Prochlorococcus (10, 11), show similar patterns of cell size and genome minimization. Conditions favoring picophytoplankton growth, such as increased stratification, less mixing, and reduced nutrient concentrations in ocean surface waters, are predicted climate change outcomes, and thus picoeukaryote dynamics may be useful ecosystem indicators.

We sequenced the nuclear genomes of Micromonas isolates RCC299 and CCMP1545 (Table 1 and figs. S2 and S3) (12). These isolates are from distant ocean provinces and fall into distinct phylogenetic clades that can co-occur (Fig. 1) (12, 13) but are generally considered a single species (Micromonas pusilla). Transmission electron microscopy revealed no morphological differences (12), and 18S ribosomal DNA (rDNA) identity was high (97%). Surprisingly, only 90% of their 10,056 (RCC299) and 10,575 (CCMP1545) predicted genes (table S1) were shared (Fig. 2A). In contrast, Ostreococcus lucimarinus and O. tauri share 97% of cataloged genes (12), and yeast genera can share ∼95% of homologs (14). The divergence we observed between the Micromonas isolates supports their classification as distinct species.

Fig. 2.

Comparison of Mamiellales gene complements. (A) Venn diagram comparing RCC299 and CCMP1545, O. tauri and O. lucimarinus gene complements. Circle sizes roughly represent relative numbers of genes in each genome. (B) Proportions of genes within eukaryotic orthologous groups (KOGs) and without KOG placement for the gene pools: unique, genes in one Micromonas species only and not the other Mamiellales (proportions shown are for RCC299; see fig. S14 for CCMP1545 proportions); shared, genes in both Micromonas species but neither Ostreococcus species; and core, genes found in the 4 Mamiellales genomes. (C) Phylogenomic profiling for core, shared, and unique genes as a percentage of the gene pool affiliated (≥50% bootstrap values) with different lineages.

Table 1.

Characteristics of the Micromonas genomes.

View this table:

Synteny, GC content, and codon usage pointed to a shared evolutionary history for RCC299 and CCMP1545 but underscored their genomic divergence [supporting online material (SOM) text S1]. Each genome contained a region that had 14% lower than average GC content, composing 7% (RCC299) and 8% (CCMP1545) of the genome (figs. S3 and S4), which also had higher transcriptional activity (SOM text S1). Similar regions in Ostreococcus (8, 9) form smaller genome proportions. DNA alignment between RCC299 and CCMP1545 low-GC regions was poor, protein colinearity was absent, and codon usage was different, in contrast to normal GC chromosomes (figs. S4 to S6).

Two major evolutionary themes emerged from our analyses. First, the common ancestor of the Mamiellales had already undergone genomic reduction, highlighted by their organellar genomes (SOM text S2, fig. S7, and tables S2 to S4). Second, Micromonas appeared to be less derived than Ostreococcus, rendering insights into the genetic composition of the proto-prasinophyte (the common ancestor of plants and prasinophytes) and specialization in extant species. Most “core” nucleus-encoded genes (genes common to the four Mamiellales genomes) were found to have known functions (Fig. 2, A and B) in key pathways (SOM text S3 to S6, tables S5 to S9, and fig. S8), such as photosynthesis, and included seleno-proteins (SOM text S3 and table S10). A significant proportion of genes grouped with land plants (Fig. 2C). Core genes branching with chromalveolates (mostly diatoms and brown algae) (Fig. 2C) presumably reflected losses (or extensive divergence) in other green lineage organisms and red algae or perhaps horizontal gene transfer (HGT).

The proto-prasinophyte features we discovered in Micromonas included transcription factors that probably belong to the “basal green toolkit” (SOM text S7, figs. S9 to S11, and table S11). For example, early-branching land plants encode most higher-plant transcription factor families except for the YABBY family (15), which was therefore posited to be evolutionarily associated with the development of leaves. However, we found YABBY in Micromonas, although it is absent from Chlamydomonas and Ostreococcus, indicating that it was part of the basal toolkit (fig. S11). We also found diversified homeodomains (fig. S12 and table S12) that are relevant to the evolution of green regulatory networks.

Although prasinophytes are often considered asexual, our observations indicated that the proto-prasinophyte was sexual. First, meiotic-specific and non-meiotic representatives of the RECA-RAD51, TOP6A/SPO11, and MUTS gene families were found (SOM text S5 and table S13). Second, the low-GC regions showed features of sex chromosomes, including RWP-RK transcription factor family genes (SOM text S7 and table S14). Third, numerous Mamiellales genes encoded hydroxyproline-rich glycoproteins (HRGP) (SOM text S6, table S15, and fig. S13), which are cell-wall components in Chlamydomonas and plants (16). Like the many carbohydrate-active enzymes (SOM text S6 and table S17), this was unexpected because cell walls have not been observed in Micromonas or Ostreococcus (Fig. 1A, inset) (4). In Chlamydomonas, one HRGP gene set is expressed only after sexual fusion to produce a thick adhesive zygote wall (17). Micromonas may behave similarly. Collectively, these data indicate the occurrence of sexual differentiation and the formation of a resistant life-cycle stage.

Fourteen percent of genes were shared between RCC299 and CCMP1545 but not with Ostreococcus (Fig. 2, SOM text S3 and S8, table S18, and fig. S14). Shared enzymes for the synthesis and remodelling of peptidoglycan in the plastid provided new insight into the evolutionary history of the ancestral cyanobacterial endosymbiont (SOM text S6) (18, 19). The shared genes also showed more rapid evolutionary rates than core genes (fig. S15), indicating that they escaped constraints acting on the Mamiellales core but still probably play important roles, given their presence in both isolates. Moreover, a larger proportion of “unique” genes (used here to mean genes mutually exclusive between RCC299 and CCMP1545) branched with opisthokont or bacterial lineages (Fig. 2C), which is consistent with acquisition by means of HGT. Many were of unknown function (Fig. 2B) but may provide useful indicator information. Following early genome reduction, fundamentally different selection/acquisition processes acting on the unique genes appear to have promoted differentiation.

Marked differences in nutrient transport were seen as compared with that in other green-lineage organisms. Between the Micromonas species, 52 of the 59 transporter gene families common to land plants were present as well as several transporter gene families found in marine chromalveolates but not in other green-lineage members (SOM text S9 and table S19). Both Micromonas spp. had more transporter families represented and higher numbers of transporters than Ostreococcus, although CCMP1545 was missing specific transporter gene families, including some related to nitrogen uptake (SOM text S9 and table S19). These differences possibly reflected environmental parameters; for instance, RCC299 is from highly oligotrophic waters, in which nutrient scavenging is essential.

We explored other genomic features related to competition and mortality that influence community structure (SOM text S10 to S13 and figs. S16 to S18). Two types of carbon-concentrating mechanisms (CCM) were identified (SOM text S12 and figs. S17 and S18) that can alleviate CO2 limitation during blooms. The more unusual Micromonas CCM, a C4-like carbon fixation pathway, includes a nicotinamide adenine dinucleotide phosphate–dependent malic-enzyme (NADP-ME) that is targeted to the plastid lumen, an atypical localization that probably reduces CO2 leakage (SOM text S12). Because C4-like pathways have now been identified in the four Mamiellales genomes and in diatoms (SOM text S12), they may represent a fairly basic necessity rather than a rare form of optimization in a few taxa. Both Micromonas species appeared to have more robust defenses against heavy-metal toxicity and reactive-oxygen species (SOM text S13 and table S20) than Ostreococcus. The larger Micromonas genome sizes may thus facilitate broader physiological response capabilities than the Ostreococcus genomes.

We found few (CCMP1545) (table S21) or no (RCC299) recognizably functional transposable elements (TEs). Most eukaryotes, including Ostreococcus (9), contain many TEs, and TE content is positively correlated with genome size above an ∼10 Mb threshold (20, 21). Any relic or degenerate TEs in Micromonas had low similarity to known TEs, and structural features of class II elements were not found. GC bias was thought responsible for the high proportion of TEs in the low-GC regions of Ostreococcus and for loss of synteny in these regions (9). However, the low-GC regions of Micromonas, although rearranged (fig. S5), had few simple repeats, contained only potential relic TEs, and showed high transcriptional activity (theoretically facilitating TE insertion) (SOM text S1), which suggests TE activity/propagation is actively hindered.

We discovered intronic repeat sequences in CCMP1545 that were absent from RCC299 and other published genomes (SOM text S14, tables S22 and S23, and figs. S19 to S22). These abundant introner elements (IEs) were located within introns, extended nearly to donor and acceptor sites (Fig. 3 and fig. S21), and lacked known TE characteristics (22). RCC299 genes generally had fewer introns than IE-bearing CCMP1545 homologs (Fig. 3), and CCMP1545 had a higher overall intron frequency (Table 1). The 9904 IEs fell into four heterogeneously distributed subfamilies (fig. S22 and table S22), making up 9% of the genome. We also found IEs in Sargasso Sea metagenome data (23) that have flanking coding domains with a high similarity to CCMP1545 but lower similarity to RCC299. Micromonas 18S rDNA sequences in the same metagenome data belong to uncultured clade M_IV (Fig. 1A) (13). Given the extent of genome reduction, the abundance of IE suggests that they are functional or resistant to purging.

Fig. 3.

Depiction of Micromonas orthologs with and without IEs. Single-exon (horizontal dark green bars represent exons) RCC299 [JGI protein identification (ID) 84234, chromosome 8] corresponds to a multi-exon gene in CCMP1545 (JGI protein ID 70142, scaffold 11). Different IE elements are shown (red and orange) within introns (horizontal light green lines). Diagonally oriented green lines show syntenic relationships by connecting exons with >70% nucleotide identity [minimum 100 base pairs (bp)]. Purple (RCC299, reversed orientation) and blue (CCMP1545) curves and peaks represent 16-nucleotide oligomer frequencies.

Putative RNA interference (RNAi) components also differed between the Micromonas species (SOM text S4 and table S6). Only RCC299 had an argonaute-encoding gene. A version of argonaute is also found in Chlamydomonas and plants but not Ostreococcus. DEAD box and SDE3 gene analyses provided circumstantial evidence for a diverged RCC299 RNA helicase. Argonaute can act to combat TE invasion (24), which is notable given that RCC299 had no recognizable TEs or IEs.

Both Micromonas spp. had putative thiamine pyrophosphate (TPP) riboswitches, untranslated mRNAs that regulate gene expression by means of metabolite binding (25, 26). These were not associated with homologous genes nor with known thiamine-biosynthesis–related genes, except for N-myristoyltransferase 1 (NMT1) (Table 2 and SOM text S15). CCMP1545 riboswitches were located at both gene ends (Fig. 4A), an arrangement never before seen, and formed two divergent groups: 5′ riboswitches shared with Ostreococcus and 3′ riboswitches shared with RCC299 (Fig. 4B). A conserved 3′ riboswitch was shared between folate receptor (FOLR)–like (RCC299) and SSSF-P (CCMP1545), even though these genes were not held in common; yet Ostreococcus also had SSSF-P and a 5′ riboswitch (Fig. 4A). Only one of the seven Micromonas riboswitches was associated with a multi-exon gene (FOLR-like). Thus, it appears that the putative riboswitches in Micromonas act akin to bacterial riboswitches and lack the spliceosomal functions that evolved in other eukaryotes (26).

Fig. 4.

TPP riboswitch arrangements. (A) High nucleotide identity of 3′ riboswitch sequences (yellow profiles) associated with FOLR-like (pink; RCC299 only) and SSSF-P (blue; CCMP1545) and identity between CCMP1545 and Ostreococcus 5′ riboswitches (white profiles) associated with SSSF-P homologs (blue). Plant riboswitches are often located in 3′UTRs (25), whereas bacterial and fungal riboswitches are often located in 5′UTRs. CCMP1545 has them in both positions. The downstream gene (purple) is a putative dihydrouridine synthase conserved in the four Mamiellales genomes. (B) Predicted secondary structure of FOLR-like–associated riboswitch showing the positions that are conserved among a range of organisms, particularly plants (yellow background), and a conserved position in all known plant riboswitches but not conserved in Micromonas (pink boxed U). Nucleotides adjacent to P2, P4, and P5 regions reflect differences in the CCMP1545 SSSF-P 3′ riboswitch (blue) and CCMP1545 SSSF-F 5′ riboswitch (brown). Differences in the more variable P1 and P3 are not marked in order to maintain the figure's simplicity.

Table 2.

Genes with associated TPP riboswitches in RCC299, CCMP1545, and Ostreococcus (both O. tauri and O. lucimarinus). The position of the riboswitch relative to the gene is indicated in the columns headed “Riboswitch.” DC, domain containing; NF, not found with protein-protein basic local alignment search tool (BLASTP) or protein-nucleotide six-frame translation (TBLASTN). See SOM text S15 for gene descriptions. Protein IDs refer to JGI genome browser protein IDs.

View this table:

Deficiencies in the thiamine-biosynthesis pathway (27, 28) were notable (SOM text S15). However, comparison with other lineages indicated the Micromonas riboswitch-containing genes represent ancient thiamine-pathway components. We identified TPP riboswitches associated with SSSF-P in SAR11 bacteria, which also lack classical thiamine-biosynthesis genes (10), and with SSSF-F in Chlamydomonas and Volvox. The functional importance of the gene-riboswitch associations is supported by the same gene-riboswitch pairings being found in these disparate lineages (SOM text S15).

The Micromonas genomes reveal features of the ancestral algae that initiated the billion-year trajectory of the green lineage and the greening of Earth. Their divergence, combined with acquisition strategies that are consistent with HGT, highlight the dynamic nature of marine protistan evolution and provide a springboard for unraveling functional aspects of phytoplankton populations. The challenge now is to identify biogeochemically important features within this natural diversity and apply them in assessing ecological transformations caused by environmental change.

Supporting Online Material

www.sciencemag.org/cgi/content/full/324/5924/268/DC1

Materials and Methods

SOM Text

Figs. S1 to S22

Tables S1 to S25

References

References and Notes

View Abstract

Navigate This Article