Genome Streamlining in a Cosmopolitan Oceanic Bacterium

See allHide authors and affiliations

Science  19 Aug 2005:
Vol. 309, Issue 5738, pp. 1242-1245
DOI: 10.1126/science.1114057


The SAR11 clade consists of very small, heterotrophic marine α-proteobacteria that are found throughout the oceans, where they account for about 25% of all microbial cells. Pelagibacter ubique, the first cultured member of this clade, has the smallest genome and encodes the smallest number of predicted open reading frames known for a free-living microorganism. In contrast to parasitic bacteria and archaea with small genomes, P. ubique has complete biosynthetic pathways for all 20 amino acids and all but a few cofactors. P. ubique has no pseudogenes, introns, transposons, extrachromosomal elements, or inteins; few paralogs; and the shortest intergenic spacers yet observed for any cell.

Pelagibacter ubique, strain HTCC1062, belongs to one of the most successful clades of organisms on the planet (1), but it has the smallest genome (1,308,759 base pairs) of any cell known to replicate independently in nature (Fig. 1). In situ hybridization studies show that these organisms occur as unattached cells suspended in the water column (1). They grow by assimilating organic compounds from the ocean's dissolved organic carbon (DOC) reservoir, and can generate metabolic energy either by a light-driven proteorhodopsin proton pump (2) or by respiration (3). The marine planktonic environment is poor in nutrients, and the availability of N, P, and organic carbon typically limits the productivity of microbial communities. P. ubique is arguably the smallest free-living cell that has been studied in a laboratory, and even its small genome occupies a substantial fraction (∼30%) of the cell volume. The small size of the SAR11 clade cells fits a model proposed by Button (4) for natural selection acting to optimize surface-to-volume ratios in oligotrophic cells, such that the capacity of the cytoplasm to process substrates will be matched to steady-state membrane transport rates.

Fig. 1.

Number of predicted protein-encoding genes versus genome size for 244 complete published genomes from bacteria and archaea. P. ubique has the smallest number of genes (1354 open reading frames) for any free-living organism.

Surprisingly, this genome appears to encode nearly all of the basic functions of α-proteobacterial cells (Table 1). The small genome size is attributable to the nearly complete absence of nonfunctional or redundant DNA and the paring down of all but the most fundamental metabolic and regulatory functions. For example, P. ubique falls at the extreme end of the range for intergenic DNA regions, with a median spacer size of only three bases (Fig. 2). Intergenic DNA regions vary considerably among bacteria and archaea, even including parasites that have small genomes (5). No pseudogenes, phage genes, or recent gene duplications were found in P. ubique.

Fig. 2.

Median size of intergenic spacers for bacterial and archaeal genomes. Inset shows expanded view of range for organisms with the smallest intergenic spacers.

Table 1.

Metabolic pathways in Pelagibacter.

Pathway Prediction
Glycolysis Uncertain
TCA cycle Present
Glyoxylate shunt Present
Respiration Present
Pentose phosphate cycle Present
Fatty acid biosynthesis Present
Cell wall biosynthesis Present
Biosynthesis of all 20 amino acids Present
Heme biosynthesis Present
Ubiquinone Present
Nicotinate and nicotinamide Present
Folate Present
Riboflavin Present
Pantothenate Absent
B6 Absent
Thiamine Absent
Biotin Absent
B12 Absent
Retinal Present

To further explore this trend, we investigated paralogous gene families by means of BLAST clustering with variable threshold limits. The genome had the smallest number of paralogous genes observed in any free-living cell (Fig. 1) (fig. S1). A steep slope in the decline of potential paralogs with increasing gene pairwise similarity threshold, relative to other organisms, suggested that the few paralogs present in P. ubique are descended from relatively old duplication events, and that steady evolutionary pressure has constrained the expansion of gene families in this organism (fig. S2). Furthermore, there was no evidence of DNA originating from recent horizontal gene transfer events. The presence of DNA uptake and competence genes (PilC, PilD, PilE, PilF, PilG, PilQ, comL, and cinA) in the genome suggests that P. ubique has the ability to acquire foreign DNA. These data are consistent with the hypothesis that cells in some ecosystems are subject to powerful selection to minimize the material costs of cellular replication; this concept is known as streamlining (5).

Several hypotheses have been used to explain genome reduction in prokaryotes, particularly in parasites, which have the smallest cellular genomes known. The relaxation of positive selection for genes used in the biosynthesis of compounds that can be imported from the host, together with a bias favoring deletions over insertions in most or all bacteria, appear to account for genome reduction in many parasites and organelles (5). The streamlining hypothesis assumes that selection acts to reduce genome size because of the metabolic burden of replicating DNA with no adaptive value. Under this hypothesis, it is presumed that repetitive DNA arises when mechanisms that add DNA to genomes—for example, recombination and the propagation of self-replicating DNA (e.g., introns, inteins, and transposons)—overwhelm the simple economics of metabolic costs. However, evolutionary theory predicts that the probability that selection will act to eliminate DNA merely because of the metabolic cost of its synthesis will be greatest in very large populations of cells that do not experience drastic periodic declines (6).

The streamlining hypothesis has been used to explain genome reduction in Prochlorococcus, a photoautotroph that reaches population sizes in the oceans that are similar to those of Pelagibacter (79). Prochlorococcus genomes range from 1.66 to 2.41 million base pairs (Mbp). Many organisms with reduced genomes, including some pathogens, also have very low G:C to A:T ratios (10) (fig. S3), which can be attributed to biases in mutational frequencies, but alternatively might convey a selective advantage by lowering the nitrogen requirement for DNA synthesis, thereby reducing the cellular requirement for fixed forms of nitrogen (7). N and P are both proportionately important constituents of DNA that are frequently limiting in seawater. The P. ubique genome is 29.7% G+C. Of four complete Prochlorococcus genome sequences, the two that lack the DNA repair enzyme 6-0-methylguanine-DNA methyltransferase also have very low G:C to A:T ratios. In the absence of this enzyme, the extent of accepted G:C to A:T mutations increases; however, the P. ubique genome encodes this enzyme, which suggests that other factors are the cause of its low G:C to A:T ratio.

Annotation revealed a spare metabolic network encoding a variant of the Entner-Duodoroff pathway, a tricarboxylic acid (TCA) cycle, a glyoxylate bypass, and a typical electron transport chain (Table 1). Anapleurotic pathways for cellular constituents, other than five vitamins, appeared to be complete, but genes that would confer alternate metabolic lifestyles, motility, or other complexities of structure and function were nearly absent. Conspicuous exceptions were genes for carotenoid synthesis, retinal synthesis, and proteorhodopsin. P. ubique constitutively expresses a light-dependent retinylidine proton pump and is the first cultured bacterium to exhibit the gene that encodes it (2). The genome also contained genes for type II secretion (including adhesion) and type IV pilin biogenesis. Examination of gene distributions among metabolic categories (fig. S4) supported the conclusion that genome reduction in P. ubique has spared genes for core proteobacterial functions while reducing the proportion of the genome devoted to noncoding DNA. Relative to other α-proteobacterial genomes, the proportions of P. ubique genes encoding transport functions, biosynthesis of amino acids, and energy metabolism were high (table S3).

The sheer size of Pelagibacter populations indicates that they consume a large proportion of the labile DOC in the oceans. The global DOC pool is estimated to be 6.85 × 1017 g C (11), roughly equaling the mass of inorganic C in the atmosphere (12). Examination of the P. ubique genome revealed that about half of all transporters, and nearly all nutrient-uptake transporters, are members of the ATP-binding cassette (ABC) family (table S1). ABC transporters typically have high substrate affinities and therefore provide an advantage at the cost of ATP hydrolysis. Inferred transport functions included the uptake of a variety of nitrogenous compounds: ammonia, urea, basic amino acids, spermidine, and putrescine. Broad-specificity transporters for sugars, branched amino acids, dicarboxylic and tricarboxylic acids, and a number of common osmolytes (including glycine betaine, proline, mannitol, and 3-dimethylsulfoniopropionate) were found in the genome. Autoradiography with native populations of SAR11 has demonstrated high uptake activity for amino acids and 3-dimethylsulfoniopropionate (13). Hence, efficiency is achieved in a low-nutrient system by reliance on transporters with broad substrate ranges (14) and a number of specialized substrate targets, in particular, nitrogenous compounds and osmolytes.

The genome encoded two sigma factors, the heat shock factor σ32 and a σ70 (rpoD), but no homolog of rpoN, the gene for the nitrogen starvation factor σ54 (table S2). Only four two-component regulatory systems were identified, three of which match the only two-component regulatory systems in Rickettsia (15). The presence of homologs to PhoR/PhoB/PhoC, NtrY/NtrX, and envZ/OmpR suggested regulated responses to phosphate limitation, N limitation, and osmotic stress. The only additional two-component system, RegB/RegA, has been implicated in the regulation of cellular oxidation/reduction processes in phototrophic α-proteobacteria (16). A gene encoding a ferric iron uptake regulator was also present.

In its simplicity the P. ubique genome is unique among other heterotrophic marine bacteria, such as Vibrio sp. (17), Pseudoalteromonas (18), Shewanella (19), and Silicibacter (20), which have considerably larger genomes (4.0 to 5.3 Mbp) and global regulatory systems that enable them to implement a variety of metabolic strategies in response to environmental variation. We hypothesize that P. ubique makes use of the ambient DOC field (21), whereas heterotrophic bacterioplankton with larger genomes are poised to rapidly exploit pulses of nutrients (22) at the expense of replication efficiency during the intervening periods (23). This hypothesis is consistent with the observation that P. ubique has a single ribosomal RNA (rRNA) operon and a low growth rate (0.40 to 0.58 cell divisions per day) that does not vary in response to nutrient addition. In contrast, heterotrophic marine bacteria with large genomes have some of the highest recorded growth rates and are very responsive to nutrient concentration.

Like some other α-proteobacteria and especially archaea, HTCC1062 has an alternate thymidylate synthase for thymine synthesis, thyX (24). As in other strains that lack the most common thymidylate synthase (thyA) but have thyX, HTCC1062 also lacks the dihydrofolate reductase folA (25). Evidence suggests that the gene encoding thyX can substitute for folA (24). A full glycolytic pathway was not reconstructed because of the confounding diversity of glycolytic pathways (26). Five enzymes in the canonical glycolytic pathway were not seen, including two key enzymes involved in allosteric control: phosphofructokinase and pyruvate kinase. An enzyme thought to substitute for pyruvate kinase (27), known as PPDK (pyruvate-phosphate dikinase), was found. Some but not all of the enzymes for the nonphosphorylated Entner-Duodoroff pathway, considered more ancient than canonical glycolysis (26, 28), were detected, as well as a complete pathway for gluconeogenesis, also considered more ancient than canonical glycolysis (29). Sugar transporters with best BLAST hits to maltose/trehalose transport were found, so presumably a complete glycolytic pathway does function in this cell.

Whole-genome shotgun (WGS) sequence data from the Sargasso Sea segregated at high similarity values, relative to other α-proteobacteria and proteobacteria, in a BLASTN analysis of the P. ubique genome (fig. S4). Sequence diversity prevented Venter et al. (19) from reconstructing SAR11 genomes from the Sargasso Sea WGS data set, although SAR11 rRNA genes accounted for 380 of 1412 16S rRNA genes and gene fragments they recovered (26.9%), and the library was estimated to encode the equivalent of about 775 SAR11 genomes. Three Sargasso Sea contiguous sequences (contigs) that were long (5.6 to 22.5 kb) and highly similar to the P. ubique genome were analyzed in detail. Genes on these contigs were syntenous with genes from the P. ubique genome, with amino acid sequence identities ranging from 68 to 96% (fig. S5). Phylogenetic analysis of four conserved genes from these contigs (those encoding RNA polymerase subunit B, Fig. 3; elongation factor G, fig. S6; DNA gyrase subunit B, fig. S7; and ribosomal protein S12, fig. S8) showed them to be associated with large, diverse environmental clades that branched within the α-proteobacteria. We hypothesize that evolutionary divergence within the SAR11 clade and the accumulation of neutral variation are the most likely explanations for the natural heterogeneity in SAR11 genome sequences.

Fig. 3.

Maximum likelihood phylogenetic tree for the gene encoding RNA polymerase subunit B. Sequences represented by accession numbers are environmental sequences from the Sargasso Sea (19). The sequence indicated by a star is part of the 5.7-kb contig IBEA_CTG_2159647 that is part of a conserved gene cluster also present in Pelagibacter ubique. Numbers indicated by solid arrowheads represent amino acid percentage identity to the Pelagibacter gene. For comparison, the identity between two species of Mesorhizobium is also indicated (open arrowhead). Bootstrap support (100 maximum-likelihood replicates) is indicated for the major clades (* if less than 50).

Metabolic reconstruction failed to resolve why P. ubique will not grow on artificial media. When cultured in seawater, it attains cell densities similar to populations in nature, typically 105 to 106 ml–1 depending on the water sample (3). No evidence of quorum-sensing systems was found in the genome, and experimental additions of nutrients supported the results from metabolic reconstruction, which suggests that an unusual growth factor may play a role in the ecology of this organism.

P. ubique has taken a tack in evolution that is distinctly different from that of all other heterotrophic marine bacteria for which genome sequences are available. Evolution has divested it of all but the most fundamental cellular systems such that it replicates under limiting nutrient resources as efficiently as possible, with the outcome that it has become the dominant clade in the ocean.

Supporting Online Material

Materials and Methods

Tables S1 to S3

Figs. S1 to S9


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article