Globally Distributed Uncultivated Oceanic N2-Fixing Cyanobacteria Lack Oxygenic Photosystem II

See allHide authors and affiliations

Science  14 Nov 2008:
Vol. 322, Issue 5904, pp. 1110-1112
DOI: 10.1126/science.1165340


Biological nitrogen (N2) fixation is important in controlling biological productivity and carbon flux in the oceans. Unicellular N2-fixing cyanobacteria have only recently been discovered and are widely distributed in tropical and subtropical seas. Metagenomic analysis of flow cytometry–sorted cells shows that unicellular N2-fixing cyanobacteria in “group A” (UCYN-A) lack genes for the oxygen-evolving photosystem II and for carbon fixation, which has implications for oceanic carbon and nitrogen cycling and raises questions regarding the evolution of photosynthesis and N2 fixation on Earth.

Biological N2 fixation (BNF) is catalyzed by the enzyme nitrogenase, which is present in diverse Bacteria and Archaea (1). Marine BNF is particularly important in the oligotrophic open-ocean gyres, where nitrogen (N) inputs to stratified surface waters are stoichiometrically related to photosynthetic carbon (C) fixation and the vertical export of C to the deep ocean (2). Biogeochemically based estimates of oceanic N2 fixation rates are much higher than previously believed (25). Oceanic BNF was assumed to be primarily due to the filamentous cyanobacteria Trichodesmium (6, 7) and the symbiotic filamentous cyanobacteria Richelia (8, 9), until the recent discovery that oceanic unicellular cyanobacteria are important in oceanic N2 fixation (10, 11). The unicellular N2-fixing cyanobacteria were initially discovered only by amplification of nitrogenase genes and gene transcripts (mRNA) from oceanic water samples (12, 13), because they cannot be detected with traditional net collection methods and microscopy as can Trichodesmium and Richelia.

One phylogenetic group of unicellular N2-fixing cyanobacterial nitrogenase gene sequences (“group A” or UCYN-A nifH) (13) is most closely related to sequences from the marine unicellular cyanobacterium Cyanothece sp. strain ATCC 51142, a marine strain isolated from an intertidal habitat, and to those of the unicellular cyanobacterial symbiont of the diatom Rhopalodia gibba (14). UCYN-A nifH gene sequences have been reported from the Atlantic and Pacific Oceans (15), but UCYN-A cyanobacteria have not been successfully cultivated despite repeated attempts. These microorganisms express nitrogenase genes with maximum transcript abundances during the light period (11, 16). The daytime expression of nitrogenase presents an enigma, because the enzyme is inactivated by oxygen evolved during photosynthesis. Most cyanobacteria use temporal or spatial separation of photosynthesis and N2 fixation to prevent nitrogenase inactivation (17). However, we found that UCYN-A cyanobacteria have a genotype not previously known in free-living cyanobacteria and are genetically incapable of oxygenic photosynthesis, which also explains why they can fix N2 during daylight.

We initially assumed that the UCYN-A cyanobacteria would have cell diameters of 2 to 8 μm, similar to those of typical coastal (Cyanothece) and oceanic (Crocosphaera watsonii, UCYN-B) cyanobacteria (18), but discovered that the UCYN-A cells were less than 1 μmin diameter (19). Natural populations of UCYN-A cells could not be completely separated from other small phototrophic and heterotrophic populations by flow cytometry (FCM), but highly enriched cell sorts were obtained by screening cells sorted by fluorescence-activated cell sorting (FACS) for UCYN-A nifH genes with a quantitative real-time fluorescence polymerase chain reaction (QPCR) assay, Real-Time TaqMan (16, 20). We refined sort parameters (fig. S1) from those we previously used (19) to sort natural populations of UCYN-A cells from numerous North and South Pacific Ocean water samples (Fig. 1). We found that the populations are widely distributed, which indicated that our sorting strategy for UCYN-A cells was robust (table S2).

Fig. 1.

Cytogram of the forward scatter (x axis) and chlorophyll fluorescence (y axis) profile from concentrated water samples collected from a depth of 15 m at Station ALOHA (North Pacific) in late January 2008. The UCYN-A population sorted for MDA genome amplification and subsequent sequencing indicated by the red arrow.

Amplification of 16S ribosomal RNA (rRNA) genes from the sorted cells, using PCR with universal 16S rRNA primers, showed that the sorted population contained some noncyanobacterial Bacteria (such as Pelagibacter ubique) and non–N2-fixing cyanobacteria (Prochlorococcus and Synechococcus) in addition to the UCYN-A cyanobacteria (table S1). The percentage of UCYN-A cells in the defined sort region varied depending on the source of the (tables S1 and S2).

All 16S rRNA, nifH, and nifD sequences amplified from sorted cells were consistent with the UCYN-A cells being Cyanothece-like unicellular N2-fixing cyanobacteria (13) (Fig. 2 and figs. S2 and S3) even though they were much smaller. The 16S rRNA Cyanothece-like sequence was linked to the UCYN-A nifH gene by FACS-sorting single cells and multiplex gene amplification (fig. S2) of nifH and 16S rRNA genes from individual cells in which the positive single-cell QPCR reaction mixtures for nifH were used as a template for the second 16S rRNA amplification (21). Reactions that were negative for UCYN-A nifH were also negative for unicellular N2-fixing cyanobacterial 16S rRNA (fig. S2).

Fig. 2.

Phylogenetic tree of North and South Pacific Ocean UCYN-A full-length 16S rRNA nucleotide gene sequences showing the relationship of UCYN-A to Cyanothece sp. ATCC 51142 and other unicellular cyanobacteria. The arrow indicates the UCYN-A group.

A water sample collected from a depth of 15 m at the North Pacific Ocean long-term monitoring site Station ALOHA in late January 2008 (during Hawaii Ocean Time-series cruise HOT 199) indicated the presence of a high proportion (51%) of UCYN-A cells (Fig. 1). We used this population for metagenomic analysis, using Titanium sequencing technology from 454 Life Sciences (21). Multiple displacement amplification (MDA) (21) of DNA from approximately 5000 sorted cells was used to generate a genomic shotgun library that we estimate gave at least 10-fold genomic coverage of the 2- to 3-Mb UCYN-A genome in approximately 400,000 sequence reads (assuming that roughly one-third of the reads in the run were not UCYN-A). 177,834 sequence reads were most similar to the N2-fixing unicellular cyanobacteria Cyanothece sp. ATCC 51142, Cyanothece sp. CCY 0110, or C. watsonii WH 8501, on the basis of a BLAST analysis (21), whereas only 40,593 sequences had best BLAST hits to Prochlorococcus proteins, and 96,341 sequences were most similar to proteins from other microorganisms, including noncyanobacterial Bacteria. The 16S rRNA sequences also showed that the library was dominated by UCYN-A DNA sequences (Fig. 3), and they agreed with the QPCR data for UCYN-A cyanobacteria. The sequence read library represented good coverage of the UCYN-A genome because it contained the entire nitrogenase gene cluster on one assembled contig (Fig. 4) (21). The nitrogenase gene arrangement and composition were very similar to those of Cyanothece sp. ATCC 51142 and of the R. gibba symbiont (Fig. 4). Most of the proteins in the common cyanobacterial genome core (22) were identified by BLAST. Comparison of sequences to the cyanobacterial genome core (21) indicated that at least 79% of the core cyanobacterial genome had been sampled (fig. S4).

Fig. 3.

Phylogenetic affiliation of 16S rRNA sequences from metagenome sequences obtained from the station ALOHA 15-m sample. Sequences matching 16S rRNA were first retrieved by comparing the metagenome reads against those of the Ribosomal Database Project. Matching sequences were then assigned phylogenetically by BLASTn comparison to the nonredundant database at the National Center for Biotechnology Information (NCBI).

Fig. 4.

Comparison of the UCYN-A nif gene cluster to those of Cyanothece sp. ATCC 51142 and the R. gibba symbiont. The nif cluster was located on a large (135,849 bp) contig assembled from 5161 sequence reads. The figure shows the synteny of approximately 23 kbp of coding sequences for the nif gene clusters of UCYN-A (contig 02671, NCBI accession no. FJ170277), a N2-fixing endosymbiont of R. gibba (NCBI accession no. AY728387), and the circular chromosome of Cyanothece sp. ATCC 51142 (EMBL accession no. CP000806). Genes with full-length matches of amino acid sequence are shown in dark gray, genes with no matches are shown in light gray, and noncoding DNA is shown in white.

Although the core genes and all the nif genes were present, no candidate UCYN-A sequences were found that corresponded to C fixation, C concentration, or photosystem II (PSII) and associated pigments (such as phycoerythrin or phycobiliprotein linker). However, good coverage of candidate UCYN-A PSI genes was obtained (Fig. 5). In comparison, we had equal coverage of Prochlorococcus PSI and PSII genes, although they were a much smaller component of the sequence library (Fig. 5). However, we did detect a complete PSI psaA gene, found on one single assembled contig [5788 base pairs (bp) assembled from 948 sequence reads], followed immediately by 1857 bp of the psaB gene, which clustered with sequences from other unicellular N2-fixing cyanobacteria (fig. S3). No unicellular N2-fixing cyanobacteria phycoerythrin or phycocyanin genes (or associated linkers) were found in the metagenome, which is consistent with the observation that UCYN-A cyanobacteria were not detected by phycoerythrin fluorescence in FCM (19). Additional PCR experiments readily amplified the UCYN-A PSI genes found on the genomic contig, but no C fixation or PSII genes were obtained. We obtained further evidence that UCYN-A lack PSII genes by amplifying photosynthetically critical genes using degenerate PCR primers (21). From multiple samples of sorted cells, we successfully amplified UCYN-A psaA fragments with DNA sequences >98% identical to those genes found on the genomic contigs. However, attempts to amplify genes for the PSII D1 protein (psbA) as well as the RuBisCO large subunit yielded only sequences that were >98% identical to those of Prochlorococcus or >94% identical to those of Synechococcus species, and none that were similar to those of unicellular diazotrophs, as expected for UCYN-A sequences. The absence of any candidate UCYN-A PSII genes is clear, given the random and complete coverage of the sequencing effort, the high percentage of genome core coverage, the lack of amplification when degenerate primers were used, and the random location of photosynthesis genes in cyanobacterial genomes (23). Although it has been shown that some cyanobacteria can uncouple PSI from PSII using either organic (24) or inorganic (25) substrates as electron donors, this is the first observation of a cyanobacterium that completely lacks the PSII apparatus. This conclusion is striking because there are no reports to our knowledge of free-living cyanobacteria that are not oxygenic phototrophs.

Fig. 5.

Photosystem genes identified from sequencing of FACS-sorted environmental samples. The distribution of the number of sequence reads having top BLAST matches to individual photosystem genes in metagenomes of UCYN-A and Prochlorococcus is shown. Candidate UCYN-A photosystem sequences were identified by screening for reads having top BLAST matches to photosystem genes in Cyanothece, Crocosphaera watsonii WH8501, and Synechocystis, which are phylogenetically related to UCYN-A and hence collectively represent a proxy of the UCYN-A metagenome. The inset shows the distribution of the total number of PSI and PSII genes identified in metagenomes of UCYN-A, Prochlorococcus, and marine Synechococcus.

The UCYN-A phylogeny and poor pigmentation (dim chlorophyll fluorescence) resemble those seen in observations of the R. gibba symbiont (14). Although the R. gibba symbiont sequences are no closer to UCYN-A sequences phylogenetically than to those of other cyanobacteria in the unicellular N2-fixing cyanobacterial lineage (Fig. 2 and fig. S3), the symbiont of R. gibba also appears to have lost photosynthetic capabilities and pigmentation (14, 26). In the R. gibba symbiont, some of the PSII proteins (psbC and psbD) have become pseudogenes (14, 26). We did not detect UCYN-A cells in association with large cells, neither in size-fractionation filtration nor by FCM, and it appears to be a free-living organism.

Our work shows that oceanic UCYN-A cyanobacteria are missing the entire PSII apparatus, although the PSI apparatus appears to be intact (Fig. 5). It is possible that UCYN-A cyanobacteria are photoheterotrophic cells that generate adenosine triphosphate with PSI and are not able to fix C by the Calvin-Benson-Basham cycle as oxygenic photoautotrophs do. It is unclear how PSI functions in the absence of PSII. The UCYN-A strain, like the R. gibba symbiont, fixes N2 during the daytime, judging by its nitrogenase gene expression pattern (11). The lack of a functional PSII in the UCYN-A cells means that nitrogenase will not be poisoned by oxygen evolved from photosynthesis. The lack of oxygenic photosynthesis in UCYN-A cells has implications for C and N cycling in the oceans, as well as for the evolution of photosynthesis and N2 fixation. UCYN-A cyanobacteria overlap in size and fluorescence characteristics with non–N2-fixing microbial populations in the open ocean, including Prochlorococcus, Synechococcus, and other Bacteria, making them difficult to detect and quantify in the oceanic picoplankton. However, unicellular cyanobacteria may have a substantial impact on the N budget (10, 11), particularly in this case, because N2 fixation is not linked to C fixation. It is critical to determine the global importance of N2 fixation by this unusual photoheterotrophic cyanobacterial group in order to better constrain the global ocean N budget.

Supporting Online Material

Materials and Methods


Figs. S1 to S4

Tables S1 and S2

References and Notes

View Abstract

Navigate This Article