The cyanobacterium Prochlorococcus has divergent light-harvesting antennae and may have evolved in a low-oxygen ocean.

Marine picocyanobacteria of the genus Prochlorococcus are the most abundant photosynthetic organisms in the modern ocean, where they exert a profound influence on elemental cycling and energy flow. The use of transmembrane chlorophyll complexes instead of phycobilisomes as light-harvesting antennae is considered a defining attribute of Prochlorococcus Its ecology and evolution are understood in terms of light, temperature, and nutrients. Here, we report single-cell genomic information on previously uncharacterized phylogenetic lineages of this genus from nutrient-rich anoxic waters of the eastern tropical North and South Pacific Ocean. The most basal lineages exhibit optical and genotypic properties of phycobilisome-containing cyanobacteria, indicating that the characteristic light-harvesting antenna of the group is not an ancestral attribute. Additionally, we found that all the indigenous lineages analyzed encode genes for pigment biosynthesis under oxygen-limited conditions, a trait shared with other freshwater and coastal marine cyanobacteria. Our findings thus suggest that Prochlorococcus diverged from other cyanobacteria under low-oxygen conditions before transitioning from phycobilisomes to transmembrane chlorophyll complexes and may have contributed to the oxidation of the ancient ocean.

The free-living, planktonic, unicellular cyanobacterium of the genus Prochlorococcus is the smallest and most abundant photosynthetic organism on Earth (1). Together with their closest phylogenetic relatives of the genus Synechococcus, they are collectively known as the marine picocyanobacteria (≤2 μm in cell diameter) and are significant contributors to primary production in the modern ocean (2). Distinguishing characteristics of Prochlorococcus, vis-a-vis other marine cyanobacteria, include an extremely small size (<1 μm in diameter) and the possession of transmembrane divinyl chlorophyll a and b complexes as their main photosynthetic light-harvesting antennae instead of the membrane-bound phycobilisomes that are present in most other cyanobacteria (1, 3). These features are readily discernible with flow cytometry and are widely utilized to selectively quantify Prochlorococcus in marine samples (1). The coding potential and diversity of Prochlorococcus have been studied in numerous phylogenetically distinct isolates and hundreds of single-cell–amplified genomes (SAGs) (4). Based on their physiology, genetic characteristics, and environmental niches, the genus is broadly divided into high-light (HL) and low-light (LL) adapted ecotypes (5). The current understanding is that light, temperature, and nutrients—including iron—play integral roles in the ecology and evolution of Prochlorococcus (6–10).

Prochlorococcus thrives in and is best known from sunlit and well-oxygenated, nutrient-poor, tropical and subtropical waters. However, Prochlorococcus is also the dominant photoautotroph in the nutrient-rich, oxygen-depleted waters of major anoxic marine zones (AMZs) in the Arabian Sea and the eastern tropical North (ETNP) and South Pacific (ETSP) (11, 12), where photosynthetic eukaryotes are not found. In these systems, Prochlorococcus can drive a cryptic oxygen cycle (13), by which aerobic metabolic processes can exist under apparent anoxic conditions due to the oxygen produced through oxygenic photosynthesis. Phylogenetic analysis of environmental sequences of the 16S to 23S ribosomal RNA (rRNA) internal transcribed spacer (ITS) region has revealed that the majority of the AMZ Prochlorococcus belong to two distinct phylogenetic lineages, originally termed LL V and LL VI (12) and renamed here as AMZ I and AMZ II. These lineages correspond to uncultured basal groups within the Prochlorococcus genus, for which there is no public genetic information except for their inferred capacity for nitrate utilization based on limited shotgun metagenomic sequencing data (14). Thus, further details on their coding potential and evidence for anaerobic or microaerobic metabolisms remain unconstrained for these AMZ lineages. Such information is important for understanding the origin and structure of the photosynthetic apparatus (15, 16), the evolution of marine pelagic picocyanobacteria (17), the functioning of ancient and modern AMZs (18), and the differential modes of metabolic coupling between Prochlorococcus and other AMZ microorganisms.

Results and Discussion

To explore the genetic makeup of these uncultured picocyanobacteria from AMZ waters, we analyzed 128 SAGs from water samples of the secondary fluorescence maximum in the eastern Pacific Ocean off northern Chile and Mexico (Fig. 1, Table 1, and ). The geochemical characteristics of these AMZ waters (18) include undetectable levels of oxygen, no accumulation of sulfide or ammonium, the presence of significant levels of major nutrients, and, particularly of nitrite, a product of the anaerobic process of dissimilative nitrate reduction. Cells with optical characteristics typical of either Prochlorococcus (red fluorescence only) or Synechococcus (red and orange fluorescence) (Fig. 1) were sorted for SAG generation and sequencing using previously described techniques (19). The AMZ SAG assemblies had an estimated completeness of up to 75%, an estimated genome size of ∼2.5 Mbp, and a guanine–cytocine content of ∼50% (). These characteristics are more similar to those previously observed for the Prochlorococcus LL IV and marine Synechococcus lineages than for HL Prochlorococcus (6, 20).

Fig. 1.

Representative biogeochemical and flow-cytometry profiles and phylogenetic tree of unicellular cyanobacteria with Prochlorococcus from AMZs. (A) Dissolved oxygen concentration (blue), in vivo chlorophyll a fluorescence in relative units (dark green), and Synechococcus-like (orange) and Prochlorococcus-like (light green) cell abundance at Station 6 in the AMZ of the ETNP (Table 1). The gray area corresponds to depths with <0.5 μmol ⋅ kg−1 O2. (B) Fluorescence versus light scatter cytograms (in relative units) of cyanobacteria-like particles in the secondary deep-chlorophyll maximum. The upper panel shows particles with red fluorescence, a proxy for chlorophyll a-containing cells, and the lower panel shows particles with orange fluorescence, a proxy for phycoerythrin-containing cells. (C) Maximum-likelihood tree determined by phylogenetic inference using 49 concatenated ribosomal proteins: Prochlorococcus (collapsed genomes of recognized ecotypes in green), Synechococcus (collapsed genomes of marine subclusters in orange), and AMZ Prochlorococcus (in bold). PCC 7002 is the marine onshore Synechococcus sp. strain PCC 7002, and PCC 6803 is the freshwater Synechosystis sp. strain PCC 6803. The circles on each node represent support values higher than 50% (n = 1,000 iterations). Characteristic traits of the different cyanobacteria lineages examined are given in colored lines. The dashed line indicates that only a few members of the Synechococcus subcluster have that trait.

Table 1.

Sampling location and hydrographic characteristics of waters from which samples for single-cell genomics were collected

Cruise	Station	Latitude (°N)	Longitude (°W)	Date (dd/mm/yy)	Depth (m)	Temperature (°C)	Salinity	CTD oxygen (µmol/kg)	Phosphate (µmol/kg)	Nitrate + nitrite (µmol/kg)	Prochlorococcus-like (104 cells/mL)	Synechococcus-like (104 cells/mL)
MV1015	1	−20,083	70,800	19/11/10	53	12.874	34.731	3.3	2.72*	17.85*	1.04	0.37
NH1315	6	18,920	104,891	19/07/13	100	13.995	34.781	1.5	2.52	20.44	60.8	4.3

Values from 50 m in separate cast.

Prochlorococcus SAGs affiliated with the uncharacterized AMZ I lineage were recovered with the traditional red-fluorescence-only cell sorting approach. However, SAGs affiliated with the previously uncharacterized AMZ II lineage were obtained only from cells that exhibited red and orange fluorescence, which is traditionally associated with Synechococcus. Moreover, we discovered a previously unknown Prochlorococcus lineage, termed here as AMZ III, among the cells with both red and orange fluorescence. According to phylogenetic analyses of individual and concatenated universal single-copy marker genes, AMZ I is a sister clade to LL IV and presents two subclades, termed here as AMZ IA and AMZ IB, with AMZ II occupying a basal position to the LL IV and AMZ I lineages (Fig. 1 and ). This tree topology is consistent with the established ITS phylogeny (8, 12). Furthermore, the newly discovered AMZ III lineage is located at the base of the Prochlorococcus radiation, thus representing the extant Prochlorococcus relatives closest to marine Synechococcus identified to date. The phylogenetic grouping of the AMZ Prochlorococcus SAGs was further corroborated by their genomic similarity ().

An analysis of the individual SAGs (Fig. 2 and ) revealed that all three AMZ lineages encode genes for chlorophyll b synthase pcCao (21) and lycopene ε-cyclase crtL (22) but lack the gene dvr encoding for the divinyl chlorophyll a reductase (23). This coding potential contributes to the characteristic pigment traits of all Prochlorococcus, such as divinyl chlorophyll a (instead of the monovinyl form present in most of the other cyanobacteria, including marine Synechococcus), divinyl chlorophyll b, and α-carotene (3, 24). This genomic information is consistent with the abundance of these diagnostic pigments in AMZ waters of the Arabian Sea and eastern North Pacific (11) as well as of the eastern South Pacific (). However, the AMZ II and AMZ III SAGs were also found to contain genes encoding for complete synthesis of phycobilisomes, coinciding with the orange fluorescence of these two lineages (Fig. 1). Specifically, AMZ II and AMZ III SAGs encoded genes for the synthesis of the phycobiliproteins allophycocyanin, phycocyanin, and phycoerythrin (I and II), as well as the corresponding linkers, and lack the chlorophyll light-harvesting pcb genes typical of Prochlorococcus (Fig. 2 and ). Furthermore, AMZ II and AMZ III SAGs encoded a chlorophyll-binding protein of the iron-stress–induced (IsiA) type, which is typical of marine Synechococcus and is phylogenetically distant from the Prochlorococcus-typical pcb genes (25), including those in AMZ I SAGs (). Likewise, we found the kaiA gene in AMZ II and AMZ III (Fig. 2 and ), a critical component of the circadian clock KaiABC system (26) that has not been found in Prochlorococcus before (27, 28). Thus, our results indicate that the replacement of phycobilisomes for divinyl chlorophyll a and b complexes and the loss of the kaiA gene occurred within the Prochlorococcus diversification and support previous observations regarding genetic remnants of such evolutionary events still being present in some Prochlorococcus lineages (28, 29). As shown above, these evolutionary steps appear to have occurred after Prochlorococcus acquired the capacity to synthesize divinyl chlorophyll b (Fig. 1).

Fig. 2.

Presence–absence of selected genes in AMZ Prochlorococcus SAGs (see for details).

One of the major challenges modern cyanobacteria experience when living in anoxia and temporary darkness is the synthesis of tetrapyrrole pigments, as oxygen is required in several of their shared enzymatic steps. These biomolecules have important functions, including electron transfer, oxygen binding, and light absorption, with chlorophyll a being essential for photosynthesis (30). Certain cyanobacteria, however, have alternative or complementary enzymes for pigment biosynthesis that work under oxygen-limited conditions (31). In particular, three genes, hemN, acsF, and ho, encode secondary versions of enzymes involved in heme, chlorophyll, and phycocyanobilin biosynthesis, respectively. The gene hemN encodes the oxygen-independent version of the enzyme that catalyzes the biosynthesis of protoporphyrinogen IX, the common precursor of Chl-a, hemes, and bilins; acsF encodes the cyclase that converts Mg-Proto monomethyl ester to divinyl protochlorophyllide during the synthesis of Chl-a; and ho encodes a heme oxygenase involved in bilin synthesis. These genes are clustered together in the genomes of the freshwater cyanobacterium Synechocystis PCC 6803 and the euryhaline coastal marine cyanobacterium Synechococcus PCC 7002. In both isolates, such genes are highly expressed under oxygen-limited conditions but not under fully oxic conditions (32, 33). We identified the gene cluster acsF-ho-hemN in SAGs of the three AMZ lineages and—with a similar genomic configuration—also in marine Synechococcus sp. strains RS9917, WH 8101, and CB0101 (Figs. 2 and 3). This gene cluster has not been previously reported in any of the prior studies of Prochlorococcus and marine Synechococcus (31). Phylogenetic analyses revealed that the gene encoding the AcsF cyclase branches closest to homologs in the LL Prochlorococcus and Synechococcus strains RS9917, WH 8101, and CB0101 but not with homologs in other marine Synechococcus nor with those in HL Prochlorococcus ecotypes (Fig. 3). The cluster-forming ho and hemN genes branch in a similar way as acsF, except that no homologs from LL Prochlorococcus are grouped closely with them (Fig. 3 ).

Fig. 3.

Synteny and phylogeny of three genes involved in tetrapyrrole biosynthesis under oxygen-limited conditions. (A) Synteny of three orthologous genes involved in chlorophyll (acsF), phycocyanobilin (ho), and heme (hemN) biosynthesis across selected cyanobacterial lineages. The gene arsR encodes a putative transcriptional regulator, and yjbL encodes an unknown protein. Maximum-likelihood phylogenetic trees of (B) acsF-, (C) ho-, and (D) hemN-predicted amino acid sequences. The colored boxes mark the gene versions that appear organized in a cluster and should work under oxygen-limited conditions; lineages appearing in A are in bold. Conserved residues for several functions are given in . Filled circles on each node correspond to support values higher than 50% (n = 1,000 iterations).

The phylogeny and conserved gene arrangement shared between freshwater and marine cyanobacteria suggests that Prochlorococcus may have evolved before the ocean became fully oxygenated. In this scenario, ho and hemN would have been lost as Prochlorococcus adapted to an oxic ocean, but acsF would have been retained in the LL lineages (Fig. 3 ). The alternative scenario that the low-oxygen pigment biosynthesis trait was acquired after Prochlorococcus evolved in an oxic ocean cannot be totally discarded. However, a detailed analysis of contigs containing the genes for pigment biosynthesis (as well as those for the light-harvesting antenna) using previously described approaches (14, 34, 35) did not provide any evidence of recent horizontal gene transfer or of hallmark viral signatures. If Prochlorococcus evolved in an oxygen-depleted ocean before it could adapt to fully oxygenated oligotrophic surface waters, the replacement of phycobilisomes for transmembrane divinyl chlorophyll complexes may have been driven by the low levels and narrow spectral characteristics of the light available at depth (36). Moreover, continuous changes in the light field at depth due to variable optical conditions of surface waters or due to internal waves could have allowed for the coexistence of lineages with different light-harvesting antennae, as is currently observed. However, in the proposed scenario, only those Prochlorococcus with chlorophyll complexes would have been able to colonize the high-oxygen surface waters.

Metagenomic fragment recruitment () showed that AMZ I, II, and III lineages are found only in oxygen-depleted oceanic waters. Moreover, AMZ I outnumbers II and III, which is consistent with the flow cytometry data (11, 12) indicating that red-only fluorescent cells dominate in AMZ waters. The presence of indigenous lineages with a specific gene cluster for pigment synthesis under oxygen-limited conditions (Fig. 3) is thus consistent with the designation of Prochlorococcus “AMZ” ecotypes. Moreover, these ecotypes, which inhabit nutrient-rich waters, have a complete gene repertoire for the use of ammonium, urea, nitrite, and nitrate (Fig. 2 and ) as nitrogen sources. On the other hand, according to the analysis of public metatranscriptomic data (), at least some of the photosynthesis genes discussed above are being actively transcribed in AMZs. However, targeted gene expression studies over the light–dark periods will be required to address this issue fully.

Other genotypic properties that appear to be specific to AMZ ecotypes and that may confer them adaptations to an anoxic environment include the presence of a hybrid-cluster protein (hcp) gene and cognate fnr regulator of the anaerobic Cpr/Fnr family as well as a gene ytfE encoding an iron-sulfur cluster repair protein YtfE (Fig. 2 and ). These three genes are in the same genomic neighborhood in AMZ I and AMZ II, but ytfE is in a different location in AMZ III. The HCPs and YtfE are associated with anaerobic nitrosative stress (e.g., ref. 37), while HCPs are also essential in protein S-nitrosylation by nitric oxide (38), a fundamental redox-based cellular signaling mechanism. Additionally, AMZ Prochlorococcus SAGs encode enzymes involved in glycerol oxidation (glpK and glpD) and alcohol fermentation (adh) (Fig. 2). These additional features indicate that AMZ Prochlorococcus may be capable of supplementing their carbon and energy requirements through the use of organic compounds, as previously suggested for other lineages (39, 40), to survive under dark anoxic conditions.

Marine planktonic picocyanobacteria likely emerged in the Neoproterozoic 1,000 to 542 Mya (17), roughly at the time of the oxygenation of the deep ocean (41). We have shown that extant lineages inhabiting AMZs are basal to the Prochlorococcus genus, can have phycobilisomes as light-harvesting antenna, and carry ancestral adaptations to low-oxygen conditions. These findings suggest that Prochlorococcus emerged in low-oxygen environments and may have contributed to sustaining aerobic metabolisms and providing a significant carbon source through oxygenic photosynthesis, in the ancient, anoxic upper ocean, as they do today (13). The evolutionary history of Prochlorococcus thus appears intimately related to the availability of oxygen in the ocean.

Materials and Methods

Sampling and Hydrographic Data Collection.

Sampling was carried out during Cruise MV1015 (DOI: 10.7284/903808), the “Biogeochemical Gradients: Role in Arranging Planktonic Assemblages (BiG RAPA) Expedition” (https://hahana.soest.hawaii.edu/cmoreserver/cruises/big_rapa/plan.htm), at Station 1 in the ETSP and during Cruise NH1315 (DOI: 10.7284/903869) at Station 6 in the ETNP (Table 1). For Cruise MV1015, hydrographic data and seawater samples were collected with a rosette/conductivity-temperature-depth (CTD) system, consisting of Niskin bottles and a CTD profiler (Sea-Bird SBE 911plus) with dissolved-oxygen, fluorometer, and transmissometer sensors. For Cruise NH1315, hydrographic data and seawater samples were collected with a pump profiling system, similar to the one previously used in AMZ waters (13).

Picocyanobacteria Cell Count by Flow Cytometry.

For each site, duplicate samples of 1 mL raw seawater were fixed in a final concentration of 1% (vol/vol) glutaraldehyde at room temperature, in the dark, for 10 min and then fast frozen in liquid nitrogen and stored at −80 °C until analysis. In the laboratory, samples were thawed on ice and passed through a 60 μm mesh net, and cell abundance was determined with a FACSCalibur flow cytometer (Becton Dickinson) by using red and orange fluorescence as proxies for cells containing chlorophyll a and phycobiliproteins, respectively (42).

AMZ Prochlorococcus Single-Cell Sample Collection.

For the MV1015 (ETSP) cruise, water samples for single-cell sorting were collected from the Niskin bottles and prefiltered through a 60 μm mesh-size net. Then, ∼4,000 cells of Prochlorococcus-like cells were sorted on board into 1 mL of sterile glycerol–Tris-EDTA (TE) buffer (43) with an InFlux flow cytometer (BD Biosciences). Sorting was performed gating on putative Prochlorococcus cells in the red emission channel (excited by a 488 nm laser), using forward scattered light as a proxy for particle size (42). These presorted Prochlorococcus were cryopreserved in liquid nitrogen and then stored at −80 °C. For the NH1410 (ETNP) cruise, water samples were collected from the Pump Profiling System, filtered through a 60 μm mesh size, and stored in glycerol–TE buffer at −80 °C until cell sorting.

SAG Sequencing, Assembly, Gene Prediction, and Annotation.

Single-cell sorting, whole-genome amplification, 16S rRNA real-time PCR screening, and PCR product sequencing were performed at the Bigelow Laboratory for Ocean Sciences’ Single Cell Genomics Center (SCGC, http://scgc.bigelow.org/), as described previously (19). For the ETSP SAGs, an Illumina standard shotgun library was constructed, and samples were sequenced utilizing the HiSeq 2000 platform at Canada’s Michael Smith Genome Sciences Centre. Artifactual sequences were filtered off the raw data and draft SAGs were assembled using SPAdes 3.5. SAG contigs of >1 kbp were decontaminated using ProDeGe (44), and sequences with lengths between >200 bp and <2 kbp were considered whenever they had >80% identity (>75% alignment coverage) with Prochlorococcus reference genomes (). Quality filtered sequences of <200 bp were discarded. The decontaminated SAG sequences were uploaded for gene prediction and annotation to the Integrated Microbial Genomes & Microbiomes (IMG/MER) system (45). Genome completeness was estimated by the presence of 114 single-copy conserved bacterial phylogenetic markers among the predicted protein-coding genes of each SAG (46). The ETNP SAGs were sequenced at SCGC as described previously (19). Gene predictions, functional annotation, manual curation, and pathway reconstruction for all individual SAGs were performed within the IMG/MER platform (https://img.jgi.doe.gov) (45). The presence of genes of interest was determined by their functional annotation, according to their assigned Kyoto Encyclopedia of Genes and Genomes ID. Genomic synteny (Fig. 3 and ) were plotted with Trebol (https://inf.imo-chile.cl). Sequence identity between genomic sequences was computed using BLASTn with a bit-score cutoff of 50. The contigs were checked for evidence of recent horizontal gene transfer as in ref. 14 and for viral signatures using VirSorter (34) and VirFinder (35).

Retrieval of AMZ SAGs Phylogenetic Marker Sequences.

Prochlorococcus ITS sequences were retrieved from reference genomes () and utilized as BLASTn queries to retrieve those from the Prochlorococcus AMZ SAGs. Then, transfer RNA genes were removed and the identity was determined for all ITS sequences that ended with >600 bp. The same approach was adopted with the rbcL gene, encoding the RuBisCO large subunit and the petB gene, a component of the cytochrome b6f complex. Average nucleotide identity (ANI) analysis () was performed using FastANI (47).

Identification and Conserved Residues of Cluster-Forming Tetrapyrrole Biosynthesis Genes.

The gene cluster acsF-ho-hemN, associated with pigment biosynthesis under oxygen-limited conditions (31), was identified in individual Prochlorococcus AMZ SAGs within IMG/MER. Conserved residues in AcsF2, HO2, and HemN2 () were visually detected and manually annotated according to information provided in the literature (48–50). Though alignments were also analyzed with PROSITE (https://prosite.expasy.org), no additional relevant domains were found.

Phylogenetic and Phylogenomic Analyses.

Predicted amino acid sequences were aligned using MAFFT (51) version 7.271. Maximum-likelihood phylogenetic and phylogenomic inferences were performed using IQ-TREE (52) version 1.6.12. The models of sequence evolution were selected with ModelFinder (53) according to the Bayesian information criterion. The best fit models for single-gene phylogenies are given in the respective figure legends. Branch support values were computed using the Shimodaira–Hasegawa approximate likelihood ratio test within IQ-TREE. For phylogenomic reconstructions, concatenated amino acid–aligned sequences of 49 ribosomal proteins (Fig. 1) and 137 single-copy genes () were used (54). The best fit model for the tree in Fig. 1 was LG+R10, and LG+I+G4 was the best fit model for the tree in .

Fragment Recruitment.

The recruitment of sequence fragments from global ocean metagenomes () was performed against the contigs associated with the AMZ Prochlorococcus ecotypes obtained in the ETNP and ETSP. Fragment recruitment onto Prochlorococcus LL genomes was included for comparison (). Details about the procedure can be found in .

Transcript Recruitment.

The recruitment of transcripts encoding pigment-biosynthesis and photosynthetic proteins from public AMZ metatranscriptomes () was carried out against the AMZ SAGs and Prochlorococcus and marine Synechococcus genomes (). The details about the procedure can be found in .