Report

Potential for Chemolithoautotrophy Among Ubiquitous Bacteria Lineages in the Dark Ocean

See allHide authors and affiliations

Science  02 Sep 2011:
Vol. 333, Issue 6047, pp. 1296-1300
DOI: 10.1126/science.1203690

Abstract

Recent studies suggest that unidentified prokaryotes fix inorganic carbon at globally significant rates in the immense dark ocean. Using single-cell sorting and whole-genome amplification of prokaryotes from two subtropical gyres, we obtained genomic DNA from 738 cells representing most cosmopolitan lineages. Multiple cells of Deltaproteobacteria cluster SAR324, Gammaproteobacteria clusters ARCTIC96BD-19 and Agg47, and some Oceanospirillales from the lower mesopelagic contained ribulose-1,5-bisphosphate carboxylase-oxygenase and sulfur oxidation genes. These results corroborated community DNA and RNA profiling from diverse geographic regions. The SAR324 genomes also suggested C1 metabolism and a particle-associated life-style. Microautoradiography and fluorescence in situ hybridization confirmed bicarbonate uptake and particle association of SAR324 cells. Our study suggests potential chemolithoautotrophy in several uncultured Proteobacteria lineages that are ubiquitous in the dark oxygenated ocean and provides new perspective on carbon cycling in the ocean’s largest habitat.

The dark ocean (below 200 m) contains an active and metabolically diverse microbial assemblage that is responsible for the majority of marine organic carbon mineralization (1). In addition to heterotrophic microbial activity, autotrophic carbon assimilation may be significant not only in oxygen minimum zones (OMZs) and anoxic basins (24), but also throughout the oxygenated water column (57). In the North Atlantic, the dark ocean’s primary production is 20 to 40% of the photic zone’s exported production and may be of the same order of magnitude as the dark ocean’s heterotrophic production (6). It has been generally assumed that the predominant type of autotrophy in the dark ocean is CO2 fixation through the 3-hydroxypropionate/4-hydroxybutyrate cycle, driven by the archaeal marine group I Crenarchaea and fueled by ammonia oxidation (8, 9). However, archaeal nitrification may be insufficient to support the measured inorganic carbon fixation rates in some regions of the North Atlantic (6, 10, 11). Thus, as yet unidentified microbial lineages and energy sources may be responsible for a significant fraction of carbon fixation in the dark ocean. The paucity of representative pure cultures has hindered experimental and genomic studies of the majority of bacterioplankton from the deep ocean so far.

We used single-cell sorting and DNA sequencing to identify the predominant bacterial lineages in the mesopelagic that contain genes for inorganic carbon fixation (12). This approach can link phylogenetic [such as small subunit ribosomal RNA (SSU rRNA)] and metabolic marker genes by sequencing them from the same cell obtained directly from its environment, without the need for cultivation (1317). Using single-cell sorting and whole-genome amplification, we generated libraries of prokaryotic single amplified genomes (SAGs) from surface and mesopelagic samples from the South Atlantic and North Pacific Subtropical Gyres (figs. S1 and S2). By polymerase chain reaction (PCR) and subsequent sequencing of the SSU rRNA genes, we identified 502 mesopelagic and 236 surface-ocean prokaryote SAGs (Fig. 1 and Table 1). Microbial community composition, as determined by SAG libraries, differed between the surface and the mesopelagic zones within each gyre. In contrast, microbial composition was similar between samples collected at similar depths at the two stations, especially for the mesopelagic (Fig. 1), in agreement with past studies employing more traditional molecular techniques (18, 19). Overall, our SAG libraries captured the majority of previously described uncultivated marine lineages and opened the door for their genomic study (Fig. 1 and figs. S3 to S11).

Fig. 1

Proportions of bacterial and archaeal SAGs retrieved from surface and mesopelagic stations, identified by SSU rRNA sequence analysis. Alphaprot., Alphaproteobacteria; Deltaprot., Deltaproteobacteria.

Table 1

Summary of bacterial and archaeal SAG PCR screening results.

View this table:

Using PCR, we screened mesopelagic SAGs for several key genes mediating carbon fixation and chemolithotrophic energy production (table S1) and found that Deltaproteobacteria cluster SAR324, Gammaproteobacteria clusters ARCTIC96BD-19 and Agg47, and some Oceanospirillales possess genes for both carbon fixation and sulfur oxidation (Fig. 2). PCR did not detect genes central to the reductive tricarboxylic acid cycle. However, sequences (cbbL and cbbM) coding for the large subunit of ribulose-1,5-bisphosphate carboxylase-oxygenase (RuBisCO), a key enzyme mediating the Calvin-Benson-Bassham (CBB) cycle, were recovered from 47% of SAR324 (the first record of RuBisCO in Deltaproteobacteria), 25% of Gammaproteobacteria, and 12% of all mesopelagic bacteria SAGs. These gene frequencies represent minimal estimates and may be higher within the bacterioplankton. Several inherent limitations in the PCR-based SAG screening approach applied here could lead to an underestimation of these frequencies, including incomplete whole-genome amplification, PCR primer mismatches, and PCR template secondary structures. For comparison, recA-normalized abundances of RuBisCO genes recovered from several dark ocean (≥500 m) metagenomes and a metatranscriptome at station ALOHA were 6 to 26% and 33%, respectively (table S2). Together, SAR324, ARCTIC96BD-19, Agg47, and Oceanospirillales made up 31 and 34% of all bacterial SAGs from the South Atlantic and North Pacific mesopelagic, respectively. This is consistent with previous findings of high abundances of these lineages in the dark ocean in various geographic regions (18, 20, 21), which is indicative of their significance in biogeochemical processes.

Fig. 2

(A) Phylogenetic tree of Gammaproteobacteria and Deltaproteobacteria SAR324 SSU rRNA sequences of mesopelagic SAGs (in color) recovered from the South Atlantic and North Pacific Subtropical Gyres. SAGs for which draft genomes were generated are marked with asterisks. The tree was inferred using maximum likelihood in the RAxML program. Bootstrap values (1000 replicates) ≥ 50% are indicated at nodes. Displayed are only SAGs in which chemoautotrophy-related genes were detected. (B) Occurrence and type of carbon fixation and sulfur oxidation pathway genes recovered from SAGs.

Genomic sequencing of two ARCTIC96BD-19 and two SAR324 SAGs (tables S7 to S9) recovered complete RuBisCO operons in all four SAGs (tables S10 to S13). Phylogenetic analysis of gene sequences and the presence of protein active sites, RuBP binding sites, and RuBisCO sequence motifs confirm that these are bona fide RuBisCO genes, in contrast to RuBisCO-like systems found in certain bacteria and archaea (fig. S12 and S13) (22). Nonfunctional genes are rare in oligotrophic bacterioplankton (23, 24). Thus, the presence of RuBisCO genes in the mesopelagic is likely to be indicative of carbon fixation via the CBB cycle by several indigenous bacterial lineages and could account for at least some of the RuBisCO proteins observed below the photic zone (25). A closely related cDNA sequence from 500 m at station ALOHA (26) provides evidence for RuBisCO gene expression in situ (fig. S12). Microautoradiography linked with catalyzed reporter deposition fluorescence in situ hybridization (MAR-CARD-FISH) performed on prokaryotes from an oceanic station in the North Atlantic (fig. S1) confirmed that at least one of the RuBisCO-containing groups, SAR324, assimilated inorganic carbon in situ (Fig. 3, A and B). The SAR324 cells constituted 6 to 17% of all prokaryotes, and 3 to 21% of SAR324 cells exhibited detectable bicarbonate assimilation throughout the meso- and bathypelagic (Fig. 3C). The bicarbonate-active SAR324 cells made up 20.4 ± 13.3% (n = 5 samples) of total bicarbonate-positive 4′,6-diamidino-2-phenylindole–stained cells, suggesting that SAR324 cells may be significant contributors to the dark ocean's chemoautotrophy.

Fig. 3

Micrographs demonstrating bicarbonate uptake and aggregation around particles by Deltaproteobacteria SAR324. (A) Overlay of transmitted-light and epifluorescence micrographs showing SAR324 cells labeled by CARD-FISH and silver grains indicating bicarbonate uptake by SAR324 (dark gray dots). Particles are also visible (faint grayish background). (B) Single-layer transmitted light micrograph for better visibility of the silver halos detected by microautoradiography and particle aggregates. (C) Vertical profiles of the percent of SAR324 cells out of total prokaryote cells, the percent of total SAR324 cells positive for dissolved inorganic 14C uptake, temperature, dissolved oxygen, and salinity in the North Atlantic (58.60°N, 39.71°W) during the Dutch Geotraces-1 cruise (RV Pelagia, 5 May 2010).

RuBisCO sequences similar to those obtained from our SAGs have been detected in other dark ocean regions, including bacterial artificial chromosomes (BACs) from the Monterey Bay OMZ (~700 to 800 m) (figs. S1 and S12 to S14 and tables S3 to S6). The arrangement of the cbb operon in BAC clones was similar to that of other RuBisCO operons (27), and typical protein active sites and the RuBisCO motif were present in large-subunit sequences (fig. S13). Recently, multiple RuBisCO-containing BAC clones were recovered from aerobic but subphotic depths (200 m) of Monterey Bay (28). A time series study of the top 200 m indicates the highest cbbM abundance at subphotic depths, and no correlation with the concentration of chlorophyll, indicating decoupling from direct photosynthesis (fig. S15). The high similarity of RuBisCO genes found by this and other studies in various geographic regions suggests that SAR324 cells and Gammaproteobacteria represented in our SAG libraries may be the predominant microbial groups encoding these genes in the dark ocean.

To identify potential energy sources for chemoautotrophy in the oxygenated dark ocean, we screened mesopelagic prokaryote SAGs by PCR for genes involved in sulfur, ammonia, and nitrite oxidation (table S1). We recovered amoA from the majority of mesopelagic crenarchaeal marine group I SAGs from the South Atlantic (60%) and North Pacific (81%), supporting previous reports that this crenarchaeal group is capable of nitrification (7, 9). Although no ammonia or nitrite oxidation genes were detected in any bacterial SAGs, we found several genes involved in dissimilatory sulfur oxidation (Fig. 2). Sulfur oxidation genes were found only in those bacterial taxonomic groups that also contained RuBisCO, and the same association was found in multiple cells originating from two geographically distant locations. Partial sequences coding for lineage I and II adenosine 5′-phosphosulfate (APS) reductase (aprA) were recovered from the Deltaproteobacteria SAR324 and Gammaproteobacteria lineages ARCTIC96BD-19 and Agg47 (fig. S16A). Lineage I and II aprA genes are affiliated with sulfur-oxidizing bacteria (SOB) and are phylogenetically distinguishable from APS reductase involved in sulfate reduction and assimilatory APS pathways (29). The aprA sequences from Gammaproteobacteria SAGs clustered with SOB lineage I, whereas SAR324 aprA sequences formed a distinct cluster within SOB lineage II, closely related to Chlorobi sequences (fig. S16A). Reverse-type dissimilatory sulfite reductase (rdsrA) genes were recovered only from SAR324 SAGs and were most similar to sequences from Chlorobi (fig. S16B). The consistent co-occurrence of RuBisCO, aprA, and rdsrA genes in the same bacterial lineages suggests the potential use of dissimilatory sulfur oxidation for energetic support of autotrophic carbon fixation in these microbes.

Additional sulfur oxidation genes were identified from whole-genome analysis of ARCTIC96BD-19 and SAR324 SAGs, including genes supporting thiosulfate oxidation, adenosine 5′-triphosphate (ATP) sulfurylase, and sulfate permease (tables S10 to S13). The SAR324 cells also contained genes potentially supporting the oxidation of carbon monoxide and methane and a suite of genes found in methylotrophic pathways (table S14). Together, these data suggest a flexible and opportunistic metabolic lifestyle of the SAR324 lineage.

Because these RuBisCO-containing bacterial lineages are found throughout the dark oxygenated water column (Fig. 3C and fig. S2), it is not immediately apparent what the sources of reduced sulfur compounds are that would be required for chemolithotrophic energy production. Reduced inorganic sulfur is rare outside the vicinity of hydrothermally active or upwelling regions (2, 30). Possible other sources include proteins, osmolytes (such as dimethylsulfoniopropionate), and particles, the hot spots of microbial activity that may contain anoxic microniches (5, 31, 32). Supporting the latter, we detected a suite of motility and adhesion genes in SAR324 cells (table S15) and confirmed their common association with particles in situ (Fig. 3A). Likewise, Agg47 cells have also been found to be preferentially on particles (33). Recently, an active but cryptic sulfur cycle was reported in the OMZ waters off the coast of Chile, where sulfide produced from sulfate reduction is immediately oxidized and therefore does not accumulate in the water column (4). The ARCTIC96BD-19 lineage is phylogenetically close to the SUP05 cluster, which dominates many OMZ regions, suggesting the possibility of a similar sulfur dynamic mediated by the former lineage even in well-oxygenated waters.

All four ARTIC96BD-19 and SAR324 genomes possessed multiple ATP-binding cassette (ABC) transporters for sugar transport and accessory periplasmic components, as well as several amino acid and oligopeptide transporters (table S14), often in tandem, indicating co-regulation. This suggests a likely mixotrophic lifestyle for these RuBisCO-containing lineages.

We have demonstrated that several uncultured Proteobacteria lineages that are indigenous and abundant in the dark oxygenated ocean are likely mixotrophs and have the potential for autotrophic CO2 fixation, coupled to the oxidation of reduced sulfur compounds. Some of these dark ocean bacteria may also be methylotrophs, using reduced single-carbon compounds as energy sources for growth. These previously unrecognized metabolic types of dark ocean bacteria may play an important role in global biogeochemical cycles, and their activities may in part reconcile current discrepancies in the dark ocean's carbon budget.

Supporting Online Material

www.sciencemag.org/cgi/content/full/333/6047/1296/DC1

Materials and Methods

Figs. S1 to S19

Tables S1 to S15

References

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. Acknowledgments: This work was supported by NSF grants EF-826924 and OCE-0821374 (R.S. and M.E.S.), a grant from the Maine Technology Institute (Bigelow Laboratory), U.S. Department of Energy (DOE) JGI 2010 Microbes Program grant CSP77 (R.S. and M.E.S.), the David and Lucille Packard Foundation (C.M.P.), the Gordon and Betty Moore Foundation (E.F.D.), the Dutch Science Fund-Earth and Life Sciences (G.J.H.), the Austrian Science Fund-FWF and European Science Foundation EuroEEFG project MOCA (G.J.H.), and Marie Curie project ARCADIA (G.J.H.). Work conducted by the DOE Joint Genome Institute is supported by the Office of Science of the DOE under contract no. DE-AC02-05CH11231. We thank chief scientist S. Curless, the officers and crew of the RV Ka'imikai-O-Kanaloa, and the HOT team for sample collection at station ALOHA; S. Sievert of the Woods Hole Oceanographic Institute for providing PCR primer information; J. Heywood of Bigelow for collecting South Atlantic field samples; the crew of the RV Point Lobos; V. Rich for initial screening of Monterey Bay libraries; and F. Chavez, R. Michasaki, and T. Pennington for providing Monterey Bay oceanographic data. SSU rRNA and metabolic gene sequences have been deposited in GenBank with the following accession numbers: SSU rRNA, HQ675122 to HQ675859; cbbL, HQ675043 to HQ675069; cbbM, HQ675070-HQ675086; aprA, HQ675087-HQ675118; and rdsrA, HQ675119 to HQ675121. Whole-genome sequence data have accession numbers AFHZ00000000 (AAA001-B15), AFIB00000000 (AAA001-C10), AFHY00000000 (AAA007-O20), and AFIA00000000 (AAA240-J09). Raw sequences were deposited in the GenBank Short Read Archive under accession numbers SRA029592 and SRA035467 (AAA001-B15), SRA029604 and SRA035394 (AAA001-C10), SRA029593 and SRA035468 (AAA007-O20), and SRA029596 and SRA035470 (AAA240-J09).
View Abstract

Stay Connected to Science

Navigate This Article