Arcobacter peruensis sp. nov., a Chemolithoheterotroph Isolated from Sulfide- and Organic-Rich Coastal Waters off Peru

The Peruvian shelf waters sampled onboard the RV Meteor Research Expedition M93 in February and early March 2013 (Fig. 1; see also Table S1 in the supplemental material) were characterized by high chlorophyll concentrations and the extreme depletion of dissolved oxygen (<5 μM) below a depth of 10 m and nitrate below a depth of 30 m (27). Dissolved hydrogen sulfide was present in the anoxic waters that covered the entire near-shore Peruvian shelf between 12°S, 78.3°W and 13.3°S, 77°W (Fig. 1). Bottom water sulfide concentrations were typically 5 to 10 μM in the near-shore waters and at station U1a reached up to 23 μM (Fig. 1 and 2a). Potential denitrification rates at station U1a were the greatest in the redoxcline (6.5 ± 0.4 μM N day⁻¹) and decreased within the deeper sulfidic zone (0.9 ± 0.1 μM N day⁻¹). In contrast, dark carbon fixation rates remained constant at 2.8 ± 0.2 μM C day⁻¹ throughout the redoxcline and into the deep sulfidic bottom waters. The nitrogen loss and carbon fixation rates measured on the shelf were similar to the rates in the Peru-Chile and Namibia shelf waters and indicated active chemolithoautotrophic activity, consistent with the findings of previous studies (6, 26, 28).

Bacteria belonging to the genus Arcobacter, as identified and enumerated using the fluorescent in situ hybridization (FISH) probe Arc94, were also detected throughout these near-shore, highly productive, sulfide-rich Peru upwelling bottom waters. At near-shore station U1a, the nitrate-sulfide redoxcline supported a large Arcobacter population, exceeding 10⁶ cells ml⁻¹, that comprised 25% of the entire microbial community (Fig. 2b). At stations containing <10 μM dissolved sulfide in the bottom waters, Arcobacter was still present, but relative cell abundances were typically <3% of the microbial community, or below 10⁵ cells ml⁻¹ (Fig. S1). Offshore and in the absence of dissolved sulfide, Arcobacter abundances dropped below <0.1% of the microbial community, just at or below the limits of detection. Microbial 16S rRNA gene diversity analysis of station U1a redoxcline waters, using clone library and amplicon sequencing (pyrosequencing) techniques, identified common OMZ-occurring sulfide-oxidizing bacteria, such as the SUP05 (^UThioglobus perditus, where the superscript “U” indicates an uncultivated taxon [27]) clade within the Gammaproteobacteria, as well as uncultured Sulfurovum and Arcobacter spp. within the Epsilonproteobacteria (Fig. 2c). Arcobacter abundances at station U1a, based on both catalyzed reporter deposition (CARD)-FISH cell counts and 16S rRNA gene abundances from amplicon pyrosequencing analysis (Fig. 2b and c), peaked at the middle of the broad redoxcline at 40 m, consistent with reports elsewhere of Arcobacter spp. occurring principally at oxic-sulfidic or nitrate-sulfidic redoxclines (4, 20, 42, 43).

We isolated an Arcobacter bacterium from station U1a under anaerobic conditions using natural seawater amended with sulfide and nitrate. The isolate, Arcobacter sp. strain PSE-93, was enriched from Peruvian seawater containing sulfide and nitrate by continuous transfers until it reached 90 to 95% purity, as checked with CARD-FISH using the Arc94 probe (Fig. S2). PSE-93 had a slightly elongated morphology (0.9 μm in length and 0.6 μm in width) that was comparable to the morphology of Arcobacter spp. observed in situ. DNA extraction of the enrichment culture, followed by sequencing by use of the Pacific Biosciences RS II technology (PacBio), assembly, and polishing, resulted in 256 contigs. After the removal of short contigs (<10 kbp) from the assembly, only one contig (84-fold average coverage) that was circularized remained, yielding the complete genome of PSE-93. Analysis by the CheckM method (44), based on 338 conserved marker genes, indicated that the genome of Arcobacter PSE-93 was complete (99.6%) and free of contamination (1.0%) or strain heterogeneity (0.0%). Summary genome statistics for Arcobacter PSE-93 are shown in Table S2.

The closed circular genome (2.8 Mbp in size with a 27.8% GC content and 2,697 predicted protein-coding genes) of PSE-93 revealed that it has the potential to reduce nitrate to N₂ via a complete denitrification pathway that includes periplasmic nitrate reductase (napAB), cd₁-type nitrite reductase (nirS), nitric oxide reductase (norBC), and nitrous oxide reductase (nosZ) (Fig. 3). The presence of a cbb₃-type terminal oxidase (fixNOQP) implies that Arcobacter PSE-93 might be able to respire with oxygen. The identification of flagellum- and chemotaxis-associated genes in the PSE-93 genome indicated, furthermore, that Arcobacter might be able to position itself within the redoxcline according to the gradients of dissolved sulfide and nitrate. A gene encoding a protein with a rhodanese-like domain that may act to shuttle zerovalent sulfur in between redox centers during sulfur metabolism was found (45). The Arcobacter PSE-93 genome contained a complete periplasmic sulfide oxidation Sox pathway (soxABCDXYZ) and a sulfide-quinone reductase gene (sqr) that allows for the oxidation of various reduced sulfur compounds (including sulfide, sulfur, and thiosulfate) to sulfate. The presence of soxCD is consistent with a lack of internal elemental sulfur storage, as missing soxCD genes have been shown in other sulfide oxidizers to be correlated with intracellular sulfur storage (46). Sulfur reduction genes, such as those for polysulfide reductase, tetrathionate reductase, and sulfide hydrogenases, some of which have been identified in A. anaerophilus (10), were absent from the Arcobacter PSE-93 genome.

Key genes associated with autotrophic CO₂ fixation (e.g., ribulose-1,5-bisphosphate carboxylase/oxygenase [Rubisco] and succinyl coenzyme A [CoA] synthetase reverse trichloroacetic acid [rTCA]), which are frequently identified in sulfide-oxidizing bacteria and in other autotrophic members of the genus Arcobacter (1, 9, 10), were not found in the Arcobacter PSE-93 genome (Fig. 3), nor were genes responsible for nitrogen (N₂) fixation present. Instead, genes encoding organic matter-dependent carboxylases involved in acetate and propionate assimilation were detected, as were genes encoding carboxylases involved in fatty acid biosynthesis and anaplerotic reactions. Genes encoding organic compound transporters, e.g., TonB-dependent receptors and TRAP (tripartite ATP-independent periplasmic transporter)-type C₄ transport systems, including acetate permeases and ABC transporters associated with acetate and glucose metabolisms, were also detected in the genome. Alternative electron donors may also include formate (formate dehydrogenase) and hydrogen, for example, Ni/Fe hydrogenases, which have also been identified in marine Arcobacter species associated with unicellular protists (18).

The genome of Arcobacter PSE-93 contained six complete sets of rRNA genes (16S-23S-5S). The 16S rRNA gene sequences were almost identical and shared up to 7 mismatches (>99.53% identity). Multiple copies of rRNA genes appear to be a common feature of microorganisms within the genus Arcobacter (44) and might offer a competitive advantage because they confer a higher growth rate or competitive fitness (47). Furthermore, multiple copies of rRNA genes can lead to an overestimation of Arcobacter abundances in the environment and should be considered when interpreting 16S rRNA gene amplicon sequencing data.

We tested the ability of isolates of Arcobacter PSE-93 to oxidize sulfide and reduce nitrate on various carbon substrates under anoxic conditions. In all subsequent physiology experiments, we used plate-grown colonies (colonies grown on PY-BROTH medium [DSMZ, Germany]) obtained from the isolate described above, which were then transferred to axenic liquid medium. The identity and purity of the transferred Arcobacter PSE-93 colonies were verified by full 16S rRNA sequencing. Enough dissolved sulfide was amended to reduce all added nitrate to N₂, assuming the complete oxidation of sulfide to sulfate. In natural seawater experiments (in North Sea seawater [saNS] medium) under autotrophic conditions (i.e., in the absence of added organic carbon substrate), PSE-93 showed slightly enhanced rates of sulfide oxidation (1.29 ± 0.26 μM h⁻¹) and nitrate reduction (1.00 ± 0.04 μM h⁻¹) (Fig. 4a and d; Table 1). Arcobacter PSE-93 cell densities never exceeded 9 × 10⁴ cells ml⁻¹, and the cells exhibited negligible to nondetectable [¹³C]bicarbonate assimilation rates under these autotrophic conditions (Fig. 4d). Similarly, using synthetic seawater (Sas) medium under autotrophic conditions, cell numbers also never reached more than 3 × 10⁵ cells ml⁻¹ (Fig. S3). In both experiments, sulfide consumption occurred, and nitrate reduction rates were detectable (Fig. 4a and d; Table 1; Fig. S3). Neither sulfide nor nitrate was fully consumed; roughly half of the nitrate was reduced to N₂, while the rest of the added nitrate was reduced only to nitrite.

Complete nitrate reduction (>4.1 μM h⁻¹) to dinitrogen as well as enhanced sulfide oxidation (>1.5 μM h⁻¹) did ensue when PSE-93 was grown heterotrophically on glucose, yeast extract, or acetate (Table 1). The nitrite that formed as an intermediate was completely reduced to N₂ by the end of the experiment (Fig. 4b). Growth on acetate yielded the highest rates of sulfide oxidation (6.6 ± 0.67 μM h⁻¹) and nitrate reduction (6.2 ± 1.69 μM h⁻¹) for PSE-93 (Fig. 4b and d; Table 1). ¹³C-labeled acetate incorporation during growth on nitrate, dissolved sulfide, and acetate showed that acetate was assimilated into PSE-93 biomass at a rate of 1.55 ± 0.19 fmol C cell⁻¹ day⁻¹. Growth under these conditions yielded high cell densities of 4 × 10⁶ cells ml⁻¹ with a doubling rate of 1.4 to 1.8 per day (calculated from cell abundances and ¹³C assimilation rates). Arcobacter PSE-93 was also able to grow heterotrophically, coupling acetate oxidation to nitrate reduction (Fig. 4c and d); however, the nitrate reduction (1.3 ± 0.22 μM h⁻¹) and acetate assimilation (0.38 ± 0.06 fmol C cell⁻¹ day⁻¹) rates were over 4-fold lower, and the calculated doublings per day (0.8 to 1.0) were nearly 2-fold lower than the rate of PSE-93 growth on dissolved sulfide, nitrate, and acetate (Fig. 4c and d; Table 1). These experiments demonstrated that the Peru upwelling Arcobacter PSE-93 strain is an obligate heterotroph capable of chemolithoheterotrophic and chemoorganoheterotrophic growth. More importantly, Arcobacter PSE-93 could couple sulfide oxidation to complete denitrification when low-molecular-weight organic carbon was made available for cell biomass synthesis.

PSE-93 formed a distinct clade that was different with respect to the four recently defined Arcobacter type strain clusters (48). The closest type strains to PSE-93 were in cluster 2, which contained A. venerupis, A. suis, A. ellisii, A. cloacae, and A. defluvii (Fig. 5). To place the PSE-93 identity in relation to the Arcobacter diversity in the Peruvian upwelling, we analyzed the 16S rRNA genes recovered from station U1a. From the nearly full-length 16S rRNA amplicon sequences obtained from clone library preparations at station U1a, we identified two Peruvian Arcobacter clones. ETSP (Eastern Tropical South Pacific) clone 3 and ESTP clone 5 formed two separate phylogenetic branches (Fig. 5). Sequences affiliated with ETSP clone 5 contained Arcobacter members capable of both CO₂ fixation (e.g., A. sulfidicus and A. nitrofigilis) and acetate assimilation. In contrast, ETSP clone 3 and the nearby affiliated PSE-93 lineage contained, to the best of our knowledge, only members capable of heterotrophic growth.

Arcobacter PSE-93 was most closely affiliated with the Arcobacter clade that contained ETSP clone 3 (96.1% 16S rRNA gene identity). Within this clade, the sequence of PSE-93 was most closely related to a sequence recovered from the North Sea, that of Arcobacter sp. strain KT0913 (49), and nearly identical to that of Arcobacter sp. strain LPB0137 (the 16S rRNA gene from an environmental Korean strain). Arcobacter strain KT0913 shares a 16S rRNA gene identity value with PSE-93 above the species level (98.8%) (49); however, confirmation by comparison of the average nucleotide identity (ANI) values awaits the availability of the genome sequence of KT0913. Although Arcobacter PSE-93 enriched in our incubations was not identical to ETSP clone 3 or 5, our phylogenetic analysis suggests that PSE-93 is a close relative of ETSP clone 3, one of the most abundant Peruvian upwelling Arcobacter members. Additionally, the Arcobacter 16S rRNA genes recovered from 454 pyrosequencing (∼450 bp in length) were closely related to the clone library 16S rRNA genes, as well as to the 16S rRNA gene of the Arcobacter PSE-93 isolate (e.g., ETSP pyrotags 4, 5, and 6).

From isotope labeling experiments performed on samples obtained during RV Meteor Research Expedition M93 in Peru, we used nanoscale secondary ion mass spectrometry (nanoSIMS) to obtain single-cell measurements of the Arcobacter composition and the uptake of CO₂ in the in situ population (Fig. 6). The analysis of single-cell ¹³CO₂ assimilation rates showed that Arcobacter fixed 0.03 ± 0.01 fmol CO₂ cell⁻¹ day⁻¹. This rate was significantly lower than the measured CO₂ assimilation rates for Gammaproteobacteria clade SUP05 and other Epsilonproteobacteria (e.g., Sulfurovum spp.) of 0.19 ± 0.02 and 0.72 ± 0.03 fmol C cell⁻¹ day⁻¹, respectively (analysis of variance [ANOVA], degrees of freedom [df] = 2, P < 0.001 [considered a significant difference]; Fig. 6b). Arcobacter bacteria thus contributed to less than 1% of the overall dark CO₂ fixation at station U1a, despite the large abundance of Arcobacter (25% of the total microbial community), whereas the Gamma- and Epsilonproteobacteria accounted for nearly one-quarter of the total dark carbon fixation rates at station U1a, consistent with their abundances (Table S3). The Arcobacter spp. present in the Peru upwelling waters also differentiated themselves from the SUP05 clade and other Epsilonproteobacteria, in that Arcobacter exhibited significantly lower sulfur contents than these other groups of sulfide-oxidizing bacteria (Fig. 6c).

A. peruensis type strains: PSE-93, LMG 31510. Etymology: pe.ru.en′sis. N.L. masc. adj. peruensis, pertaining to Peru. Locality: isolated from coastal, sulfidic Peruvian shelf waters at 12.23°S, 77.18°W. Properties: Gram-negative, slightly elongated rods (0.9 μm in length and 0.6 μm in width). It grows chemolithoheterotrophically in sulfide-, nitrate-, and organic matter-rich environments in coastal upwelling water. It grows best by coupling dissimilatory nitrate reduction to dinitrogen to the oxidation of dissolved sulfide. Alternatively, heterotrophic growth on acetate in the absence of sulfide occurs, albeit at much slower doubling times. It grows aerobically on PY-BROTH medium (DSMZ, Germany) supplemented with 200 μM KH₂PO₄, 10 mM MgSO₄·7H₂O, 5 mM HEPES at a final pH of 7.8 and solidified with Bacto agar (1%, wt/vol). The median GC content of the type strain is 27.4 mol%.

Peru upwelling waters (12°S, 78.5°W and 13.5°S, 77°W) were sampled from 8 February to 4 March 2013 onboard the RV Meteor (Research Expedition M93; Fig. 1; see also Table S1 in the supplemental material). All seawater samples for this study were collected at an approximately 5-m resolution to within 5 m of the seafloor using a conductivity-temperature-depth (CTD) rosette equipped with 10-liter Niskin bottles. Oxygen was monitored with a Seabird sensor and calibrated against Niskin bottle samples by Winkler titration. Collected seawater nitrate and nitrite concentrations were determined with a QuAAtro autoanalyzer (Seal Analytical, UK) with precisions of ±0.1 μM. Sulfide concentrations were determined onboard as described by Cline (61) using 4 ml of seawater. The sulfide data in Fig. 1 were plotted with Ocean Data View software (62). For the determination of water column elemental sulfur, 50-ml samples of seawater were sampled using anaerobic techniques, fixed with 100 μl ZnCl₂ (20%, wt/wt), and stored at −20°C. Elemental sulfur was extracted from the Zn-fixed samples using a chloroform-methanol procedure (63) and measured on a Waters Acquity H-class (Waters, Japan) ultra-high-pressure liquid chromatography (UPLC) system (1.7-μm-particle-size, 2.1- by 50-mm column with a methanol eluent at a flow rate of 0.4 ml min⁻¹; Acquity UPLC BEH C₁₈) equipped with a Waters photometric diode array detector (the absorbance wavelength was set to 265 nm, with the limit of detection being 50 nM). In this procedure, elemental sulfur can derive from suspended elemental sulfur, from elemental sulfur deposited on or within cells, or from the decomposition of dissolved polysulfides (63).

One to 2 liters of seawater was collected on polycarbonate filters (0.2-μm pore size; Millipore, Germany) for genomic DNA extraction. Genomic DNA was extracted using a Qiagen AllPrep DNA/RNA kit) and quantified spectrophotometrically by use of a NanoDrop spectrophotometer (Thermo Scientific). Universal bacterial barcoded PCR primers (primers Bakt341F [CCTACGGGNGGCWGCAG] and 805R [GACTACHVGGGTATCTAATCC], targeting the V3-V4 regions [59]) were used to generate amplicons for 454 pyrosequencing (454 FLX+ Titanium chemistry; Max Planck Genome Center, Cologne, Germany [https://mpgc.mpipz.mpg.de/home/]). The PCR conditions were initial denaturation at 95°C for 5 min, followed by 25 cycles of 95°C for 40 s, 55°C for 2 min, and 72°C for 1 min and a final extension of 72°C for 7 min, with a ramp rate of 3°C s⁻¹. The PCR products were purified using a QIAquick PCR purification kit (Qiagen, Hilden, Germany). The recovered 16S rRNA sequences were trimmed to remove the primer adaptors before submission to the SILVAngs pipeline, which provided automated alignments, quality management, dereplication, and taxonomic classification (60).

The microbial diversity at station U1a was also analyzed by clone library sequencing, which produced nearly full-length 16S rRNA gene sequences. The clone library preparation procedure, including raw read quality controls and taxonomic classification using ARB software, was performed by a method similar to that described by Callbeck et al. (27). Universal bacterial primers GM3f (5′-AGA GTT TGA TCM TGG C-3′) and GM4r (5′-TAC CTT GTT ACG ACT T-3′) were used to generate 16S rRNA PCR amplicons from DNA samples taken at station U1a. The PCR conditions were initial denaturation at 95°C for 5 min, followed by 25 cycles of 95°C for 1 min, 50°C for 1 min, and 72°C for 2 min and a final extension of 72°C for 10 min, with a ramp rate of 3°C s⁻¹. DNA was visualized by gel electrophoresis. The 16S rRNA PCR amplicons were purified and then ligated into a TOPO TA vector (Invitrogen). Escherichia coli clones were picked and screened for the vector insert by PCR. The plasmids were extracted from vector-positive colonies (using a plasmid extraction kit [MoBio]), and then the plasmid vector was PCR amplified using forward and reverse M13 primers (primers M13f [5′-CCC AGT CAC GAC GTT GTA AAA CG-3′] and M13r [5′- AGC GGA TAA CAA TTT CAC ACA GG-3′]). The PCR products were purified using Sephadex (G-50 Superfine; Amersham Bioscience) and then sequenced (Sanger sequencing with a BigDye kit; Applied Biosystems). Raw sequence data were quality controlled, and vector ends were trimmed and then assembled into nearly full-length 16S rRNA contigs using Sequencher (version 4.6) software (Gene Codes Corporation, Ann Arbor, MI). The nearly full-length and partial 16S rRNA sequences recovered from the clone library preparations and from pyrosequencing were aligned by use of the SINA aligner (64) and then imported into the SILVAref115 curated 16S rRNA reference database (65) using ARB software (66). A consensus 16S rRNA gene tree of neighbor-joining (distance matrix), parsimony (Phylip DNApars program), and maximum likelihood (RAxML, version 7, software) phylogenetic trees applying various epsilonproteobacterial filters was calculated from the nearly full-length sequences (67). Partial 16S rRNA genes were added to the consensus tree using RAxML (version 7) software, applying an epsilonproteobacterial filter.

Seawater samples collected from Niskin bottles were filtered over polycarbonate filters (0.2-μm pore size; Millipore, Germany) for analysis of cell abundances and identification. The filters were fixed in paraformaldehyde (final concentration, 1 to 2% [wt/vol]) for 12 h at 4°C prior to filtration. Catalyzed reporter deposition (CARD)-FISH analysis was performed using probes targeting Arcobacter and general Epsilonproteobacteria: Arc94 (5′-TTAGCATCCCCGCTTCGA-3′) (68) and Epsy682 (5′-CGGATTTTACCCCTACACM-3′) (43). The Arc94 and Epsy682 probes were incubated in 20% formamide containing hybridization buffer at 46°C for 3 h. The total microbial community was stained with 4′,6-diamidino-2-phenylindole (DAPI), and together, DAPI-stained and Arc94- and Epsy682-hybridized cells were visualized and quantified using epifluorescence microscopy. Hybridized cells in 10 fields of view and up to 1,000 DAPI-stained cells were counted. Negative and positive controls were performed using the NON338 and EUB3381-III probes, respectively, as described previously (69).

Isotope labeling experiments to determine the rates of nitrate reduction (¹⁵N labeling) and CO₂ fixation (¹³C labeling) were performed as described previously (70). Briefly, denitrification, as well as bulk and single-cell carbon fixation rates, were determined from 12-ml Exetainer incubation experiments, as follows: in experiment 1, ¹⁵NO₃⁻ plus H¹³CO₃⁻; in experiment 2, ¹⁵NO₂⁻ plus ¹⁴NH₄⁺ plus H¹³CO₃⁻; and in experiment 3, ¹⁵NH₄⁺ plus ¹⁴NO₂⁻ and H¹³CO₃⁻. The concentrations of labeled substrates were 25 μM, 5 μM, and 5 μM for NO₃-, NO₂-, and NH₄⁺, respectively. Isotopic ratios of ¹⁵N¹⁵N, ¹⁵N¹⁴N, and ¹⁴N¹⁴N nitrogen gas were measured on a gas chromatography (GC) isotope-ratio mass spectrometer (IRMS) (VG Optima, Manchester, UK), while bulk carbon fixation rates were measured separately in ¹³C incubation experiments performed in gas-tight 4.5-liter bottles (26). The ¹³C/¹²C isotope ratio was measured on an element analyzer (EA) IRMS (FlashEA 1112 series coupled with an IRMS; Finnigan Delta Plus XP; Thermo Scientific).

In addition to the sample for IRMS measurements, extra samples from the last time point of the stable isotope incubation experiments were filtered onto gold-palladium-precoated polycarbonate filters (0.2-μm pore size; Millipore, Germany) for FISH-secondary ion mass spectrometry (SIMS) analysis. Select field of views containing hybridized cells were marked using laser dissection microscopy (DM 6000 B, Leica, Germany) and analyzed by nanoSIMS (nanoSIMS 50L; Cameca, France), which was used to measure simultaneously the single-cell isotopic composition of Arc94-hybridized cells. Secondary ions ¹²C, ¹³C, ¹⁹F, ¹²C¹⁴N, ¹²C¹⁵N, ³¹P, and ³²S were measured on 7 mass detectors. Samples were presputtered with a Cs⁺ primary ion beam of ∼300 pA, and thereafter, the instrument was tuned on a 50- by 50-μm-field-of-view raster size (yielding a mass resolution of over 8,000). Final image acquisition was performed with a primary beam between 1 and 2 pA at a 10- by 10-μm raster size (256 × 256 pixels) with a dwell time of 1 ms per pixel for 40 planes.

The ¹³C/¹²C and ³²S/(¹²C + ¹³C) ratios were calculated using look@nanoSIMS software as outlined elsewhere (71). The ¹⁹F signal (of the halogenated oligonucleotide Arc94 probe) was used to specifically identify Arcobacter hybridized cells (Fig. 6a). Single-cell carbon assimilation rates were determined using ¹³C/¹²C enrichment, the labeling percentage, the cell volume (based on a rounded rod consisting of two half spheres), and the cell carbon content (calculated using an allometric model, in which the number of femtograms of C cell⁻¹ is equal to 133.754 × V^0.438, where V is the cell biovolume [in cubic micrometers] [72]) and divided over the incubation period.

The Arcobacter culture used in this study was enriched from sulfidic station U1a (Fig. 1; Table S1) at a 50-m depth onboard the research vessel. We used sterile-filtered Peru seawater with additions of sodium nitrate and either sodium sulfite, sodium thiosulfate, or elemental sulfur as electron donors to a final concentration of 100 μM. A total of three transfers were made (with a 1% [vol/vol] inoculum) into new medium over the course of the research campaign, and samples were incubated in the dark at 14°C onboard the research vessel. Samples at ∼4°C along with additional filter-sterilized Peruvian seawater were express delivered (arriving within 2 to 3 days) to the laboratory at the Max Planck Institute for Marine Microbiology in Bremen, Germany. The enrichment cultures were maintained using 0.2-μm-pore-size filtered and autoclaved anaerobic (N₂-CO₂ atmosphere) Peruvian seawater. After autoclaving, the medium was supplemented with sulfide and nitrate (∼80 μM), inoculated (with a 1% [vol/vol] inoculum), and then incubated at close to ambient water temperatures (14°C). Sulfide and nitrate concentrations were monitored, and once the medium was consumed, the cultures were transferred to fresh medium. After five transfers (1% [vol/vol] inoculum) on nitrate and dissolved sulfide, based on the CARD-FISH screening using the Arc94 probe, Arcobacter cells made up 90 to 95% of the microbial community. Genomic DNA was then extracted from the nearly pure Arcobacter culture using a QIAamp genomic DNA kit (Qiagen, Netherlands).

Whole-genome sequencing on culture-extracted genomic DNA was performed using the Pacific Biosciences RS II technology (P4-C2 chemistry; Max Planck Genome Center, Cologne, Germany). The collected data were processed and filtered using SMRT analysis software (v.2.1). For genome assembly, SMRT analysis routine HGAP2 was applied. From the polished assembly of the sequences obtained by the use of PacBio, contigs shorter than 10 kbp were excluded, resulting in a single, long contig. For genome circularization and polishing, identical sequences on both contig ends were identified by blastn analysis and trimmed from the genome. Annotation was performed using the RAST program (73), and the annotation of the key metabolic genes was manually inspected and refined. Further analysis of the annotated genome was performed using both the RAST and Pathway tools (74). Genome completeness and contamination were estimated using the CheckM (v.1.0.11 11) tool (44) running the lineage-specific workflow using 338 marker genes.

The enriched Arcobacter strain (PSE-93) was further maintained on sterile anaerobic North Sea seawater (saNS medium) prepared in a Widdel flask, in which the medium was filter sterilized, autoclaved, and then allowed to cool under an N₂-CO₂ (90:10) atmosphere. The saNS medium was buffered to a final concentration of 2 mM HCO₃⁻. Sulfide and nitrate (∼80 μM) were amended to the medium, which was eventually dispensed into smaller serum bottles under an N₂-CO₂ atmosphere and inoculated with a 1% (vol/vol) culture from Peruvian seawater medium. The isolation of Arcobacter strain PSE-93 was performed by plating on DSMZ 1071 PY-BROTH medium (DSMZ, Germany) supplemented with 200 μM KH₂PO₄, 10 mM MgSO₄·7H₂O, 5 mM HEPES at a final pH of 7.8 and solidified with Bacto agar (1%, wt/vol). The plates were incubated at 14°C under oxic conditions. After a few weeks of incubation, single colonies were picked and restreaked on new plates until an axenic culture of PSE-93 was obtained, as confirmed by full 16S rRNA in-house Sanger sequencing and microscopic evaluation. Colonies from the axenic PSE-93 culture were thereafter used for inoculation of fresh sterile saNS medium with 50 to 100 μM nitrate and sulfide. PSE-93 was also grown on synthetic seawater (Sas) medium. Sas medium was prepared in a Widdel flask and contained, per liter of deionized water, 27.5 g NaCl, 5 g MgCl₂·6H₂O, 4.1 g MgSO₄·7H₂O, 0.66 g CaCl₂·2H₂O, and 1.02 g KCl (75). The medium was autoclaved and allowed to cool under an N₂-CO₂ (90:10) atmosphere. After cooling, sterile trace element, vitamin, and mineral solutions were added as described by Kamp et al. (75). The medium was buffered by the addition HCO₃⁻ (to a final concentration of 2 mM), and the pH was adjusted to 7.5.

For the time course experiments, the medium was dispensed anaerobically under N₂-CO₂ using Hungate techniques into sterile Duran bottles that were mounted and sealed at the top with a 50-ml glass syringe (SGE Analytical Science, Australia). Duran bottles, including the connected syringe, were filled without a headspace. In addition to the PSE-93 inoculum (1%, vol/vol), organic matter compounds glucose (1%, wt/vol), yeast extract (1%, wt/vol), and [¹³C]acetate (100 μM), as well as inorganic substrates sulfide (100 μM), nitrate (100 μM; 50% ¹⁵N labeling), and bicarbonate (2 mM; 10% ¹³C labeling), were added via a sidearm port in different combinations (Table 1). Experiments were run at 14°C, mimicking in situ conditions. Regular subsamples were taken via the sidearm port over the course of the experiment for the analysis of sulfide, nitrate, and nitrite production (1 ml subsample fixed in 500 μl 5% ZnCl₂) as well as for the analysis of labeled N₂ production (1 ml subsample in 12-ml Exetainer glass vials with a helium atmosphere and 50 μl saturated HgCl₂ solution). Additionally, 1 ml subsample was fixed in 100 μl 20% paraformaldehyde (PFA) for cell count analysis. At the end of the experiment, the remaining contents of the incubation were filtered onto a precombusted Whatman glass microfiber grade GF/F filter (GE Healthcare Life Sciences, UK) and stored at −20°C until further analysis. Samples for N₂ measurements were stored cap down in the dark at room temperature.

Dissolved sulfide concentrations in ZnCl₂-fixed samples were determined photometrically as described above. Nitrate and nitrite concentrations were determined with a CLD 60 chemiluminescence NO/NO_x analyzer (Eco Physics AG, Switzerland) after reduction to NO with acidic sodium iodide (NaI) and acidic vanadium(II) chloride, respectively (76, 77). Isotopic ratios of ¹⁵N¹⁵N, ¹⁵N¹⁴N, and ¹⁴N¹⁴N nitrogen gas were measured on a gas chromatography (GC) isotope-ratio mass spectrometer (IRMS) (VG Optima, Manchester, UK). Cell counts were obtained by flow cytometry on a BD FACSCalibur system (BD Biosciences, CA, USA) after the 2% PFA-fixed cells were stained for 20 min with SYBR green DNA stain. The instrument flow rate was calibrated with saNS medium. The sampling time was 1 min, and the background was determined with samples from the uninoculated control bottles. To quantify the amounts of ¹⁵N and ¹³C incorporated into biomass, the GFF filters were decalcified overnight, dried at 60°C for 1 h, pelletized into tin cups, and analyzed by use of a Thermo Flash EA 1112 elemental analyzer coupled to an isotopic-ratio mass spectrometer (Finnigan Delta Plus XP; Thermo Fisher Scientific, USA).

This Whole Genome Shotgun project and the 16S rRNA gene sequences (clone library) have been deposited in the DDBJ/ENA/GenBank databases under accession numbers CP032363 and MH916845 to MH916849, respectively. Water column nutrients and physical data for the RV Meteor M93 Research Expedition are available at Pangaea (https://doi.pangaea.de/10.1594/PANGAEA.860727), while station sulfur chemistry, Arcobacter cell densities, and rate process measurements have been submitted to Pangea (https://doi.pangaea.de/10.1594/PANGAEA.894324). Additional data regarding SUP05 cell densities can be found at https://doi.pangaea.de/10.1594/PANGAEA.876062.