Genomics and Ecophysiology of Heterotrophic Nitrogen-Fixing Bacteria Isolated from Estuarine Surface Water

ABSTRACT

The ability to reduce atmospheric nitrogen (N₂) to ammonia, known as N₂ fixation, is a widely distributed trait among prokaryotes that accounts for an essential input of new N to a multitude of environments. Nitrogenase reductase gene (nifH) composition suggests that putative N₂-fixing heterotrophic organisms are widespread in marine bacterioplankton, but their autecology and ecological significance are unknown. Here, we report genomic and ecophysiology data in relation to N₂ fixation by three environmentally relevant heterotrophic bacteria isolated from Baltic Sea surface water: Pseudomonas stutzeri strain BAL361 and Raoultella ornithinolytica strain BAL286, which are gammaproteobacteria, and Rhodopseudomonas palustris strain BAL398, an alphaproteobacterium. Genome sequencing revealed that all were metabolically versatile and that the gene clusters encoding the N₂ fixation complex varied in length and complexity between isolates. All three isolates could sustain growth by N₂ fixation in the absence of reactive N, and this fixation was stimulated by low concentrations of oxygen in all three organisms (≈4 to 40 µmol O₂ liter⁻¹). P. stutzeri BAL361 did, however, fix N at up to 165 µmol O₂ liter⁻¹, presumably accommodated through aggregate formation. Glucose stimulated N₂ fixation in general, and reactive N repressed N₂ fixation, except that ammonium (NH₄⁺) stimulated N₂ fixation in R. palustris BAL398, indicating the use of nitrogenase as an electron sink. The lack of correlations between nitrogenase reductase gene expression and ethylene (C₂H₄) production indicated tight posttranscriptional-level control. The N₂ fixation rates obtained suggested that, given the right conditions, these heterotrophic diazotrophs could contribute significantly to in situ rates.

IMPORTANCE

The biological process of importing atmospheric N₂ is of paramount importance in terrestrial and aquatic ecosystems. In the oceans, a diverse array of prokaryotes seemingly carry the genetic capacity to perform this process, but lack of knowledge about their autecology and the factors that constrain their N₂ fixation hamper an understanding of their ecological importance in marine waters. The present study documents a high variability of genomic and ecophysiological properties related to N₂ fixation in three heterotrophic isolates obtained from estuarine surface waters and shows that these organisms fix N₂ under a surprisingly broad range of conditions and at significant rates. The observed intricate regulation of N₂ fixation for the isolates indicates that indigenous populations of heterotrophic diazotrophs have discrete strategies to cope with environmental controls of N₂ fixation. Hence, community-level generalizations about the regulation of N₂ fixation in marine heterotrophic bacterioplankton may be problematic.

INTRODUCTION

The conversion of inert N₂ gas to ammonium (NH₄⁺) by microorganisms, known as N₂ fixation, is fundamental for biological productivity in many environments. In the tropical and subtropical oligotrophic oceans and in eutrophic freshwater lakes, pelagic N₂ fixation may supply enough N to sustain a considerable fraction of the new production (1, 2). In these environments, cyanobacteria are generally considered the main N₂-fixing organisms (diazotrophs), yet the genetic potential to fix atmospheric N₂ is distributed among a diverse array of prokaryotes, including members of the Proteobacteria, Firmicutes, and Archaea (3). Functional genes involved in N₂ fixation, especially the nitrogenase reductase-encoding gene, nifH, have been used as phylogenetic markers to identify diazotrophs and to examine their distribution in various environments (4, 5). Subsequent surveys show that, in addition to cyanobacteria, heterotrophic putative diazotrophs are widespread in marine waters (reviewed in reference 6) and may even dominate nifH gene libraries (7). Moreover, significant N₂ fixation has been documented in waters where cyanobacteria are not believed to be present or active (e.g., see references 8, to ,10). Nonetheless, the autecology and ecological importance of marine heterotrophic diazotrophs are still only poorly understood.

In order to estimate the potential contribution by heterotrophic organisms to marine N₂ fixation, cell-specific rates of N₂ fixation should be connected to the identities of the responsible cells, either directly, e.g., by nanoscale secondary ion mass spectrometry in combination with fluorescence in situ hybridization (11), or alternatively, by indirect estimates based on cell-specific rates determined in vitro in combination with in situ abundance measures. For instance, using cell-specific rates from the soil-derived Pseudomonas stutzeri strain CMT.9.A, it was recently reported that the abundance of a predominant gammaproteobacterial phylotype in the eastern tropical south Pacific Ocean could not account for local N₂ fixation (12, 13). Such calculations are, however, potentially compromised by the lack of cell-specific N₂ fixation rates of marine heterotrophic diazotrophs, which in turn is caused by a scarcity of cultivated representative species. Attempts to cultivate heterotrophic diazotrophic bacteria from marine environments have generally been few and had varying degrees of success (14–17). Recently, alpha- and gammaproteobacterial representative diazotrophs have been cultivated from pelagic waters of the estuarine Baltic Sea (18–20), but further ecophysiological and genomic investigations elucidating the regulation of N₂ fixation in relation to environmentally relevant stimuli have not yet been undertaken.

A multitude of constraints presumably regulate N₂ fixation by heterotrophic bacteria in pelagic marine systems. The reduction of N₂ to NH₄⁺ is expensive in terms of reducing power and cellular energy, requiring at least 16 mol of ATP to reduce 1 mol of N₂ and, hence, demanding considerable amounts of organic carbon (C) compounds. In addition, the nitrogenase enzyme complex is usually highly sensitive towards O₂, and N₂ fixation is therefore supposedly tightly regulated by the presence of O₂, as well as reactive N (21). Despite these high energy requirements and stringent control mechanisms, nifH gene transcripts affiliated with heterotrophic diazotrophs have been detected in low-carbon, oxic, oligotrophic surface waters (7, 22), as well as in N-replete environments (8, 23), emphasizing that the extent and regulation of heterotrophic N₂ fixation are currently poorly understood.

In this study, we report the comparative investigation of the genomic traits relating to N₂ fixation in three heterotrophic bacterial isolates from Baltic Sea surface waters: Pseudomonas stutzeri strain BAL361, Rhodopseudomonas palustris strain BAL398, and Raoultella ornithinolytica strain BAL286. Two of the three isolates (P. stutzeri BAL361 and R. palustris BAL398) have been documented to occur at densities similar to those of other dominant diazotrophic groups (20), and close relatives of all three appear to be widely distributed in marine waters. Pseudomonas-like gammaproteobacterial nifH sequences are prevalent and often dominating in sequence data sets from a multitude of marine environments, including estuarine waters and the Pacific, Atlantic, and Indian Oceans (13, 24). Alphaproteobacterial diazotrophs of the order Rhizobiales, including the genus Rhodopseudomonas, are integral parts of the diazotrophic communities not only in the Baltic Sea (20, 24) but also in the Mediterranean Sea (25), South China Sea (26), and the Atlantic and Indian Oceans (13). The third isolate (R. ornithinolytica BAL286) represents the Enterobacteriaceae. This family has been found to dominate nifH gene sequence libraries from temperate estuarine surface waters in the summer season (24), and diazotrophic members of the genus Klebsiella have been recovered from estuarine eel grass root-associated communities (27). Thus, the three bacterial isolates collectively represent a subset of widely distributed and potentially important heterotrophic diazotrophs. Furthermore, we report data on nitrogenase activity in these bacteria as a function of different concentrations of O₂, dissolved inorganic N (NH₄⁺ and nitrate [NO₃⁻]), and dissolved organic C. Lastly, we evaluate the applicability of relative nifH transcript abundances as a proxy for N₂ fixation in the isolates examined.

RESULTS AND DISCUSSION

Genomic traits.

The genomes of P. stutzeri BAL361, R. palustris BAL398, and R. ornithinolytica BAL286 were assembled into 398, 906, and 816 contigs, respectively. Based on these contigs, P. stutzeri BAL361 has a genome of 4.88 Mb comprising 4,514 coding sequences (CDSs), whereas R. palustris BAL398, and R. ornithinolytica BAL286 both have 6.13-Mb genomes comprising 5,860 and 5,720 CDSs, respectively (Table 1).

TABLE 1 .

Comparison of selected genomic features for the Baltic Sea isolates and reference genomes of “Candidatus Pelagibacter” and F. taffensis

Genomic feature	P. stutzeri BAL361	R. palustris BAL398	R. ornithinolytica BAL286	“Candidatus Pelagibacter” sp. IMC9063	F. taffensis DSM16823
Genome size (Mb)	4.88	6.13	6.13	1.28	4.63
No. of coding sequences	4514	5860	5720	1439	4132
% GC content	63	65	56	30	37
No. of RNA genes	57	49	85	35	44
No. of mobile genetic elements	4	9	16		6
Aromatic compound metabolism subsystem (no. of defined features)
Quinate degradation	1	1	4	1	1
Biphenyl degradation	3
n-Phenylalkanoic acid degradation	13
Benzoate degradation	9		4
Chloroaromatic degradation	4	5	5
Catechol branch of beta-ketoadipate pathway	8	9	8		2
Salicylate and gentisate catabolism	3	7	5	3	2
Homogentisate pathway of aromatic compound degradation	8	15	11		3
Salicylate ester degradation		2	1	1
N-heterocyclic aromatic compound degradation		2
Aromatic amine catabolism		5	7
Protocatechuate branch of beta-ketoadipate pathway			14
4-Hydroxyphenylacetic acid catabolic pathway			15
Central metacleavage pathway of aromatic compound degradation			6
Gentisate degradation				2	1
p-Hydroxybenzoate degradation		2	4
Total	49	48	84	7	9
Nitrogen metabolism subsystem (no. of defined features)
Nitrogen fixation	22	22	22
Denitrification	22	12	9
Nitrate and nitrite ammonification	26	9	20
Ammonia assimilation	13	19	12	6
Dissimilatory nitrite reduction	13
Nitrosative stress	5	2	6		2
Cyanate hydrolysis	4	8
Total	105	72	69	6	2

Consistent with the literature on the well-described P. stutzeri and R. palustris species (28–30), the genome sequences of P. stutzeri BAL361 and R. palustris BAL398 showed evidence of high metabolic versatility, including, for instance, metabolic subsystems related to diverse carbohydrate and fatty acid metabolisms, as well as RuBisCO-catalyzed C fixation and phototrophy in R. palustris BAL398. In contrast, R. ornithinolytica is not very well described, but it too seemed to be versatile, harboring genes relating to a multitude of catabolic subsystems, including mono-, di-, oligo-, and polysaccharide utilization and fermentation. All three isolates seem to have allocated a substantial part of their genomes to aromatic hydrocarbon metabolism, with 49, 48, and 84 genomic features relating to this metabolic subsystem in P. stutzeri BAL361, R. palustris BAL398, and R. ornithinolytica BAL286, respectively (Table 1). Both P. stutzeri (see reference 28 and references therein) and R. palustris (31) can use aromatic compounds as sole C sources. Little is known about the aromatic hydrocarbon metabolism of R. ornithinolytica, but it has been isolated from oil-contaminated soil (32) and can grow with benzoic acid as the sole C source (33). We can only speculate on how this relates to the ecology of these organisms and whether it influences their role as diazotrophs, but we note that the genetic capacity to degrade aromatic compounds is widespread in Baltic Sea bacterioplankton (34) and that aromatic compounds may be particularly prevalent in waters influenced by river outflow (35), like the Baltic Sea.

The number of N metabolism-related features also suggests a high degree of versatility in this metabolic subsystem in all three isolates compared to the nondiazotrophic reference strains “Candidatus Pelagibacter” sp. strain IMC9063 and Fluviicola taffensis DSM16823 (Table 1). In addition to N₂ fixation and ammonium assimilation, all three isolates seemed to be able to respire NO₃⁻ through ammonification, as well as denitrification. If these genetic subsystems are used in situ, these organisms can conceivably switch between being a source and a sink of N depending on the growth conditions. Taken together, the diverse potential metabolic strategies observed among the isolates indicate that high metabolic flexibility is a key trait in heterotrophic, N₂-fixing bacteria inhabiting Baltic Sea surface waters.

Nitrogenase reductase gene (nifH) sequences were obtained from P. stutzeri BAL361 and R. palustris BAL398 in connection with their isolation (20). Upon the isolation of R. ornithinolytica BAL286, an iron-only (FeFe) nitrogenase-associated reductase gene (anfH) sequence, similar to those known from Azotobacter vinelandii and Azomonas macrozytogenes (18), was recovered, but not a conventional molybdenum-iron (FeMo) nitrogenase-associated reductase gene (nifH) sequence. In congruence with the observation that alternative nitrogenases seem to complement the FeMo nitrogenase when Mo is depleted (36), rather than being autonomously regulated, a FeMo nitrogenase-associated reductase-encoding gene was recovered from the genome sequence of R. ornithinolytica BAL286. It was related to the nifH gene from Klebsiella pneumoniae (92% nucleotide sequence similarity; GenBank accession number V00631). R. palustris has previously been reported to harbor alternative nitrogenases (29), but this was not the case for R. palustris BAL398, confirming a previous PCR-based analysis of this strain (20).

The nifH and anfH gene sequences suggested that the isolates represented the alphaproteobacterial and gammaproteobacterial parts of the canonical nifH cluster I (37), as well as the archeal and FeFe nitrogenase-containing cluster II (Fig. 1).

FIG 1 .

Amino acid sequence-based, unrooted, neighbor-joining tree showing the affiliations of the isolates with the canonical nifH clusters (37). Color-coded lines indicate cluster affiliations as shown in the key. Pseudomonas stutzeri BAL361 represents the gammaproteobacterial part of cluster I, Rhodopseudomonas palustris BAL398 represents the alphaproteobacterial part of cluster I, and R. ornithinolytica BAL286 represents the gammaproteobacterial part of cluster I, as well as cluster II.

The genetic information related to N₂ fixation was clustered in distinct regions in all three genomes. These regions contained genes encoding the nitrogenase complexes, as well as genes involved in FeMo cofactor synthesis and transcriptional regulation. In P. stutzeri BAL361, this region consisted of 49.3 kb encoding 61 genes (Fig. 2A). This region had a GC content considerably higher than that of the rest of the genome (66.2% versus 62.6%), supporting the notion that the N₂ fixation region in P. stutzeri is part of a genomic island acquired by horizontal gene transfer (HGT) (38). Several CDSs were interspersed between the genes identified as being part of the N₂ fixation subsystem, contributing to the rather large size of the N₂ fixation region in this genome. Despite their unknown function, these CDSs seem to be conserved among organisms fixing N₂ under microaerobic and aerobic conditions, and their transcription is increased under conditions favoring N₂ fixation, suggesting that they do play a role in aerobic N₂ fixation (38).

FIG 2 .

N₂ fixation gene clusters of Pseudomonas stutzeri BAL361 (A), Rhodopseudomonas palustris BAL398 (B), and Raoultella ornithinolytica BAL286 (C). Color-coded arrows indicate the locations of coding sequences (CDSs) and their orientations. The GC contents are depicted beneath the gene clusters (blue, GC content below cluster average; green, GC content above cluster average).

The N₂ fixation region of R. palustris BAL398 was in general similar to the composition of the FeMo N₂ fixation cluster of R. palustris strain CGA009 (29), comprising 27 CDSs within a single 22.0-kb region and lacking the negative transcriptional regulator-encoding gene, nifL (Fig. 2B). And yet, no genes related to alternative nitrogenase complexes were observed, supporting the idea that these may have been acquired recently by HGT in R. palustris CGA009 (39) or lost secondarily in R. palustris BAL398.

Despite containing the genetic information of both the FeMo and the FeFe nitrogenase complexes, the N₂ fixation region of R. ornithinolytica BAL286 constituted only 28.8 kb and comprised 29 CDSs. It was hence the most condensed N₂ fixation region of the three (Fig. 2C). The part of the region comprising the FeMo nitrogenase complex was similar to that of Klebsiella pneumoniae strain M5a1 (40), but following the nifQ gene, the FeFe complex was encoded in the opposite transcription orientation. Bacteriophage-associated sequences were found just upstream from the FeFe nitrogenase gene cluster, and a distinct change in GC content occurred at the intersection between the FeMo and FeFe clusters (Fig. 2C). It seems, therefore, likely that the alternative nitrogenase complex of R. ornithinolytica BAL286 has been acquired horizontally, as in P. stutzeri BAL361, in this case by bacteriophage-mediated transfer.

It is generally believed that the congruence between nifH gene phylogeny and 16S rRNA gene phylogeny is an indication that HGT of entire nitrogenase gene clusters is a rare event (3). Transfer events within phylogenetic groupings would, however, not affect phylogenetic congruence and it seems to have happened in Pseudomonas stutzeri, as well as in other diazotrophs (39, 41, 42). R. palustris CGA009 harbors alternative nitrogenase gene clusters, which seem to have been acquired through HGT (39). This indication of an HGT event is consistent with the fact that our R. palustris BAL398 strain does not harbor any alternative nitrogenase gene clusters. In combination with our deduction that R. ornithinolytica BAL286 has acquired an FeFe nitrogenase system horizontally, we speculate that alternative, accessory nitrogenase systems may be particularly prone to HGT.

Collectively, our results show that the genetic regions comprising the N₂ fixation clusters in the three organisms examined differ in structure, composition, and complexity, presumably reflecting different ecophysiologies of the organisms. Furthermore, it appears that the genetic clusters are in a state of flux, continuously being influenced by HGT.

Growth kinetics, N₂ fixation potential, and regulation.

To ensure that C₂H₄ production in the acetylene reduction assays (ARAs) was measured on actively growing cultures, the growth of the three isolates was monitored over a 20-day period. In ARAs, the acetylene was added after 14 days, when all three isolates were still actively growing (Fig. 3). Growth under standard conditions, i.e., in carbonate-buffered microoxic diazotroph medium devoid of reactive N and containing 37 ± 4 µmol O₂ liter⁻¹ and 20 mmol glucose liter⁻¹, yielded rather slow growth with generation times of 15, 4, and 30 days for P. stutzeri BAL361, R. palustris BAL398, and R. ornithinolytica BAL286, respectively. These rates are much lower than growth by marine bacterioplankton, which generally exhibit generation times of 9 to 18 h (43).

FIG 3 .

Growth of Pseudomonas stutzeri BAL361 (A), Rhodopseudomonas palustris BAL398 (B), and Raoultella ornithinolytica BAL286 (C) in carbonate-buffered microoxic diazotroph medium under standard conditions (20 mmol glucose liter⁻¹, 37 ± 4 µmol O₂ liter^{− 1}). Cell counts were obtained from each of the three cultures daily and are depicted from day 5 to 17 as means of values obtained within 48-h periods. Error bars represent standard deviations of the means. Pink bars indicate the 24-h period when cells were exposed to acetylene in the acetylene reduction assay.

Growth in parallel cultures supplemented with dissolved inorganic N (60 µmol NO₃⁻ or NH₄⁺ liter⁻¹) was also monitored. Cell counts showed that the addition of NO₃⁻ and NH₄⁺ decreased the generation times to 3 and 19 h for P. stutzeri BAL361, 9 and 22 h for R. palustris BAL398, and 22 and 19 h for R. ornithinolytica BAL286, respectively. This suggested that the slow growth in the cultures devoid of dissolved inorganic N was due to N limitation.

When fixing N₂ under oxic conditions, aerobic and facultative aerobic microorganisms are faced with the problem of synthesizing ATP from oxidative phosphorylation while at the same time protecting the nitrogenase complex from O₂ inhibition. Azotobacter vinelandii can fix N₂ under aerobic conditions, and soil-derived Pseudomonas stutzeri strains are known to fix N₂ microaerobically (44). The genomic features of the nitrogenase gene cluster of P. stutzeri BAL361 seem to support the idea that this organism is capable of microaerophilic or aerobic N₂ fixation as well (38). Indeed, the growth of this isolate was stimulated by an increased O₂ concentration at the time of inoculation, with a generation time of 3.6 days compared to 15 days under low initial O₂ concentrations (Fig. 3A versus 4A). ARA showed that ethylene (C₂H₄) was produced at 4 to 165 µmol O₂ liter⁻¹, with the highest cell-specific C₂H₄ production rates, 0.0034 fmol C₂H₄ cell⁻¹ h⁻¹, observed at 165 µmol O₂ liter⁻¹ (Fig. 4D). Interestingly, P. stutzeri BAL361 formed aggregates 1 to 4 mm in diameter when grown under oxic conditions, as indicated by the resazurin dye signal (see Fig. S1 in the supplemental material), suggesting that these bacteria excrete extracellular polymeric substances in order to control O₂ diffusion and facilitate N₂ fixation in an oxic environment. While this is consistent with a previous observation (19), the frequency of this trait in indigenous bacterioplankton is unknown.

FIG 4 .

Growth of Pseudomonas stutzeri BAL361 (A), Rhodopseudomonas palustris BAL398 (B), and Raoultella ornithinolytica BAL286 (C) in an oxygenated version of the growth medium (≈200 µmol O₂ liter⁻¹ at the time of inoculation). Cell counts were obtained from each of the three cultures daily and are depicted from day 5 to 17 as means of values obtained within 48-h periods. Error bars represent standard deviations of the means. (D) Cell-specific ethylene (C₂H₄) production rates as a function of the O₂ concentrations measured in the cultures at the time of C₂H₄ quantification. Horizontal and vertical error bars, where present, indicate the standard deviations of the mean O₂ concentration values measured at the time of C₂H₄ quantification and of the mean C₂H₄ production values, respectively.

R. palustris BAL398 and R. ornithinolytica BAL286 showed limited or no growth at approximately 200 µmol O₂ liter⁻¹ (Fig. 4B and C); however, low-O₂ conditions stimulated N₂ fixation in these isolates (Fig. 4D). C₂H₄ production was low (≈0.001 fmol C₂H₄ cell⁻¹ h^{− 1}) in the 0 to 5 µmol O₂ liter⁻¹ range but increased at 38 µmol O₂ liter⁻¹ to 0.0067 fmol C₂H₄ cell⁻¹ h⁻¹ in R. palustris BAL398. The facultative anaerobic R. ornithinolytica BAL286 displayed the same overall pattern, reaching a peak C₂H₄ production rate of 0.0058 fmol C₂H₄ cell⁻¹ h⁻¹ in the presence of 14 µmol O₂ liter⁻¹, supporting observations that N₂ fixation in enterobacterial diazotrophs is stimulated by the presence of low concentrations of O₂ (45, 46). Hence, these diazotrophs are likely faced with the challenge of balancing the detrimental effects of O₂ on the nitrogenase complex with the need to produce sufficient energy to fuel it.

Increasing C concentrations were found to stimulate N₂ fixation in P. stutzeri BAL361. C₂H₄ production increased steadily from <0.001 fmol C₂H₄ cell⁻¹ h⁻¹ in cultures supplemented with 0.020 µmol glucose liter⁻¹ to 0.089 fmol C₂H₄ cell⁻¹ h⁻¹ in cultures with 2,000 µmol glucose liter⁻¹ (Fig. 5). A decrease in C₂H₄ production at 20,000 µmol glucose liter⁻¹ coincided with a change in the resazurin dye signal, indicating that the increased respiration had reduced the O₂ concentration to a level impairing N₂ fixation. Hence, interactions between factors controlling N₂ fixation may complicate the interpretations made from culture-based estimates of N₂ fixation. C₂H₄ production was only observed at the highest C concentration (20,000 µmol glucose liter^{− 1}) for R. palustris BAL398. Here, cell-specific C₂H₄ production rates reached 0.0044 fmol C₂H₄ cell⁻¹ h⁻¹, suggesting that this organism required more energy to sustain N₂ fixation. R. palustris is known to produce substantial amounts of H₂ when fixing N₂ (29, 47), implying a significant waste of ATP (45). This could potentially be overcome by this photoheterotroph by acquiring energy from anoxygenic photosynthesis, as suggested for Rhodopseudomonas capsulata (48), but we chose not to include light as a factor in these assays to reduce complexity. R. ornithinolytica BAL286 did increase C₂H₄ production in response to increasing glucose concentrations, and yet the cell-specific rates dropped, suggesting that other factors controlling N₂ fixation interfered in these treatments.

FIG 5 .

Cell-specific ethylene (C₂H₄) production rates measured in triplicate serum vials as a function of eight glucose concentrations. Bulk C₂H₄ production did increase in R. ornithinolytica BAL286 cultures as a function of increasing glucose concentrations (not shown), but the higher number of cells meant a decrease in the cell-specific rates. Error bars indicate standard deviations of values from triplicate serum vials. Note the double logarithmic scale.

Due to the high energy consumption by the N₂ fixation reaction, it is generally believed that N₂ fixation shuts down in response to available reactive N. All three isolates decreased N₂ fixation in response to increasing NO₃⁻ concentrations, with the C₂H₄ production rates dropping 1 to 2 orders of magnitude (Fig. 6A). N₂ fixation decreased significantly in response to increasing NH₄⁺ concentrations in P. stutzeri BAL361 and R. ornithinolytica BAL286 cultures as well, but in R. palustris BAL398 cultures, N₂ fixation increased dramatically upon the addition of 5 µmol NH₄⁺ liter⁻¹ or more (Fig. 6B), reaching the highest cell-specific C₂H₄ production rates measured in any of the ARAs (0.515 fmol C₂H₄ cell⁻¹ h⁻¹). The cell-specific C₂H₄ production rates were not significantly different between P. stutzeri BAL361 and R. ornithinolytica BAL286 in the presence of NH₄⁺ (P = 0.90), whereas the C₂H₄ production rates were significantly higher in R. palustris BAL398 cultures (P = 0.01). We therefore speculate that the nitrogenase complex has a function in addition to N acquisition in R. palustris BAL398. In fact, it was recently proposed that R. palustris may use CO₂ and N₂ fixation to maintain a balanced redox state when growing on lactate, which has the same oxidation/reduction value as glucose, which was used in this study (49). Hence, R. palustris BAL398 likely uses N₂ fixation as an electron sink when grown at suboxic conditions on glucose and NH₄⁺. This falls in line with the many spurious findings of marine N₂ fixation at NH₄⁺ concentrations of up to 200 µmol liter⁻¹ (8, 50) and highlights that regulation of N₂ fixation by reactive N is presently far from understood.

FIG 6 .

Cell-specific ethylene (C₂H₄) production rates measured in triplicate serum vials as a function of eight different concentrations of NO₃⁻ (A) and NH₄⁺ (B). (A) C₂H₄ production rates dropped 1 to 2 orders of magnitude for all three isolates when exposed to even low NO₃⁻ levels. (B) C₂H₄ production rates dropped for R. ornithinolytica BAL286 and P. stutzeri BAL361, while they increased significantly for R. palustris BAL398 compared to the rates for the other isolates when exposed to NH₄⁺ (P = 0.01). Error bars represent the standard deviations of values from triplicate serum vials, and lowercase letters indicate Tukey’s honestly significant difference (HSD) groupings. Note the logarithmic scale on the y axis.

To investigate whether nitrogenase activity was regulated on the transcriptional level, cell-specific nifH and anfH mRNA transcript abundances were determined by extracting RNA from the same cultures as those used in the ARAs at the point of C₂H₄ quantification and subsequently running reverse transcriptase quantitative PCRs (RT-qPCRs). The transcript abundances were then normalized to the flow cytometry-determined cell numbers. It was, however, not possible to identify any clear patterns in transcript abundances, and the correlations between cell-specific nitrogenase reductase gene transcript abundances and cell-specific C₂H₄ production rates were never significant (Spearman’s rank-order correlation test, α = 0.05) (see Table S1 in the supplemental material). Hence, posttranscriptional regulation is likely important here. It has previously been reported that the nitrogenase enzyme complex of Rhodospirillum rubrum, among others, is subject to posttranscriptional modifications that stringently regulate enzyme activity by covalent attachment of ADP-ribose moieties to the Fe protein (51–53). This reversible ADP-ribosylation inhibits the association of the Fe protein with the MoFe protein and acts as a second layer of control on the energy-consuming reduction process. Hence, nifH and/or anfH transcript numbers are likely not good proxies for N₂ fixation under the circumstances tested here. Furthermore, normalization to the transcripts of housekeeping genes may give a better indication of the relative levels of transcription of nifH and anfH genes, as flow cytometry data also include inactive cells and cells containing multiple chromosomes.

With some exceptions (R. palustris BAL398 in the presence of NH₄⁺), N₂ fixation was highest in cultures where O₂ concentrations were reduced (5 to 165 µmol liter⁻¹), C concentrations were high (2,000 to 20,000 µmol liter⁻¹), and reactive N was absent. Converting C₂H₄ production to N₂ fixation in these cultures using the theoretical conversion factor of 3 (54), the heterotrophic isolates fixed N₂ at rates of 0.0007 to 0.03 fmol N cell⁻¹ h⁻¹. These numbers are similar to but generally higher than the rates used in previous estimates of the contribution by heterotrophic diazotrophs to N₂ fixation in situ (13). R. palustris BAL398 and a Pseudomonas-like isolate similar to P. stutzeri BAL361 have previously been quantified at abundances of up to 4.7 × 10⁴ and 7.9 × 10⁴ cells liter⁻¹, respectively (20). Applying the measured rates, cells at these densities could contribute with N₂ fixation rates of 0.790 to 56.9 pmol N liter⁻¹ day⁻¹.

Converting the C₂H₄ production rates obtained from R. palustris BAL398 cultures subjected to NH₄⁺ additions to N₂ fixation rates as described above, significantly higher N₂ fixation rates are obtained (0.17 fmol N cell⁻¹ h⁻¹). Hence, under conditions low in O₂ and replete in C and NH₄⁺, for instance in association with fecal pellets or aggregates (55), the observed abundances of R. palustris BAL398 (4.7 × 10⁴ cells liter⁻¹ [20]) could account for N₂ fixation rates of 192 pmol N liter⁻¹ day⁻¹. Interestingly, Guerinot and Colwell (56) reported that N₂ fixation by diazotrophic isolates in seawater incubations was only quantifiable when particulate material was present, supporting the idea that these may be hot spots for N₂ fixation despite being replete in reactive N. In addition, actively N₂-fixing photosynthetic bacteria have been found to be associated with copepods in the Caribbean Sea (57) and, recently, N₂-fixing heterotrophs in the particulate size fraction of >10 µm were shown to be more abundant in oxygenated water than in suboxic waters (19).

The rates measured for diazotrophic isolates are low relative to the peak rates of estuarine surface waters (47 to 83 nmol liter⁻¹ day⁻¹ [24]) but are comparable to the rates obtained from deeper estuarine waters (0.44 nmol liter⁻¹ day⁻¹ [8]) and the lower end rates measured in Pacific or Atlantic surface waters (0.24 to 35 nmol liter⁻¹ day⁻¹ [58, 59]). The rates obtained for individual isolates under optimal laboratory conditions are likely much higher than the rates would be per cell in situ; however, it should be kept in mind that a bulk in situ rate presumably represents the commensal activity of multiple diazotrophic species in the sample. With that in mind, our reported rates for diazotrophic isolates indicate that, under the right conditions, heterotrophic diazotrophs have the potential to contribute significantly to in situ rates.

Collectively, our study documents that cultivable diazotrophs from temperate estuarine waters exhibit great metabolic versatility and that the genomic N₂ fixation-related features are conserved and yet distinct between organisms, possibly reflecting the physiology of the organism. Furthermore, our data indicate that the gene clusters encoding the nitrogenase complexes are influenced by HGT events and are in a constant state of flux, facilitating adaptation to changing environmental conditions through the introduction of alternative nitrogenases and shuffling of the genetic contexts. The high level of cell-specific N₂ fixation observed under some conditions in this cultivation-based study suggests that heterotrophic bacteria may, at least occasionally, contribute significantly to overall N₂ fixation in temperate estuarine waters, given their high in situ abundances (e.g., see references 20, 24, and 29). Regulation of N₂ fixation by O₂, C, or reactive N was not straightforward and in some cases counterintuitive. Hence, making predictions about the occurrence and scale of N₂ fixation in different environments based on such parameters is difficult. Due to these discrete mechanisms of regulation in response to different environmental factors, coupling genomic features and ecophysiological characteristics in multiple and diverse representative isolates seems to be necessary in order to make better community-level generalizations about N₂ fixation.

MATERIALS AND METHODS

The bacteria being investigated were isolated from surface waters in the Baltic Sea. P. stutzeri BAL361 (nifH nucleotide accession number KC140355) and R. palustris BAL398 (nifH nucleotide accession number KC140365) were isolated from carbonate-buffered, microoxic diazotroph enrichment cultures devoid of N that were inoculated with surface water (3 m; total depth, 459 m) from the Landsort Deep (58°36′N, 18°14′E) in March 2009 (20). R. ornithinolytica BAL286 (nifH nucleotide accession number AY972875) was isolated from a depth of 3 m (total depth, 10 m) at a station in the strait between mainland Sweden and the island of Öland (56°37′N, 16°21′E) in April 2005 using semisolid diazotroph medium (18).

Genome sequencing and comparisons.

The isolates were grown in ZoBell broth (60) to an optical density of 0.5 to 1.0 and harvested, and genomic DNA was extracted (EZNA tissue DNA kit; Omega Bio-Tek, Norcross, GA, USA). DNA from P. stutzeri BAL361 and R. palustris BAL398 was sheared using a Bioruptor (Diagenode, Liege, Belgium), and a paired-end (PE) sequencing library (~450-bp inserts) for Illumina was constructed as previously described (61) using short indexing primers (62). To generate the 100-nucleotide (nt) PE data, approximately 1/10 of one lane on an Illumina HiSeq 2000 was run. Sequences were assembled using Velvet de novo assembler version 1.2.08 with scaffolding switched off and a k-mer of 47. For R. ornithinolytica BAL286, a sequencing library was built using the Nextera XT DNA kit (Illumina, San Diego, CA) with 1 ng input DNA. This library was sequenced on the Illumina MiSeq platform as 2 × 250-bp paired-end reads. This genome sequence was assembled using CLC Genomic workbench 6.0.4 (CLC bio, Aarhus, Denmark) following standard quality trimming and adapter removal using the same software package. For P. stutzeri BAL361 and R. palustris BAL398, the N₂ fixation regions were reassembled and mapped against known reference genomes using the CLC platform. Genomes were annotated using the NCBI Prokaryotic Genome Annotation Pipeline (PGAP) and the online Rapid Annotation using Subsystem Technology (RAST) resource version 2.0 (63). The annotations of the N₂ fixation regions were revised manually using Artemis (64). The genomic features of the three sequenced isolates were compared to the genomes of two members of prevalent marine bacterial groups, i.e., (i) “Candidatus Pelagibacter” sp. IMC9063 (GenBank accession number CP002511), a SAR11 clade member related to prevalent phylotypes recovered from Baltic Sea surface waters (65) and also exhibiting 99% 16S rRNA gene similarity to a sequence recovered from Baltic Sea surface water throughout 2003 (GenBank accession number DQ270271) (66), and (ii) the genome of Fluviicola taffensis DSM16823 (GenBank accession number NC_015321), a member of the Bacteroidetes phylum exhibiting 92% 16S rRNA gene similarity to a sequence recovered from Baltic Sea surface waters during the summer and early autumn of 2003 (GenBank accession number DQ270281) (66).

Medium preparation and growth kinetics.

To investigate the N₂ fixation potential of the isolates, various growth media were tested, including the liquid N-free medium previously applied to grow R. ornithinolytica BAL286 (18), diazotroph medium RBA (DSMZ), and the carbonate-buffered microoxic diazotroph medium used to isolate P. stutzeri BAL361 and R. palustris BAL398 (20). Only the latter medium supported the growth of all three isolates and was therefore used in the subsequent ecophysiological analyses.

The growth kinetics of the isolates were examined to ensure that acetylene reduction was assessed for growing cells. First, aliquots (25 ml) of medium were distributed into 50-ml borosilicate serum vials, which were capped with butyl rubber stoppers and crimp sealed. To ensure a low O₂ concentration (37 ± 4 µmol O₂ liter⁻¹) in the medium prior to inoculation, the headspaces were replaced three consecutive times with an atmosphere containing 90% N₂, 5% CO₂, and 5% H₂ and the liquid was allowed to equilibrate with the modified atmosphere for 2 h before the O₂ concentrations were verified using a FireStingO₂ optical O₂ meter equipped with an OXR50 fiber-optic O₂ sensor (Pyroscience, Aachen, Germany). The medium was then supplemented with a 0.2 μm-filtered glucose solution (20 mmol liter⁻¹ final concentration). Second, for each isolate, cells were harvested from 2 ml of actively growing ZoBell cultures (4,000 × g for 5 min), washed twice in 1 ml phosphate-buffered saline (PBS), and suspended in 2 ml of the carbonate-buffered microoxic diazotroph medium. Serum vials containing 25 ml medium were then inoculated with 200 µl cell suspension and incubated for approximately 20 days in the dark at room temperature with shaking at 150 rpm. Third, growth in the vials was monitored by flow cytometry as described previously (67) using a FACSCanto II flow cytometer (BD Biosciences, NJ, United States).

In order to investigate the effects of lower glucose concentrations and higher O₂ concentrations on the bacterial growth, parallel incubations containing 0.2 µmol glucose liter⁻¹ or ≈200 µmol O₂ liter⁻¹ were included. The effect of reactive N on the growth of the bacteria was examined by adding 60 µmol NH₄⁺ or NO₃⁻ liter⁻¹ (final concentration) to a subset of incubations. This was added on day 12 as in the acetylene reduction assay (see below).

Acetylene reduction assay.

Nitrogenase activity was assessed using the acetylene reduction assay (68). Serum vials containing 25 ml carbonate-buffered microoxic diazotroph medium were inoculated and incubated as described above. After 14 days, the pressure was equilibrated and 10% of the headspace was replaced with laboratory-grade acetylene gas (C₂H₂; Air Liquide, Taastrup, Denmark). Incubations continued for 24 h before the C₂H₄ produced was quantified using a flame ionization detector (FID)-equipped CP9000 gas chromatograph (Chrompack, Bergen op Zoom, Netherlands). The base medium for these measurements contained 20 mmol glucose liter⁻¹ and 37 ± 4 µmol O₂ liter⁻¹, and was free of reactive N. To examine the regulation of N₂ fixation by different concentrations of C, O₂, and reactive N, one of these parameters was changed while the other two were kept constant. The effect of O₂ on N₂ fixation was investigated by regulating the O₂ concentrations of the base medium by adding 2, 4, or 8 ml of pure O₂ gas or by adding ferrous sulfate (FeSO₄) and dithiothreitol (DTT) in equivalent amounts (0.40, 0.60, 1.0, or 1.6 mmol liter⁻¹, final concentration of each compound) to triplicate vials. This produced O₂ concentrations in the medium ranging from 0 to 240 µmol O₂ liter⁻¹ at the time of C₂H₄ quantification. The effect of C was examined using base medium made with high-performance liquid chromatography (HPLC)-grade water (Sigma) instead of MilliQ. Triplicate vials contained one of eight concentrations increasing from 0.002 to 20,000 µmol glucose liter⁻¹ with log factor increments. Similarly, the effect of reactive N was examined using the following eight concentrations of either NO₃⁻ or NH₄⁺: 0, 0.1, 0.5, 1, 5, 20, 60, or 150 µmol liter⁻¹. The reactive N was added to triplicate vials on day 12 to ensure that the cultures did not reach stationary phase at the time of acetylene addition (day 14).

Nitrogenase reductase gene expression analyses.

To couple nitrogenase activity to nitrogenase reductase gene expression, cell-specific nifH and anfH RNA transcript numbers in the different treatments were determined. Following C₂H₄ production measurements, 2 ml of culture from one of each of the different treatments was centrifuged at 8,000 × g for 5 min. The pellet was dissolved in 200 µl RNAlater (Ambion; Life Technologies) and kept at −80°C until extraction. RNA was extracted using the RNeasy minikit (Qiagen) with an additional DNase treatment applied after elution. Complete digestion of DNA was verified by PCR. cDNA was synthesized using TaqMan reverse transcription reagents (Applied Biosystems) and the reverse primer nifH3 (4). Primers were designed to target the nifH genes of the three isolates and the anfH gene of R. ornithinolytica BAL286 using Primer3 (version 0.4.0, online resource) and were checked for hairpins and dimers using NetPrimer (Premier Biosoft). Twenty-microliter qPCR mixtures were made containing 1× SYBR Select master mix (Life Technologies Europe BV), 300 nM of each primer, RT-PCR-grade water, and 2 µl template. The reactions were run on an Agilent Mx3005P qPCR thermal cycler using the following temperature settings: 50°C for 2 min, 95°C for 2 min, and 40 cycles of 95°C for 15 s and 60°C for 1 min, followed by a melt curve analysis to check for unspecific PCR products. No PCR products other than the ones for the target genes were detected. Lastly, cell-specific gene expression was calculated based on the flow cytometry data obtained from the corresponding cultures.

Accession numbers.

The genome sequences were deposited as whole-genome shotgun projects at GenBank under accession numbers JXXD00000000 (Pseudomonas stutzeri BAL361), JXXE00000000 (Rhodopseudomonas palustris BAL398), and JXXF00000000 (Raoultella ornithinolytica BAL286).

SUPPLEMENTAL MATERIAL

Figure S1 

Seven focus-stacked images of a Pseudomonas stutzeri BAL361aggregate from an oxygenated culture (≈200 µmol O₂ liter⁻¹ at the time of inoculation) devoid of reactive N. The aggregate was positioned under a light microscope in a 96-well flat-bottom plate. Download

Table S1 

Spearmann’s rank-order correlations between cell-specific nifH and anfH mRNA transcript abundances and cell-specific C₂H₄ production rates. r is the correlation coefficient, and P is the significance level.