Abstract
Ammonia-oxidizing archaea (AOA) are one of the most abundant groups of microbes in the world’s oceans and are key players in the nitrogen cycle. Their energy metabolism, the oxidation of ammonia to nitrite, requires oxygen. Nevertheless, AOA are abundant in environments where oxygen is undetectable. In incubation experiments, where oxygen concentrations were resolved to the nanomolar range, we show that Nitrosopumilus maritimus produces oxygen (O2) and dinitrogen (N2). The pathway is not completely resolved, but it has nitric oxide as a key intermediate. Part of the oxygen produced is directly used for ammonia oxidation, while some accumulates in the surrounding environment. N. maritimus joins a small handful of organisms known to produce oxygen in the dark, and based on this ability, we re-evaluate their role in oxygen-depleted marine environments.
Main Text
Ammonia-oxidizing archaea (AOA) are only known to oxidize ammonia to nitrite using oxygen: NH3 + 1.5 O2 → NO2- + H2O + H+ (1, 2). Yet, AOA are highly abundant in environments with very low or even undetectable oxygen concentrations such as marine oxygen-minimum zones (OMZs) (3–5). The role of AOA in such environments is enigmatic as they have no known anaerobic metabolism. We used trace luminescence oxygen sensors (hereafter: optodes) (6) to explore the physiology of AOA at low nanomolar oxygen concentrations and functional anoxia (oxygen levels below detection) as typically found in OMZs (7, 8), discovering oxygen production by the marine AOA Nitrosopumilus maritimus SCM 1(9). Dark, non-photosynthetic, oxygen production is rare in nature with three known pathways including chlorite dismutation during perchlorate/chlorate respiration (ClO2- → Cl- + O2), detoxification of reactive oxygen species (e.g. H2O2 dismutation) and nitric oxide dismutation (2NO2- → 2NO → N2 + O2) (10). While not fully resolved, we show that the pathway of oxygen production by N. maritimus is none of these and thus novel. Given the abundance of N. maritimus in oxygen-sparse environments, anaerobic oxygen production may be common in nature. We also show that oxygen production is linked to N2 production and thereby identify a previously unknown and potentially environmentally significant N2 production pathway.
We first grew axenic cultures of N. maritimus aerobically as an ammonia oxidizer. The cultures were then sparged with argon to oxygen levels below 5μM, where the remaining oxygen was consumed by N. maritimus through continued ammonia oxidation, indicating physiologically active cells. Surprisingly, after oxygen was completely consumed, it immediately started to increase again (Fig. 1A). A series of additions of oxygen-saturated water showed the same recurring pattern: oxygen was consumed and increased directly thereafter (Fig. 1A). When no oxygen additions were made, oxygen build-up continued over hours and reached levels of 100-200nM (Fig. 1A). This pattern was observed reproducibly in multiple incubations carried out over 2 years.
In comparison, no oxygen build-up was detected in filtered abiotic controls or when cells were killed by the addition of mercuric chloride (Fig. S1) ruling out the possibility of abiotic oxygen production or intrusion of oxygen into the incubation bottle. Contamination by oxygen intrusion was further ruled out with incubations in an anaerobic chamber showing the same trend of oxygen production (Fig. S2). Involvement of medium components (e.g. HEPES, EDTA) in oxygen production was also excluded (Fig. S3), and furthermore, oxygen microelectrodes, which make use of a different oxygen measurement principle, showed the same patterns of oxygen increase as the optode measurements (Fig. S4). Incubations with N. maritimus in medium containing pyruvate showed no difference compared to incubations without pyruvate (Fig. S5). This experiment rules out oxygen production by H2O2 dismutation (H2O2 → H2 + O2) as pyruvate reacts with and removes H2O2 via an abiotic decarboxylation reaction (11).
As described in more detail below, we measured nitric oxide (NO) concentrations with a microelectrode and found that NO slightly interferes with the optode measurements of O2 (Fig. S6) but not the O2 microelectrode measurements. Therefore, when optode O2 concentrations and NO were simultaneously measured, a NO correction was applied to the O2 measurements. The correction, however, was relatively small (0-17%) and fully predictable from the NO concentrations. This correction was only applied when NO and O2 were simultaneously measured, recognizing that other optode O2 measurements may be slight overestimates (depending on the NO concentration) of the actual O2 concentration (Fig. S7).
Despite this small interference on our oxygen measurements, N. maritimus clearly produces oxygen when the culture reaches anoxia. We hypothesize that oxygen accumulates as a net balance of simultaneous oxygen production and oxygen consumption by ammonia oxidation, where plateauing oxygen concentrations over time represent a balance between these processes. To test this hypothesis, cyanide (0.5 mM) was added to the oxygen-producing culture. Cyanide inhibits oxygen respiration by the heme–copper oxygen reductase, and thus inhibits ammonia oxidation (12, supplementary information). Upon cyanide addition, and after an initial lag phase, oxygen concentrations steadily increased at rates ca. 5 times higher (65±12 nmol/L/h) than before cyanide addition (14±2 nmol/L/h) (Fig. 1B). In similar experiments where NO was also measured, O2 increase was uncoupled from NO concentration after cyanide addition (Fig. S8). Overall, these results are consistent with the hypothesis that, in the absence of cyanide, some portion of oxygen production by N. maritimus is utilized within the cells and does not accumulate into the surroundings.
We tracked the conversion of 15N-labelled ammonium to nitrite to directly explore if N. maritimus continues to oxidize ammonia while producing oxygen. In this experiment, cell cultures were washed to reduce the high nitrite background that accumulated (ca. 1 mM) during normal aerobic growth. After this, 15N-ammonium (I: 5 μM and II: 25 μM) was added as well as a small amount of 14N-nitrite to “capture” any produced 15N-nitrite from further transformations. The labelling experiments showed continued ammonia oxidation to nitrite during oxygen production (Fig. 2). These results, consistent with the cyanide addition experiments, confirmed that ammonia oxidation occurs together with oxygen production. Furthermore, the rates of ammonia oxidation in these duplicate experiments were 46 (I) and 39 nM/h (II) (table 1), requiring an oxygen production rate of 69 and 60 nM/h respectively, given the stoichiometry of ammonia oxidation (NH3 + 1.5O2 → NO2- + H2O + 1H+). Oxygen accumulated at an average rate of only 1.2 nM/h in both experiments. Our results further demonstrate that most of the oxygen produced by N. maritimus was immediately consumed through ammonia oxidation. The cell density in these incubations was 1.3*107 cells mL-1, and therefore, the average ammonia oxidation rate per cell was 3-3.5 amol/cell/h. In marine OMZs with typical AOA cell densities of about 1-10*104 cells mL−1 this would correspond to ammonia oxidation rates of 1-10 nM/d, which are in the range of anammox rates in open-ocean oxygen-minimum zones, the so far only known anaerobic ammonia oxidation process that is considered to be relevant in marine OMZs (8).
To explore for translational changes associated with oxygen production, we performed differential proteomic analysis comparing N. maritimus performing standard aerobic metabolism to its metabolism during oxygen production. The proteome translation profile during aerobic metabolism matched earlier findings (14, see supplementary information for details). However, compared to the mid-log phase during normal aerobic growth, 88 of the 1453 recovered proteins had significantly increased abundances when the culture produced oxygen, while 11 proteins were significantly decreased in abundance (P<0.05; table S1).
The plastocyanin Nmar_1665, the multicopper oxidase type 3 Nmar_1354 and the putative nitroreductase Nmar_1357, were among the most up-regulated proteins during oxygen production, suggesting that they could play a role in the oxygen-production pathway (Fig. S9). Indeed, nitric oxide processing has been suggested for the gene cluster Nmar_1352-1357 (13, 14), which would be consistent with the tight coupling of oxygen production and nitric oxide accumulation as explored below. The 17 small blue Cu-containing plastocyanins encoded by N. maritimus have a proposed function in hydroxylamine (NH2OH) oxidation (14). Of these, Nmar_1665 was the only plastocyanin with a significant upregulation under oxygen production, while all other plastocyanins were abundant under both metabolic modes. Other proteins that are either significantly up or down regulated include different transcriptional regulators: e.g members of the AsnC family (Nmar_1524 [up], Nmar_1292 [up], Nmar_1628 [down] (Fig. S9)). Archaeal AsnC regulatory proteins have a function in the regulation of central and energy metabolism, and a role in the switch between aerobic and anaerobic metabolism has been suggested (15, 16). The NADH dehydrogenase 30 kDa subunit Nmar_0278 was significantly down regulated. Otherwise, no proteins with functions in energy or carbon metabolism changed significantly in abundance.
Overall, the few significant changes in the proteome between aerobic respiration and oxygen production suggest that most proteins are needed and active under both metabolic modes. However, some of the upregulated proteins could be involved in catalyzing and regulating oxygen production in AOA.
We now explore possible metabolic pathways for dark oxygen production in N. maritimus. Of the three known pathways of dark oxygen production, we rule out perchlorate/chlorate respiration as our cultures were perchlorate/chlorate/chlorite free. Furthermore, as discussed above, we also rule out hydrogen peroxide dismutation as a source of oxygen. As N. maritimus metabolizes nitrogen and accumulates NO under normal aerobic ammonia oxidation (17), NO dismutation becomes a potential source of oxygen in our experiments. So far, NO dismutation is only known among the NC10 bacteria (10). These organisms are methane oxidizers and generate NO for dismutation to oxygen and dinitrogen (18), where the oxygen is used to oxidize methane. Because oxygen production and consumption are tightly coupled, methane-oxidizing NC10 bacteria are not known to liberate free oxygen into the environment (18).
Using NO microelectrodes, we noted that NO and oxygen production were mostly tightly coupled (Fig. S7) (although the coupling was less tight in other cases; Fig. S8)). Furthermore, when the NO scavenger PTIO (2-Phenyl-4,4,5,5-tetramethylimidazoline-1-oxyl 3-oxide) was added, oxygen production ceased (Fig. S10). Taken together, these results suggest that NO is a crucial intermediate in oxygen production.
We used 15N-labelled nitrite to further unravel the pathways of nitrogen and oxygen cycling during oxygen production by N. maritimus. In incubations with added 15N-nitrite, 30N2 was produced during oxygen production (Fig 3A, S12 and 13), while no formation of 29N2 was detected (Fig. 3B). Dinitrogen production by N. maritimus or other AOA isolates has not previously been reported. Our results furthermore show that both nitrogen atoms in the N2 originated from nitrite, with none coming from ammonium. This result was confirmed by incubations with 15N-ammonium, where no immediate conversion of 15N-ammonium to 29N2 or 30N2 was detected (Fig. 3A, B). Instead, 15N-ammonium was most likely converted to nitrite and diluted into the large existing nitrite pool in this experiment. In contrast, when 15N-labelled ammonium was added to washed cultures with a small nitrite pool (5 and 25 μM), the 15N-nitrite produced from ammonia oxidation was further converted to N2 (Fig. 2B and S11) consistent with our experiments with 15N-labelled nitrite.
Thus far we have shown that NO is a likely intermediate in oxygen production and that N2 is produced together with O2 by N. maritimus. Rates of O2 accumulation and N2 production from the different incubations shown in Figs. 2 and 3 are summarized in table 1. These results would generally be consistent with NO dismutation as a source of both N2 and O2, where NO is produced from the reduction of nitrite. In incubations with added 15N-nitrite, however, oxygen accumulation exceeded N2 production in the first 20 h (incubations 3 and 4 in Table 1, Fig. 3c, S12 and S13), demonstrating a decoupling of O2 and N2 production in this phase of the experiment. As net rates of O2 accumulation may underestimate gross rates of O2 production, as explored above, and as N2 production in our experiments is a gross production rate, there is an imbalance between the production rates of O2 and N2. Such an imbalance would be inconsistent with O2 and N2 production directly from NO dismutation. In this case the O2 accumulation rates should not exceed N2 production rates, but the imbalance we measure suggests that further intermediate(s) must exist between NO, and N2 and O2 production. We suggest that N2O may be such an intermediate, where 2NO → N2O + 0.5O2.
Indeed, in our incubations supplied with 15N-nitrite, 46N2O accumulated before 30N2 production accelerated (Fig. 3D), and the rates of N2 production and N2O accumulation taken together in the first 20 h match the O2 accumulation rates within the uncertainties (table 1). Furthermore, the dismutation of NO(aq) to O2(aq) and N2O(aq) is thermodynamically favorable (ΔG0’= −165 KJ/mol O2,). However, for this pathway to occur, an unknown N2O reductase would need to be present in the genome of N. maritimus (14). N-nitrosating hybrid formation, in which one N atom from NO2- and one from NH4+ (or an intermediate of ammonia oxidation) combine to form N2O, has been proposed as a possible source for N2O production in AOA (19), but the isotopic signature of the N2O produced in our experiments (46N2O) does not support this pathway (expected 45N2O). In incubations supplied with 15N-ammonium and a small nitrite pool (Fig. 2), N2O accumulated transiently as well (Fig. S14). In these incubations N2 production far exceeded O2 accumulation and no N2O accumulation would be required for a mass balance. This is not surprising as high rates of ammonia oxidation (Table 1, Fig 2) indicate that O2 accumulation rates in these experiments far underestimate gross rates of O2 production as explored above. This does not mean that N2O was not an intermediate in these experiments, only that these incubations did not demonstrate an initial imbalance between O2 and N2 production.
To summarize, like for NO dismutation in NC10 bacteria, O2 production in N. maritimus has NO as an intermediate and produces N2. However, unlike for NC10 bacteria, our results suggest that the pathway of O2 production employed by N. maritimus has an extra intermediate that may be N2O. A proposal for the metabolic pathway associated with oxygen production in N. maritimus is shown in figure 4. While ammonia oxidation to nitrite is accomplished by the O2 produced by N. maritimus, the conversion of nitrite to N2 requires reducing equivalents regardless of the O2 production pathway. The required electrons can partly, but not fully, be obtained from the ongoing oxidation of ammonia. Another source of electrons could be intra- or extracellular organic matter produced during normal aerobic ammonia oxidation (20) or, in situ, by dissolved organics available in the water column.
By showing that the O2 production pathway in AOA is coupled to N2 production, we also uncovered a new, potentially environmentally significant, pathway of N2 production. Furthermore, by converting ammonium through nitrite to N2, AOA perform a so far unrecognized pathway of anaerobic ammonia oxidation. 15N-tracer experiments currently performed to measure N-cycling rates in marine oxygen-depleted environments would overlook this pathway and account for its N2 production as canonical denitrification and/or anammox.
AOA are one of the most abundant groups of microbes in the global ocean and key players in the marine nitrogen cycle, also in low-oxygen environments. Thus, a widely distributed oxygen-producing pathway by AOA could have far-reaching implications for the microbial ecology and biogeochemical cycling in oxygen-depleted environments, including the possibility that some of the produced could be used by other microbial cells. The discovery of an anaerobic oxygen-producing pathway in AOA can explain the presence and role of AOA in such environments solving a longstanding enigma.
Funding
This work was supported by the Villum Foundation, Denmark (Villum Young Investigator Grant No. 25491 to BK and Villum Investigator Grant No. 16518 to DEK) and the Independent Research Fund Denmark (Grant No. 14181-00025 to DEC).
Autor contributions
B.K. and D.E.C. designed the experiments. B.K. performed the experiments and analyzed data with input from M.L, L.B., M.K., B.T. and D.E.C.. N.J. performed proteomics. B.K. and D.E.C. wrote the manuscript, with contributions and approval from all other authors.
Competing interests
The authors declare no conflict of interest.
Data and materials availability
All data is available in the main text or the supplementary materials.
Acknowledgments
We thank Annie Glud for assistance with microelectrode measurements and providing microelectrodes.
Footnotes
↵† Villum investigator