Axenic 5 L batch pre-cultures of the AOA strains N. maritimus SCM1 and N. piranensis D3C (JCM 32271, DSM 106147, and NCIMB 15115) were grown at 28°C in the dark in synthetic Crenarchaeota medium (SCM) HEPES-buffered (pH 7.8), as described by Könneke et al. (2005) and Martens-Habbena et al. (2009), modified with a 6 mM final concentration of sodium bicarbonate (Kraft et al., 2022).

The oxygen-depleted incubations were prepared by sparging the aerobically grown batch culture with argon gas (99.99%) for 45 min to reduce the oxygen concentration in the culture. The culture was sterilely transferred into 330-ml custom-made glass bottles designed to avoid oxygen intrusion with a glass capillary and a port for inserting a microsensor (Tiano et al., 2014) through a glass tube connection, using the overpressure generated in the argon-sparged culture bottle. All bottles were filled without headspace and closed with glass stoppers. The bottles were continuously stirred with glass-coated stirring bars (VWR, United Kingdom) at 300 rpm. The bottles were incubated in a water bath at 28°C in the dark. Control incubations with the custom-made bottles and killed controls with HgCl₂ have been previously reported in Kraft et al. (2022), showing no oxygen intrusion from the atmosphere.

Oxygen was monitored constantly during the incubations with trace fluorescence oxygen sensors, also referred to as optodes, with a detection limit of 0.5 nM (Lehner et al., 2015). The optodes were previously glued to the glass bottles. NO was monitored with microsensors (Unisense, Denmark), inserted into the sensor ports of the bottles, which were previously sterilized with 70% ethanol, and rinsed with autoclaved ASTM1a water. NO was observed to cause a small and predictable interference with the optodes (up to 17%). Therefore, the oxygen concentration measurements were corrected for NO interference, as in Kraft et al. (2022). All bottles, stirring bars, tube connections, and materials used for the incubation were previously autoclaved.

Two sets of experiments with ¹⁵N-stable isotope compounds were used for the identification of the intermediates in dinitrogen and oxygen production via NO dismutation.

For the first experimental setup, batch cultures of N. maritimus and N. piranensis were grown aerobically with ¹⁵N-labeled ammonium (¹⁵NH₄⁺) until it was completely oxidized to ¹⁵NO₂⁻, and thus cultures contained a pool of 1 mM of ¹⁵NO₂⁻ (late exponential phase). Prior to the incubation, more than 500 μM of ¹⁴NH₄⁺ was added to the culture to ensure the survival (ammonia oxidation) of the cultures during the experiment and to capture traces of ¹⁵NH₄⁺ that could have remained in a large pool of ¹⁴NH₄⁺. The incubations under oxygen depletion were set up, as described in the section “oxygen-depleted incubations.” The sets of replicates (at least 3 bottles of 330 mL each per incubation) were spiked with 1.2 μM of unlabeled N₂O (⁴⁴N₂O) after 30 h in the case of N. maritimus and with 3 μM after 6 h and 1.2 μM after 30 h in the case of N. piranensis. Three incubation replicates were kept without the addition of N₂O as a control. A killed control was performed by adding mercury chloride to the incubation.

In the second set of experiments, an aerobically grown batch culture of N. piranensis was maintained with ¹⁴NH₄⁺ until it was completely oxidized to ¹⁴NO₂⁻. Then, the batch culture contained a pool of approximately 1 mM of ¹⁴NO₂⁻ (the late exponential phase). Prior to the incubation, more than 500 μM of ¹⁴NH₄⁺ was added to the culture to ensure the survival (ammonia oxidation) of the cultures during the experiment. The oxygen-depleted incubation was started, as described in the section “oxygen-depleted incubations.” One set of replicates (at least three bottles of 330 mL each) was spiked with 40 nM of ¹⁵N-labeled N₂O (⁴⁶N₂O) at 8 h and with 90 nM at 42 h. Three incubation replicates were kept without the addition of ⁴⁶N₂O as a control.

Samples were collected with gas-tight syringes (Hamilton, United States) that were connected to stainless steel needles (Ochs, Germany) through the capillaries of the incubation bottles. When collecting the samples, the volume collected was simultaneously replaced with deoxygenated sterile culture media to avoid headspace formation in the incubation bottle. The samples were collected in 3-ml gas-tight exetainers, headspace-free, and preserved with 50 μL of saturated HgCl₂ solution. The isotopic signature of N₂ and N₂O was analyzed by coupled gas chromatography–isotope ratio mass spectrometry (GC-IRMS) on a Thermo Delta V Plus isotope ratio mass spectrometer (Dalsgaard et al., 2012). Total N₂O concentrations were analyzed using a gas chromatograph (GC-TRACE1300, Thermo Scientific) equipped with an electron capture detector. Concentrations were plotted as the average of at least three replicates, with error bars representing the standard deviation. Rates were calculated from the change in concentration over time, with r² > 0.9.

The AOA strains N. maritimus and N. piranensis were previously observed to conduct NO dismutation upon oxygen depletion (Kraft et al., 2022; Hernández-Magaña et al., 2023), in which they produced oxygen and ultimately N₂ from NO₂⁻. Transient accumulation of N₂O in both strains was reported in the cited publications, suggesting that AOA can produce N₂O under oxygen depletion and further reduce it to N₂. To assess the role of N₂O as an intermediate in NO dismutation by AOA and, therefore, the AOA’s potential to reduce N₂O, we performed incubations under oxygen depletion with pure cultures of N. maritimus and N. piranensis. The first set of incubations was started with a pool of ¹⁵NO₂⁻ and spiked with 1.2–1.5 μM of unlabeled nitrous oxide (⁴⁴N₂O) at 30 h for both AOA strains and additionally with 3 μM of ⁴⁴N₂O at 6 h only for N. piranensis.

A striking decrease in the total N₂O concentration was observed after the spike in all the incubations. N₂O consumption was especially fast within the first 3 h after the spike (Figure 1). For example, N. piranensis consumed on average 497 nM/h in the first 3 h after the 6-h spike and 198 nM/h after the 30-h spike. Overall, strikingly fast N₂O consumption after the spikes was consistently observed in all the incubations. After this first fast decrease in N₂O, N₂O consumption slowed down. Then, N. maritimus had the highest consumption rate of 48 nM/h, followed by the incubation of N. piranensis after the spike at 6 h, which had a rate of 41 nM/h. Finally, the same strain after the spike at 30 h consumed all spiked N₂O in approximately 30 h at a rate of 28 nM/h. The accumulation of ⁴⁶N₂O from ¹⁵NO₂⁻ started within the first hours of the oxygen-depleted incubations, followed by a linear production of ³⁰N₂ (Figure 2, controls). In N. maritimus incubations, the production of ³⁰N₂ increased at approximately 20 h, while in N. piranensis N₂ production was linear from the beginning of the oxygen-depleted incubation. Another subtle difference between the strains was the transient accumulation of N₂O, which was maintained throughout the whole incubation period for N. maritimus. For N. piranensis, the N₂O accumulation started quickly after oxygen depletion, reaching its maximum within the first 20 h and decreasing almost totally after 40 h of oxygen depletion. Despite these differences in accumulation patterns between strains, there is consistency in the transient accumulation of ⁴⁶N₂O and in the formation of ⁴⁶N₂O only from ¹⁵NO₂⁻ via NO, which is consistent with previous observations (Kraft et al., 2022; Hernández-Magaña et al., 2023).

If N₂O is a free intermediate in the NO-dismutation pathway (a product of the NO-dismutation step), which is reduced to N₂ and not a byproduct, an increase in the pool of ⁴⁴N₂O over ⁴⁶N₂O (⁴⁴N₂O spike) would lead to an increase in ²⁸N₂ production instead of ³⁰N₂ production compared to the control incubations, in which only ⁴⁶N₂O is available. Thus, the reduction of N₂O from a pool enriched with ⁴⁴N₂O would be observed as a slowing of ³⁰N₂ accumulation. For the incubation with N. maritimus in which ⁴⁴N₂O was spiked at 30 h (Figure 1A), ³⁰N₂ accumulation stopped until the spiked ⁴⁴N₂O was consumed (55 h) and then ³⁰N₂ accumulation started again (Figure 2A), demonstrating the direct reduction of N₂O to N₂ and the role of N₂O as an intermediate in the NO-dismutation pathway. For the incubations with N. piranensis, a similar pattern in ³⁰N₂ production was observed after the spike of ⁴⁴N₂O at 6 h of the incubation (Figure 2C). In the case of the spiked incubations of N. piranensis at 30 h, the effect of the ⁴⁴N₂O spike on the ³⁰N₂ production was more difficult to notice in the averaged trend (Figure 2B) and easier to distinguish in the trends of the individual replicates (Supplementary Figure S1A). The replicate with the fastest total N₂O consumption (Supplementary Figure S1B) was the only replicate with no visible effect on the ³⁰N₂ production after the 30-h spike (Supplementary Figure S1A), suggesting that the N₂O pool was consumed too fast to capture the N₂O produced from nitrite. Additionally, no production of ³⁰N₂ or ⁴⁶N₂O or consumption of N₂O after a spike of ⁴⁴N₂O was detected in the killed control with N. maritimus (Supplementary Figure S2), indicating that the consumption of N₂O was performed by active cells of AOA.

Complementary incubations to the previous ones were performed to further explore the ability of AOA to reduce N₂O to dinitrogen. N. piranensis was selected based on the observations in previous incubations that pointed toward a faster N₂O turnover during NO dismutation. For these incubations, the concentration of spiked N₂O was reduced so that the total N₂O concentration remained in the range in which N. piranensis was previously observed to accumulate, to better simulate the conditions under which the reduction of N₂O to N₂ naturally takes place. To track the outcome of the small spikes of N₂O in this set of incubations, the ¹⁵N-labeled compound was N₂O and not nitrite. Oxygen-depleted incubations were performed with a batch culture with a pool of ¹⁴NO₂⁻. After 8 h, 40 nM of ⁴⁶N₂O was spiked into the incubation bottles. The added ⁴⁶N₂O was completely consumed approximately 24 h after the spike at a rate of approximately 1.5 nM/h (Figure 3A). After 42 h of incubation, a second addition of ⁴⁶N₂O was made, this time aiming for a final concentration of approximately 90 nM. The ⁴⁶N₂O added was rapidly consumed again the second time at a rate of approximately 3.5 nM/h.

³⁰N₂ production was only observed in the replicates spiked with ⁴⁶N₂O (Figure 3B) and was within the expected range. After the second spike, up to 106 ± 2 nM of ³⁰N₂ was produced by the end of the incubation (Figure 3B), indicating a complete conversion of the spiked ⁴⁶N₂O to N₂. The measured ⁴⁶N₂O spike was 38 ± 9 nM in the first spike and 89 ± 4 nM in the second spike. Thus, the incubation received a total of 127 ± 9 nM ⁴⁶N₂O. Every time a sample was collected, the volume was replaced with anoxic sterile medium (see Materials and Methods), leading to a dilution of the added ⁴⁶N₂O and the produced ³⁰N₂. Taking this into account, the expected concentration of ^15-15N compounds at the end of the incubation was 106 nM (see Supplementary material), consistent with the ³⁰N₂ accumulated by the end of the incubation. To summarize, the spiked ⁴⁶N₂O was completely reduced to and recovered as ³⁰N_2.

In the controls that did not receive any ¹⁵N-labeled compounds, accumulation of neither ³⁰N₂ nor ⁴⁶N₂O was observed. In these incubations, the accumulation of unlabeled N₂O started in all replicates at the beginning of the incubation and continued until the spike. After the ⁴⁶N₂O spike, the total N₂O concentration (⁴⁴N₂O + ⁴⁶N₂O) was slightly higher in the spiked replicates compared to the controls. Overall, the total N₂O concentration remained within the natural range in which N. piranensis would normally accumulate (Supplementary Figure S3). Taken together, the results of both sets of incubations present solid evidence for N₂O turnover by AOA under oxygen depletion and support its role as an intermediate in the NO-dismutation metabolic pathway.

In order to resolve trends in oxygen consumption and accumulation, oxygen concentrations were measured during all incubations with sensors that can resolve oxygen concentrations in the nanomolar range (Lehner et al., 2015). In all incubations, the oxygen was respired within the first few minutes after the transfer to the incubation bottles. For N. maritimus, shortly after the oxygen was depleted, oxygen started to accumulate, coupled with NO accumulation (Figure 4A). This pattern of oxygen accumulation has been reported previously by Kraft et al. (2022) for N. maritimus and for other AOA species, including N. piranensis, by Hernández-Magaña et al. (2023). When samples were collected, despite the precautions taken (see methodology), the sampling was always accompanied by a small intrusion of oxygen, hereafter referred to as oxygen pulses. Immediately after the oxygen pulses caused by the sampling, oxygen was respired until depletion and oxygen accumulation started again, which was consistently within the nanomolar range and coupled with the transient N₂O accumulation and N₂ production described in the previous section.

In addition to the overall oxygen trends described, there was some variability between strains and among incubations. In the incubations with N. maritimus, sometimes there was a decrease in oxygen accumulation toward the end of the incubations (Figure 4A and Supplementary Figure S4A), as also observed by Kraft et al. (2022). NO accumulated coupled with oxygen accumulation, especially within the first 20 h of the incubation, reached its highest concentration during this time. Like oxygen, NO accumulation decreased toward the end of the incubation (Supplementary Figure S4B). Oxygen was still consumed following oxygen pulses, indicating the culture’s activity. The cessation of oxygen accumulation should not necessarily be interpreted as a lack of oxygen production, but most likely as a more efficient use of it, as in the case of the methane oxidizer Ca. Methylomirabilis oxyfera, which internally consumes all the oxygen produced via NO dismutation (Ettwig et al., 2010).

In the case of N. piranensis, variability in oxygen accumulation trends between different culture batches was observed. In some incubations, oxygen accumulated after its consumption (Figure 4B), but in other cases, no oxygen accumulation was observed (Figures 4C,D). However, NO, N₂O, and N₂ accumulation from NO₂ were still observed during these incubations. A possible explanation for the lack of oxygen accumulation while the production of N₂ continued is that NO dismutation continued and the produced oxygen was used more efficiently, as mentioned above. A possible contamination of the culture, which could also lead to the consumption of oxygen produced during incubation, was excluded by fluorescence microscopy.

To test whether N. piranensis consumed oxygen more efficiently and thus prevented its accumulation, in some sets of incubations, 0.5 mM of potassium cyanide was added to inhibit oxygen respiration by heme–copper oxygen reductases (Wilson et al., 1994). Indeed, after cyanide addition, oxygen accumulated rapidly (Figure 4D and Supplementary Figure S8A), confirming that oxygen was still being produced but was consumed directly, preventing the accumulation of detectable oxygen concentrations.

Oxygen and dinitrogen production through NO dismutation has been observed so far in several different marine and terrestrial AOA isolates, including N. maritimus (Kraft et al., 2022) and N. piranensis (Hernández-Magaña et al., 2023), which we selected to study the pathway in more detail. In the proposed NO-dismutation pathway, the product of ammonia oxidation, NO₂⁻, is reduced to NO, which is dismutated. The proposed products of the dismutation step are N₂O and O₂. N₂O is then further reduced to N₂, making it an intermediate in the NO-dismutation pathway. Transient N₂O accumulation was observed from the beginning of the incubations, followed by N₂ accumulation when N. maritimus and N. piranensis were exposed to oxygen depletion (this study, Kraft et al., 2022; Hernández-Magaña et al., 2023). The production of N₂O and oxygen from NO dismutation by AOA is a substantial difference from the other known NO-dismutation metabolism by Ca. M. oxyfera. This bacterium directly produces N₂ and oxygen via NO dismutation without N₂O as an intermediate (Ettwig et al., 2010). By using a combination of oxygen-depleted incubations with different ¹⁵N-labeled compounds and N₂O spikes with different isotopic signatures, we confirmed that N₂O is the direct intermediate of NO dismutation by AOA. Furthermore, the investigated AOA isolates not only turn over N₂O from NO dismutation but rapidly reduce externally supplied N₂O to N₂.

Although the production and accumulation of N₂O by AOA had been previously reported (Santoro et al., 2011; Löscher et al., 2012), they had been mainly attributed to hybrid formation from ammonia oxidation products and nitrite in oxic incubations (Stieglmeier et al., 2014; Kozlowski et al., 2016; Hink et al., 2017). The N₂O produced in the mentioned studies showed a hybrid isotopic signature, suggesting that one of the N atoms originated from hydroxylamine and one from NO₂. A recent study by Wan et al. (2023), using dual-isotope labeling, assessed multiple N₂O formation mechanisms by N. maritimus and suggested ammonia as the main source of N atoms in N₂O under oxic conditions. The same study also found that the production of N₂O from nitrite only occurred by hybrid formation when ammonia and oxygen were present. Under the oxygen concentrations used by Wan et al. (2023), the ⁴⁶N₂O formation from ¹⁵NO₂⁻ was negligible, and the authors suggested that the production of N₂O by NO dismutation in AOA is restricted to anoxia. In our experiments, the isotopic signature of the N₂O accumulated by N. maritimus and N. piranensis upon oxygen depletion (⁴⁶N₂O) indicates that the origin of both N atoms in N₂O is the pool of ¹⁵NO₂⁻, suggesting that under oxygen depletion, NO₂⁻ is the only source of N atoms for N₂O formation, which is consistent with the observations by Kraft et al. (2022). It is worth highlighting that oxygen depletion is required for NO accumulation and, consequently, for NO dismutation to take place. At higher oxygen concentrations, NO would not accumulate to the concentrations observed in the incubations presented here because it reacts with oxygen via autooxidation, producing NO₂⁻ (Ford et al., 1993; Hickok et al., 2013).

During the incubations with a pool of ¹⁵NO₂⁻, N. maritimus and N. piranensis accumulated oxygen, NO, and ⁴⁶N₂O and produced ³⁰N₂. In the incubations of N. piranensis with a pool of ¹⁴NO₂⁻, the spiked ⁴⁶N₂O resulted in the accumulation of ³⁰N₂, with N₂O being the only source of ¹⁵N atoms to form dinitrogen. The quick consumption of the spiked N₂O shows that N. maritimus and N. piranensis quickly turn over the N₂O pool to N₂ when exposed to oxygen depletion. This evidence supports the role of N₂O as an intermediate in the NO-dismutation pathway. To the best of our knowledge, this is the first time that direct N₂O reduction to dinitrogen by AOA has been shown in physiology experiments.

When exposed to anoxia, NO dismutation is advantageous for AOA because it constitutes an alternative pathway to sustain energy generation and provides alternative electron acceptors and oxygen that can sustain ammonia oxidation at nanomolar ranges (Kraft et al., 2022). While the NO-dismutation reaction is electron-neutral, the other N-conversion steps in the pathway require electrons. The reduction of NO₂⁻ to NO requires one electron per molecule of NO produced, and the reduction of N₂O to N₂ requires two electrons per molecule of N₂ produced. While the electrons could be partly supplied by ammonia oxidation, the source of the remaining electrons has yet to be discovered. Potential electron donors are organic compounds naturally accumulated in the culture medium during aerobic cell growth (Bayer et al., 2019, 2022).

If AOA were capable of using alternative electron donors other than ammonia, N₂O could serve as the sole electron acceptor under anoxia. The rapid conversion of N₂O to N₂ in the two AOA isolates investigated here supports this possibility. Metabolic activity and growth with N₂O as the only electron acceptor are common in many different denitrifying and non-denitrifying microorganisms, with a NosZ N₂O reductase (Mania et al., 2016; Conthe et al., 2018; Lycus et al., 2018; Read-Daily et al., 2022). Given the incubation times in the present study, cell growth was not expected to be observed. During aerobic ammonia oxidation under oxic and optimal conditions, AOA grow at a relatively slow rate. Generation times of N. maritimus and N. piranensis are at a minimum of 19 and 27 h, respectively (Qin et al., 2017; Bayer et al., 2019). Growth rates under anoxic conditions are expected to drop. Therefore, to detect cell growth when AOA perform NO dismutation under oxygen depletion, future research should explore alternatives to cell counts, such as using activity proxies like the incorporation of specifically labeled substrates (Musat et al., 2012; Hatzenpichler et al., 2020).

Although nitrite reduction to NO is most likely performed by the NirK nitrite reductase (Bartossek et al., 2010; Kozlowski et al., 2016), the enzymes responsible for NO dismutation and the further reduction of N₂O by AOA remain to be identified. No genes encoding potential NO dismutases or N₂O reductases have been identified in the genomes of N. maritimus, N. piranensis, or other AOA species (Walker et al., 2010; Qin et al., 2020). All known nitrous oxide reductases belong to the NosZ family. However, the existence of N₂O reductases outside of this family has been proposed multiple times, as reduction of N₂O has been observed in pure microbial cultures that lack a NosZ enzyme (Arciero et al., 2002; Fernandes et al., 2010). Recently, the cytochrome P450 was suggested to be involved in the production of N₂O via NO reduction by the AOA Nitrosocosmicus oleophilus MY3, based on N₂O production measurements coinciding with higher expression of the cytochrome (Jung et al., 2019). These observations were made under oxic conditions and low pH (5.5), in contrast to the conditions used in the current study (oxygen depletion and media HEPES buffered at a pH of 7.6). In Ca. M. oxyfera, quinol-dependent NO reductases (qNORs) have been identified as putative NO dismutases encoded by the nod gene (Ettwig et al., 2010, 2012; Zhu et al., 2019), but AOA do not possess these genes. To date, the potential for NO dismutation followed by N₂O reduction to N₂ by AOA would therefore be overlooked in comparative genomic analyses.

In physiological studies, NO dismutation by AOA would also have been easily overlooked because AOA cultures were not exposed long enough to oxygen depletion, and nitrogen compounds were studied with lower-resolution methods. The lowest oxygen concentrations examined in previous physiological studies of AOA were approximately 1 μM in the headspace (0.1%) (Qin et al., 2015, 2017), and ammonia oxidation was no longer detectable with colorimetric assays. These oxygen concentrations greatly exceed the concentrations at which we observed NO dismutation and oxygen accumulation. At the oxygen concentrations of the present study, the ammonia oxidation rates are in the range of 40 nM/h and would only be detectable by using ¹⁵N-tracers, as shown by Kraft et al. (2022). Furthermore, the rates of N₂ production via NO dismutation are also low and would not be detectable without the use of ¹⁵N-tracers.

In oxygen-depleted incubations in which oxygen accumulation was observed in all replicates, oxygen accumulation often decreased toward the end of the incubation, similar to previous observations (Kraft et al., 2022; Hernández-Magaña et al., 2023). The quick oxygen respiration after oxygen pulses, despite oxygen not being accumulated, indicates respiratory activity of the AOA cells (Figure 4). Moreover, the reduction of N₂O and constant production of N₂ in these incubations continued despite the apparent lack of oxygen accumulation. In the specific case of N. piranensis, oxygen did not accumulate in some incubation bottles (Figures 4C,D). After cyanide addition, a rapid increase in oxygen concentration was observed (Figure 4D and Supplementary Figure S7A). These observations, taken together, suggest that the cultures most likely continued to produce oxygen via NO dismutation, but all of it was utilized immediately, leading to no accumulation at detectable concentrations. Therefore, the lack of oxygen accumulation does not imply a lack of activity or absence of oxygen production, but most likely a more efficient usage of the oxygen produced, which prevents accumulation from being detected.

These observations point toward a change in the efficiency of the coupling between oxygen production and its use: at the beginning of the incubation, oxygen is produced faster than it is used, and later the production and consumption processes become more tightly coupled, or in the case of N. piranensis, some culture batches may have a faster response to oxygen depletion, which leads to a tighter coupling between oxygen production and consumption. Tightly coupled oxygen production and consumption takes place in cultures of the methane oxidizer Ca. M. oxyfera: no oxygen accumulates during NO dismutation, as it is immediately utilized intracellularly for methane oxidation (Ettwig et al., 2010). In Ca. M. oxyfera, the detection of oxygen produced via NO dismutation was only possible after the inhibition of the oxygen-consuming methane mono-oxygenase complex (pMMO) by acetylene (Ettwig et al., 2010, 2012; Wu et al., 2011), which is comparable to our observation of oxygen accumulation after cyanide addition.

Whether AOA are capable of producing oxygen via the NO-dismutation pathway in the environment is still unknown and challenging to detect, as any trace of oxygen produced in the environment would be immediately used by the microbial community (Garcia-Robledo et al., 2017), and because the isotopic signature of ¹⁵N-tracer methods to detect NO dismutation is indistinguishable from denitrification. Therefore, it is important to perform investigations in environmental settings to unveil the potential influence of AOA activity on the oxygen and nitrogen metabolism of natural communities in oxygen-depleted ecosystems.