Clarifying Microbial Nitrous Oxide Reduction under Aerobic Conditions: Tolerant, Intolerant, and Sensitive

Some bacteria can reduce N2O in the presence of O2, whereas others cannot. It is unclear whether this trait of aerobic N2O reduction is related to the phylogeny and structure of N2O reductase. ABSTRACT One of the major challenges for the bioremediation application of microbial nitrous oxide (N2O) reduction is its oxygen sensitivity. While a few strains were reported capable of reducing N2O under aerobic conditions, the N2O reduction kinetics of phylogenetically diverse N2O reducers are not well understood. Here, we analyzed and compared the kinetics of clade I and clade II N2O-reducing bacteria in the presence or absence of oxygen (O2) by using a whole-cell assay with N2O and O2 microsensors. Among the seven strains tested, N2O reduction of Stutzerimonas stutzeri TR2 and ZoBell was not inhibited by oxygen (i.e., oxygen tolerant). Paracoccus denitrificans, Azospirillum brasilense, and Gemmatimonas aurantiaca reduced N2O in the presence of O2 but slower than in the absence of O2 (i.e., oxygen sensitive). N2O reduction of Pseudomonas aeruginosa and Dechloromonas aromatica did not occur when O2 was present (i.e., oxygen intolerant). Amino acid sequences and predicted structures of NosZ were highly similar among these strains, whereas oxygen-tolerant N2O reducers had higher oxygen consumption rates. The results suggest that the mechanism of O2 tolerance is not directly related to NosZ structure but is rather related to the scavenging of O2 in the cells and/or accessory proteins encoded by the nos cluster. IMPORTANCE Some bacteria can reduce N2O in the presence of O2, whereas others cannot. It is unclear whether this trait of aerobic N2O reduction is related to the phylogeny and structure of N2O reductase. The understanding of aerobic N2O reduction is critical for guiding emission control, due to the common concurrence of N2O and O2 in natural and engineered systems. This study provided the N2O reduction kinetics of various bacteria under aerobic and anaerobic conditions and classified the bacteria into oxygen-tolerant, -sensitive, and -intolerant N2O reducers. Oxygen-tolerant N2O reducers rapidly consumed O2, which could help maintain the low O2 concentration in the cells and keep their N2O reductase active. These findings are important and useful when selecting N2O reducers for bioremediation applications.

and the production of reactive oxygen radicals upon transient exposure to oxygen (7). This process could also contribute to the sensitivity of N 2 OR to oxygen at the enzyme level. In addition to the effect on the enzyme itself, O 2 can also influence the transcription of the nos cluster. The O 2 -sensing transcription regulators, such as FNR and NNR, as well as small RNA, can suppress the transcription of nos (8,9).
While the impact of O 2 on microbial N 2 O reduction has been well documented, some denitrifying bacterial strains have been reported to reduce N 2 O in the presence of O 2 (i.e., aerobic N 2 O reduction) (10,11). However, the ecophysiology of aerobic N 2 O reduction remains largely unclear. Questions that remained unanswered include whether the O 2 sensitivity of N 2 OR is related to their structure and how widely aerobic N 2 O reducers occur in the N 2 OR phylogeny.
There are two distinct clades (clade I and II) for nosZ, which is the key functional gene of N 2 OR (12). Genomic differences between the two clades are associated with nos cluster organization, the translocation pathway, and co-occurrence with other denitrifying genes (13). Several studies have reported the physiological differences between the two clades. Yoon et al. (14) report that clade II bacteria (Dechloromonas aromatica and Anaeromyxobacter dehalogenans) showed high affinities to N 2 O but lower maximum reduction rates than those of clade I bacteria (Stutzerimonas stutzeri, formerly known as Pseudomonas stutzeri [15], and Shewanella loihica). In contrast, Suenaga et al. (3) found that the N 2 O reduction biokinetics could not be used to distinguish the clade I bacteria (S. stutzeri and Paracoccus denitrificans) and clade II bacteria studied (Azospira spp.). Nevertheless, it is still unclear how clade I and II N 2 O reducers behave in the presence of O 2 .
Therefore, the objectives of this study were to (i) characterize the oxygen sensitivity of various N 2 O reducing bacteria, (ii) classify N 2 OR based on their oxygen sensitivity, and (iii) examine the relationships between N 2 OR oxygen sensitivity, nosZ phylogeny (clade I versus clade II), and the predicted N 2 OR structures.

RESULTS
Michaelis-Menten kinetics of aerobic and anaerobic N 2 O reduction. By fitting the N 2 O reduction results normalized by the optical density (OD) at 600 nm wavelength to the Michaelis-Menten model, we obtained the maximum rate (V max ) and Michaelis constant (K m ) values for various N 2 O-reducing strains under aerobic and anaerobic conditions. A wide range of V max for nitrous oxide reduction rates was observed. Under anaerobic conditions, bacteria with clade I N 2 OR generally exhibited faster N 2 O reduction than those with clade II N 2 OR ( Fig. 1 and 2). Under anaerobic conditions, S. stutzeri TR2 (clade I N 2 OR) ( Fig. 1-B1) had the highest V max (8.37 6 0.81 mM/s/OD), whereas G. aurantiaca T-27 (clade II N 2 OR) ( Fig. 2-C1) had the lowest V max (0.13 6 0.02 mM/s/OD). This general trend in kinetics, however, may not extend beyond the studied strains, especially given the diversity of microorganisms harboring clade II NosZ (12).
The ability to reduce N 2 O in the presence of O 2 varied by strain, and there was no overall trend between the tested strains with clade I and II N 2 OR. For example, S. stutzeri TR2 ( Fig. 1 The transition points from aerobic to anaerobic N 2 O reductions (i.e., the change of the slopes between two linear rates) were clearly observed after oxygen was depleted for all tested strains, except for G. aurantiaca T-27. For G. aurantiaca T-27, the N 2 O reduction rate gradually changed depending on the oxygen concentration ( Fig. 2-C2). In order to further investigate the different oxygen inhibition kinetics observed for G. aurantiaca, nonlinear least square fitting with multiple variables was used to determine the inhibition constant (K i ). The noncompetitive inhibition model was found to best describe the changing V max against various O 2 and N 2 O concentrations (see Fig. S1 in the supplemental material), with a K i value of 7.86 6 1.69 mM O 2 .
The fitted K m values for anaerobic N 2 O reduction ranged from 1.85 6 1.25 mM (for D. aromatica) to 11.14 6 6.04 mM (for S. stutzeri TR2). The K m values of aerobic N 2 O reduction for P. denitrificans and S. stutzeri TR2 and ZoBell strains did not significantly differ from those of anaerobic N 2 O reduction (Student's t test, P . 0.05). This finding indicates that the affinity of clade I N 2 OR tested did not change with and without the presence of O 2 . Classification of oxygen sensitivity of N 2 O reduction. Based on the microsensor analysis, a broad range of N 2 O reduction kinetics was observed under aerobic and anaerobic conditions. As we plotted the extrapolated anaerobic and aerobic V max values ( Fig. 3A), three distinct types of responses to oxygen were found in the studied strains, as follows: (i) strains with V max not affected by oxygen, including S. stutzeri ZoBell and TR2, are classified as oxygen tolerant; (ii) strains with much lower aerobic V max than anaerobic V max , including P. denitrificans, A. brasilense, and G. aurantiaca, are classified as oxygen sensitive; and (iii) strains that have no N 2 O reduction activity when oxygen is present, including P. aeruginosa and D. aromatica, are classified as oxygen intolerant. NosZ phylogeny seems to be not associated with the classification of oxygen sensitivity. Moreover, the half-saturation coefficients for N 2 O under anaerobic and aerobic conditions agree with previously reported observations. Bacteria harboring clade II NosZ generally have lower K m values than those with clade I NosZ, suggesting differentiating ecological niches for these two groups of N 2 Oreducing bacteria (14).
NosZ amino acid sequence similarities among the strains. The NosZ amino acid sequences of the strains studied were compared to examine whether the observed differences in oxygen sensitivity originate from the differences in the enzyme structures. Strains investigated in this study cover a variety of classes, including Alphaproteobacteria (A. brasilense and P. denitrificans) and Gammaproteobacteria (S. stutzeri and P. aeruginosa) for those having clade I NosZ and Betaproteobacteria (D. aromatica) and Gemmatimonadetes (G. aurantiaca) for those having clade II NosZ. Based on the NosZ phylogenetic analysis, clade I and clade II NosZ were clearly separated (Fig. 4), similar to the previous report (16). The two S. stutzeri strains, of which both showed oxygen-tolerant N 2 O reduction, shared a high similarity in the NosZ amino acid sequences (92.6%) (see Fig. S4 in the supplemental material). However, P. aeruginosa PAO1, which showed oxygen-intolerant N 2 O reduction, also has similar NosZ amino acid sequences to S. stutzeri (77.5% with the ZoBell strain and 79.7% with the TR2 strain). NosZ of oxygen-sensitive N 2 O reducers (P. denitrificans, A. brasilense, and G. aurantiaca) and oxygen-intolerant N 2 O reducers (P. aeruginosa and D. aromatica) were not clustered with each other. In addition, we could not identify amino acid residues that appeared specific to each of the oxygen-tolerant, -sensitive, and -intolerant groups.
Multiple sequence alignment showed that the candidate ligands of Cu A and Cu Z centers were found in all NosZ sequences (see Fig. S3 in the supplemental material). The Cu Z catalytic site contains seven histidine ligands which were all conserved in the proposed Cu Z center among clade I and clade II (Fig. S3). The candidate ligands of Cu A (two cysteines NosZ structural similarities. To identify the structural differences between oxygentolerant, -sensitive, and -intolerant NosZ, we predicted the enzyme structures based on the NosZ sequences by using Alphafold2 (17) with the ZoBell NosZ (18) as a query structure. We obtained high-confidence NosZ structures, as evaluated based on the sequence coverage and predicted per-residue confidence measure (pLDDT) scores from AlphaFold, with conserved Cu A and Cu Z catalytic domains (see Fig. S2 in the supplemental material). Slight structural differences were seen between clade I and II NosZ as measured by the Dali Zscores, whereas no differences were seen between NosZ structures from oxygen-tolerant, -sensitive, and -intolerant strains. The Z scores for all clade I NosZ against the reference ZoBell NosZ were $59.8. In addition, the predicted structures for all clade I NosZ showed the root mean square deviation (RMSD) value of ,2.0 and had no structurally dissimilar amino acid residues of longer than 80 amino acids (aa) to the reference NosZ. In contrast, the Z scores for the NosZ of D. aromatica and G. aurantiaca (clade II) were 50.8 and 49.3, respectively. Poor matches with the query sequence were obtained for the clade II NosZ with RMSD values of .2.0 and structurally dissimilar amino acid residues of .80 aa. Most of the structural heterogeneity was observed in the C and N terminals.

DISCUSSION
Biological N 2 O reduction is generally believed to occur under strictly anaerobic conditions. The oxygen sensitivity of N 2 O reduction can be explained by (i) the transcriptional regulation of nos and (ii) the inactivation of N 2 OR by molecular oxygen. The transcription of nosZ can be regulated directly or indirectly by O 2 -sensing transcriptional regulators. For instance, the transcription of nosZ is directly regulated by fumarate and nitrate reductase protein (FnrP) in response to oxygen depletion in P. denitrificans (19). P. aeruginosa also has similar FNR-type sensing regulators; the cascading regulation of anaerobic regulator of arginine deiminase and nitrate reductase (ANR) and dissimilatory nitrate respiration regulator (DNR) indirectly controls the synthesis of N 2 OR (20). Another potential explanation of oxygen sensitivity points to the inactivation of N 2 OR upon exposure to oxygen. The N 2 OR isolated under aerobic and anaerobic conditions exhibited various redox and spin states of copper in active sites. Under limited exposure to oxygen, the enzyme shifted in electron paramagnetic resonance spectra but retained its N 2 O-reducing activity (21). In contrast, aerobic incubation caused loss of copper content and inactivation of the catalytic site. Inactivation of N 2 OR by oxygen was also reported due to irreversible confirmation changes. A sulfur atom binding to the active site of N 2 OR isolated from S. stutzeri ZoBell was lost during aerobic enzyme isolation, leading to irreversible inactivation (6). However, these two mechanisms do not explain the occurrence of O 2 -tolerant N 2 O reducers. The cells used for the microsensor experiments were incubated under anaerobic conditions with the addition of nitrite or N 2 O to induce the expression of N 2 OR. One exception is for G. aurantiaca T-27 T . This strain was incubated under aerobic conditions, as G. aurantiaca T-27 T is an obligate aerobic bacterium that can express nosZ in the presence of O 2 (11,22). The same cell cultures were used for aerobic and anaerobic N 2 O reduction rate measurements; therefore, the initial level of N 2 OR expressed in the cells should be the same between the two conditions (i.e., aerobic versus anaerobic N 2 O reduction). Consequently, the transcriptional regulation of nos is not contributing to the O 2 tolerance during N 2 O reduction of each tested strain.
In addition, the structures of NosZ, including the active sites, were highly similar between O 2 -tolerant, -sensitive, and -intolerant N 2 O reducers. Based on the structural similarity and the presence of conserved residues in the active sites, all of the active sites of NosZ and copper cofactors examined most likely receive similar inhibitory effects upon exposure to oxygen (6,21). Despite similar N 2 O respiration and bioenergetics in clade I and clade II NosZ, other accessory proteins encoded by the nos cluster are expected to function differently (23). These auxiliary processes could be involved in the maintenance and repair of NosZ, with detailed mechanisms remaining unclear.
Another mechanism that may explain the observed occurrence of O 2 -tolerant N 2 O reduction is the scavenging of O 2 in the cells. A whole-cell assay (as opposed to the assay done with isolated enzymes) was used in this study to calculate the N 2 O and O 2 consumption rates. When both N 2 O and O 2 are present, facultatively anaerobic bacteria (e.g., denitrifiers) usually prioritize the respiration of O 2 over N 2 O because aerobic respiration is more favorable from both bioenergetic and kinetic perspectives (24). A rapid O 2 consumption rate can potentially lower the in situ O 2 concentration in the periplasm, where N 2 OR is located. From a simplified estimation shown in the supplemental materials, an O 2 consumption rate of 1 mM/s/OD can cause a significant decrease in O 2 concentration across cell membranes. When the O 2 respiration rate is comparable to the O 2 diffusion rate that replenishes dissolved oxygen in the periplasm, the local oxygen minimum could protect N 2 OR from inhibition in O 2 -tolerant N 2 O reducers. From the tested strains, we indeed observed that bacteria with higher oxygen consumption rates generally have greater oxygen tolerances (Fig. 3B). A threshold of O 2 consumption rate could potentially exist, where a lower rate could not emulate the diffusion rate of O 2 sustaining an anaerobic zone for N 2 OR. Such a protection mechanism could be analogous to the respiration of O 2 in Azotobacter protecting O 2 -sensitive nitrogenase (25).
Our results have some implications for N 2 O removal applications. N 2 O-reducing bacteria, including some of the strains examined in this study, have been used for N 2 O mitigation in natural and engineered systems (2). For instance, bioaugmentation of S. stutzeri TR2 to denitrifying activated sludge has been demonstrated to mitigate N 2 O emissions (26,27). Azospirillum brasilense strains were also used as a microbial inoculant for N 2 O mitigation in soil (28). Nevertheless, engineering applications of biological N 2 O mitigation face major challenges, including the oxygen sensitivity of N 2 O reduction due to the coexistence and fluctuations of dissolved oxygen and N 2 O concentrations commonly observed in natural and engineered systems. Based on the classification of O 2 tolerance in this study, kinetic parameters can be used as selection criteria for microorganisms in environmental applications. Oxygen-tolerant N 2 ORs were identified only in S. stutzeri in this study. S. stutzeri also exhibited some interesting kinetics when both electron acceptors (O 2 and N 2 O) are present. The TR2 strain showed preferred N 2 O respiration over oxygen, contrary to predictions based on electron supply rate to the electron transport chain (29). In addition, the ZoBell strain can reduce N 2 O fast and in the presence of O 2 , making it promising for N 2 O bioremediation applications. Besides N 2 O reduction rates, microorganisms with low K m values, such as P. denitrificans and D. aromatica, could be useful in scavenging low concentrations of dissolved N 2 O. It is important to note, however, that the kinetics and O 2 sensitivity of N 2 O reducers can be influenced by environmental factors, such as the type of organic carbons (30) and temperature (31). Therefore, when selecting appropriate N 2 O reducers for engineering applications, their N 2 O reduction kinetics and O 2 sensitivity should be measured under environmentally relevant conditions.
These strains, except for D. aromatica RCB and G. aurantiaca T-27 T , were grown on R2A agar plates amended with 10 mM acetate and 5 mM nitrite under aerobic conditions. After 48 h of incubation at 30°C, single colonies were picked and transferred to 10 mL of R2A broth with 10 mM acetate and 5 mM nitrite. Each liquid culture was incubated in a sealed tube with an N 2 atmosphere at 30°C until harvested during the exponential growth phase. D. aromatica RCB was grown on Trypticase soy agar (TSA) supplemented with 5% defibrinated sheep blood under anaerobic conditions at 30°C for 10 days. Single colonies were transferred to 10 mL of R2A broth supplemented with 20 mM lactate and incubated under a 1.39% N 2 O atmosphere (in N 2 ) at 30°C until harvested. G. aurantiaca T-27 T was grown on R2A agar under aerobic conditions. Single colonies were transferred to 10 mL of R2A broth and aerobically incubated at 25°C until harvested. The addition of nitrite inhibited the growth of G. aurantiaca, which was expected to have an incomplete denitrification pathway (32).
Microsensor experiments. Bacterial cultures were harvested during the early to mid-exponential growth phase as determined by the optical density at 600 nm (OD 600 ) measurement. Cultures were washed twice with a sterile 10 mM piperazine-N,N9-bis(2-ethanesulfonic acid) (PIPES) buffer (pH 7.5) and resuspended in a PIPES buffer supplemented with 10 mM sodium acetate. The cell suspensions were purged with a gas mix of N 2 O (1.39%, vol/vol) in N 2 for 10 min to achieve targeted levels of dissolved N 2 O concentrations (300 mM). The cell suspensions were then diluted with PIPES buffer to the desired concentration (;10 6 CFU/mL; OD 600 , ;0.1) and transferred to a double chamber containing mini stirrer bars (Unisense, Aarhus, Denmark) (see Fig. S5 Table S1 in the supplemental material). The OD 600 of the cell suspension was recorded at the end of each microsensor test. At least three independent microsensor measurements were done for each strain.
The measured concentrations of N 2 O and O 2 were averaged over time intervals of 100 to 1000 s depending on the duration of microsensor tests. This step is useful to minimize the noise generated by the microsensors. Linear rates for N 2 O consumption were extrapolated within each time interval. The Michaelis-Menten plots were then constructed using the rates and corresponding N 2 O concentrations. A nonlinear least square method with the Levenberg-Marquardt algorithm (33) was used for curve fitting on Origin 2021 (version 9.8.0.200) to determine kinetic parameters, including the maximum rate (V max ) and the Michaelis constant (K m ). Similarly, V max for O 2 was linearly extrapolated from O 2 concentrations measured by the microsensor.
Bioinformatics and comparative protein structure modeling. The NosZ sequences of the selected strains (GenBank accession numbers WP_011287329, EHY76008, BAM68548, NP_252082, QEL93987, WP_156798935, and Q51705 for D. aromatica RCB, S. stutzeri ZoBell, S. stutzeri TR2, P. aeruginosa PAO1, A. brasilense Sp7, G. aurantiaca T-27, and P. denitrificans JCM 21484, respectively) were retrieved from National Center for Biotechnology Information (NCBI; https://www.ncbi.nlm.nih.gov). Multiple sequence alignment and phylogenetic tree construction were done using the neighbor-joining method without distance correction by using Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/). NosZ structures were predicted through the nondocker implementation of AlphaFold2 version 2.1.1 via the Minnesota Supercomputing Institute (MSI). The NosZ sequence of the selected strains was used as the input with the default prediction parameters to run on a Linux environment. The best-predicted protein models were selected for each sequence and loaded into PyMOL (Schrödinger, Inc., New York, NY). All models were colored based on their predicted local distance difference test (pLDDT) that are stored in the B-factor fields of the PDB files. All predicted structures were compared against each other using DaliLite.v5 (http://ekhidna2.biocenter.helsinki.fi/dali) (34).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 5.5 MB.