Introduction

Coping with rising levels of the potent greenhouse gas nitrous oxide (N2O) in the atmosphere calls for the development of mitigation strategies to reduce N2O accumulation and emission in soil management and wastewater treatment (WWT). The presence and activity of N2O-reducing organisms in fertilized soils and WWT plants, such as bacteria and archaea harboring nosZ-type genes, may be key in such mitigating strategies (Thomson et al. 2012). Nitrous oxide reductase (N2OR), the enzyme encoded by the nosZ gene, is a terminal reductase present in some microbial respiratory electron transport chains (ETC) that catalyzes the only microbial reaction known to consume N2O, converting it to innocuous N2 (which constitutes 79% of the Earth’s atmosphere). Although N2O reduction is generally associated to denitrifying organisms, many N2O reducers lack reductases other than N2OR (i.e., nitrate-, nitrite-, or nitric oxide-reductase; Hallin et al. 2018). However, most, if not all, denitrifiers—and presumably N2O reducers—are facultative aerobes, having the terminal oxidases necessary for O2 respiration (van Spanning and Richardson 2007).

Based on what is known on the biochemistry of model organisms like Paracoccus denitrificans, N2O and O2 respiration presumably share the core of the ETC (Chen and Strous 2013), with electrons branching out to O2 (via cytochrome oxidases), N2O (via N2OR), or other NOx (in denitrifying N2O reducers) depending on electron acceptor availability. It is a common notion that, when both N2O and O2 are available, N2O reducers will consume O2 preferentially over N2O (and other N oxides; Shapleigh 2013). Even though N2O is a stronger electron acceptor than O2 in terms of thermodynamics, a number of authors have shown that N2O respiration is energetically less efficient than aerobic respiration, resulting in lower biomass growth yields per substrate (Koike and Hattori 1975; Stouthamer et al. 1982; Beun et al. 2000). We cannot rule out the existence of a more energy-efficient N2O reduction process (Conthe et al. 2018a), considering the broad phylogenetic diversity of N2O reducers and our limited knowledge regarding non-denitrifying N2O reducers in particular. However, given the growth yields reported in literature, it would make evolutionary sense for microorganisms to favor aerobic respiration over the respiration of N compounds to optimize energy conservation in the cell. Intriguingly, the physical mechanism directing electrons to O2 preferentially over other N compounds, when both electron acceptors are available, remains unclear.

Regulatory systems on a transcriptional or post-transcriptional level have been shown to shut down denitrification in the presence of oxygen in a variety of organisms (Zumft 1997). For instance, the NosZ protein of Paracoccus denitrificans and Pseudomonas stutzeri is inhibited by O2 in vitro (Coyle et al. 1985; Alefounder and Ferguson 1982), which could be a form of allosteric regulation in vivo. It has also been proposed that N2OR is—for reasons unknown—less competent than the cytochrome oxidases involved in respiration of O2 in the “competition” for electrons in the ETC (Qu et al. 2015). Nevertheless, diverse studies have reported the occurrence of denitrification in the presence of O2 (termed aerobic denitrification; Chen and Strous 2013 and references therein). Regarding N2O reduction more specifically, a significant degree of N2OR transcription and activity has been found under aerated conditions (Körner and Zumft 1989; Qu et al. 2015).

From a greenhouse gas mitigation point of view, it is interesting to study O2 and N2O mixotrophy—or the capability of microorganisms to simultaneously respire O2 and N2O—in order to understand how frequent oxic-anoxic shifts during nitrogen removal from wastewater, in space or time, may affect the N2O-reducing capacity of activated sludge. WWTP design and operation vary greatly, but universal questions to address are, e.g., (a) if N2OR activity can persist in aerated zones consuming nitrification-derived N2O potentially minimizing greenhouse gas emissions or (b) if, on the contrary, N2OR is relatively less active than the other NOx reductases in the presence of O2, leading to N2O accumulation in the aerobic-anoxic transition zones.

We explored O2 versus N2O respiration in a continuous enrichment culture selected and grown with N2O as the sole electron acceptor and fully characterized—in terms of stoichiometry and community composition—in a previous study (Conthe et al. 2018b). The culture had been found to be composed of a relatively simple microbial community dominated by Dechlorobacter-like Betaproteobacteria. In this study, operation of the chemostat was continued and the N2O-limited steady-state conditions were intermittently interrupted to perform short-term batch experiments in situ, with varying concentrations of N2O, O2, or both N2O and O2 simultaneously, to determine (i) whether O2 is, in fact, preferentially consumed over N2O when both electron acceptors are available, (ii) under which O2 concentrations (if any) N2O consumption can take place, and (iii) to begin to unravel the mechanism governing the electron flow in the ETC to O2 or N2O.

Materials and methods

Chemostat operation

Following the work presented in Conthe et al. (2018b), a microbial enrichment using acetate as a carbon and energy source and exogenous N2O as the sole electron acceptor was maintained under N2O-limiting conditions in a continuous culture at 20 °C, pH 7, and a dilution rate of 0.026 ± 0.001 h−1. The reactor set-up, operation, sampling, and medium composition are described in detail in Conthe et al. (2018b, c). One hundred percent pure N2O gas diluted in Argon gas was fed to the chemostat at a total flow rate of 200 ml/min and the offgas from the reactor was recirculated at a rate of 700 ml/min, resulting in an incoming N2O concentration of roughly 0.30%. The stability of the culture in terms of conversion rates and microbial community composition was monitored by regular sampling of the broth and biomass and via online monitoring of the acid (1 M HCl) dosing (a proxy for acetate consumption in the system) and offgas composition.

Batch experiments

The steady-state conditions of the culture were briefly interrupted on different operation days in order to perform batch experiments in situ and determine the maximum conversion rates of the enrichment under non-limiting conditions (Figure S1). The medium and effluent pumps were switched off and the gas supply rates of O2 (from a bottle of pure O2) and/or N2O were modified to achieve different electron acceptor concentrations within the system in random steps. Two main types of batches were performed: (1) supplying a single electron acceptor—either N2O or O2—at different concentrations or (2) supplying N2O and O2 simultaneously, keeping the N2O gas supply rate constant and varying that of O2. Additionally, we performed a batch test in which a constant O2 gas supply rate was maintained while varying that of N2O as well as short batch tests with either NO3− or NO2− to assess the denitrifying capacity of the culture. Note that gas recirculation was maintained during the experiments, causing an apparent delay between the conversions in the chemostat and the offgas concentration values measured. To avoid acetate depletion, a concentrated solution of sodium acetate was added to the broth at the start of the experiments and the 1 M HCl solution used for pH control during continuous operation was replaced by 1 M acetic acid for the duration of the experiment. For the batch tests with NO3− and NO2−, these compounds were supplied as 1 M KNO3 or 1 M KNO2.

Analytical procedures

Samples from the reactor for analysis of acetate and NH4+ were immediately filtered after sampling (0.45-μm pore size poly-vinylidene difluoride membrane, Merck Millipore, Carrigtohill, Ireland). Acetate was measured with a Chrompack CP 9001 gas chromatograph (Chrompack, Middelburg, The Netherlands) equipped with an HP Innowax column (Agilent Technologies, Santa Clara, CA, USA) and a flame ionization detector. Ammonium, NO3−, and NO2− concentrations were determined spectrophotometrically using cuvette test kits (Hach Lange, Düsseldorf, Germany). For the estimation of biomass concentration, the volatile suspended solids (VSS) concentration was determined by centrifuging 0.2 L of the enrichment, drying the pellet overnight at 105 °C, and then burning the pellet at 550 °C for 2 h to determine the ash content. Additionally, the optical density of the culture (at a wavelength of 660; OD660) was monitored. Concentrations of N2O, N2 and CO2, Argon, and O2 in the headspace of the reactor were measured online via mass spectrometry (Prima BT, Thermo Scientific). The dissolved O2 concentration in the broth during the batch tests with O2 was measured with two types of oxygen sensors: a Clark electrode calibrated in the range of 0–20.8% and an optical oxygen probe calibrated in range 0–2% (Presens, Regensburg, Germany).

Calculations

Elemental and electron balances during steady state were set up as described in Conthe et al. (2018a, b, c). During the batch tests, the conversion rates (r, in mol h−1) for O2 and N2O were calculated from the measured ingoing and outgoing gas composition and the argon supply rate (see Figures S2–S6 and Tables S2–S6 for details). The average biomass concentration value for each experimental step was derived from the ammonium uptake rates (see for example Figure S4b) and used to calculate the corresponding biomass specific rates (q, in mol CmolX−1 h−1). A standard and constant biomass composition of CH1.8O0.5N0.2 (Roels 1980). The qO2 and qN2O obtained for each step were plotted against the corresponding concentration of dissolved O2 or N2O in the broth in order to determine the qmax and Ks of the enrichment for O2 and N2O. The concentration of dissolved O2 was obtained experimentally with the DO probes while the concentration of dissolved N2O was estimated given a kLaN2O of 180 h−1—obtained by scaling the experimentally derived kLaO2 (Janssen and Warmoeskerken 1987) and deriving the corresponding KLabroth and KLaheadspace assuming a tbroth of 6 s (1800 and 50 h−1, respectively). A Monod model fitting the results was obtained by minimizing the sum of squared errors using the Microsoft Excel software.

The thermodynamic efficiency of metabolic growth using acetate as an electron donor and O2, N2O, or NO3− as an electron acceptor can be interpreted by the Gibbs free energy (ΔG01) dissipated per C mole of biomass growth or per electron-equivalent used for respiration. These values were calculated based on Kleerebezem and van Loosdrecht (2010) and using the thermodynamic values found in Thauer et al. (1977)—please refer to Table S7 for more details.

DNA extraction and 454 amplicon sequencing of 16S rRNA gene

The taxa-based community composition of the enriched culture during the period of operation presented in this study was determined by 454 amplicon sequencing of the 16S rRNA gene following the procedure described in Conthe et al. (2018a, b, c) and the sequences are available at NCBI under BioProject accession number PRJNA413885.

Results

Continuous operation and microbial community composition of the N2O-reducing enrichment

A culture enriched from activated sludge using acetate as a carbon source and electron donor and exogenous N2O as the sole electron acceptor was studied for a total period of 155 days (> 100 volume changes) in a chemostat under electron acceptor (N2O) limiting conditions (Figure S1). The start-up and characterization of the enrichment during the first 70 days of operation, in terms of conversion rates, stoichiometry, and microbial community composition, are described in Conthe et al. (2018b). During the subsequent period reported here, the conversion rates and corresponding biomass yields remained consistent with the previous period, characterized by steady-state growth on acetate oxidation coupled to N2O reduction to N2 (Tables 1 and 2). Furthermore, 454 amplicon sequencing of the 16S rRNA gene of the microbial community confirmed the continued prevalence of a Dechlorobacter-like OTU (Figure S1), transiently co-occurring (around day 100) with two other closely related OTUs classified as Azonexus and uncultured Rhodocyclaceae.

Table 1 Average biomass-specific conversion rates during steady state and the batch experiments
Full size table
Table 2 Experimentally determined biomass yields per mole of electron donor or per mole of electron equivalents respired during growth with N2O, NO3−, and O2 as an electron acceptor and corresponding Gibbs free energy dissipation values based on these yields
Full size table

O2 vs. N2O batch tests: affinity and yields

Batch experiments with varying supply rates of either N2O or O2 were performed on days 106 and 132, respectively (Fig. 1). The maximum biomass specific conversion rates of N2O (\( {q}_{\mathrm{N}2\mathrm{O}}^{\mathrm{max}} \)) and acetate were identified by increasing the N2O supply rate to non-limiting conditions. The \( {q}_{\mathrm{N}2\mathrm{O}}^{\mathrm{max}} \) values identified were roughly fourfold higher than the actual biomass specific conversion rates during steady state (Table 1). When exposed to varying concentrations of O2, the culture was able to switch to aerobic respiration in the order of seconds. The maximum O2 reducing capacity (\( {q}_{\mathrm{O}2}^{\mathrm{max}} \)) was comparable to N2O respiration when expressed per mole electron accepted. NO3− and NO2− reducing capacities were much lower compared to N2O or O2 (< 15% of the maximum N2O or O2 reduction rate; Table 1).

Fig. 1
figure 1

Offgas data from the batch experiments with varying concentrations of a N2O; day 106, b and O2; day 132. For the experiment with O2, the dissolved oxygen concentration (DO) was measured both with a Clark electrode (DO_1) and an optical sensor (DO_2). The affinity of the culture for N2O and O2 was determined from these experiments (see Fig. 5). The asterisk mark time points at which acetate had been depleted and was added to the culture

Full size image

Plotting the biomass-specific electron transfer rate (qe−) at different dissolved O2 (DO) or N2O concentrations, we could determine the apparent Ks for O2 or N2O by fitting a Monod model to the data (Fig. 2). Given the confidence intervals, the absolute value for this parameter could not be identified accurately, but the results demonstrate clearly that the Ks value for O2 is 1 or 2 orders of magnitude smaller compared to Ks-N2O. The maximum biomass-specific conversion rate of O2 (\( {q}_{\mathrm{O}2}^{\mathrm{max}} \)) was roughly two times lower than that of N2O (\( {q}_{\mathrm{N}2\mathrm{O}}^{\mathrm{max}} \)) per mole of electron acceptor but the conversion rates expressed as electron equivalents (\( {q}_{\mathrm{e}}^{\mathrm{max}} \)) were comparable for both processes, since double the electrons are taken up during the reduction of O2 to H2O compared to N2O to N2.

Fig. 2
figure 2

Biomass specific transfer rates of electron equivalents (qe-) as a function of the electron acceptor concentration (either N2O, in black, or O2, in gray), along with the fitting Monod model (with the corresponding qe-max and Ks parameters). The inset is an enlargement of the graph at low O2/N2O concentrations. The rates presented were obtained from the experiments shown in Fig. 2

Full size image

The biomass yields per mole of electron donor (determined from the steady-state growth on N2O in the chemostat, and from the batch experiments with O2 as the sole electron acceptor) are presented in Table 2.

Simultaneous O2 and N2O batch tests

Batch experiments with excess N2O and varying concentrations of O2, supplied simultaneously, were performed on days 110 and 155 (Figs. 3 and 4). The maximum electron transfer rate (\( {q}_{\mathrm{e}}^{\mathrm{max}} \))—combining the electron transfer capacities of N2O and O2—summed up to a value comparable with the \( {q}_{\mathrm{e}}^{\mathrm{max}} \) found during the N2O- or O2-only experiments. N2O reduction to N2 co-occurred with aerobic respiration only at relatively low concentrations of O2 (Fig. 3d). The experiments performed on days 110 and 155 differed regarding the O2 concentration range at which N2O reduction could co-occur (roughly < 4 and < 1.5 μM O2 on days 110 and 155, respectively) but, nevertheless, N2O reduction in the presence of O2 contributed to no more than a small fraction of the total electron acceptor capacity (generally < 20% of qe—tot; Fig. 4). An additional batch experiment on day 113, with a constant supply of O2 and a varying supply of N2O, also showed that N2O reduction was undetectable in the presence of relatively high concentrations of O2 (≈5 μM; Fig. 3c).

Fig. 3
figure 3

Offgas data from the batch experiments with excess N2O and varying concentrations of O2 on a day 125 and b day 155. The dissolved oxygen concentration (DO) was measured both with a Clark electrode (DO_1) and an optical sensor (DO_2). The biomass specific electron transfer rates to either N2O or O2 during these experiments are shown in Fig. 4. The asterisk marking the last two steps of the batch experiment on day 155 indicates the culture ran out of NH4+ for growth, and thus the rates during these steps was not considered. c Offgas data of batch experiment with excess O2 and varying concentrations of N2O on day 113. Detailed data from these experiments can be found in the Supplementary Materiasl—Tables xxx–xxx and Figures xxx to xxx. d Detailed view of one of the steps from the batch experiment depicted in (a) showing the simultaneous consumption of O2 and N2O, and subsequent production of N2

Full size image
Fig. 4
figure 4

Biomass specific transfers rate of electron equivalents (qe-) (a) to O2 and b N2O and c percentage of total electrons being shuttled to N2O vs. O2 at varying O2 concentrations during the batch tests on day 125 (in black) and day 155 (in gray). The Monod model of O2 consumption in the absence of N2O (shown in Fig. 3) is included in (a) for comparison. The inset in (c) is an enlargement of the graph at low O2 concentrations

Full size image

Discussion

Aerobic respiration was distinctly favored over N2O respiration in the enrichment despite the fact that the culture had been operated for an extensive number of generations with N2O as only electron acceptor. Upon a sudden change in supply from N2O to O2, the culture readily switched to O2 respiration and, when both electron acceptors were available, N2O reduction was only observed at relatively low concentrations of O2 (< 4 μM = 0.13 mg O2/L). Under conditions of electron acceptor excess (N2O and/or O2), growth in the system was likely limited by the electron supply rate to the electron transport chain (see Fig. 5) and not by the capacity of N2OR or O2 reductases. This was inferred from the fact that the maximum electron acceptor capacity of the culture was comparable for N2O and O2 respiration (i.e., \( {q}_{{\mathrm{e}}^{-}\mathrm{N}2\mathrm{O}}^{\mathrm{max}} \) ≈ \( {q}_{{\mathrm{e}}^{-}\mathrm{O}2}^{\mathrm{max}} \)), and could be due to kinetic limitations in acetate uptake, acetate oxidation in the citric acid cycle, or in some shared component of the ETC itself.

Fig. 5
figure 5

Simplified representation of the proportional distribution of electrons (e−) in the electron transport chain (ETC) during batch tests with only N2O (open arrow) versus batch tests with the simultaneous addition of O2 and N2O (black arrows) showing that there is a preferential shuttling of electrons to O2R than to N2OR. This simplified schematic is based on the assumptions that (i) both enzymes share a common electron pool (/quinone pool) and (ii) that all cells have a similar electron distribution among terminal reductases (whereas it would be possible for the majority of cells to switch fully to aerobic respiration, and a small fraction to continue respiring N2O)

Full size image

The overall electron transfer capacity during the simultaneous respiration of N2O and O2 (i.e., \( {q}_{{\mathrm{e}}^{-}\mathrm{TOT}}^{\mathrm{max}} \)) was comparable to \( {q}_{{\mathrm{e}}^{-}\mathrm{N}2\mathrm{O}}^{\mathrm{max}} \) or \( {q}_{{\mathrm{e}}^{-}\mathrm{O}2}^{\mathrm{max}} \)2. This suggests that “aerobic N2O respiration” (by analogy to aerobic denitrification) generally occurs if the electron supply rate to the ETC exceeds the electron accepting capacity of the O2 reductases. In other words, N2O respiration complements aerobic respiration primarily when O2 is limiting. Nonetheless, our results indicate that, under O2-limiting conditions, N2O reducers can use O2 and N2O mixotrophically as proposed by Chen and Strous 2013 (Fig. 5). We cannot exclude heterogeneity in electron acceptor use within the population in our bioreactor leading for example to most of the culture respiring O2 and a side population reducing N2O. Under the microscope, we did not observe formation of aggregates or biofilms which could create anoxic niches in spite of the O2 supply (data not shown), yet oxygen gradients and anoxic microzones could still form around suspended cells if O2 diffusion rate is slower than the respiration rate. Nevertheless, with the strong sparging and mixing conditions imposed on the culture, we would expect that most cells would be exposed to comparable environmental conditions.

The Ks values of the enrichment culture were in the same range as the Km values reported for purified N2OR and different O2 reductases in literature, i.e., in the μM range for N2O and nM range for O2 (Pouvreau et al. 2008 and references therein, Yoon et al. 2016). The relatively high KS,N2O (two orders of magnitude higher than for O2) is noteworthy in a culture presumably well-adapted to N2O-limiting conditions. Also the observation that, even after a prolonged absence of O2 in the environment, the cellular machinery specific for aerobic respiration (i.e., cytochrome oxidases) was constitutively present (in contrast to NO3− and NO2− reductases). According to these results, the preferential use of O2 over N2O in natural systems could be attributed to a difference in affinity (μmax/Ks) for O2 and N2O.

With regard to efficiency of N2O respiration versus O2 respiration, our chemostat enrichment cultures corroborate studies in literature (Koike and Hattori 1975; Stouthamer et al. 1982; Beun et al. 2000) and predictions based on our knowledge of the ETC in model denitrifiers (Chen and Strous 2013): with biomass yields per mole of acetate during growth with N2O (or NO3) roughly 1/3 lower than yields during O2 respiration (Table 2). The relatively low growth yields on N2O imply that N2O reduction to N2 is, thermodynamically, a very inefficient process with high energy dissipation. Thus, ensuring the maximization of energy conservation during microbial growth may be the evolutionary driver behind the preferential flow of electrons to O2 over N2O.

We cannot provide a conclusive answer regarding which cellular mechanism governs the preferential use of O2 in the presence of excess N2O observed. However, the instantaneous switch from N2O to O2 respiration suggests that the preference for O2 over N2O is regulated at the metabolome level and is independent from transcriptional regulation, e.g., by control of enzyme activity, like allosteric inhibition of N2OR, or simply a higher affinity of O2 reductases for the electrons coming from a common quinone pool.

Translated to the environmental conditions in a WWT plant, the results from this study suggest that oxic-anoxic transitions are unlikely to result in N2O emissions associated to denitrification as a result of N2OR inhibition by O2 since the enrichment culture readily switched back and forth between O2 and N2O respiration. This implies that (a) either N2OR is not directly inhibited by O2 in vivo or (b) inhibition is readily reversible once O2 is depleted.

On the other hand, the fact that aerobic respiration is so strongly favored over N2O respiration would make it a challenge to exploit the N2O sink capacity of activated sludge in the aerated/nitrification zones of WWT plants. The range in which significant N2O consumption co-occurred with O2 consumption in our experiments was narrow: roughly up to 1.5–4 μM O2, i.e., 0.05–0.13 mg O2/L, presumably below common DO values in the aerated tanks of WWTP (Tchobanoglous and Burton 2002). The very high affinity for oxygen minimizes the range of dissolved oxygen concentrations in which O2 and N2O respiration could occur in parallel. However, a beneficial difference in full-scale systems compared to our enrichment, in terms of avoiding N2O accumulation, may be that mass transfer limitation induced oxygen limitation within the activated sludge flocs provide anoxic zones, prone to N2O reduction, even when O2 is present in the bulk liquid (Picioreanu et al. 2016). This, together with the fact that N2O is much more soluble than O2, could perhaps be exploited to enhance the N2O sink capacity of activated sludge.