Mechanisms of transient nitric oxide and nitrous oxide production in a complex biofilm

Abstract

Nitric oxide (NO) and nitrous oxide (N₂O) are formed during N-cycling in complex microbial communities in response to fluctuating molecular oxygen (O₂) and nitrite (NO₂⁻) concentrations. Until now, the formation of NO and N₂O in microbial communities has been measured with low spatial and temporal resolution, which hampered elucidation of the turnover pathways and their regulation. In this study, we combined microsensor measurements with metabolic modeling to investigate the functional response of a complex biofilm with nitrifying and denitrifying activity to variations in O₂ and NO₂⁻. In steady state, NO and N₂O formation was detected if ammonium (NH₄⁺) was present under oxic conditions and if NO₂⁻ was present under anoxic conditions. Thus, NO and N₂O are produced by ammonia-oxidizing bacteria (AOB) under oxic conditions and by heterotrophic denitrifiers under anoxic conditions. NO and N₂O formation by AOB occurred at fully oxic conditions if NO₂⁻ concentrations were high. Modeling showed that steady-state NO concentrations are controlled by the affinity of NO-consuming processes to NO. Transient accumulation of NO and N₂O occurred upon O₂ removal from, or NO₂⁻ addition to, the medium only if NH₄⁺ was present under oxic conditions or if NO₂⁻ was already present under anoxic conditions. This showed that AOB and heterotrophic denitrifiers need to be metabolically active to respond with instantaneous NO and N₂O production upon perturbations. Transiently accumulated NO and N₂O decreased rapidly after their formation, indicating a direct effect of NO on the metabolism. By fitting model results to measurements, the kinetic relationships in the model were extended with dynamic parameters to predict transient NO release from perturbed ecosystems.

Introduction

Nitric oxide (NO) and nitrous oxide (N₂O) are produced and consumed by catabolic reactions of bacteria involved in the biogeochemical N-cycle. These reactions are fostered by increased anthropogenic N input into the environment leading to a steadily increasing atmospheric N₂O concentration. This is of environmental concern, because the infrared radiative forcing potential of N₂O is ∼200 times that of CO₂, which makes N₂O a potent greenhouse gas (Stein and Yung, 2003). Moreover, NO and N₂O are involved in a set of catalytic reactions that transform ozone to molecular oxygen (O₂) in the stratosphere (Crutzen, 1979).

Denitrification and nitrification are generally considered to be the two main processes responsible for the formation of NO and N₂O (Stein and Yung, 2003). Heterotrophic denitrification is the respiratory, sequential reduction of nitrate (NO₃⁻) or nitrite (NO₂⁻) via NO and N₂O to N₂ (Zumft, 1997). The key enzymes in denitrification are nitrite reductase (Nir), nitric oxide reductase (Nor) and nitrous oxide reductase (Nos). NO levels in heterotrophic denitrifiers are well regulated, independent of NO₃⁻ and NO₂⁻ concentrations and are in the range of low nanomolar concentrations (Goretski et al., 1990). NO consumption in heterotrophic denitrifiers might be mediated by widespread NO detoxifying enzymes, such as flavohemoglobins (Hmp or Fhp) and flavorubredoxin (NorVW), or respiratory NorB that can reduce NO to N₂O (Rodionov et al., 2005).

Nitrification is the aerobic oxidation of ammonium (NH₄⁺) performed by different groups of microorganisms. Aerobic NH₄⁺ oxidation (Aox) to NO₂⁻ is performed by ammonia-oxidizing bacteria (AOB) or archaea (Arp and Stein, 2003; Konneke et al., 2005). In a next step, NO₂⁻ is oxidized aerobically to NO₃⁻ (Nox) by nitrite-oxidizing bacteria (NOB). Several studies have demonstrated the production of NO and N₂O by pure cultures of AOB (Kester et al., 1997; Lipschultz et al., 1981; Shaw et al., 2006), but the mechanism is not completely understood. Generally two different pathways are inferred. First, the activity of hydroxylamine oxidoreductase (HAO) converts hydroxylamine (NH₂OH) to NO₂⁻ and releases small amounts of NO and N₂O (Hooper, 1968). Second, the activity of nitrifier-encoded Nir and Nor reduces NO₂⁻ to NO and N₂O in a process termed nitrifier denitrification (Poth and Focht, 1985; Bock et al., 1995; Schmidt et al., 2004). In both pathways, O₂ and NH₄⁺ are required to form NH₂OH, the electron donor for NO₂⁻ reduction.

NO and N₂O turnover has also been studied in complex microbial communities. Studies in soils and nitrifying granules revealed that denitrification and nitrification are the contributing microbial pathways. Commonly, NO and N₂O accumulation increased with decreased O₂ and with increased NO₂⁻ and NH₄⁺ concentrations (Conrad, 1996; Colliver and Stephenson, 2000; Kampschreur et al., 2008).

Change in environmental conditions leads to transient production of NO and N₂O in pure cultures of AOB and heterotrophic denitrifiers (Kester et al., 1997; Bergaust et al., 2008), as well as in mixed microbial communities (Kampschreur et al., 2008; Morley et al., 2008). This transient production can lead to high concentrations and might thus contribute significantly to NO and N₂O emissions from various habitats. Despite the importance of transient NO and N₂O formation, the coupling and regulation of the responsible pathways remain poorly understood in complex microbial communities. This is primarily because the experiments on release of NO and N₂O from natural samples commonly rely on analysis of the headspace volume or the bulk solution, which have no spatial and low temporal resolution. However, local conversion rates and limited transport in aggregated or attached microbial communities lead to stratification and to microenvironments that are different from the bulk solution. Thus, high spatial resolution measurements of NO, N₂O and O₂ are required to distinguish between the contribution of aerobic (Aox) and anaerobic (heterotrophic denitrification) processes to NO and N₂O emission from stratified ecosystems where nitrification and denitrification co-occur. In addition, high temporal resolution measurements during system perturbation are a powerful method for unraveling complex sets of processes and to obtain insights into the coupling of different pathways.

The objective of this study was to assign NO and N₂O production to nitrifying or denitrifying processes occurring in a complex biofilm. Furthermore, we aimed to quantify the influence of O₂, of NO₂⁻ fluctuations and of metabolic state on the dynamics of the transient NO and N₂O formation. We conducted microsensor measurements with high spatial and temporal resolution to characterize in situ microenvironmental conditions, quantify the rates of the relevant processes and follow NO and N₂O formation upon perturbations. Furthermore, we developed a novel metabolic model that allowed numerical simulations of the measured NO transitions. On the basis of this model, we propose mechanisms that explain transient turnover of NO and N₂O by AOB and associated heterotrophic denitrifiers in the studied biofilm.

Materials and methods

Biofilm growth

Biofilms were grown in an aerated flow cell (∼800 ml) using Tygon tubing as a surface for biofilm growth. The inoculum was obtained from a sewage treatment plant (Seehausen, Bremen) and was fed with medium at a flow rate of 1 ml min⁻¹. Biofilms with a thickness of 0.4–0.7 mm developed within 2–3 months in a medium consisting of 10 mM NH₄Cl and trace elements at final concentrations of 3 μM Na₂-EDTA, 1.5 μM FeSO₄, 77 nM H₃BO₄, 100 nM MnCl₂, 160 nM CoCl₂, 20 nM NiCl₂, 2.4 nM CuCl₂, 100 nM ZnSO₄ and 30 nM Na₂MoO₄ in tap water at pH 7.4. One month before the measurements, the medium was changed to a phosphate-buffered artificial freshwater medium containing 17 mM NaCl, 2 mM MgCl₂, 0.9 mM CaCl₂, 6.7 mM KCl and 1.5 mM KH₂PO₄/K₂HPO₄ (pH 7.5), supplemented with 400 μM NH₄Cl and trace elements as stated before. For microsensor measurements, small pieces of the biofilm-covered tubing were transferred into a smaller flow cell (∼80 ml) placed in an aquarium. The aquarium served as a reservoir for 1.7 l of aerated artificial freshwater medium (with or without NH₄Cl) that recirculated through the small flow cell at a flow rate of 3 ml s⁻¹ to create a constant flow of ∼0.2 cm s⁻¹ above the biofilm.

Experimental design

After biofilms adjusted to the conditions in the small flow cell for 1–2 days, steady-state microprofiles of O₂, pH, NH₄⁺, NO₂⁻, NO₃⁻, N₂O and NO were measured in the presence of 400 or 0 μM NH₄Cl and at varying O₂ and NO₂⁻ concentrations in the overlying water. The metabolic response of the biofilms was studied by changing the conditions in the aquarium in the following sequence: (1) starting condition with O₂ at air saturation, (2) switching to low O₂ (∼3% air saturation) by purging the medium with N₂, (3) addition of 3 mM NaNO₂ with O₂ at air saturation and (4) switching to low O₂ (∼3% air saturation) in the presence of 3 mM NaNO₂. The response to the addition of 3 mM NaNO₂ at low O₂ was investigated in separate experiments. Transient concentration changes of NO, N₂O and O₂ were monitored inside the biofilm upon switching the conditions until a new steady state was reached. The recirculated medium was sampled regularly to test for NO₃⁻ and NO₂⁻ accumulation from NH₄⁺, stability of NH₄⁺ concentrations, pH and temperature. NO₃⁻ and NO₂⁻ accumulated only in the presence of NH₄⁺, and reached maximum concentrations of approximately 30 and 5 μM, respectively. NH₄⁺ did not decrease below 370 μM. The temperature was 25–26 °C, and the pH was 7.2–7.3.

Microsensor measurements

Concentrations of O₂, N₂O and NO were measured with amperometric microsensors, whereas liquid-ion exchange microsensors were used for pH, NH₄⁺, NO₂⁻ and NO₃⁻ measurements. Microsensors were prepared and calibrated as previously described (de Beer and van den Heuvel, 1988; Revsbech, 1989; de Beer et al., 1997; Andersen et al., 2001; Schreiber et al., 2008). Vertical concentration profiles were measured with the microsensor mounted on a 3-axis micromanipulator (MM 33; Märzhäuser, Wetzlar, Germany). The vertical axis was motorized for μ-positioning (VT-80 linear stage; Micos, Germany, equipped with a 3564-K-024-BC motor, Faulhaber Group, Schönaich, Germany), and measurements were controlled by μ-Profiler software (www.microsen-wiki.net). The microsensor tip was adjusted manually to the sample surface with the help of a dissection microscope (Stemi SV 6; Carl Zeiss AG, Oberkochen, Germany).

Diffusive fluxes across the liquid–biofilm interface were calculated from the concentration gradients multiplied by the molecular diffusion coefficient D, as previously described. Values used for D were 2.34 × 10⁻⁹ m² s⁻¹ for O₂, 1.98 × 10⁻⁹ m² s⁻¹ for NH₄⁺, 1.86 × 10⁻⁹ m² s⁻¹ for NO₂⁻, 1.92 × 10⁻⁹ m² s⁻¹ for NO₃⁻, 2.36 × 10⁻⁹ m² s⁻¹ for N₂O and 2.21 × 10⁻⁹ m² s⁻¹ for NO (Broecker and Peng, 1974; Li and Gregory, 1974; Zacharia and Deen, 2005).

Metabolic modeling of NO production

We developed an N-cycle model that couples processes involved in the production and consumption of NO, O₂ and NO₂⁻ in the biofilm (Figure 1). We assumed that NH₄⁺ was aerobically converted to NO₂⁻ by Aox, and NO₂⁻ subsequently converted to NO₃⁻ by NO₂⁻ oxidation (Nox). Anaerobic conditions favor NO₂⁻ consumption by heterotrophic denitrification (hD). This results in the formation of NO, which is consumed in a sequential step by heterotrophic denitrification (hD-NO). Nitrifier denitrification by NH₄⁺-dependent AOB (niD) produced NO aerobically from NO₂⁻. Subsequently, NO is consumed by nitrifier denitrification (niD-NO). In addition, the model incorporated chemical NO oxidation with O₂ to NO₂⁻ (chem) and O₂ consumption by heterotrophic respiration (hR).

Figure 1

Schematic representation of the metabolic model of NO turnover in a complex biofilm. Pathways and compounds marked by solid black lines were calculated by the numerical model, whereas those marked with gray dashed lines were not calculated. Italic text next to the arrows indicates the pathways: Aox, ammonium oxidation; Nox, nitrite oxidation; niD, NO production by nitrifier denitrification; niD-NO, NO consumption by nitrifier denitrification; hD, NO production by heterotrophic denitrification; hD-NO, NO consumption by heterotrophic denitrification; hR, heterotrophic oxygen respiration; chem, chemical NO oxidation. Additional mechanisms that influence the rate of the respective pathway, with the compounds that affect those mechanisms in parentheses, are indicated by T, threshold; T̃, extended threshold; Dyn, dynamic function and Sh, shift function (see text and Equations (4), (5), (6) and (7) for details).

This model was implemented numerically in MATLAB (The MathWorks, Inc., Natick, MA, USA, the code is available at www.microsen-wiki.net) to describe the kinetic, metabolic and mass-transfer control of NO, O₂ and NO₂⁻ in the biofilm. We assumed that the biofilm was laterally homogeneous and transport was governed by diffusion. Thus, the dynamics of a solute with concentration C (mol m⁻³) in and above the biofilm was described by a one-dimensional diffusion-reaction equation

Here, C denotes concentration of NO, NO₂⁻ or O₂, D_C is the corresponding diffusion coefficient in water (m² s⁻¹), which was assumed to be constant throughout the biofilm and R_C (mol m⁻³ s⁻¹) is the net reaction rate at which the solute is produced or consumed. Both C and R_C were explicit functions of time t and depth z, with z>0 and z<0 corresponding to the biofilm and overlying water, respectively. The reactions for production and consumption of NO, O₂ and NO₂⁻, as shown in Figure 1, were stochiometrically balanced and used to express the net reaction rates R_C for NO, O₂ and NO₂⁻ as

Each individual rate was an explicit function of the concentrations of the solutes involved in the process and is described in detail in the Supplementary Table S1.

Kinetic control of the reaction rates Rⁱ was described by the Michaelis–Menten law,

where K_Cⁱ (mol m⁻³) is the affinity constant of the process i to substrate C, v_maxⁱ (mol m⁻³ s⁻¹) is the maximum rate of the process i and M denotes the Michaelis–Menten function describing the relationship between affinity and concentration without v_max. The affinity constants were obtained from the literature and the maximum rates were derived from the measured steady-state fluxes at the liquid–biofilm interface (as in Tables 1 and 2) divided by the biofilm thickness. The maximum rates of NO-consuming processes (niD-NO under oxic conditions and hD-NO under anoxic conditions) were determined by assuming that both processes were coupled to the respective production process. This allowed subtraction of the net NO production rate from the rate of the NO-producing process, which was determined from NO₂⁻ consumption.

Table 1 Fluxes of measured solutes through the liquid–biofilm interface determined from microprofiles in a biofilm with the medium containing 400 μM NH₄Cl

Table 2 Maximum activity of selected processes in the biofilm under different conditions

Metabolic control of the reaction rates was implemented by combining information available from pure culture studies with postulated mechanisms based on data presented in this work. First, maximum activity of NO production (R^hD) and NO consumption (R^hD-NO) by heterotrophic denitrification occurred at micro-oxic to anoxic conditions, that is, at O₂ below a certain threshold concentration ${\Theta }_{{\text{O}}_{\text{2}}}^i$. This was implemented by multiplying the maximum reaction rates, v_max^hD and v_max^hD-NO, with a threshold function

where δ_Cⁱ represents the width of the concentration interval over which the threshold function changes from 1 to 0 (see Supplementary Figure S1). Furthermore, NO production by nitrifier denitrification (R^niD) was allowed only at low O₂ or high NO₂⁻ concentrations, which was achieved by multiplying the corresponding v_max^niD value with an extended threshold function (see Supplementary Figure S2)

Threshold values were chosen by biological reasoning and by matching the concentration dynamics observed in the measurements.

Second, our experimental data suggested that after the O₂ concentration has decreased below a certain threshold, and when NO₂⁻ was simultaneously present in sufficient amounts, the rate of NO production by heterotrophic denitrification, R^hD, increased slowly with time (Figure 3d). This mechanism was implemented by further multiplying the v_max^hD value with the dynamic function

where t₀ is the time at which O₂ decreased below a threshold value ${\Theta }_{{\text{O}}_{\text{2}}}^{\text{hD}}\text{-2}{\text{Δ}}_{{\text{O}}_{\text{2}}}^{\text{hD}}$ whereas NO₂⁻ was simultaneously present above 1 μM. The value of Δt=400 s was estimated by fitting the measured dynamic increase of NO in the presence of NO₂⁻ after the conditions in the overlying water were switched from oxic to suboxic (Figure 3d). In contrast, the dynamic increase of heterotrophic denitrification toward its maximum rate was accelerated to Δt=4 s if NO was present above Θ_NO^hD=0.32 μM. This assumption is based on reported evidence that NO serves as a signal for the expression of denitrification genes (Zumft, 2002, 2005).

Third, a shift mechanism (Sh) was implemented to model the instantaneous increase in NO concentration after NO₂⁻ was added under oxic conditions (Figure 4a). Reasoning for this mechanism was based on the assumption that HAO function is impaired by NO₂⁻, leading to the release of NO that is reported to be an HAO-bound intermediate (Arp and Stein, 2003). This was implemented by removing a fraction f of the NO₂⁻ production rate by Aox from $R_{{\text{NO}}_2^{\text{-}}}$ in Equation (2c) and adding it to the total net NO production rate R_NO in Equation (2a) as a function

Here, t₀ is the time when NO₂⁻ reached a threshold concentration of ${\Theta }_{{\text{NO}}_2^{\text{-}}}^{\text{Sh}}$ = 200 μM. The fraction of the shifted R^Aox decreased exponentially with time, resembling an adjustment of AOB metabolism after perturbation. The values of f=0.55 and Δt=200 s were estimated from the experimental data (Figure 4a).

The time-dependent diffusion-reaction equations (Equation (1)) were solved for all solutes using boundary conditions: (1) solute concentrations were fixed to the concentrations in the overlying water, C_w, at the top of the diffusive boundary layer (DBL), that is, C(−z_DBL)=C_w, and (2) the diffusive flux at the base of the biofilm was set to zero, that is, ∂C/∂z(z_B)=0, where z_DBL and z_B denote the thickness of the DBL and the biofilm, respectively. Experimental perturbations that resulted in O₂ decrease and NO₂⁻ increase in the medium were implemented by varying C_w over time. Experiments performed in the absence of NH₄⁺ were modeled by excluding all processes from NO, NO₂⁻ and O₂ turnover that require NH₄⁺ as electron donor, namely, Aox, nitrifier denitrification (niD) and NO consumption by nitrifier denitrification (niD-NO).

All parameters of the model are listed in Supplementary Table S1. When available, they were adjusted within a biologically reasonable range of values reported in the literature; otherwise, they were adjusted to match the experimental data presented in this work. In the paper, the model is used to interpret and discuss the experimental findings.

Results

Performance of the biofilm in steady state

For all compounds, the microprofiles showed either production or consumption within the entire biofilm. Stratified zones of production and consumption were not apparent (Figure 2 and Supplementary Figure S4). Thus, the overall performance of the biofilm was estimated from the fluxes across the liquid–biofilm interface (Table 1). Maximum potentials of Aox, Nox, nitrifier denitrification and heterotrophic denitrification were determined by creating the conditions such that only the process of interest occurred, and coupled processes were inhibited (Table 2).

Figure 2

Averaged steady-state microprofiles of NO (a and c) and N₂O (b and d) in a complex biofilm. Microprofiles were measured in artificial freshwater medium containing 400 μM NH₄Cl with 5 μM NO₂⁻ (filled symbols) or with 3 mM NO₂⁻ (open symbols) and during aeration (a and b) or N₂ purging (c and d) of the medium. The dashed line represents the biofilm surface. Horizontal bars represent standard errors (number of profiles is given in Table 1).

The biofilm was fully oxic when the medium was aerated (Supplementary Figure S4A). In the presence of O₂, NH₄⁺ was completely converted to NO₃⁻ with minor accumulation of NO₂⁻ (Supplementary Figure S4D), indicating that Aox and Nox occurred at similar rates (Tables 1 and 2). The coupling between Aox and Nox was not affected by addition of 3 mM NO₂⁻. This was indicated by a similar 1:2:1 stoichiometry of NH₄⁺, O₂ and NO₃⁻ fluxes in both the absence and the presence of NO₂⁻ (Table 1). However, nitrifier denitrification was induced by addition of NO₂⁻ in the presence of NH₄⁺ and O₂. Nitrifier denitrification was indicated by the fact that the gross NO₂⁻ uptake, calculated from net NO₂⁻ uptake from the medium and NO₂⁻ production by Aox, exceeded the maximum NO₂⁻ consumption potential of Nox (Table 2). The remaining NO₂⁻ was reduced by AOB, with a rate that was ∼20% of the NO₂⁻ production rate of Aox. Consumption of NH₄⁺ and O₂ was slightly elevated in the presence of NO₂⁻ (Table 1).

In the absence of NH₄⁺, the potential for heterotrophic processes was detectable, which were probably performed at the expense of reduced organic carbon present in the biofilm. Under oxic conditions, heterotrophic respiration of O₂ accounted for ∼15% of the total O₂ consumption. In the absence of O₂, NO₂⁻ consumption by heterotrophic denitrification was ∼50% of the NO₂⁻ consumption by NOB (Tables 1 and 2).

The effects of NO₂⁻ and O₂ on the formation of NO and N₂O in NH₄⁺-containing medium are summarized in Table 1 and Figure 2. In the presence of NH₄⁺ and high NO₂⁻, NO and N₂O were produced regardless of the O₂ concentration in the medium. In contrast, at low NO₂⁻, NO production was observed only under anoxic conditions, whereas N₂O production was low regardless of O₂. N₂O concentrations in the biofilm were an order of magnitude higher than NO concentrations, with NO ranging from <0.02 to 0.35 μM and N₂O from 0.35 to 5.4 μM. In the presence of high NO₂⁻, the yields were 0.007 mol NO per mol NH₄⁺ and 0.022 mol N₂O per mol NH₄⁺. At low NO₂⁻, the N₂O yield was reduced to 0.015 mol N₂O per mol NH₄⁺ (Table 1). In the absence of NH₄⁺ and NO₂⁻, NO and N₂O fluxes were negligible regardless of O₂. However, in the absence of NH₄⁺ and presence of NO₂⁻, NO and N₂O were produced, but only under anoxic conditions (Supplementary Figure S6). The resulting fluxes were in the same range as those observed in the presence of NH₄⁺ and NO₂⁻ under anoxic conditions.

Transient NO and N₂O formation in response to O₂ and NO₂⁻ changes

Upon the start of N₂ purging, O₂ decreased gradually in the biofilm until anoxic conditions were reached. The transient phase lasted ∼7 min in the presence of NH₄⁺ and ∼12 min in the absence of NH₄⁺ (Supplementary Figure S3). During this transition, highly dynamic concentration changes of NO and N₂O were detected with microsensors positioned in the biofilm at 200 μm depth (Figures 3a–d).

Figure 3

(a–d) Time series of measured NO (black line) and N₂O (gray line) concentrations with microsensors inserted in the biofilm at 200 μm depth. Purging of the medium with N₂ started at t=0 min. (e–h) Time series of NO derived from the model shown in Figure 1. In each row, the boundary conditions and perturbations were implemented in the model such that they corresponded to the conditions applied during the measurement. Different stages of the model are shown, including (1) the model governed by kinetics and thresholds only (T and T̃; Equations (4) and (5); filled circles), (2) the model implementing the dynamic function (Dyn, Equation (6)) on heterotrophic denitrification (open circles), (3) the model implementing the dynamic function controlled by NO concentration (open triangles), (4) and the model implementing NO loss by diffusion after stopping all processes when the peak NO concentration was reached (filled triangles). White background indicates oxic, shaded areas indicate the transient phase from oxic to anoxic and gray areas indicate anoxic conditions. The medium composition with respect to NH₄⁺ and NO₂⁻ is depicted at the top-left of each row.

Decreasing O₂ concentrations in the presence of NH₄⁺ caused a transient accumulation of NO and N₂O, which decreased to a new steady state after anoxic conditions were reached (Figures 3a and b). Although the accumulation was more pronounced at high NO₂⁻ concentration, the final steady-state levels were on average comparable to those observed at low NO₂⁻ (see also Figure 2c). Control measurements showed that in the absence of NH₄⁺ and NO₂⁻, NO and N₂O were neither produced nor consumed during the decrease of O₂ concentration (Figure 3c). In contrast, the absence of NH₄⁺ at high NO₂⁻ concentration resulted in slow formation of NO and N₂O, starting shortly before anoxic conditions were reached (Figure 3d).

When NO₂⁻ was added under oxic conditions, NO concentration increased within less than a minute from below the detection limit to 1.2 μM, after which it decreased within 20 min to a new steady state. This was observed only if NH₄⁺ was present (Figure 4a). In the absence of O₂, concentrations of NO increased upon the addition of NO₂⁻. The presence of NH₄⁺ did not influence the final NO steady-state concentrations, but affected the kinetics of its formation. NO formed slowly in the absence of NH₄⁺. In contrast, the presence of NH₄⁺, which resulted in low concentrations of NO₂⁻ and NO₃⁻ in the medium, caused an instantaneous increase of NO from slightly elevated levels (Figure 4b).

Figure 4

(a and b) Time series of NO measured with a microsensor inserted in the biofilm at 200 μm depth. 3 mM NO₂⁻ was added at t=0 min to the medium containing 400 μM NH₄⁺ and ∼5/30 μM NO₂⁻/NO₃⁻ (black line) or to the medium that did not contain NH₄⁺ and NO₂⁻/NO₃⁻ (gray line). White and gray backgrounds indicate oxic and anoxic conditions, respectively. (c and d) Time series of NO derived from the model (see Figure 1). In each row, the boundary conditions and perturbations were implemented in the model such that they corresponded to the conditions applied during the measurement. Different symbols depict different stages of the model. In panel (c), this includes the model governed by kinetics and thresholds only (filled circles), the model additionally implementing a shift function (Sh) that resulted in the production of NO instead of NO₂⁻ by Aox (open triangles), the model implementing NO loss by diffusion after stopping all processes when the peak NO concentration was reached (filled triangles) and the control condition where all NH₄⁺-dependent processes were switched off (open circles). In panel (d), this includes the model governed by kinetics and thresholds only (filled circles), with the dynamic function added (open triangles) and the model in the absence of NH₄⁺-dependent processes governed by kinetics and thresholds only (closed triangles) or with the dynamic function included (open circles).

Discussion

Regulation of steady-state NO and N₂O production by ammonium oxidation under oxic conditions and by heterotrophic denitrification under anoxic conditions

NO and N₂O formation within a complex N-cycling community could be mediated by processes, such as Aox, Nox, heterotrophic denitrification or anaerobic oxidation of NH₄⁺ (anammox) (Freitag et al., 1987; Stein and Yung, 2003; Kartal et al., 2007). Measuring the in situ activities and microenvironmental conditions enabled us to assign concomitant NO and N₂O formation to active processes.

Ammonium oxidation in steady state

AOB require NH₄⁺ and O₂ to form NO and N₂O either by the HAO pathway or by nitrifier denitrification. The present data showed that NO and N₂O formation under oxic conditions depended on the presence of NH₄⁺. NO and N₂O formation did not occur upon addition of NO₂⁻ when NH₄⁺ was absent (Figure 4a, Supplementary Figures S6A and S6B). These experiments showed that under oxic conditions, NO and N₂O were formed by AOB, but not by NOB or (aerobic) heterotrophic denitrifiers. Previous studies emphasized the dependence of nitrifier denitrification on reduced O₂ and elevated NO₂⁻ concentrations (Lipschultz et al., 1981; Poth and Focht, 1985; Kester et al., 1997; Beaumont et al., 2004a; Kampschreur et al., 2008). Our experiments showed that nitrifier denitrification and simultaneous NO and N₂O formation occurs at high NO₂⁻ concentrations even if O₂ concentrations are high (Figures 2a and 4a and Supplementary Figure S4A). This was previously observed for Nitrosomonas europaea and Nitrosospira spp. (Shaw et al., 2006). Moreover, it has been shown that denitrifying enzymes (NirK and NorB) in N. europaea are expressed under fully oxic conditions (Beaumont et al., 2004a, 2004b).

In the model (Figure 1), NO production by nitrifier denitrification was kinetically controlled by O₂ and NO₂⁻. The resulting steady-state NO concentration was kinetically controlled by the affinity of NO consumption by nitrifier denitrification to NO. However, NO production was only measured at high NO₂⁻ concentrations or under micro-oxic conditions. This indicates that NO accumulation under those conditions was controlled by NO production. To limit NO production in the model to high NO₂⁻ and low O₂ conditions, we implemented an extended threshold function (T̃, Equation (5), Supplementary Figure S2). This allowed an independent influence of O₂ and NO₂⁻ on NO production with a restricted maximum rate, excluding additive effects when both conditions were present at the same time.

Heterotrophic denitrification in steady state

Earlier studies have reported the ability of AOB pure cultures to produce NO under anoxic conditions (Kester et al., 1997; Schmidt et al., 2001) and reported that in a nitrifying mixed culture, NH₄⁺ affected anaerobic NO formation in a concentration-dependent manner (Kampschreur et al., 2008). Conversely, we found that under anoxic conditions, NO and N₂O formation did not depend on the presence of NH₄⁺ and was only observed when NO₂⁻ and NO₃⁻ were present. This showed that heterotrophic denitrifiers, but not O₂-depending AOB, were responsible for NO and N₂O formation under anoxic conditions. The dependence of NO and N₂O formation on NO₂⁻ was confirmed under anoxic conditions in the absence of NH₄⁺ (Supplementary Figures S6C and S6D). In the presence of NH₄⁺, nitrification leads to accumulation of approximately 5 μM NO₂⁻ and 30 μM NO₃⁻ in the medium before N₂ purging. NO₃⁻ and NO₂⁻ served as electron acceptors for heterotrophic denitrification under subsequent anoxic conditions. This explains the formation of NO and N₂O under anoxic conditions by heterotrophic denitrification in the presence of NH₄⁺ but not in its absence. Anammox could oxidize NH₄⁺ and reduce NO₂⁻ under anoxic conditions, and could account for NO and N₂O formation. However, the biofilms grew under oxic conditions, which hamper growth of anammox bacteria (Strous et al., 1997). Furthermore, NO₂⁻ uptake of the biofilm under anoxic conditions was similar in the presence and the absence of NH₄⁺ (Tables 1 and 2), indicating that anammox did not contribute to additional NO₂⁻ reduction.

In the model (Figure 1), Michaelis–Menten kinetic dependence of heterotrophic denitrification on NO₂⁻ allowed NO production in the presence of NO₂⁻, whereas threshold functions (T, Equation (4), Supplementary Figure S1) for O₂ restricted the process to anoxic conditions. The measured steady-state NO concentrations can be modeled (compare Figures 2a,c with Figures 3e,f and 4c,d) by kinetically controlling its accumulation with low K_m values for the NO consumption pathways, as has been described elsewhere (Betlach and Tiedje, 1981). As a result, steady-state NO concentrations under anoxia were more or less independent of NO₂⁻ concentrations, even though NO₂⁻ concentrations varied over ∼3 orders of magnitude, corresponding to observations in pure culture studies (Goretski et al., 1990). In addition, the K_m value of heterotrophic denitrifiers for NO₂⁻ was very low in the model, resulting in a minor increase of NO production due to the additional NO₂⁻, because NO₂⁻ consumption was already saturated at low NO₂⁻ concentrations. In contrast, measured N₂O, the sequential product of NO reduction, showed a marked increase in the presence of high NO₂⁻ concentrations. Thus, in heterotrophic denitrifiers, accumulation of NO may not occur at high NO₂⁻ concentrations due to efficient reduction of NO to N₂O.

Furthermore, the model (Figure 1) showed that chemical NO consumption under oxic conditions was always several orders of magnitude lower than the biological rate, and contributed insignificantly to the total NO loss. Chemical production of NO from acidic decomposition of NO₂⁻ did not play a role, because the biofilm pH did not drop below 6.5 (Supplementary Figures S5A and S5B).

A recently described model was able to assign NO accumulation in a nitrifying reactor to nitrifier denitrification by implementing kinetic control and by modeling different production scenarios independent from each other (Kampschreur et al., 2007). In our model, the use of threshold functions affected by O₂ and NO₂⁻ concentrations in addition to Michaelis–Menten kinetics enabled us to consider all possible NO-producing pathways simultaneously under varying conditions.

Perturbations of active ammonia oxidation and heterotrophic denitrification cause instantaneous NO and N₂O formation

The time series measurements showed that NO and N₂O reached transient maxima within the biofilm upon decreasing O₂ and adding NO₂⁻. This phenomenon was also observed in studies with pure cultures of AOB and heterotrophic denitrifiers (Kester et al., 1997; Bergaust et al., 2008), as well as with mixed microbial communities (Kampschreur et al., 2008; Morley et al., 2008). However, these observations are not completely understood. We observed transient maxima of NO and N₂O only when either Aox or heterotrophic denitrification was actively performed before O₂ and NO₂⁻ changed. We conclude that the instantaneous formation of NO and N₂O is caused by perturbation of the active pathways on the enzyme level, resulting in an imbalance of the fine-tuned mechanisms that maintain NO homeostasis. NO and N₂O dynamics following perturbation were similar, demonstrating that they are sequentially produced by AOB and heterotrophic denitrifiers. The model can be extended to N₂O concentrations if K_m values for N₂O reduction are available.

Heterotrophic denitrification in transient states

NO₂⁻ additions were followed by instantaneous NO formation, but only if NO₂⁻ and NO₃⁻ were present at low concentrations and if O₂ was absent before the perturbation (Figure 4b). Under such conditions, only heterotrophic denitrification can cause NO formation. The instantaneous formation of NO upon NO₂⁻ addition shows that NO-producing and NO-consuming enzymes are directly affected by NO₂⁻ and that de novo synthesis of enzymes cannot explain the dynamics. For example, it was demonstrated previously that NO₂⁻ can directly inhibit Nor (Kucera et al., 1986; Kucera, 1992), which would lead to accumulation of NO. In addition, kinetics of NO₂⁻- and NO-reducing enzymes might allow NO accumulation depending on the NO₂⁻ concentration. Instantaneous NO production was modeled by assuming that full expression of heterotrophic denitrification potential requires the absence of O₂ and the presence of NO₂⁻ (Zumft, 2005). Under conditions allowing full expression of heterotrophic denitrification, kinetics explained the instantaneous increase of NO upon addition of NO₂⁻ in the model (Figure 4b, black line and Figure 4d, filled circles).

In contrast, measured NO increased slowly if the conditions before the perturbation did not allow full expression of denitrification (that is, the presence of O₂ or absence of NO₂⁻; Figure 3d, black line and Figure 4b, gray line). We assumed that in these cases, removal of O₂ (Figure 3d) or addition of NO₂⁻ (Figure 4b) results in the expression of denitrification enzymes. The subsequent slow NO increase was modeled by implementing a dynamic function (Dyn; Equation (6)) in addition to the kinetic control (Figures 3h and 4d open circles). Modeling without the dynamic function resulted in an instantaneous increase of NO for transients upon O₂ removal and NO₂⁻ addition (Figure 3h, filled circles and Figure 4d, filled triangles), because in this case the expression of heterotrophic denitrification was assumed to be constitutive. This shows that the requirements for the expression of enzymes for heterotrophic denitrification, and their slow expression when all conditions are met, need to be included when modeling transient NO accumulation.

Ammonium oxidation in transient states

The measurements showed that NO formed instantaneously if O₂ was removed or NO₂⁻ was added, but only if NH₄⁺ was available to allow active Aox before the perturbation. AOB form NO and N₂O by the HAO pathway or by nitrifier denitrification (Arp and Stein, 2003). Both pathways might contribute to NO and N₂O formation, but cannot be separated from each other by our experiments. Our model showed that the instantaneous transient increase of NO caused by AOB upon NO₂⁻ addition (Figure 4a) cannot be explained solely with NO production by nitrifier denitrification. Here, NO increased slowly to steady state because of the delayed diffusion of NO₂⁻ into the biofilm matrix (Figure 4c, filled circles). The transients upon NO₂⁻ addition were satisfactorily modeled by implementing a shift function (Sh; Equation (7)), which resulted in the production of NO instead of NO₂⁻ from NH₄⁺ when NO₂⁻ concentrations increased above a certain threshold $\text{(}{\Theta }_{{\text{NO}}_2^{\text{-}}}^{\text{Sh}}=200\text{ μM)}$ (Figure 4c, open triangles). The successful use of this function supports our hypothesis that NO formation by AOB occurs if their active metabolism is disturbed by NO₂⁻ that possibly impairs the smooth functioning of HAO (Arp and Stein, 2003).

AOB-dependent transient increase of NO during O₂ decrease in the model resulted from a micro-oxic threshold imposed on NO production by nitrifier denitrification $\text{(}{\Theta }_{{\text{O}}_2}^{\text{niD}}=40\text{ μM)}$ (Figures 3e and f, filled circles). The increase of NO during micro-oxic conditions was not counteracted by NO consumption due to nitrifier denitrification (niD-NO) because O₂ disappeared before the K_m value for NO (0.6 μM) was reached. Rather, the onset of anoxia resulted in the decline of NO production by nitrifier denitrification (niD) because of kinetic limitations by the substrate O₂. The modeled results suggest that nitrifier denitrification metabolism is fully expressed under oxic conditions as reported by Beaumont et al. (2004a), resulting in instantaneous NO formation upon reduced O₂ concentrations. This increase cannot be counteracted until NO accumulates to concentrations equal to the K_m value of the NO consumption pathway.

Direct regulation of NO and N₂O decrease after its transient accumulation

NO and N₂O decreased to a new steady-state level after they reached peak concentrations upon O₂ and NO₂⁻ concentrations were changed. NO concentrations were always one order of magnitude below N₂O, indicating that regulation of potentially cytotoxic NO (James, 1995) is more critical than that of nontoxic N₂O, and that the K_m value of N₂O reduction is higher than that of NO reduction. The decrease of NO within minutes after the accumulation of NO indicates that genetic regulation cannot explain the dynamics. Instead, different metabolic pathways governed NO turnover after the conditions changed, or enzymes were directly affected after the peak was reached, for example, by inhibition.

NO decrease after O₂ removal

The pathways governing NO turnover switched from nitrifier denitrification to heterotrophic denitrification between oxic and anoxic conditions (Figures 3a and b). Directly after nitrifier denitrification stops, heterotrophic denitrification cannot instantaneously start, as the enzymes must be expressed. Therefore, modeling with the dynamic function (Dyn; Equation (6)) showed a transient NO minimum when reaching anoxia, because NO was lost from the biofilm by diffusion and increased slowly thereafter (Figures 3e and f, filled triangles and open circles). However, we did not measure such a minimum, indicating that NO production continued directly after the transition. Possible explanations are that either nitrifier denitrification continued or the dynamic of expression of heterotrophic denitrification (Dyn, Equation (6)) was enhanced in the transient phase in response to NO. Modeling showed that the later scenario was not biologically feasible, because the expression of heterotrophic denitrification needed to be enhanced to be 100-fold (Δt=4 s instead of 400 s) faster than was calculated for the onset denitrification upon reduced O₂ (Figures 3e and f, open triangles and Figure 3h, open circles). Alternatively, NO production by nitrifier denitrification might continue for a short time at anoxia with oxidizing compounds stored in the form of NH₂OH or oxidized cytochromes.

NO decrease after NO₂⁻ addition

O₂ concentrations remained stable upon the addition of NO₂⁻, which prevented switching between Aox and heterotrophic denitrification that could affect NO. Thus, the instantaneous decrease of NO after reaching the peak concentration (Figures 4a and b) might be regulated by direct effects on the enzymes involved in NO turnover within the bacteria that produce NO. This is supported by earlier studies, which showed that Nir and Nor can be inhibited by high NO concentrations in heterotrophic denitrifiers (Dhesi and Timkovich, 1984; Carr and Ferguson, 1990; Koutny and Kucera, 1999).

The model (Figure 1) suggests that the transient NO maximum after the addition of NO₂⁻ was due to an imbalance of the AOB metabolism (Sh, Equation (7)). Moreover, the decrease of the measured concentration from the maximum was slower than the modeled decrease by diffusion only (Figure 4a, black line and Figure 4c, filled triangles), indicating that NO production continued during the decrease. The model showed that if nitrifier denitrification is the only NO-producing process under oxic conditions, NO increased slowly after NO₂⁻ was added (Figure 4c, filled circles). Therefore, we propose that the direct formation of NO instead of NO₂⁻ continues with a decreasing rate after it becomes active. This implies that NO or NO₂⁻ concentrations directly affect the enzymes of AOB, resulting in the regulation of the metabolic imbalance.

Significance to NO and N₂O formation in the environment

The extent of transient NO and N₂O accumulation upon perturbation strongly indicates that it can significantly contribute to emissions of gaseous N oxides into the atmosphere, especially from habitats exposed to fluctuations in O₂ and inorganic N compounds. It is possible that the large uncertainties about the sources in the global N₂O budget (Stein and Yung, 2003) are linked to the contribution of fluctuating emissions from ecosystems, which are not normally considered during measurements (for example, measurements in chambers often impede exposure to environmental fluctuations).

Fluctuations in environmental conditions that affect O₂ and NO₂⁻ concentrations and thus lead to NO and N₂O production may occur in soils as a result of drying and wetting, or by a variable fertilizer input. Furthermore, estuaries are exposed to a fluctuating N input through precipitation and fertilizer runoff from land. Large oceanic volumes are influenced by mixing of water masses with different O₂ and NO₂⁻ concentrations. For example, massive accumulation of N₂O was observed in the upper layer of the Arabian Sea during severe upwelling-induced hypoxia on the western Indian shelf (Naqvi et al., 2000). Under these conditions, the upper layer of the water body is especially exposed to O₂ fluctuations caused by O₂ input through wave action. Hence, the fact that oceans are still considered a relatively minor source of N₂O in the global budget might be an underestimate linked to difficulties in measuring and modeling N₂O emissions during frequently occurring, short-term environmental perturbations in marine waters. Presently, a linear, empirical-derived framework is employed in large-scale N₂O emission models (Jin and Gruber, 2003; Duce et al., 2008). The characteristics of the metabolic model, such as thresholds, dynamic function and shift function, developed in this study to describe transient production mechanisms, may represent an important step toward more mechanism-based modeling.

Conclusions

In conclusion, characterization of microenvironmental conditions is required to determine the source of NO and N₂O production in complex, stratified environments. The presence of O₂ determines whether NO and N₂O are produced by AOB or by heterotrophic denitrifiers. Interestingly, NO and N₂O formation by AOB does not require micro-oxic conditions if NO₂⁻ is present in high concentrations (mM range). On the other hand, NO production, but not N₂O production, by heterotrophic denitrifiers is almost independent of NO₂⁻ concentrations. The high temporal resolution achieved by microsensors allowed the measurement of highly dynamic NO and N₂O formation following the change in O₂ and NO₂⁻ concentrations. Interpretation of the results with a metabolic model showed that Aox and heterotrophic denitrification need to be actively performed to respond to perturbations in O₂ and NO₂⁻ with instantaneous NO and N₂O production. The resulting transient accumulation is counteracted within minutes by regulating the NO turnover in the biofilm. This occurs either because a different pathway becomes active or because enzymes involved in the production process are affected directly. At steady state, NO concentrations are regulated by kinetic control of the consumption processes. In a complex environment, massive formation of NO and N₂O may occur if a metabolically active N-cycling community is exposed to an external perturbation.