Chemolithoautotrophic AOB (ammonia-oxidizing bacteria) form a crucial component in microbial nitrogen cycling in both natural and engineered systems. Under specific conditions, including transitions from anoxic to oxic conditions and/or excessive ammonia loading, and the presence of high nitrite (NO2−) concentrations, these bacteria are also documented to produce nitric oxide (NO) and nitrous oxide (N2O) gases. Essentially, ammonia oxidation in the presence of non-limiting substrate concentrations (ammonia and O2) is associated with N2O production. An exceptional scenario that leads to such conditions is the periodical switch between anoxic and oxic conditions, which is rather common in engineered nitrogen-removal systems. In particular, the recovery from, rather than imposition of, anoxic conditions has been demonstrated to result in N2O production. However, applied engineering perspectives, so far, have largely ignored the contribution of nitrification to N2O emissions in greenhouse gas inventories from wastewater-treatment plants. Recent field-scale measurements have revealed that nitrification-related N2O emissions are generally far higher than emissions assigned to heterotrophic denitrification. In the present paper, the metabolic pathways, which could potentially contribute to NO and N2O production by AOB have been conceptually reconstructed under conditions especially relevant to engineered nitrogen-removal systems. Taken together, the reconstructed pathways, field- and laboratory-scale results suggest that engineering designs that achieve low effluent aqueous nitrogen concentrations also minimize gaseous nitrogen emissions.
Lithotrophic AOB (ammonia-oxidizing bacteria) such as Nitrosomonas europaea produce nitric oxide (NO) and nitrous oxide (N2O) both as part of O2-sufficient nitrogen metabolism  and during O2 limitation [2–5], often resulting from transient switching between oxic and anoxic conditions or varying nitrogen loads [6–8]. One mode of NO and N2O production by AOB is termed nitrifier denitrification and involves sequential reduction of nitrite (NO2−) to NO and N2O via nitrite reductase and nitric oxide reductase respectively [9,10]. Nitrifier denitrification is active during aerobic conditions and is also especially promoted during O2-limiting conditions and high NO2− concentrations. The second mode of N2O production by AOB involves oxidation of hydroxylamine (NH2OH) to NO by HAO (hydroxylamine oxidoreductase)  possibly via nitroxyl (HNO, or NO− as it commonly exists at physiological pH values around 7.0) and subsequent reduction to N2O potentially by either isolated or concerted activity of cytochrome c′-beta and sNOR, a haem–copper nitric oxide reductase (as summarized in [10,12]). This mode, which is termed nitrification-dependent NO and N2O production, is active only under oxic conditions. NH2OH is in turn produced from the oxidation of ammonia (NH3) by AMO (ammonia mono-oxygenase) (as summarized in Figure 1). In addition, it has been proposed that NO and N2O can also be produced by chemodenitrification and chemical oxidation of NH2OH, also with HNO as an intermediate [13–15].
Electron-transport pathway in N. europaea
In terms of their genomic inventory, AOB indeed contain nirK and norCBQD gene homologues that encode a periplasmic copper-containing nitrite reductase (NirK) and nitric oxide reductase (Nor) respectively  (as shown in Figure 1). NirK could speed up NH2OH oxidation by channelling electrons from the cytochrome pool to NO2− (to form NO) and thus play a facilitative role in NH3 oxidation itself . Reduction of NO2− to NO by NirK activity could also be essential to confer tolerance to NO2− in N. europaea . In addition, AOB possess the inventory to alternatively convert NO into N2O, using cytochrome c′-beta (encoded by cytS) and a haem–copper nitric oxide reductase, sNOR (encoded by a four-gene cluster, norSY-senC-orf1) or they employ cytochrome P460 (encoded by cytL) to comproportionate NO and NH2OH to NO2− [10,12]. The role of c′-beta and P460 in mitigating NO toxicity has been shown in other bacteria [18,19], but not in AOB .
Given the ever-expanding knowledge of the genomic inventory of AOB-related NO and N2O production, better predictions of the triggers for the production of these gases by AOB can potentially be made. Such predictions can be especially helpful to understanding and minimizing NO and N2O emissions from systems where nitrogen cycling is expected to contribute to these emissions, for instance engineered biological wastewater-treatment systems designed and operated for BNR (biological nitrogen removal).
Briefly, BNR involves the sequential oxidation of NH3 (the principal form of nitrogen present in sewage streams) to NO2− or nitrate (NO3−) via nitrification and the sequential reduction of NO2− or NO3− to dinitrogen (N2) via denitrification. Nitrification is promoted in BNR systems by providing sufficient aeration and alkalinity . Denitrification is promoted in non-aerated bioreactors with the added introduction of organic electron donors such as methanol in some cases . Thus, in a BNR system, AOB (and the remaining members of the microbial community therein) are subjected continuously to cyclic anoxic–oxic conditions, which are imposed to sustain both nitrification and denitrification in a single process train. Additionally, the influent flow and concentrations of wastewater also change diurnally, in response to human water-use patterns and corresponding wastewater-discharge rates. On the basis of recent studies, the conditions relevant to BNR bioreactors, which have been implicated in NO and N2O production, are transient anoxic–oxic disturbances, high NO2− concentrations and variations in influent NH3 loading [6,8,21–24].
The main focus of the present review is to construct, using a genome-informed approach, the set of biological pathways that are likely to drive NO and N2O production by lithotrophic AOB in response to conditions commonly encountered in engineered wastewater-treatment systems. On the basis of the genomic inventory, conditions typical of BNR wastewater bioreactors as well as experimental data, the pathways have been systematically and conceptually reconstructed (Figures 2–5). In these Figures, the baseline conditions are depicted by black arrows. Increased or decresed contribution of a given process to the overall nitrogen flux is shown by thick green or broken red arrows respectively. The chemical production of NO from NH2OH via HNO is not captured in these pathways.
Oxidative and reductive nitrogen transformations in AOB
Impact of transient anoxic–oxic conditions
Impact of varying influent NH3 loading
Impact of O2 limitation
As a baseline condition, it is assumed that, during the NH3 oxidation cascade, there is minimal NH2OH accumulation and that the main product is NO2−. Low but finite (non-zero) amounts of NO and N2O are produced from this scenario, both from nitrifier-denitrification and from nitrification-dependent reactions (Figure 2).
Transient anoxic–oxic conditions
Upon the imposition of anoxia (and exacerbated by high NO2− concentrations), it is expected that NO2−-reduction activity (resulting in NO formation) and nirK expression would increase (Figure 3A, and as demonstrated by recent studies [8,25,26]). The reducing equivalents for NO production are presumably derived intracellularly from the cytochrome pool . During entry into strict anoxia, which also results in near cessation of NH3 oxidation or NH2OH oxidation activity, the initial production of NO from NO2− reduction might be supported by an endogenous reducing equivalent pool via cytochrome oxidation. Additionally, subsequent reduction of the NO produced to N2O maybe limited by reducing equivalent availability, the higher operating redox potential for the AOB Nor compared with NirK, differential regulation of NirK and Nor or possibly the higher affinity of Nir for electrons over Nor. Indeed, production of only NO by AOB under strict anoxic conditions has been shown [7,26,27], and the absence of N2O formation during strict anoxic conditions parallels reduction in norB mRNA concentrations . N2O formation is only observed upon recovery to oxic conditions, when the reducing equivalents from NH3 or NH2OH oxidation are available again (as modelled in ). Nevertheless, the true source of reducing equivalents for NO and N2O formation via the NO2− reduction remains to be determined experimentally.
Upon recovery to oxic conditions in continuously NH3-fed systems, the oxidation of the accumulated NH3 results in high specific substrate consumption rates (q) close to the maximum value (qmax). In such a case, NH2OH accumulates transiently (R. Yu and K. Chandran, unpublished work), potentially necessitating its oxidation to NO in addition to NO2− to prevent self-inhibition. The NO thus formed is reduced to N2O via the alternative nitric oxide reductases, c′-beta or sNOR. Therefore, in this sense, N2O production by AOB is directional and caused by the recovery from strict anoxic conditions to oxic conditions rather than by imposition of anoxia . Additionally, the mutually distinct production of NO and N2O during anoxic and oxic conditions by AOB is different from their always coupled production in organoheterotrophic bacteria .
Transient overloading and oxidation of NH3
The imposition of excessive NH3 loads to an AOB bioreactor triggers a higher oxidation rate of NH3 (and potentially also a higher amo gene expression, as suggested by ), which could in turn result in NH2OH accumulation. In response, NH2OH oxidation to NO and its reduction to N2O are likely to occur (Figure 4A). During O2-sufficient conditions, it is unlikely that nitrifier denitrification is induced over and above the baseline conditions. Therefore the initial N2O production is likely to be due in large part to NH2OH oxidation and NO reduction. Upon the conversion of the accumulated NH2OH or upon achieving a higher rate of NH2OH conversion into NO2− (by more cell synthesis), it is also possible that the higher NO2− concentrations could drive nitrifier denitrification to produce N2O. Thus the overall production of N2O during transient overloading of NH3 is driven by both oxidative and reductive formation of NO. In addition, we have observed a steady subsidence in N2O emissions upon repeatedly imposing NH3 pulse loadings to an N. europaea chemostat (Figure 6), also pointing further to adaptive responses in N2O production.
Adaptive response in N2O emissions from a chemostat pure culture of N. europaea which was subjected to 1-h feed-pulses each day for 18 days
Sustained O2-limiting conditions
During O2-limiting conditions, when there is a steady supply of reducing equivalents either from NH3 or NH2OH oxidation, N2O and NO formation has indeed been observed  and could be especially increased by high NO2− concentrations  (Figure 5). However, during sustained long-term O2 limitation, the amount of N2O and NO produced is lower than that produced during transient anoxic–oxic disturbances [8,23] or even transient O2 limitation from O2-sufficient conditions , pointing to microbial adaptation to limiting O2 as well.
Given that N2O production in AOB is inherently linked to the NH3- and NH2OH-oxidation or NO2−-reduction rates, the use of N2O emissions as an ‘early warning’ indicator of nitrification process upsets due to O2 limitation and NH3 overloading has been proposed [31,32]. A recent study has also pointed to a link between N2O production in AOB and qmax . In the light of these findings, it is perhaps essential to explore differences, if any, between different lithotrophic AOB due to differences in substrate affinity coefficients for NH3 or qmax. For instance, oligotrophic AOB such as Nitrosomonas oligotropha or Nitrosospira spp., attain qmax at much lower NH3 concentrations compared with AOB, which prefer higher NH3 concentrations, such as Nitrosomonas eutropha. Thus it could be conceptualized that, everything else considered equivalent, N. oligtropha or Nitrosospira spp. have a higher propensity for NO and N2O production at lower NH3 concentrations, compared with N. europaea or N. eutropha. Production of N2O by Nitrososipra spp. via nitrifier denitrification has indeed been shown . Nonetheless, the specific pathways of NO and N2O production by different AOB remain to be compared.
Furthermore, whereas the studies reviewed in the present paper provide good insights into potential pathways at work in NO and N2O production by lithotrophic NH3-oxidizing bacteria, they still do not address the additional complexity of microbial adaptation to repeated transient or sustained imposition of limiting dissolved O2 concentrations, possibly combined with factors such as high NO2− concentrations. Indeed, in a recent study, a protracted transient in N2O and NO production as well as the mRNA concentrations of nirK and norB was observed upon switching from non-limiting dissolved O2 concentrations to limiting dissolved O2 concentrations (which secondarily led to high NO2− concentrations, ~500 mg/l NO2−-N). However, microbial adaptation to limiting dissolved O2 and high NO2−-N was observed over a period of 80 days as the gas concentrations and the mRNA levels of nirK and norB decreased , but not entirely to pre-transient levels. Therefore more studies that look at longer-term adaptation are needed at the whole-cell and molecular levels.
It must also be acknowledged that the presence of chemo-organoheterotrophic denitrifying bacteria or anammox (anaerobic ammonium oxidation) bacteria in activated sludge could potentially reduce the magnitude of N2O emissions by conversion into N2, under anoxic conditions. Reduced N2O emissions in reactor zones or even entire BNR plants where simultaneous nitrification and denitrification occur have indeed been documented . Lower N2O emissions have also been observed from the anammox-enriched stage of a full-scale wastewater-treatment facility compared with the upstream nitritation stage, presumably more enriched in AOB .
Minimizing N2O emissions from engineered nitrogen-removal systems
From a broader perspective, estimates of N2O emissions from wastewater treatment using traditional USEPA (U.S. Environmental Protection Agency) methodology  still constitute the equivalent of nearly 900000 additional vehicles on the road every year (calculated based on USEPA estimates of wastewater-treatment- and automobile-derived N2O emissions). On the basis of recent nationwide monitoring studies conducted in Australia, The Netherlands and the U.S.A., the actual measured N2O emissions from sewage-treatment facilities are generally higher than IPCC (Intergovernmental Panel on Climate Change) estimates (in some cases by about two orders of magnitude [21,24,35]). This is likely because nitrification has been ignored as a source of N2O in BNR systems and wastewater-treatment plants in general by the IPCC and USEPA estimation approaches . Indeed, it has been observed that aerated zones (with non-limiting dissolved oxygen concentrations) in BNR bioreactors, where nitrification presumably dominates nitrogen cycling, N2O emissions are about two orders of magnitude higher than from non-aerated zones, wherein denitrification dominates nitrogen cycling . If the new emissions measurements are extrapolated linearly and uniformly to the number of vehicular equivalents, then biological wastewater treatment contributes as much as 90 million vehicles in the U.S.A. alone, which now constitutes a potential concern.
Nonetheless, on the basis of potential mechanisms and triggers of N2O emissions from AOB, as discussed above, it is possible to minimize N2O emissions, while achieving desired water quality. Indeed, a consistent theme to emerge from field-scale studies is that wastewater-treatment plants that achieve high degrees of nitrogen removal in the aqueous phase also minimize N2O emissions in the gaseous phase, often below the IPCC and USEPA estimates [21,24,35]. In practice, high degrees of nitrogen removal from wastewater can be achieved by engineering treatment plants to consistently achieve complete nitrification and denitrification, irrespective of intrinsic perturbations or variations in influent nitrogen loading. Additionally, it has been suggested that N2O emissions credits might be a potential vehicle for well-designed and operated BNR facilities to be rewarded for high levels of nitrogen removal from the wastewater stream as well as minimizing N2O emissions .
The rapidly accumulating knowledge of the genomic inventory of AOB provides significant insights into the potential pathways of their NO and N2O production. When the inventory is considered in conjunction with environmental conditions imposed upon AOB (using field and laboratory data), it becomes possible to reconstruct likely pathways that are involved in NO and N2O production under these conditions. In the framework of engineered wastewater-treatment systems, the very configurations and operating conditions employed to achieve BNR from the aqueous phase may give rise to NO and N2O emissions in the gaseous phase. Nevertheless, given the parallels between poor treatment and excessive N2O emissions, there is the opportunity to simultaneously minimize both aqueous and gaseous nitrogen pollution via what can be termed environmentally sustainable engineering design of sewage-treatment processes.
ICoN2 and the NCycle16: The 2nd International Conference on Nitrification (ICoN2) and the 16th European Nitrogen Cycle (NCycle16) Linked Independent Meetings held at Hotel Val Monte, Berg en Dal, The Netherlands, 3–7 July 2011. Organized and Edited by Mike Jetten (Radboud University, Nijmegen, The Netherlands) and David Richardson (University of East Anglia, Norwich, U.K.).
We thank Kira Schipper and Udo van Dongen for conducting the cyclic ammonia pulse experiments presented in Figure 6.
This work was supported by a National Science Foundation CAREER award, the Water Environment Research Foundation and a TU-Delft Visiting Faculty grant (to K.C.), the Natural Sciences and Engineering Research Council of Canada (to L.Y.S.), the U.S. National Science Foundation (to M.G.K.). Collaboration between the authors is an activity under the umbrella of the Nitrification Network (http://nitrificationnetwork.org) funded by the U.S. National Science Foundation [grant number EF-0541797].