The sludge used in this study was collected from an AME-D culture that was enriched for more than 1 year in a batch bioreactor with continuous CH₄ supply. The sludge was pre-incubated in a nitrite-containing medium under anoxic conditions in the dark at 25°C for 3 days to eliminate residual organic carbon sources. Then the sludge was centrifuged at 3,300 g for 5 min and the supernatant was discarded. The sludge pellet was resuspended and washed in sterile phosphate buffer solution (pH = 6.8) for three times to further remove extracellular organic carbons. Subsequently, it was centrifuged at higher speed of 14,000 g for 10 min to ensure complete precipitation of suspended cells. The experiments were initiated when no organic carbon was present in the supernatant. The organic carbon was determined with a Total Organic Carbon (TOC) analyzer (MultiN/C3100, Analytikjena, Germany). After all residual organics were removed, the sludge was re-suspended in the basal medium (Liu et al., 2014) at a mixed liquor suspended solid (MLSS) concentration of 17,735 mg/L. The basal medium contained the following components in 1 liter (L) distilled water: 1,250 mg KHCO₃, 50 mg KH₂PO₄, 300 mg CaCl₂·2H₂O, 200 mg MgSO₄·7H₂O, 345 mg NaNO₂, 1.0 mL acidic trace element solution and 1.0 mL alkaline trace element solution. Constituents of the acidic and alkaline trace element solution were described by Liu et al. (2014). Inorganic nitrogenous compounds were prepared from NaNO₂ to result in a concentration of NO2--N in the basal medium of 70 mg/L. Ammonium (NH4+) and NO3- were also supplied at low concentrations, 0.07 and 3.82 mg/L, respectively, to serve as nitrogen sources for microbes in the AME-D system. The pH of the medium was 7.2.

Batch experiment was conducted using 10 mL of the prepared sludge and 20 mL of the basal medium in a 150 mL glass vial. Freshly prepared mixture of the basal medium and the sludge was sparged with CH₄ (99.99%) for 5 min. The vial was crimp-sealed with a butyl rubber stopper. Different volumes of CH₄ (0, 6, 24, and 60 mL) were withdrawn from the headspace of the vials with a gas-lock syringe. Afterwards, the same volume of pure O₂ (99.99%) was injected into the headspace of vials. The final compositions of O₂ and CH₄ in the gas mixtures of each treatment were shown in Table 1. Un-inoculated media was used as blank controls to test for leakage and non-biological chemical transformations. Triplicate samples were incubated at 28°C in shaking incubator at 180 rpm in the dark for 60 h. CH₄ and nitrogen (NO3--N, NO2--N and NH4+-N) consumptions and intermediate production were monitored in each treatment.

The NO3--N, NO2--N, and NH4+-N concentrations were determined by ultraviolet spectrophotometry (UV-5300PC, METASH, China). The measurement of volatile suspended solid (VSS) was performed according to standard methods (American Public Health Association, American Water Works Association, Water Pollution Control Federation, 1989). In this study, nitrite removal expressed as mmol nitrite consumed per gram of biomass per day i.e., mmol NO2--N/gVSS/d. To quantify CH₄ content in the headspace of each vial, a sample of 0.5 mL was removed from the headspace of the vial with a gas-lock syringe. The sample was analyzed with a gas chromatograph (GC) equipped with a thermal conductivity detector under the conditions previously described by Zhang et al. (2014). Methane oxidation activity was expressed as CH₄ consumed per gram of biomass per day i.e., mmol CH₄/gVSS/d (Wang et al., 2008). NO concentration was measured by GC-mass spectrometry (6890N GC-5973MS, Agilent, USA) with the methods described by Leone et al. (1994). N₂O concentration in the vial headspace was determined by a GC (GC-14B, Shimadzu, Japan) equipped with an electron capture detector and a Porapak Q column maintained at 330 K. Intermediate (methanol, formaldehyde, formate, acetate, and other potential organics) concentrations were determined with a high performance liquid chromatograph using the methods described by Thalasso et al. (1997).

Sludge samples were analyzed for DNA before and after incubation. Three independent DNA extractions of each treatment were performed from 30 mg of sludge using a FastDNA SPIN Kit for Soil (MP Biomedical, LLC, Ohio, USA). The extraction was performed according to the manufacturer's instructions. The concentrations and the quality of DNA samples were measured with a Nanodrop analyzer (Thermo Scientific, Wilmington, DE, USA). Extracted DNA was stored at −20°C prior to subsequent analyses.

The abundance of aerobic methanotrophs and denitrifiers was estimated through quantitative PCR (q-PCR). Due to the low level of mmoX-harboring methanotrophs in the enrichment and high enough Cu concentration (2 μM) in the medium to inhibit the expression of mmoX gene (Takeguchi et al., 1997), the function of sMMO encoded by mmoX gene was considered to be negligible. Therefore, only pmoA gene encoding a subunit of pMMO was used to investigate the abundance variation of aerobic methanotrophs in this study. For denitrifiers, the nirK gene encoding copper nitrite reductase and the nirS gene encoding cytochrome cd₁-containing nitrite reductase were used as the biomarkers. The quantification was based on the intensity of SYBR Green dye fluorescence, which can bind to double-stranded DNA. Standard curve for each gene were generated using a 10-fold dilution series of the linearized plasmid standard (10⁻¹–10⁻⁶ ng) ranging from 10⁸ to 10³ copies. Each qPCR assay (25 μL) included 12.5 μL of 2 × SYBR Premix Ex Taq (Takara, Dalian, China), 1 μL of each forward and reverse primer (20 μM), either 1 μL of template DNA or the standard vector plasmid of the clones grown as single cellular suspension. The optimized thermal conditions and primers used for each gene can be viewed in the Supplementary Table 1. All real-time PCR assays were performed in triplicate for each sample in a Bio-Rad CFX1000 Thermal Cycler. All PCR runs included negative controls that did not contain DNA templates. The gene copy numbers were determined by comparing threshold cycles obtained in each PCR run with those of known standard DNA concentrations. Standard curves were obtained using serial dilutions of linearized plasmids containing cloned pmoA, nirK, and nirS genes.

All data are presented as means and standard deviations. Analysis of variance and least significant difference (LSD) tests at the 5% level were used to determine the statistical significance of different treatments. Any differences with p ≥ 0.05 were not considered as statistically significant. The relationships between nirK/nirS gene copies and the nitrite removals at four treatments were tested with linear regression analyses using SPSS 20.0 for Windows (SPSS Inc., Chicago, IL).

Methane oxidation activity and concentrations of extracellular metabolites (methanol, formaldehyde, formate, and acetate, etc.) were determined under four different O₂:CH₄ ratios. As shown in Figure 2, methane oxidation activity substantially decreased from 277.80 mmol/gVSS/d to 21.06 mmol/gVSS/d when the O₂:CH₄ ratio was increased from 0 to 1 (p < 0.05). With the exception of treatment 1 at an O₂:CH₄ ratio of 0, a similar trend was observed for qPCR data of the pmoA gene. The pmoA gene abundance decreased almost an order of magnitude (from 9.36 × 10⁹ to 1.63 × 10⁹ copies per g dry biomass) when the O₂:CH₄ ratio was increased from 0.05 to 1 (Figure 2). The three primary metabolites observed in the bulk media were formaldehyde, acetate and citrate. Their concentrations went up from 42 to 76 μg/L (formaldehyde), 11 to 45 μg/L (acetate) and 0 to 28 μg/L (citrate), respectively, when O₂:CH₄ ratio was increased from 0 to 1 (Figure 3). Methanol, formate and butyrate were considered as three trace metabolites and their concentrations were lower than 1.20 μg/L (Supplementary Figure 1). However, methanol were not detected in the treatment 1 in the absence of O₂.

The concentrations of NO2--N, NH4+-N and NO3--N were measured at the beginning and the end of incubation. Results showed that concentrations of NH4+-N and NO3--N decreased slightly by the end of the experiment and were detected in all final samples (Supplementary Table 2). There was a jump in the nitrite removal from 0.53 mmol NO2--N/gVSS/d to 7.32 mmol NO2--N/gVSS/d when the O₂:CH₄ ratio was increased from 0 to 0.25 (Figure 4). However, the nitrite removal decreased by 53.8% as the O₂:CH₄ ratio was increased from 0.25 to 1 (p < 0.05). The amount of reduced nitrogen released as NO was very low (0.13–0.32 μM; Table 2). In addition, the percentage of the reduced nitrogen emitted as N₂O decreased from 37.96 to 12.30% when the O₂:CH₄ ratio increased from 0 to 0.25 (Table 2). This percentage at the O₂:CH₄ ratio of 1 was about 2 times higher than that at the O₂:CH₄ ratio of 0.25, while this difference was insignificant (p > 0.05; Table 2). The highest nitrite removal and the low percentage of NO-N and N₂O-N in total reduced nitrogen were observed at the O₂:CH₄ ratio of 0.25. Based on the results, the O₂:CH₄ ratio of 0.25 was proposed as the optimal ratio for this AME-D system.

Nitrite reduction is catalyzed by nitrite reductase which are found in two different forms: copper nitrite reductase encoded by nirK gene and cytochrome cd₁-containing nitrite reductase encoded by nirS gene (Wang et al., 2014). Investigating the difference in nirK and nirS gene abundance might provide further evidences for the variation of nitrite removal. The abundance of nirK gene was more than doubled as the O₂:CH₄ ratio was increased from 0 to 0.25 (from 3.47 × 10¹⁰ copies/gVSS up to 7.91 × 10¹⁰ copies/gVSS). However, it descended to 6.23 × 10¹⁰ copies/gVSS as the O₂:CH₄ ratio was further raised to 1. In contrast, copy numbers of nirS gene displayed no significant change (a slight increase from 1.82 copies/gVSS to 2.34 × 10¹¹ copies/gVSS) when O₂:CH₄ ratio was increased from 0 to 0.25. However, it went down significantly to 7.10 × 10¹⁰ copies/gVSS at higher O₂:CH₄ ratio of 1. In addition, the linear correlation analysis revealed that nirK gene copies were positively correlated with nitrite removal (r² = 0.8639, p = 0.0707), whereas nirS gene copies had a slightly negative correlation with the nitrite removal (r² = 0.0121, p = 0.8899; Figure 5). If the nitrite removal was considered as the representation of the denitrifying conditions in the corresponding treatment, the higher nitrite removal indicated the better denitrifying conditions in this treatment. Thereby, the linear coorelation analyses suggest that nirK-type denitrifiers might be more responsive to the denitrifying conditions than nirS-type denitrifiers (Yoshida et al., 2010).

The results from the batch experiment showed that the peak nitrite removal (7.32 mmol NO2--N/gVSS/d) was obtained at an O₂:CH₄ ratio of 0.25. As NH4+-N and NO3--N are the preferred microbial nitrogen sources (Modin et al., 2007), their presence throughout the incubation period (Supplementary Table 2) indicated that nitrite was negligibly consumed as a nitrogen source for assimilation. Therefore, nitrite removal from the media was through dissimilatory, i.e., denitrification. Additionally, the low percentage of the consumed NO2--N emitted as N₂O-N (12.30%) and NO-N (0.20%, Table 2) at the O₂:CH₄ ratio of 0.25 indicated that enough carbon sources provided by the methanotrophs may allow denitrifiers to reduce almost 87.5% of NO2--N through complete denitrification. At a lower O₂:CH₄ ratio of 0.25, it was consistent with our hypothesis that nitrite removal were not inhibited by O₂, and were stimulated by the increased carbon provided by methanotrophic metabolism. However, nitrite removal was less effective at higher O₂:CH₄ ratios with higher oxygen concentrations, which was corroborated by the lower abundance of denitrifying genes (Figure 4).

The conclusion that nitrite removal was stimulated by carbon sources released from aerobic methane oxidation at a O₂:CH₄ ratio lower than or equal to 0.25 could be supported by the variation of intermediate concentrations among four treatments. Concentrations of accumulated intermediates were higher and higher in the bulk media with the O₂:CH₄ ratios increasing from 0 to 0.25 (Figure 3). It may be attributed to the increased O₂ concentration in these treatments. Morinaga et al. (1979) and Costa et al. (2001) demonstrated that the O₂ level would significantly impact the consumption and production of metabolites in methane metabolism. However, the effect was not unambiguously determined. Morinaga et al. (1979) observed formaldehyde accumulation under oxygen-limited conditions, whereas Costa et al. (2001) discovered that formaldehyde accumulated under oxygen-excessive conditions. Our results indicated that the increased oxygen level promoted intermediates accumulation in methane metabolism. Low level of available carbon sources for denitrifiers in Treatment 1 resulted in the lowest nitrite removal. Once O₂ was largely induced in the headspace in Treatment 2 and 3, concentrations of intermediates increased in the liquid bulk. The observation of simultaneous higher nitrite removal and lower ratio of accumulated N₂O-N:consumed NO2--N indicated that complete denitrifying activity was improved by these accumulated intermediates.

The ability of denitrifiers to resist the inhibition of O₂ at the optimal O₂:CH₄ ratio may be due to two reasons. Firstly, the abundance of nirK-harboring denitrifiers was the highest among the four treatments and these denitrifiers are able to tolerate higher O₂ levels (Desnues et al., 2007). Secondly, the sludge aggregation/granulation in anoxic micro-environments can lessen the O₂ exposure of the denitrifiers. As shown in Figure 6, at the O₂:CH₄ ratio of 0.25, suspended biomass formed granule-like agglomerates with an average diameter of about 2 mm, while such effect did not occur for the O₂:CH₄ ratios of 0, 0.05, and 1. Sludge aggregation has also been observed in an AME-D system with an optimized O₂ level that had the highest nitrogen removal rate (Thalasso et al., 1997). It can be therefore concluded that sludge aggregation at an optimal O₂:CH₄ ratio could improve nitrite removal. The spatial distribution of microorganisms and O₂ level within the aggregates should be investigated in the future.

With regard to the effect of the O₂:CH₄ ratio on denitrification, Sun et al. (2013) observed a similar phenomenon that the O₂:CH₄ ratio affected nitrogen removal of AME-D process in a membrane biofilm bioreactor (MBfR). Their optimal O₂:CH₄ ratio of 1.5 was considerably higher than the one in this study. It is likely that the spatial arrangement of the microbial community within a well-developed biofilm would allow for greater tolerance to O₂, as compared to a suspended culture. They speculated that greater metabolite excretion by the methanotrophs improved nitrate removal performance initially, and that excessive O₂ caused the significant drop in the denitrifying rate when O₂:CH₄ ratio increased to 2.0. The results from this study corroborate the trend that Sun et al. (2013) observed.

Methane oxidation activity dramatically decreased when O₂:CH₄ ratio increased from 0 to 1 (Figure 2). It was probably due to the large change in methane availability. Li et al. (2014) observed a similar trend with methanotrophic activity of landfill cover soils. The methane oxidation activity at O₂:CH₄ ratio of 0.25 (4 × 10⁴ ppmv O₂/2 × 10⁵ ppmv CH₄) was almost 4-fold higher than that at O₂:CH₄ ratio of 4.00 (2 × 10⁵ ppmv O₂/5 × 10⁴ ppmv CH₄) (Supplementary Table 3). In their research, methane oxidation activity was much more sensitive to CH₄ than O₂, and a drop of CH₄ concentration would result in a simultaneous decrease of methane oxidation activity. Additionally, copy numbers of the pmoA gene decreased as the methane concentration decreased (Baani and Liesack, 2008; Li et al., 2014). In the current research, CH₄ concentration declined from 100 to 50% and O₂ concentrations rose from 0 to 50% for the ratios tested. This suggests that a decrease in methane oxidation activity with a substantial increase in O₂:CH₄ ratio is plausible. However, it is unexpected that the high pmoA gene copy numbers and methane oxidation activity were observed in Treatment 1 in the absence of O₂. Rechecking the gaseous composition of the headspace in this treatment at the beginning of the incubation, it was discovered that trace amount (0.012%) of O₂ can still be detected after the sludge was sparged with pure CH₄ (99.99%). This trace level of O₂ may result in the observed CH₄ consumption. However, further investigations are still required to focus on examining if anaerobic methane oxidation contributed to CH₄ consumption when this trace amount of O₂ was depleted.

Acetate, citrate, and formaldehyde were detected as three primary compounds detected during the AME-D process, while methanol, formate, and butyrate occurred in trace quantities. It is difficult to postulate which were the main substrates for the denitrifiers as their levels of consumption relative to production were not known. However, thermodynamic analysis of AME-D process may provide useful information for speculating the actual metabolic pathway (all of the following thermodynamic analyses are based on aerobic methane oxidation coupled with complete denitrification).

Methanol is considered as the most effective intermediate for denitrification according to the review of the AME-D process (Zhu et al., 2016). The reactions (Equations 3–5) related to energy production contained in AME-D process using NO2- as the denitrifying electron acceptor were shown in Table 3. These equations are based on one electron equivalent (eeq). Assuming that at least X (<1) mol of methanol is needed by aerobic methanotrophs for their requirement of cell synthesis and maintenance when one mole of CH₄ is oxidized to methanol, the remaining part of methanol can be used for denitrification as the electron donor. An energy-balanced equation for the AME-D process can be described as Equation (6).

Where ε, is the energy transfer efficiency a value of 0.37 (McCarty, 2007). [ΔG_Eq.(3) – 0.5ΔG_xy] denotes the net energy production for the oxidation from CH₄ to methanol after considering the energy input for the mono-oxygenase and the required reducing equivalent, which equals to 47.26 kJ/eeq. ΔG_m represents the minimum maintenance energy requirement with a value of about 8.1 kJ/mol oxidized CH₄ (Modin et al., 2007). The exhaustive description of calculation processes for [ΔG_Eq.(3) – 0.5ΔG_xy] and ΔG_m is presented by Zhu et al. (2016). From Equation (6), the value derived for X is 0.40. Because the energy required by cell synthesis has not been considered during the above calculation, the theoretical maximum quantity of methanol used by denitrifiers was 0.60 mol. This means that the maximum proportion of methanol that can be captured by denitrifiers is 60%. The overall reaction of AME-D process in the presence of NO2- was described as Equation (7), a combination of the proposed value and Equations (3–5).

All six substrates detected in this study were potential carbon sources for denitrifiers. In order to understand which substrates were likely the functional intermediates, thermodynamic analysis was performed based on the chemical data associated with the O₂:CH₄ ratio of 0.25. During the process of thermodynamic derivation, all of organics detected in the bulk liquid were individually chosen as the trophic link of aerobic methane oxidation and denitrification. After several iterations, the same general equation (Equation 8) for the AME-D process, which was in agreement with CH₄ and NO2- consumption was obtained through three different approaches:

Based on the above analysis, it was evident that the percentage of carbon flow from methane to denitrification (4.34–6.52%) was much lower than the ideal flow (60%). Further improvement in the carbon flow is vital to enhance the AME-D denitrification rates.

The effect of the O₂:CH₄ ratio was demonstrated to significantly impact nitrogen removal during the AME-D process through contribution of the carbon metabolites generated by the methanotrophs and oxygen inhibition. To date, most studies have considered only the individual impact of O₂ on the apparent nitrogen removal rate of the AME-D process (Werner and Kayser, 1991; Thalasso et al., 1997; Modin et al., 2010), whereas the combined effect of O₂ and CH₄ and associated mechanisms are rarely investigated. It is necessary to address the impact of O₂:CH₄ ratio on CH₄ and NO3-/NO2- metabolism in the AME-D process, rather than the tendency of methane oxidation rates and denitrifying activities under different gaseous environments. The knowledge will allow a better understanding of the specific roles of the O₂:CH₄ ratio in CH₄ and NO3-/NO2- metabolism. It is expected that this will contribute to well-founded strategies that will improve nitrogen removal, one of bottle-necks in the application of the AME-D process.

In this study, the optimal O₂:CH₄ ratio for denitrification was found to be 0.25. At this point, denitrifying activity reached the highest level of 7.32 mmol NO2--N/gVSS/d. When the O₂:CH₄ ratio was below the optimal ratio, nitrite removal was improved with the increased O₂:CH₄ ratio, presumably due to an increase in available substrates released by aerobic methanotrophs. Methanol, butyrate and formaldehyde were thermodynamically speculated as the main active intermediates of the AME-D process. When the O₂:CH₄ ratio was above the optimal ratio, nitrite removal was presumably inhibited by the excessive O₂. These results indicate that adjusting the O₂:CH₄ ratio can improve the cooperation between aerobic methanotrophs and denitrifiers to obtain better nitrogen removal performance using the AME-D process.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.