Aerobic denitrification as an N2O source from microbial communities

Abstract Nitrous oxide (N2O) is a potent greenhouse gas of primarily microbial origin. Oxic and anoxic emissions are commonly ascribed to autotrophic nitrification and heterotrophic denitrification, respectively. Beyond this established dichotomy, we quantitatively show that heterotrophic denitrification can significantly contribute to aerobic nitrogen turnover and N2O emissions in complex microbiomes exposed to frequent oxic/anoxic transitions. Two planktonic, nitrification-inhibited enrichment cultures were established under continuous organic carbon and nitrate feeding, and cyclic oxygen availability. Over a third of the influent organic substrate was respired with nitrate as electron acceptor at high oxygen concentrations (>6.5 mg/L). N2O accounted for up to one-quarter of the nitrate reduced under oxic conditions. The enriched microorganisms maintained a constitutive abundance of denitrifying enzymes due to the oxic/anoxic frequencies exceeding their protein turnover—a common scenario in natural and engineered ecosystems. The aerobic denitrification rates are ascribed primarily to the residual activity of anaerobically synthesised enzymes. From an ecological perspective, the selection of organisms capable of sustaining significant denitrifying activity during aeration shows their competitive advantage over other heterotrophs under varying oxygen availabilities. Ultimately, we propose that the contribution of heterotrophic denitrification to aerobic nitrogen turnover and N2O emissions is currently underestimated in dynamic environments.


Reactor operation
Table S1.Measured average substrate loading and steady-state conversion rates of the low-(R4) and high-frequency (R32) reactors.Overall rates refer to rates estimated over the total duration of an oxic/anoxic cycle, and considers the average of three effluent concentrations (beginning and end of oxic phase, and end of anoxic one).Only for the gaseous compounds (CO2 and N2O) individual rates for each phase (oxic and anoxic) were measured on top of the overall rates.O2 was only added and consumed in the oxic phase, yet an "overall" rate was also calculated by averaging the aerobic O2 consumption over the entire cycle duration (eq.S9) for further balancing purposes.Overall C-mmol/h 0.68 ± 0.02 0.82 ± 0.01 -0.69 ± 0.02 -0.83 ± 0.01 a Calculated from the NH4 + consumption rates.

Loading
b Always 0 in the effluent.k L a determination.The oxygen volumetric mass transfer coefficient (kLa) was determined to calculate the oxygen transfer rate in the oxic phase.The kLa of R4 and R32 were determined under identical conditions as the enrichments (500 rpm stirring, 400 mL/min gas flow), but with water instead of biomass.The O2 transfer rates were determined by following the dissolved oxygen concentration during the sparging of air (400 mL/min) in anoxic water.The kLa was obtained by fitting the integrated mass transfer equation to the dissolved O2 concentration profile over time, with CO2 * the solubility of O2 at 20°C: The obtained kLa values were 36.2 (R4) and 37.0 h -1 (R32).

Calculation of consumption and production rates
Consumption and production rates of all dissolved and gaseous compounds were measured or estimated in the oxic and anoxic phases, and overall (combined oxic and anoxic).Consumption rates are negative and production rates are positive.Overall consumption and production rates in the liquid.Consumption and production rates of NO3 -, NO2 -(Cin = 0), NH4 + and the organic compounds acetate, propionate and butyrate (Cout = 0) were calculated from a mass balance: (eq.S2) with Ri the molar rate (mmol•h -1 ), Fi the influent and effluent flow rates (L•h -1 ), and Ci the concentration of compound i (mmol•L -1 ).Cout was the average of three effluent measurements (taken at the beginning and end of the oxic phase, and end of the anoxic phase).The sample for Cin was taken directly at the entry point of the reactors, yet differences with stock feed solution remained negligible throughout the experimental period.The flow rates were the average of the measured flow rates during the entire operation.Linear error propagation was applied to determine the standard deviation in the rates (eq.S3), using the standard deviations of Fin, Fout, and Cout (deviation between the three measurements).Linear error propagation of a function f dependent on multiple variables (x, y, …): • σ y 2 + ⋯ (eq.S3) with σf, σx, and σy the standard deviations of f, x, and y, respectively, and ∂f/∂x and ∂f/∂y the partial derivatives of f with respect to x and y, respectively.Overall, aerobic, and anaerobic consumption and production rates in the gas phase.A script was written in RStudio to calculate the overall, and separate aerobic and anaerobic N2O and CO2 rates from continuous measurements recorded every minute.The molar gas flow leaving the reactor was calculated for each time point based on the constant influent volumetric gas flow rate (400 mL•min -1 ) and the measured temperature and atmospheric pressure: With Ngas the molar gas flow rate (mmol•h -1 ), Patm the atmospheric pressure (mbar), FV,gas the volumetric gas flow rate, R the ideal gas constant (L•mbar•K -1 •mmol -1 ), and T the reactor temperature (K).The molar gas fractions were normalized to the zero measurement before further calculations, by subtracting the corresponding value measured for the zero concentration.For each of the gases, the molar flow rates in the off-gas were calculated at each time point from measured gas fractions and the total molar gas flow rate: The N2O rate was normalized per mole of nitrogen: (eq.S6) with Ni the molar gas flow rates (mmol•h -1 ), Ngas the molar gas flow rate (mmol•h -1 ), and yi the molar fractions of each compound in the off-gas.The accumulation rates at every time point were calculated with the following mass balance: (eq.S7) The average fraction of CO2 in the influent air was 450 ppm.The dataset containing the rates at every minute was split in oxic and anoxic periods, with the oxic period defined for time points with DO > 1% (0.08 mg O2•L -1 ).Daily average aerobic, anaerobic, and overall rates were calculated with the corresponding dataset.The standard deviation of these averages was taken as the uncertainty in the rates.Overall and aerobic consumption and production rates of oxygen.The O2 consumption rates during the oxic phase were calculated from the dissolved oxygen measurements during maximum aeration periods (> 20% O2 in the off-gas and dissolved oxygen > 70%): (eq.S8) With kLa the experimentally measured transfer coefficient (h -1 ), HO2 the Henry coefficient for O2 (0.001283 mmol•L -1 •mbar -1 ), Patm the atmospheric pressure (mbar), yO2 the O2 molar fraction in the off-gas, DO the measured dissolved oxygen, and V the broth volume (L).The aeration over-capacity in the reactors was determined by comparing the maximum O2 transfer rate from the gas to the liquid (equivalent to the maximum possible O2 microbial respiration rate) to the actual O2 respiration rates.The maximum possible O2 transfer rates, i.e. the maximum microbial respiration capacity, would be achieved when the DO is 0, so eq.S8 can be simplified into eq.S9.These rates were calculated to be 7.5-fold higher than the actual O2 respiration rates, reflecting the aeration over-capacity in the reactors.
(eq.S9) Daily averages were calculated and taken for further calculations.The standard deviation of these averages were taken as the uncertainty of the rates.The "overall" consumption rate of O2 for further electron balancing purposes over an entire cycle was taken as the weighted average of the aerobic and anaerobic (=0) rates: R i overall = t aerobic 24 • R i aerobic + t anaerobic

24
• R i anaerobic (eq.S10) with taerobic and tanaerobic (h) the total time in one day in which the dissolved oxygen was above or below 1%, respectively.
Overall respiratory electron flow to nitrogen oxides and O 2 .The absolute and relative overall flows of electrons from organic electron donors to the electron acceptors NO3 -and O2 were calculated from the overall rates considering four and five electrons for the conversion of NO3 -to N2O and N2, respectively, and four electrons for the reduction of O2 to H2O (Table S3).NO3 -and O2 were the sole electron acceptors and NH4 + fully sustained biomass growth (detailed in the following section), minimizing NO3 -assimilation.Thus, both substrates account for the entirety of the catabolic electron flow.
Table S3.Absolute (mmol e -/h) and relative (%) overall electron flows from organic carbon to the electron acceptors NO3 - and O2 in the low-(R4) and high-frequency (R32) reactors.The electron flows were calculated from the NO3 -and O2 consumption and the N2O accumulation rates.

R32
(mmol e -/h) 2.8 ± 0.3 4.7 ± 0.6 % NO 3 -/ Total 56 ± 4% 39 ± 4% % O 2 / Total 44 ± 4% 61 ± 4% Biomass production rates.The biomass concentration was estimated from NH4 + measurements and carbon balances.In the studied system, ammonia oxidation was fully inhibited via continuous ATU addition, thus the assimilation into biomass (0.2 N-mol/C-mol) was the sole NH4 + consumption process.The biomass production rate was calculated as follows: R X = |R NH 4 + 0.2 ⁄ | (eq.S11) The carbon balance included only the organic carbon substrates (acetate, propionate, and butyrate), CO2 and biomass, as no other products were detected in the HPLC.Therefore, the biomass production rate could also be directly calculated according to the following equation: R X (Cmmol • h −1 ) = |R Ace + R Pro + R But + R CO2 | (eq.S12) For all calculations, an empirical biomass formula of CH1.8N0.2O0.6 was used.The biomass concentration (CX in C-mmol•L -1 ) was then estimated from the production rate (RX) and the flow rate (Fout in L•h -1 ): C X = R X /F out (eq.S13) Both methods resulted in similar estimations, showing that the biomass concentration and its production rate can be determined through either one of the methods (Figure S5).The biomass rates and concentrations based on NH4 + measurements were used for further calculations.Error propagation was applied to determine the standard deviation in the rates, using the standard deviations of Fout, and NH4 + , organic carbon, and CO2 rates.the organic substrate and CO2 measurements.Overall carbon, nitrogen, and electron balances.Mass balances were performed using the consumption and production rates averaged over the steady-state period to ensure that all substrates and products were recovered.The overall carbon balance was calculated from the consumption and production rates of acetate, propionate, butyrate, biomass (estimated from the NH4 + rates), and CO2: The nitrogen compounds involved in the nitrogen balance would be NH4 + , biomass, NO3 -, NO2 - , NO, N2O, and N2.All compounds were measured (or estimated, in the case of biomass) except N2.NO and NO2 -accumulation was absent or negligible throughout the entire experiment, so we could assume that the missing nitrogen was recovered as N2, representing full denitrification from NO3 -: (eq.S15) Based on all calculated and estimated rates, an electron balance was calculated.
(eq.S16) Uncertainty of the balances were calculated through linear error propagation from the standard deviations of the respective rates (eq.S3).
Table S4.Overall carbon, and electron balances over the entire steady-state period of the low-(R4) and high-frequency (R32) oxic/anoxic cycling denitrifying reactors.
Separate aerobic and anaerobic rates were calculated for all compounds continuously measured in the gas (N2O and CO2) or liquid phase (O2).In turn, grab samples for the quantification of all other compounds were less sensitive to the small concentration changes occurring during each phase, so consumption and production rates could not be determined directly with high confidence.Instead, aerobic and anaerobic rates were calculated from the overall mass balance and the phase-specific N2O, CO2, and O2 rates as detailed below.In short, as the overall balances closed (Table S4), all biological processes taking place in the controlled environments of the reactor are known.Also, the overall rates are the sum of the aerobic and anaerobic ones weighted by their corresponding time fractions.The calculation of aerobic and anaerobic conversion rates detailed below were based on carbon, nitrogen, and electron balances, so one needs to know which processes occurred in the reactor broth in each phase.Specifically, from the closed carbon and electron balances (Table S4) we know that denitrification occurred, with N2O and N2 as end-products.We do not know, however, which fraction of this conversion occurred in the oxic and anoxic phases.To determine this, three scenarios were considered (Figure S6).The rationale underlying these scenarios is briefly explained: 1) Scenario 1 was developed based on past literature.Aerobic denitrification was widely considered to be absent or negligible under fully oxic conditions.Nevertheless, we measured aerobic production of N2O in our reactors, which means that at least part of the NO3 -was aerobically converted to N2O.For this scenario, we assumed that this was the only fraction of NO3 -converted aerobically, with the remaining converted under anoxic conditions.2) The aerobic and anaerobic electron balances did not close in scenario 1, which means that our assumption was incorrect.3) Based on the electron gaps observed in scenario 1, we developed scenario 2. In this case, we rationalized that the missing or surplus of electrons in scenario 1 must belong to a "blind" amount of NO3 -aerobically reduced to N2.In other words, for scenario 2, we considered that part of the NO3 -was aerobically converted to N2O (directly measured, same as scenario 1) and an additional part was converted to N2 (estimated).4) The estimations made in scenario 2 were validated with measurements, so we could confidently estimate the aerobic and anaerobic NO3 -consumption rates.5) Even though scenario 2 seems to accurately describe the microbial conversions in our reactors, we considered the possibility of PHA accumulation in the anoxic phase.Scenario 3 was developed to evaluate if the potential PHA accumulation would affect the estimated aerobic and anaerobic NO3 -consumption rates in scenario 2. We concluded that even large amounts of PHA accumulation would not affect the estimated rates.Detailed calculations performed in each scenario are also explained: • Scenario 1: no aerobic conversion NO 3 -to N 2 , only to N 2 O. From literature, it is known that aerobic denitrification is not commonly observed in a denitrifying microbial community, at least not at a significant rate.However, in this study, significant N2O production was observed during the aerated periods.So, at first, the aerobic NO3 - consumption rate was assumed equal to the observed N2O production, excluding any N2 production: The anaerobic NO3 -consumption rate was calculated from the measured overall rate and the supposed aerobic rate, knowing that the overall consumption rate comes from a balance between the aerobic and the anaerobic rates (eq.S10).The N2 production rate in the anoxic phase was calculated from the NO3 -and the N2O rates (eq.S15).Similarly to NO3 -, the NH4 + consumption rates could not be determined in each phase individually.Therefore, differently from the overall mass balance approach, the biomass production rates in each phase were derived from the carbon mass balances (eq.S12).The validity of this estimation was proven above (Figure S5).The NH4 + consumption rate was then estimated from the biomass production rate (eq.S11).The electron balance (eq.S16) and electron gap were then calculated for both the oxic and anoxic phases (Table S6): e − gap = R eD − R eA (eq.S18) Electron gap (e-mmol•h -1 ) 2.1 ± 0.7 -5.0 ± 2.3 0.7 ± 0.8 -3.2 ± 1.4 The electron balances did not close in either of the phases in both reactors.The balances show an underestimation of electrons accepted in the oxic phase and an overestimation in the anoxic phase.This suggests that more NO3 -was reduced in the oxic phase than accounted for, whereas an excess NO3 -reduction was accounted for in the anoxic phase.Scenario 2: yes aerobic conversion of NO 3 -to both N 2 and N 2 O -closing the electron mass balance.From the electron balances in the previous scenario and the symmetric electron gaps of the two phases, it was hypothesized that the surplus of reduced NO3 -accounted for in the anoxic phase was actually reduced in the oxic phase.New aerobic and anaerobic NO3 -consumption rates were estimated by closing the respective electron gaps, assuming full conversion of NO3 -to N2 (5 e -transfer): (eq.S19) Linear error propagation was applied to estimate the standard deviations of the derived rates (eq.S3).With the estimations of the aerobic and anaerobic NO3 -conversion rates, we estimated the overall NO3 -consumption rate (eq.S10).Considering that we also have the actual measured value for this rate, we could validate the calculation of the aerobic and anaerobic NO3 -consumption rates by comparing the recalculated overall NO3 -consumption rates in scenario 2 (eq.S10) to the measured rates (Figure S7, panel B).Therefore, we confidently estimated the aerobic and anaerobic NO3 -consumption rates as 0.48 ± 0.14 and 1.13 ± 0.54 N-mmol/h (R4) and 0.17 ± 0.17 and 1.34 ± 0.31 Nmmol/h (R32), respectively.From the total aerobic electron flow, 36±7% and 11±11% went to denitrification.These values were validated with direct calculations from measured concentration profiles throughout each phase, including the NO2 -accumulation rates (supplementary Figures S8-9).• Scenario 3: yes aerobic conversion of NO 3 -to both N 2 and N 2 O, and simultaneous PHA accumulation.Cyclic conditions may select for populations accumulating storage compounds, such as polyhydroxyalkanoates (PHAs).We assessed the potential impact on the estimated aerobic NO3 -consumption rates in scenario 2 of PHA accumulation in the anoxic phase and its subsequent consumption in the oxic period.Biomass contains 4.2 electrons per carbon, whereas polyhydroxybutyrate (PHB, the most common form of PHA) contains 4.5 electrons per carbon, so changes in the electron balance were minimal.Assuming that 50% of the biomass growth in the anoxic phase was actually PHA accumulation, the estimated aerobic and anaerobic NO3 - consumption rates were 0.50 ± 0.14 and 1.09 ± 0.53 N-mmol/h (R4) and 0.20 ± 0.18 and 1.29 ± 0.30 N-mmol/h (R32), nearly identical to scenario 2. Therefore, our conclusions would remain unchanged even in the case of significant PHA accumulation.Confirmation of the aerobic denitrification rates through concentration profiles.The aerobic denitrification rates obtained in scenario 2 using mass balances, and reported in the main text, were confirmed with cycle measurements performed after 43-44 days of operation (Figure S8).The aerobic nitrate and nitrite net accumulation rates were determined by calculating the slope of the linear regression of the concentration profiles (Figure S8) and multiplying by the broth volume.For R32, the rates measured during two cycles were averaged.The error of these rates was assumed to be the error in the slope (R4) or the standard deviation of replicates (R32).The nitrate consumption rates were calculated from these accumulation rates and the influent rate (Table S1): R NO 3 − ,cons = R NO 3 − ,accum − R NO 3 − ,in (eq.S20) There was no nitrite in the influent but it was continuously produced from nitrate reduction, so the net consumption rates were equal to the accumulation rates added to the nitrate consumption rates (RNO2-,prod = RNO3-,cons).R NO 2 − ,cons = R NO 2 − ,accum − R NO 2 − ,prod (eq.S21) The obtained aerobic NOx -denitrification rates were 0.47±0.07(R4) and 0.33±0.16mmol-N•h - 1 (R32), similar to the rates obtained through mass balances with eq.S19.The percentage of aerobic electrons used in denitrification vs. O2 respiration, 33±4% (R4) and 27±9% (R32), were also similar to the values obtained through mass balances (Figure S9).The calculated values were determined from relatively small fluctuations in nitrate concentrations, so they were only used to validate the values obtained through the alternative method (mass balances).NH4 + oxidation activity tests were performed with biomass extracted from R4 and R32, in the presence and absence of the NH4 + oxidation inhibitor ATU.A negative control replaced biomass with water and a positive control contained biomass from an enriched nitrifying microbial community.The nitrifying culture at pH 7 with a biomass concentration of 0.04 gVSS/L was enriched from activated sludge in a 2 L continuously-stirred tank reactor for 53 days, with an HRT of 4.2 days, using NH4 + as energy source (supplied at 34 NH4 + -N mmol/d), bicarbonate as carbon source, and O2 as electron acceptor (provided as air at 500 mL/min).The nitrifying biomass was centrifuged and the pellet was resuspended in PBS buffer and added to rubber sealed bottles (filled with air) to prevent excessive evaporation.The bottles were incubated overnight in a shaker at room temperature after addition of 10 mg-N/L NH4 + to start the batches.NH4 + consumption and NO2 -and NO3 -production were observed only in the positive control.NH4 + concentrations increased in the experiments with biomass from R4 and R32, indicating biomass decay.Identical concentration profiles between experiments performed with or without ATU further confirm the absence of NH4 + oxidation activity in R4 and R32.

Oxic/anoxic cycling in 5 reference Dutch WWTPs
The exposure frequency of activated sludge to oxic/anoxic cycles in wastewater treatment plants (WWTPs) cannot exactly be determined, but estimations were made for different WWTPs using flow rates and tank volumes.The hydraulic residence time in each of the tanks was determined (Figure S19): anaerobic (no O2, no NOx), anoxic (no O2), facultative (can function as anoxic or aerobic tank, according to the treatment needs), and aerobic (with O2).The sludge residence time in the anoxic tanks varied between 11 and 142 minutes, whereas this was 13-155 min for the aerobic zones.The biomass that passes through the settler experiences approximately one oxic/anoxic transition per day, which is equivalent to 15-42 transitions per sludge retention time (SRT, equivalent to the cell generation time, normally 15-20 days in a WWTP).If cells remain in a recycling loop between the aerobic and anoxic zones they can experience up to 9-36 transitions per day, i.e. 132-756 switches per SRT (Table S12).Similarly to the activated sludge in the WWTPs, the biomass in our reactors experienced 4 (R4) and 32 (R32) oxic/anoxic transitions per day, equalling 8 and 64 transitions within one SRT.Experiments with even higher frequency of oxic/anoxic transitions within one SRT should be performed to assess the extent of aerobic denitrification in the highest frequency ranges observed in WWTPs.

Figure S1 .
Figure S1.Daily average anoxic (top, grey) and oxic (bottom, blue) N2O production rates in the low-(R4) and highfrequency (R32) reactors.The shaded areas are the standard deviation of the daily averages, representing the fluctuation of N2O rates within each day.

Figure S2 .
Figure S2.Headspace wall-growth cleaning events (vertical lines) did not affect the profile of the oxic (blue) and anoxic (grey) N 2 O emissions in the low-(R4) and high-frequency (R32) reactors.

Figure S3 .
Figure S3.Concentration profiles during one or two oxic/anoxic cycles after 43 (R4) or 44 (R32) days of operation.The nitrite (symbols) and dissolved oxygen concentrations (blue area), as well as the N2O production rate measured every minute (black line) are represented.In this case, nitrite accumulated in the anoxic phase in R4 and in the oxic phase in R32.

Figure S5 .
Figure S5.Highly comparable biomass concentrations and production rates in the low-(R4) and high-frequency (R32) reactors estimated with two different methods.Panel A: Biomass concentration over time in both the low-and highfrequency reactors, expressed as g/L.Panel B: average biomass production rates during the steady-state.The estimated concentrations and rates were determined from the NH4 + measurements (light grey) and the carbon balance (dark grey), i.e.

Figure S6 .
Figure S6.Schematic representation of the three scenarios considered to calculate the aerobic and anaerobic conversion rates of soluble substrates.The compounds/conversions included in the electron balance in each scenario in the anoxic (grey) and oxic (blue) conditions are represented.All scenarios considered the organic carbon © as electron donor and biomass (X) production as electron sink.Under oxic conditions, all scenarios considered O2 reduction to H2O.Under anoxic conditions, all scenarios considered full denitrification from NO3 -to N2.The different conversions considered for each scenario under oxic conditions were: (1) partial denitrification of NO3 -to N2O, no PHA pool; (2) full denitrification of NO3 -to N2, no PHA pool;(3) full denitrification of NO3 -to N2 with consumption of a PHA pool generated under anoxic conditions.The different conversions considered under anoxic conditions were: (1) and (2) no PHA pool; (3) PHA accumulation.We had experimental measurements of the N2O and CO2 production and O2 consumption rates in each phase, in addition to the overall consumption and production rates of all compounds.

Figure S7 .
Figure S7.Nitrate consumption rates in the oxic (blue) and anoxic (grey) phases of the low-(R4) and high-frequency (R32) reactors.Panel A: comparison between the predicted rates according to scenario 1 (light, assuming no aerobic N2 production) and scenario 2 (dark, assuming yes aerobic N2 production).Panel B: The measured and estimated overall nitrate consumption rates were also compared to validate the calculations.

Figure S8 .
Figure S8.Nitrate and nitrite concentration profiles used to confirm the aerobic denitrification rates.Nitrate concentrations (grey symbols), expected nitrate concentrations if there was no consumption (grey lines), and nitrite concentrations (blue symbols) during the anoxic (grey area) and oxic (blue area) periods are represented.The measurements were performed after 43 (R4) and 44 (R32) days of operation.

Figure S9 .
Figure S9.Percentage of aerobic electron flow used in denitrification in both reactors, as calculated through concentration profiles measured on a single day and overall steady-state mass balances (scenario 2 in Figure S6).

Figure S10 .
Figure S10.Ammonium, nitrite, and nitrate concentration profiles during ammonium oxidation activity tests with biomass extracted from low-(R4, yellow) and high-frequency (R32, blue) reactors, alongside a negative (water) and a positive control (nitrifying mixed culture).Batches were performed with 10 mg NH4 + -N/L, in the presence or absence of ATU.

Figure S11 .
Figure S11.Heatmap with gene presence (grey) and protein expression (coloured) of the nitrogen metabolism, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the oxic phase of R4.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S12 .
Figure S12.Heatmap with gene presence (grey) and protein expression (coloured) of the nitrogen metabolism, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the anoxic phase of R4.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S13 .
Figure S13.Heatmap with gene presence (grey) and protein expression (coloured) of the respiratory chain and ROS-protection pathway, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the oxic phase of R4.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S14 .
Figure S14.Heatmap with gene presence (grey) and protein expression (coloured) of respiratory chain and ROS-protection pathway, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the anoxic phase of R4.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S15 .
Figure S15.Heatmap with gene presence (grey) and protein expression (coloured) of the nitrogen metabolism, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the oxic phase of R32.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S16 .
Figure S16.Heatmap with gene presence (grey) and protein expression (coloured) of the nitrogen metabolism, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the anoxic phase of R32.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S17 .
Figure S17.Heatmap with gene presence (grey) and protein expression (coloured) of respiratory chain and ROS-protection pathway, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the oxic phase of R32.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S18 .
Figure S18.Heatmap with gene presence (grey) and protein expression (coloured) of respiratory chain and ROS-protection pathway, represented as relative abundance of the total proteome, of all MAGs (ordered from high to low abundance in the metagenome) at the end of the anoxic phase of R32.Right bar charts: total relative abundance of each MAG in the metaproteome.

Figure S19 .
Figure S19.Hydraulic residence time in tanks with different conditions in five different Dutch WWTPs, representing the time that the sludge experiences those conditions.The anaerobic tanks do not contain O2 nor nitrogen oxides, the anoxic tanks have no oxygen, the aerobic tanks are aerated with air and the facultative tanks can function either as anoxic or aerobic tanks.

Table S5 .
Summarized explanation of how the overall, aerobic, and anaerobic rates were determined in scenario 1.

Table S6 .
Electron balances and gaps in the oxic and anoxic phases of the low-frequency and high-frequency reactors, assuming the exclusive conversion of NO3 -to N2O under oxic conditions.

Table S7 . Characteristics of the draft genomes recovered from R4 ordered from high to low abundance (top 10 + others):
Genbank accession number, genome completeness, contamination,

Table S8 . Characteristics of the draft genomes recovered from R32 ordered from high to low abundance (top 10 + others):
Genbank accession number, genome completeness, contamination,

Table S9 .
Reference KO-numbers of the genes from the nitrogen metabolism.

Table S12 .
Number of oxic/anoxic transitions experienced by the biomass in our reactors and in five different WWTP configurations (FigureS19), in one day and within one sludge retention time (SRT).