Effect of free ammonia inhibition on NOB activity in high nitrifying performance of sludge

The inhibition of free ammonia (FA) on nitrite-oxidizing bacteria (NOB) was investigated using an enriched NOB community with high nitrifying performance. A continuous-flow reactor was operated for the enrichment of the bacterial community. High-throughput sequencing analysis showed that Nitrospira (NOB) using in batch experiments was extended from 4.78% to 12.08% during the under continuous-flow operation for 27 days. For each batch experiments, an ammonia injection at the start-up resulted in the desired initial FA concentration (at pH = 8.1–8.2, T = 25 °C), and a continuous ammonia feeding stream allowed for a relatively stabilized FA levels as much as the initial one. Results indicated that FA inhibition on NOB was not instantaneous but occurred gradually at a certain reaction time. Low concentrations of FA (18.08–24.95 mg L−1) had a limited inhibition on NOB with increasingly high nitrate production rates, whereas high FA levels (36.06–50.66 mg L−1) exerted a significant negative impact on the NOB. Also, strong adaptation happened in these high levels of FA inhibition on NOB, which resulted in an overall low NOB activity during the whole aerobic operation.


Introduction
Traditional nitrication is a two-step process commonly utilized in wastewater treatment plant (WWTP). These steps are performed by two different autotrophic bacterial population. First, ammonia is oxidized to nitrite by the ammonia-oxidizing bacteria (AOB) and subsequently, the nitrite is converted to nitrate by nitrite-oxidizing bacteria (NOB). Recently, partial nitrication (ammonia oxidation to nitrite) has drawn much attention as it could reduce the oxygen and organic compound demand by 25% and 40%, respectively, compared with nitrication and denitrication. 1 The success of the process depends on the elimination of NOB, thereby resulting in nitrite as the primary nal product. 2 Therefore, attempts have been made to devise a microbiological method for AOB enrichment and NOB washout to enable optimal partial nitrication. 3,4 Some previous works have proved that temperature, pH and dissolved oxygen (DO) are responsible for nitrite build-up. [5][6][7] Nevertheless, the direct utilization of temperature (30-35 C) changes is not feasible in municipal WWTP due to high energy consumption and capital investment. Especially in winter, when the temperature outside is almost close to 0 C, it is hard to maintain the chamber temperature at 30-35 C for effective partial nitrication. It has been suggested that high DO concentration improves the ammonia oxidation rate. 8 Therefore, maintaining a low DO concentration (<0.5 mg L À1 ) for partial nitrication fails to work in highly efficient nitrogen removal systems. Means to maintain high pH and ammonia levels for partial nitrication in WWTP consume huge material resources, thus incurring high operation costs. Therefore, a better solution is proposed using a method for entire NOB washout by bacteria screening from biological system, resulting in the sole existence of AOB. 9,10 Such sludge used in WWTP treating municipal and high-strength wastewater, achieves saving of operation and infrastructure costs. It is unnecessary to deliberately control low-level of DO, which enables to improve the efficiency of partial nitrication in wastewater treatment.
Washing out the entire NOB population by bacteria screening depends on inhibitors, free ammonia (FA) 11,12 and free nitrous acid (FNA). 13,14 Traditionally, FA is the rst choice for nitrite build-up and NOB washout. In recent years, some researchers have proposed a new strategy to increase bioenergy recovery from wastewater or sludge treatment using FA. [15][16][17][18] They demonstrate that pre-treatment of the activated sludge with FA for 24-40 h could reduce sludge yields by 20%. At the same time, the NOB activity could be decreased to some extent without any adverse impact on the reactor performance. [19][20][21] Consequently, it is imperative that the characteristics and regulations of FA inhibition are discussed.
It is known that the inhibitory effect of FA exerts severe pressure on NOB activity. 22 Although FA is known to slow down AOB activity, NOB is considered to be much more sensitive to FA than AOB. 23 Anthonisen et al. 13 suggested that inhibition of NOB begins at a FA concentration of 0.1-1 mg L À1 , whereas the inhibition threshold for AOB was 10-150 mg L À1 . FA concentrations of 4 mg L À1 and 6 mg L À1 have been reported to inhibit the NOB in two different studies. 24,25 In addition, Abeling et al. 26 reported that NOB was inhibited when FA reached 1-5 mg L À1 .
The data mentioned above clearly shows that the threshold FA concentration for NOB inhibition varies. However, these FA values were calculated from the initial ammonia concentrations without considering the nitrication rate and ammonia consumption. Hence, the variation in FA concentration is because of the consumption of ammonia under different nitrication rates. Thus, the initial FA concentration should not be considered the real inhibitory value. Very few reports focus on exploring relatively stabilized FA inhibition of NOB with high nitrifying performance in a certain time.
In this paper, a series of batch experiments were designed to establish the relationship between relatively stable FA inhibitory concentrations and NOB response. An enriched NOB culture with high nitrate production rate was used, taken from the continuous-ow reactor with 99% conversion ratio of ammonia to nitrate. The NOB response was assessed systematically by monitoring NOB activity and nitrate production.

Continuous-ow operation to enrich NOB
A continuous-ow reactor with a working volume of 95 L was employed to selectively grow NOB for 62 days. Nitrifying sludge from the Gao Bei Dian Wastewater Treatment Plant (WWTP) in Beijing, China was used as the inoculum. This reactor was fed with stock solutions of NH 4 Cl (nitrogen source), KH 2 PO 4 (phosphorus source) and Na 2 CO 3 (pH buffer, alkalinity source) as well as a mineral stock solution detailed in Table 1. The 1 mL trace element stock solution contained 0.5 g L À1 ZnSO 4 $7H 2 O, 0.5 g L À1 MnCl 2 $4H 2 O, 0.4 g L À1 CoCl 2 $6H 2 O, 0.4 g L À1 CuSO 4 $5H 2 O, 0.2 g L À1 NiCl 2 $6H 2 O and 0.05 g L À1 Na 2 MoO 4 -$2H 2 O. The hydraulic retention time (HRT) was maintained at 5 h. Temperature and pH inside the reactor were set at 25 AE 1 C and 7.0 AE 0.1, respectively, whereas DO was controlled in the range of 1.0-1.5 mg L À1 . Throughout the operation period, 99% of the incoming ammonia was oxidized to nitrates. Mixed liquor suspended solids (MLSS) was calculated as 2438 mg L À1 on the rst day of the continuous-ow culture, which increased up to 4675 mg L À1 on day 62.  Fig. 1 was used to determine the inhibitory effects of FA. The reactor was fed with NH 4 Cl and Na 2 CO 3 solutions from two 250 mL feeding bottles using peristaltic pumps that controlled the ow rates. This setup offered continuous streams of ammonia and Na 2 CO 3 , respectively. The pH adjusted automatically at 8.1 AE 0.1 and was controlled by the ow rate of the Na 2 CO 3 feeding stream. The temperature was controlled at 25 AE 1 C using a thermostatic heater covering the reactor exterior. Continuous aeration was supplied through a perforated tube using an air pump. The dissolved oxygen (DO) concentration was monitored through a DO probe and maintained at 1.0-1.5 mg L À1 by an air owmeter. The mixture was stirred mechanically at a speed of 200 rpm during the aerobic reaction. The sampling port on top of the reactor allowed for sample collection and feed addition.

Experimental operations.
A total of eight batch experiments were performed in a continuous ammonia feeding mode for 3 h each in an aerobic condition in the fermenter. 3.5 L of mixed liquor was removed from the parent continuous-ow reactor in batches across different time points as the nitrate production rate (NPR 0 ) reached various levels ( Table 2). Ammonia was added in two ways: ammonia was injected into the fermenter at the start-up and a consistent ammonia feed was supplied throughout the operational phase. The specics of the batch tests are as follows: (i) Before each experiment, two sludge samples with same characteristics were collected from the continuous-ow reactor simultaneously. Their MLSS was consistent with that in continuous-ow reactor. All samples were washed thrice with  14 which is less than the inhibitory concentration of 0.02 mg L À1 . Therefore, free nitrous acid (FNA) had negligible inuence on the reaction. (iv) On reaching 3 h, the heater, stirrer, aeration devices, pH controller and NH 4 Cl feeding pump were turned off in each batch reaction. The mixed liquor was discarded from the reactor. As the continuous-ow reaction progressed, the NPR 0 of the sludge improved gradually. When the NPR 0 reached a high value, fresh sludge was withdrawn from the continuous-ow reactor to investigate the inhibitory effect of high FA levels.

Analytical methods
The inuent and effluent samples were withdrawn from the continuous-ow reactor on a daily basis and analyzed immediately for NH 4 + -N, NO 2 À -N and NO 3 À -N. Mixed liquor suspended solids (MLSS) was also measured inside the continuous-ow reactor on a daily basis. All analyses were performed according to standard methods. 27 In the batch experiments, NH 4 + -N, NO 2 À -N and NO 3 À -N were measured for the calculation of FA, FNA and nitrate production rate. DO and pH were monitored in the continuous-ow reactor using programmable logic controller, whereas an installed Mettler-Toledo GmbH DO Probe, (Mettler-Toledo, Switzerland) and pH meter (Mettler-Toledo CH-8902, Switzerland) were used in the batch tests, respectively.
In the batch experiments, concentration of FA and FNA were calculated using eqn (1) and (2) proposed by Anthonisen et al. (1976) 13 : where NPR 0 is the nitrate production rate measured for Blank I, was zero) and t is the reaction time (a constant of 1 h). FA has a strong negative effect on the biosynthesis process of NOB resulting in the reduction of NOB activity. Hence, the formulae for the calculation of NOB activity in batch experiments were established as follows: where [NO 3 À -N] t is the residual NO 3 À -N concentration at times 0.5, 1, 1.5, 2, 2.5 and 3 h, [NO 3 À -N] tÀ1 is the residual NO 3 À -N concentration half an hour before [NO 3 À -N] t was measured (0, 0.5, 1, 1.5, 2 and 2.5 h) and t is the 0.5 h reaction time. It should be noted that NOB activity was 100% at 0 h.

Continuous-ow performance
Long-term monitoring of the reactor performance revealed that complete nitrication was achieved. As illustrated in Fig. 2, inuent ammonia was increased in steps from an initial concentration 65 to 812 mg L À1 . Aer reaction time of 5 h, the concentration of ammonia was 2.01 mg L À1 , while that of nitrite was a negligible 0.47 mg L À1 in the effluent. Thus, the ammonia was almost completely oxidized into nitrates (over 99%) and the calculated nitrate production rate (the nitrate production rate in this reactor was assessed using the difference of nitrate in the inuent and effluent per hour) was only 12.51 mg (L h) À1 . Low nitrate production rate in the initial phase indicated that the activity of NOB was less and inappropriate for FA inhibition experiments. However, when the inuent ammonia was continually improved, it was almost completely transformed into nitrate (99% of ammonia into nitrate transformation ratio during the operation phase), thus reecting the enhancement of NOB activity. Besides, the output of nitrite was maintained 0.24-0.56 mg L À1 through the 62 days. When the nitrate production rate reached 72.04 mg (L h) À1 on day 35, the mixed liquor was collected for FA inhibition experiments. With the increase of nitrate production rate in the following days, the sludge was periodically removed for use as the bacterial source for FA inhibition experiments. Additionally, MLSS stepped up from the initial 2438 mg L À1 to a nal value of 4675 mg L À1 implying enrichment of the NOB population. High-throughput sequencing analysis indicated that the microbial culture of the enriched NOB was Nitrospira, which increased by approximately 5.6 times from 2.14% (day 1) to 12.08% (day 62). A signicant increase of the NOB fraction (4.78%) was detected on day 35, when the rst batch experiment was initiated. As the continuous-ow proceeded, the NOB activity and ratio increased progressively to reach 154.06 mg (L h) À1 and 12.08%, respectively, when the nal batch experiment was conducted on day 62. Thus, the enriched NOB taken from day 35 to 62 triggered a more evident consequence of the NOB response upon FA.

Effect of low FA concentration on NOB activity
Several tests were performed on the samples with varied NPR 0 under different FA levels (details in Table 2) in order to investigate the biocidal effect of FA on the activated sludge. Fig. 3 reects the variation of NOB activity in samples with different NPR 0 values aer being exposed to different FA concentrations. FA levels were maintained in a relatively stable range (approximately 18.08-18.83, 20.13-20.86, 22.13-22.81, 24.14-24.93 mg L À1 ) through a continuous mode of ammonia feeding and an ammonia injection at the beginning. The continuous ammonia feeding mode was to ensure the ammonia to nitrate transformation so that the initial ammonia was estimated for FA concentration instead of participating in the transformation of ammonia into nitrate. Therefore, the ammonia concentration was maintained stable during the whole aerobic period resulting in relatively level-off proles of the calculated FA concentration. The subtle uctuation in the FA proles was because the ammonia (assessed for FA levels) was oxidized into nitrate more or less. Since the oxidation of ammonia into nitrate was based on NOB activity, the nitrate production proles clearly indicated the inhibitory impact of FA on NOB. It should be noted that NOB activity was 100% at 0 h as there was no inuence of FA. As shown in Fig. 3(a), the FA range of 18.08-18.83 mg L À1 (Test 1) was selected to investigate the effect on NOB with NPR 0 of 72.04 mg (L h) À1 . The NO 3 À -N production prole ascended sharply to 104.86 mg L À1 upto 1.5 h following by a sudden transition to a slow increase to 112.28 mg L À1 . Similarly, NOB activity decreased gradually from 100% at 0.5 h to 92.45%, while an unexpected drop happened from 8.85% to 4.61%. This observation correlated well with previous studies, 28-30 which reported that FA had a negative impact on NOB activity. It could be hypothesized that NOB suppression was not instantaneous but occurred gradually aer 1.5 hours. Surprisingly, the percentage of NOB activity decreased intensely reaching 4.61% at the end of the reaction. The results indicated that the FA concentration ranging from 18.08 to 18.83 mg L À1 appeared to be the threshold for the sludge at 72.04 mg (L h) À1 . When the NPR 0 value reached 101.57 mg (L h) À1 (Test 1) as shown in Fig. 3(b), more FA (20.13-20.86 mg L À1 ) was introduced resulting in the rapid improvement of NO 3 À -N production before 1 h accompanied by a relatively gradual increase in the subsequent hours. This trend could be due to the fall of FA to the bottom range before 1 h. However, a slight increase of FA level aer 1 h allowed for considerable suppression of NOB activity. Therefore, NOB activity was reduced only to 96.48% in 1 h, while an abrupt drop in NOB activity from 96.48% to 22.74% was observed in 1-3 h. In comparison, the NOB activity at NPR 0 of 115.78 and 128.05 mg (L h) À1 corresponding to FA levels of 22.13-22.81 (Test3) and 24.14-24.93 (Test 4) mg L À1 , respectively, showcased a similar behaviour as well. The NOB activity was not reduced instantly by FA but occurred gradually. Notably, the NO 3 À -N production at the end of the aerobic reaction period showed an upward trend from Tests 1 to 4 (112. 29 As can be seen in Fig. 4(a), NO 3 À -N accumulated gradually reaching a concentration of 185.41 mg L À1 when inhibition by FA became apparent in the range of 36.06-36.73 mg L À1 (Test 5). Reduction NOB activity to 78.43% occurred at 0.5 h, which was likely due to the strong inhibitory effect of FA at high-levels. In the following hour, the NOB activity began to decline rapidly from 0.5 to 1 h, followed by a slower fall later. Similar result was obtained for the inhibition of NOB by FA at 40.06-40.81 mg L À1 concentration, which is depicted in Fig. 4(b) (Test 6). However, NO 3 À -N production reached only up to 128.05 mg L À1 at the end of the reaction. Furthermore, the NOB activity was estimated as 37.36% at 0.5 h and reduced to 15.93% to a limited extent, which was mainly because of the intense inhibition by FA present at high levels of 40.06-40.81 mg L À1 . In addition, when the mixed liquor with an initial NPR 0 of 143.35 mg L À1 was collected from the continuous-ow culture on day 57 (Test 7), NO 3 À -N accumulation was increased to 63.05 mg L À1 in the nal reaction phase. Fig. 4(c) demonstrates the NOB performance under a higher FA level of 44.03-44.96 mg L À1 . NOB activity declined signicantly to 24.17% aer 0.5 h. Subsequently, a slow descent in the NOB activity led to 8.37%, which represented a suppressed and inactive NOB response due extreme inhibition by such high levels of FA. Interestingly, a further increase of FA concentration to 50.04-50.66 mg L À1 (Test 8) did show further inhibitory effect on NOB activity. As shown in Fig. 4(d), the sludge with the highest NPR 0 of 154.06 mg (L h) À1 produced minimal NO 3 À -N compared to the other 7 experiments accumulating only 44.35 mg L À1 at the end of the reaction. Moreover, the NOB activity dropped suddenly down to 13.76% at 0.5 h and then gradually to 5.31%. According to these results, NOB was inhibited intensely with incremental FA levels in the higher range despite the NPR 0 increasing only on a small scale. Notably, the high FA concentration at the beginning of the reaction had signicant inhibitory effect NOB activity at the startup (0.5 h) in all four experiments discussed above. However, the biomass had started to adapt to the strong inhibition by high in initial stage, thus resulting in the sustenance of NOB activity i.e. ammonia to nitrate transformation through the whole aerobic reaction period. This was in agreement with Wong et al., 31 whose results suggested that NOB adapted to FA above 40 mg L À1 and could resist the inhibition.

Microbial diversity analysis
High-throughput sequencing was used to analyze the microbial diversity of the seed sludge and the cultivated sludge for comparison. A coverage 0.96 was achieved in the three samples indicating that the result exclusively represented the microorganisms present in the sludge samples.  to 0.05. The most feasible explanation for this phenomenon is that the microbial abundance and diversity declined while enriching the sludge for nitrifying bacteria. Meanwhile, nitri-ers had been predominant among the microbial communities in the sample. In other words, the performance of the system and richness of the specically functional population were improved with a reduction in the microbial diversity as previously reported. 32,33 The structure of bacterial community at the genus level is presented in Fig. 5. The seed sludge showed that the Aridibacter, Ferruginibacter and Thauera were the predominant genera as depicted in Fig. 5(a). Their relative abundances were 5.25%, 2.88% and 2.85%, respectively. They belonged to the phyla Acidobacteria, Bacteroidetes and Proteobacteria, respectively, which were the main components of the activated sludge in the wastewater treatment system. 34,35 Nitrosomonas (affiliated to b-Proteobacteria phylum, called ammonia-oxidizing bacteria) and Nitrospira (a distinct phylum, called nitrite-oxidizing bacteria) were detected at 0.43% and 2.14%, respectively, in municipal wastewater treatment. 36,37 These implied that the abundance of the specic functional species (Nitrospira) was very low. Hence, the seed sludge was not suitable for investigating the inhibitory effect of FA. With the increase of continuous-ow operation, the Nitrospira improved up to ratios of 4.78% and 12.08%, respectively, at days 35 and 62. Tests 1 to 8 were conducted using increasingly enriched Nitrospira from 4.78% to 12.08%. The results were consistent with this showing excellent biological nitrication performance i.e. high nitrate production rates ranging from 72.04 mg (L h) À1 in Test 1 to 154.06 mg (L h) À1 in Test 8. Thus, the negative impact of FA on Nitrospira (NOB) is distinct and persuasive. As represented in Fig. 5(b) and (c), the ratio of Nitrosomonas (AOB) increased from 0.54% to 24.42% during the operation period of 62 days compared to the seed sludge that contained only 0.43%. The increase in Nitrosomonas was 45 times, while Nitrospira was enriched 5.6 folds compared to the seed sludge. This is in agreement with the previous reports that NOB grew more slowly than AOB. [38][39][40] At the same time, other biological communities were gradually reduced and washed out of the continuous-ow reactor allowing the enrichment of Nitrosomonas (AOB) and Nitrospira (NOB) communities.

Proposed role of FA inhibition on NOB washout
The compound (FA) had been demonstrated to be the most effective inhibitor for NOB washout. 41 Previous studies clearly showed that AOB had a much higher level of tolerance to FA compared to NOB, which may have contributed signicantly to the elimination of NOB from the biological system. However, considering that wastewater treatment should be economic and environment friendly, it is impossible to employ high levels of FA in municipal WWTP. Thus, FA may provide a feasible solution for the washout of NOB in the lab scale.
Based on the above results, two main aspects about FA inhibition on NOB could be concluded. Firstly, the reaction time of FA inhibition was of supreme importance, as the FA inhibition on NOB was not instantaneous but occurred gradually. As presented in Fig. 3 and 4, the NOB activity declined gradually as the reaction progressed. In addition, lower and higher levels of FA had signicantly varied inhibitory effects on NOB. Lower-level of FA (18.13-24.93 mg L À1 ) had a limited inhibitory effect, whereas NOB could develop resistance to some extent towards higher levels of FA (36.58-50.66 mg L À1 ). It has been reported previously that FA concentration up to 200 mg L À1 could suppress NOB activity severely. 42,43 Also, an elevated level of FNA could be supplied for further inhibition of NOB growth resulting in the complete elimination of NOB. 44 Therefore, combination of high-levels of inhibitors FA and FNA could attain the desired NOB washout. It is imperative to investigate the combined effect of FA and FNA ion the inhibition and NOB washout in subsequent experiments.
It should be mentioned that the NOB culture used in this study coexists with AOB as evident from the high-throughput sequencing, which detected Nitrosomonas (AOB) levels of 0.54% and 24.42% at days 35 and 62 compared to 0.43% in the seed sludge. The existence of AOB in the system was inevitable while studying the effect of FA inhibition on NOB. Thus, the response of AOB has not been captured in the data from this study. However, the ammonia fed into the fermenter could transform into nitrate only because of the cooperative activity of AOB and NOB populations. Consequently, the AOB community  Paper seems to have still maintained a relative high activity under these FA concentrations. The results (FA inhibition on NOB) obtained may be valid for achieving NOB washout in the presence of AOB. This FA approach could be used to guide AOB enrichment and complete NOB washout in high nitrifying performance systems for efficient and cost-saving partial nitri-cation in municipal WWTP.

Conclusions
The feasibility of a continuous-ow conguration was demonstrated for the enrichment of NOB with high performance by conducting a series of experiments of FA inhibition. Highthroughput sequencing implied that Nitrospira (NOB) used in the batch experiments increased from 4.78% to 12.08% during the continuous-ow operation period corresponding to the high nitrate production rate from 72.04 to 154.06 mg (L h) À1 . For each batch test, the initial FA generation was through the ammonia injection at the beginning (at pH ¼ 8.1-8.2, T ¼ 25 C) of the culture and the maintenance of these FA levels by a continuous ammonia feeding stream throughout the aerobic reaction period. Results showed that the negative impact of FA on NOB was not immediate but occurred progressively. Low levels of FA (18.08-24.95 mg L À1 ) were found to have limited effect on NOB with increasingly high nitrate production rates. However, much higher levels of FA (36.06-50.66 mg L À1 ) had a strong inhibition on NOB leading to the drastic reduction of activities. Additionally, the NOB was capable of adaption to FA inhibition and thus, the microorganisms were able to remain active at high concentrations of FA. Notably, 18.08-18.83 mg L À1 FA concentrations is likely to be the threshold value for the sludge at 72.04 mg (L h) À1 . It could be suggested that FA inhibitor was effective in NOB suppression and could be applied as an inhibitor for NOB washout where partial nitrication is desired.

Conflicts of interest
There are no conicts to declare.