Immobilizing partial denitrification biomass and redox mediators to integrate with the anammox process for nitrogen removal

In this study, immobilizing partial denitrification biomass and redox mediators to integrate with the anammox process for nitrogen removal was investigated. Three redox mediators (RMs), namely, 2-methyl-1,4-naphthoquinone (ME), anthraquinone (AQ) and 1-dichloroanthraquinone (1-AQ) were catalyzed to reduce nitrate to only nitrite by denitrification to integrate with the anammox process for nitrogen removal. First, our experimental results showed that there were 35.8, 42.2 and 53.0 mg-N L−1 nitrite accumulation values with the addition of ME, AQ and 1-AQ, respectively, at the dose of 75 µM by the denitrification process at C/N = 2, which were 25.6%, 48.2% and 86.1% higher than that of the control without the addition of any RMs. Nitrate reductase activities were higher than that of nitrite reductase affected by RMs, which was the main reason for nitrite accumulation and further maintenance of the anammox process. Second, owing to the stable nitrite production by the partial denitrifying biomass with the addition of 1-AQ, the nitrogen removal rate of the reactor that integrated the partial denitrification and anammox process reached 1788.36 g-N m−3 d−1 only using ammonia and nitrate as the influent nitrogen resource in the long-term operation. Third, the 16S rDNA sequencing results demonstrated that Yersinia frederiksenii and Thauera were the primary groups of the denitrifying biomass, which were considered the dominant partial denitrification species.


Introduction
The use of anaerobic ammonium oxidation (anammox) bacteria is a potential biological nitrogen removal technology, which employs nitrite and ammonium as the electron acceptor and donor. 1,2 Owing to the low organic carbon requirements and little sludge production, the anammox process has attracted much attention for its signicant advantages compared to the traditional nitrication/denitrication nitrogen removal process, and it is widely studied for its promising application as an alternate process to the nitrogen removal processes in recent years. [3][4][5][6] Partial nitrication has been widely studied for an application as an anammox pretreatment technology. Due to the inhibition of free nitrous acid, high oxygen, high pH, free ammonia, high temperature and alkali on nitrite oxidizing bacteria, many studies have been reported for controlling the partial nitrication process. [7][8][9] Zeng et al. realized that nitrite could accumulate if the DO concentration was controlled at 0.3-0.7 mg L À1 . 10 Also, mainstream partial nitrication/anammox was proposed as pivotal for a more sustainable treatment of municipal wastewater. 11,12 Hoekstra et al. demonstrated that the total nitrogen removal rate was 0.223 AE 0.029 kg N (m 3 d) À1 during stable process operations at summer temperatures (23.2 AE 1.3 C). 11 However, unsatisfactory accumulation could occur from the partial nitrication process on account of the easy recovery of the nitrite-oxidizing bacteria from oxygen limitation when DO was increased during continuous aeration. In addition, Zeng et al. realized that a low DO concentration could bring down the activities of the aerobic ammonia-oxidizing bacteria, resulting in low nitrite accumulation by the partial nitrication process. 13 Evidently, it was not easy to obtain stable nitrite accumulation and control the further oxidation to nitrate by partial nitrication.
Actually, the accumulation of nitrite during the denitrication process has been demonstrated frequently and drawn much attention in recent years, which can offer another way to supply an electron donor for anammox. Kalyuzhnyi et al. reported the anammox process with autotrophic denitrifying conditions using sulphide as an electron donor for the production of nitrite from nitrate within an anaerobic biolm. 14 Recently, a high nitrite accumulation via denitrication sludge was cultivated and maintained in a long-term operation. [15][16][17] Cao et al. demonstrated that nitrite could be produced in a partial denitrifying upow sludge bed reactor equipped with gas automatic circulation. 15 The peak value of the nitrite accumulation reached 102.6 mg L À1 . However, there were so many studies showing that the nitrite production rate was relatively low with the partial denitrication application, and it could not meet the requirement of anammox for nitrogen removal. 18,19 Therefore, a more robust nitrite production with partial deni-trication should be developed.
Recently, the effects of redox mediators (RMs) were studied on the anaerobic degradation of inorganic and organic contaminants. 20,21 Also, many options focused on the effects of RMs on nitrogen removal by the denitrication process. Aranda-Tamaura et al. proved that 1-dichloroanthraquinone (1-AQ) could accelerate the nitrate reduction rate by sulfur autotrophic denitrication. 22 Guo et al. demonstrated that anthraquinone (AQ) immobilized by calcium alginate could catalyze the denitrication process. It was well known that the nitrogen removal performance of denitrication bacteria would show different tendencies with different kinds of RM addition. 23 The addition of RMs might expedite the electron transfer between the electron acceptors and donors. Kelso et al. found that the competition between nitrate reductase (NR) and nitrite reductase (NIR) would result in the accumulation of nitrite during the denitrication process. 24 Furthermore, our previous study proved that the unbalanced NR and NIR activities affected by RMs were considered the main reason for nitrite accumulation, and further lead to different nitrogen removal performances with different RM additions. 25 Hence, the discrepant effects of RMs on different enzymes might lead to different substrate converting rates, and the accumulation of nitrite accumulation by the denitrication process. Thus, partial denitrication controlling by redox mediators integrating the anammox process for nitrogen removal was possible.
The main objective in this study was to investigate the feasibility of nitrite accumulation by partial denitrication with RM addition. Moreover, the mediated partial denitrication with RMs by immobilizing technology would couple with the anammox process for nitrogen removal.

Inoculum and substrate medium
Seed denitrication sludge was taken from a full-scale municipal wastewater treatment plant (Qingshanhu Wastewater Treatment Plant, Nanchang, China). The sludge was cultivated using glucose as the only electron donor in a lab-scale, up-ow anaerobic sludge blanket (UASB) reactor (2.5 L) year round. The residual nitrogen was mainly in the form of nitrite at a concentration of about 20 mg L À1 . During the cultivation period, NO 3 À -N was supplied in the form of KNO 3 . The substrate medium was described by Yin et al. 25 The anammox biomass used for experiments was obtained from a laboratoryscale anammox upow column reactor. The anammox bacteria of KSU-1 strain (AB057453.1) accounted for about 70-75% of the total biomass in seed biomass by FISH observation. The composition of the trace mineral medium has been previously described by Kang et al. 26 The incubation medium consisted of trace mineral elements, KHCO 3 acting as the only carbon source, ammonium in the form of (NH 4 ) 2 SO 4 and nitrite supplied by NaNO 2 . All regents were purchased from the Sinopharm group chemical reagent co., LTD.

Analytical methods
Water samples were ltered through lters for the determination of the NH 4 + -N, NO 2 À -N and NO 3 À -N concentrations. The

Measurement of NR and NIR activity
2 g (wet weight) denitrication biomass was obtained from the reactor for the crude enzyme activity measurement. The crude enzyme extraction procedure was achieved as described by Qiao et al. 28 The protein concentration was measured using bovine serum albumin as a standard described by Bradford. 29 The NR and NIR activities were measured by the methods according to Kataoka et al. 30 and Yin et al. 31 The reducing rate of nitrate and nitrite was characterized by the NR and NIR activities. The NR and NIR enzyme activities were dened as the reduction of 1 mmol of nitrate and nitrite per minute.

Batch experiments
Three representative redox mediators, including 2-methyl-1,4naphthoquinone (ME), AQ and 1-AQ, were used. All kinds of RMs had one or more six-membered cyclic diketones with two carbonyl groups containing two double bonds. Fig. S1 † illustrates the structure of the three RMs. To avoid the toxicity of RMs on the anammox biomass reported by Qiao et al., 28 the RMs and denitrication biomass were immobilized in entrapping beads. The immobilization method was described by Qiao et al. 28 Batch experiments were carried out in 120 mL serum bottles loaded with 100 mL substrate medium. To remove the redundant nitrogen element, anammox biomass obtained from the UASB reactor was washed three times with mineral medium. Then, both entrapping beads and anammox sludge were transferred to the serum bottles. In addition, single denitrication batch experiments were only put into entrapping beads in the serum bottles to explore the nitrite accumulation of the denitrifying biomass. The temperature was maintained at 37 AE 1 C, while the initial pH was adjusted to 7.5. The serum bottles were blown with nitrogen gas to achieve anaerobic conditions. The initial ammonia and nitrate concentrations were set at about 50 and 100 mg-N L À1 , and the reaction time lasted 6 h in each integrating test. However, the initial nitrate concentration was set at 100 mg-N L À1 in the single denitrication batch experiments. Another substrate medium used in the batch experiments was the same as that of the lab-scale UASB reactor. Glucose and KHCO 3 were the carbon sources for denitrication and anammox biomass. Water samples were taken every 2 h and immediately stored in a refrigerator at 4 C for analysis. All batch tests were carried out in triplicate.

Continuous experiments
The working volumes were about 0.3 L with a height of 15 cm and an inner diameter of 5 cm. Four reactors contained 15 g (wet weight) anammox and 30 g entrapping bead (immobilization denitrication biomass and appropriate RMs). At the beginning of the reactor start-up, the inuent NH 4 + -N and NO 3 À -N doses were 50 mg L À1 and 100 mg L À1 . The initial hydraulic retention time (HRT) was 6 h, corresponding to the nitrogen-loading rate (NLR) at 600 g-N m À3 d À1 . Four reactors were continuously fed with the same media, and the inuent was purged with 99.5% N 2 to maintain dissolved oxygen (DO) below 0.5 mg L À1 . The inuent pH was adjusted to 7.0 AE 0.2 by dosing 2 M HCl, and the temperature was maintained at 35 AE 1 C using a water bath. Fig. 1 shows the schematic diagram of the continuous experiments.
2.6 DNA extraction, amplication of 16S rRNA genes, cloning and sequencing Total DNA was extracted from enriched culture samples (approximately 0.3 g) with a PowerSoil® DNA Isolation Kit (Mobio, USA), as described previously. 32,33 The 16S rRNA gene was amplied using the universal forward primer 27F (5 0 -AGAGTTTGATCMTGGCTCAG-3 0 ) and reverse primer 1492R (5 0 -TACGGYTACCTTGTTACGACTT-3 0 ). 34 The PCR procedure was as follows: initial denaturation at 94 C for 5 min and 35 cycles consisting of denaturation at 94 C for 1 min, primer annealing at 53 C for 1 min and extension at 72 C for 90 s, followed by a nal elongation step of 72 C for 7 min. The PCR products were analyzed on a 1% (w/v) agarose gel and puried using an AxyPrep™ DNA Gel Extraction kit (Axygen Biosciences, Hangzhou, China). PCR products were ligated into a pMD 18-T vector (TaKaRa Bio Inc., Dalian, China), and transformed into Escherichia coli DH5a cells. Plasmid insert-positive recombinants were selected using 5-bromo-4-chloro-3-indolyl-b-D-galactopyranoside (X-Gal) and isopropyl-b-D-thiogalactopyranoside (IPTG) LB indicator plates with 100 mg mL À1 ampicillin. The primer RV-M was used to sequence plasmid inserts by BGI. The nucleotide sequences were compared to those from the GenBank using a BLAST search of the National Center for the Biotechnology Information server. The sequences in this study and reference sequences retrieved from GenBank were aligned with the CLUSTAL_X program. Phylogenetic trees were conrmed by further analysis using MEGA soware version 5.0 with the neighbor-joining (NJ) method.

Results and discussion
3.1 Effects of ME on the nitrogen removal rate Fig. 1a-c showed nitrogen variations with time by the immobilizing denitrication biomass and various ME dosing concentrations integrated with the anammox process. The effects of ME on the nitrite accumulation by the single deni-trication process are depicted through Fig. 1d. As shown in Fig. 1a, the ammonia concentration of control decreased from 50.8 to 33.3 mg L À1 in the absence of inuent nitrite. In the experiments, the nitrite resource only could come from nitrate reduction by denitrication biomass, which was similar to our previous research. However, it was regrettable that the nitrite accumulation by normal denitrication was too little for the anammox process. Because of the underproduced nitrite product in control tests, the ammonium concentration consumption by anammox decreased within the rst 4 h of reaction, but still stabilized aer another 2 hours. Additionally, NO 3 À -N presented a declining trend with almost complete consumption within 6 and 5 h, respectively (Fig. 1c). Remarkably, the nitrate concentration decreased faster from 100 to 8.78, 6.65 and 0 mg L À1 at the ME dosing concentrations of 25, 50 and 75 mM compared with the control of 16.4 mg L À1 within 5 h reaction time. However, in comparison with the control at 20.5 mg L À1 , the peak value of nitrite accumulation reached 28.5, 31.5 and 35.8 mg L À1 at the ME dosing concentrations of 25, 50 and 75 mM, as shown in Fig. 1d. Thus, the maximum ammonia removal efficiency by anammox was 58.5% with 75 mM ME addition benetting from much more nitrite accumulation of denitrication, which was 69.8% higher than that of the control (34.5%). However, the ammonia removal efficiency of the anammox bacteria showed a decreasing tendency when the ME concentration increased to 100 mM, causing the decline of nitrite production by denitrication. Evidently, the ammonia removal efficiency of anammox was related to the nitrite accumulation of denitrication, which was affected by ME. The nitrate consumption and nitrite production of the denitrication biomass with 100 mM ME addition may be due to the original toxicity of the RMs to bacteria with a high level dose.
There was low nitrite accumulation on the batch experiments existing in both anammox and the denitrication biomass with different doses of ME addition, for the reason that there might be not enough nitrite production supplying anammox and the denitrication reaction at the same time (Fig. 1b).

Effects of AQ on the nitrogen removal rate
As described in Fig. 2, the nitrogen removal results with different AQ dose additions practically showed similar nitrate consumptions, but different ammonia removal patterns. Fig. 2c presents the residual nitrate at 0, 1.32, 9.42, 20.8 mg L À1 with 25, 50, 75, 100 mM AQ addition, respectively, corresponding to 16.4 mg L À1 NO 3 À -N of the control remaining aer 5 h reaction time. These results demonstrated that an appropriate AQ addition could enhance the nitrate consumption by the deni-trication biomass. Nonetheless, it appeared to have almost exactly the same ammonia removal performance, meaning the identical anammox activities with different AQ dose additions, including that of the control (Fig. 1a). The AQ addition did not lead to more nitrite accumulation of the denitrication process as shown in Fig. 1d, which might be the main reason for the stagnation of the anammox reaction on account of insufficient substrates. The exact reason for different nitrite accumulation performances of the denitrication reaction at different kinds of RMs might be in connection with the effect of RMs on the key enzymes, such as the NR or NIR enzymes (discussion later). In addition, the low nitrite concentration detection in the tests with mixed bacteria indicating the nitrite production by deni-trication was hardly lack for anammox biomass within the reaction time, as shown in Fig. 1b. (Fig. 1a). Conspicuously, it was found that the supreme ammonia removal efficiency reached 63.6% with 75 mM 1-AQ addition, which was 83.80% higher than that of the control group. Identically, the higher RM dose addition (100 mM) could inhibit the ammonia removal performance by anammox bacteria resulting from the least nitrite accumulation by denitrication biomass. As illustrated in Fig. 3c, the nitrate concentration decreased profoundly in general, as with AQ particularly at a dosing 1-AQ concentration beyond 50 mM. Nitrate at a 1-AQ concentration of 50, 75 and 100 mM was almost removed completely, but remained 28.5 and 20.8 mg L À1 at a concentration of 0 and 25 mM aer 4 h reaction time. Obviously, the RM addition with a certain concentration range accelerated the nitrate conversion by the denitrication biomass. Importantly, the nitrite production could be improved at a certain 1-AQ concentration due to the sharp nitrate consumption, as described in Fig. 3d. The peak nitrite accumulation reached 42.18 mg L À1 with a 75 mM 1-AQ addition, which was 1.73-fold as much as that of the control in the denitrication experiments. Thus, owing to the much greater nitrite amount resulting from denitrication, the higher ammonia removal by the anammox reaction would appear. In our study, for highlighting the effects of RMs, the NO 3 À -N concentration of the inuent only was 100 mg L À1 , so that a particularly high nitrite accumulation could not appear, similar to the study of Cao et al. 15 Our preliminary research made a point that RMs could really affect the activities of key enzymes resulting in the dissimilarity of the NR and NIR activity, which might be the main reason for nitrite accumulation during the denitrication process. 31 The effects of 1-AQ on the denitrication biomass and key enzyme activities could support the viewpoint again. The NR activity of 64.1 mM-N mg protein À1 $min À1 before RM addition was a little different from the NIR activity of 52.26 mM-N mg protein per min. The RM addition enlarged the gap of the NR and NIR activities. With 75 mM 1-AQ addition, the NR (108.46 mM-N mg protein per min) and NIR (71.46 mM-N mg protein per min) activities were enhanced 69.1% and 36.7% higher than those of the control, as depicted in Fig. S2. † Undoubtedly, nitrite can be produced excessively by partial denitrication, owing to the discrepancy between the enhanced NR and NIR activities with a certain 1-AQ dose addition, and might be used as an electron acceptor stabilized by anammox bacteria to oxidize ammonium.

Effects of the C/N ratio on the nitrogen removal rate with 1-AQ addition
Many reports in the literature proved that a suitable C/N is a key for achieving nitrite accumulation by partial denitrication. There would be nitrate residual if the carbon source was limited, and the accumulated nitrite could be reduced when the carbon source was abundant. Thus, two parallel experiments would be established to explore the effects of the C/N ratio on the nitrogen removal performance by mixed bacteria and a single denitrication biomass with 75 mM 1-AQ addition. As shown in Fig. 4d, a signicant nitrite accumulation was observed in each test during denitrication. The maximum nitrite accumulation amount in the single denitrication experiments was 53.0 mg L À1 at C/N ¼ 2 corresponding to 20.8, 35.8 and 32.2 mg L À1 at C/N ¼ 0.5, 1 and 4, respectively. Due to the too slow nitrate conversion, the derisory carbon source (such as C/N ¼ 0.5) also could bring about the decient nitrite production. Based on the huge gap in the nitrite accumulation by denitrication with different C/N ratios, the ammonia removal performance showed the diverse tendency. For instance, when the C/N ratio was adjusted to 0.5, the ammonia removal efficiency was only 36.18%. However, ammonia could be removed much more quickly with the C/N ratio increased at a certain range. The topmost ammonia removal efficiency was 78.6% at the C/N ratio of 2.0, which were 2.17 and 1.49 fold higher than that of the C/N ratio of 0.5 and 1.0. Interestingly, the sufficient carbon source at C/N ¼ 4 conversely reduced more nitrite by denitrication with the lower ammonia removal efficiency (Fig. 3a-c). Heterotrophic denitrication has been widely used for nitrogen removal from wastewater containing nitrate, but it needs external carbon sources and produces high amounts of waste sludge. 35,36 In our study, nitrate only reduced to nitrite when the C/N ratio ¼ 2, but a partial denitrication brought in less residual sludge compared with the complete denitrication.

Comparison of the nitrogen removal performance with different RMs on continuous experiments
Four identical upow xed-bed column reactors, R1 (the control reactor, without RM addition), R2 (with the ME addition of 75 mg L À1 ), R3 (with the AQ addition of 75 mg L À1 ) and R4 (with the 1-AQ addition of 75 mg L À1 ) were applied for continuous experiments with the external carbon source (glucose) as the organic carbon resource (C/N ¼ 2). For enough nitrite production by partial denitrication, R1 without any RMs was still showing the same ammonia removal rate as other reactors by anammox during the initial startup period (Fig. 5A). Only three days later, the effluent nitrite of R1 and R2 were decreased to 0 quickly, although the nitrate removal efficiency of R2 with ME addition was 26.1% higher than that of the control (Fig. 5B and  C). These results demonstrated that the ME addition could accelerate both nitrate and nitrite removal by denitrication biomass so that it was not suitable for mediating nitrite accumulation by partial denitrication. Unlike R1 and R2, the effluent nitrite of R3 and R4 with AQ and 1-AQ addition reached 28.2 and 35.7 mg L À1 in the initial startup period, which could ensure adequate nutrition for the anammox biomass in the continuous experiments. Gradually, the ammonia removal efficiency of R1 increased slowly from 39.4% to 66.6% due to a deciency of nitrite within 21 days incubation. However, the ammonia removal efficiency of R3 and R4 by the anammox reaction showed a more pronounced upward tendency. On day 21, the ammonia removal rate of R4 with 1-AQ addition reached 198.0 g N m À3 d À1 , corresponding to the ammonia removal efficiency of 100%, which was 50.1% higher than that of the control (132.0 g TN m À3 d À1 ). Apparently, the more excellent ammonia removal performance of anammox was relevant to RMs (AQ and 1-AQ) promoting the abundant nitrite accumulation by denitrication. Also, benetting the distinguished anammox reaction, the TN removal rate of R3 (565.7 g TN m À3 d À1 ) and R4 (599$3 g TN m À3 d À1 ) was enhanced by 10.6% and 17.2% compared with R1 (511.6 g TN m À3 d À1 ).
Then, the nitrogen loading rates of all reactors were increased by shortening HRT with the constant inuent substrate concentrations to investigate the long-term effects of RMs on the integrating process. In these experiments, the anammox reaction of the integrating process was the main research object, but not denitrication. Thus, HRT was adjusted on the basis of the effluent ammonia concentration. At the end of this experiment, the NRRs of R3 and R4 reached 1733.2 and 1788$4 g TN m À3 d À1 , which was about 27.4 and 31.5% higher than that of R1 (1360.2 g TN m À3 d À1 ) when the NLRs of all reactors increased to 1801.2 g TN m À3 d À1 on day 188. The results of this study proved that nitrite was produced stably with a partial denitrication application with RM addition, which revealed an alternate efficient way to treat nitrogen wastewater by integrating with the autotrophic anammox process.

Biomass identication
Considering the different forms of two functional bacteria, 36 samples of dispersed sludge and 28 samples of embedding pellets taken from R4 were identied via 16S rDNA sequencing. The anammox biomass of Candidatus Kuenenia sp. was the primary group, which accounted for approximately 69.4% (25/ 36) of the total biomass in the dispersed sludge of the reactor, as illustrated in Table 1 and Fig. S3. † In addition, some kinds of denitrifying biomass accounted for about 16.7% of the whole biomass community, such as Yersinia frederiksenii, Flavobacteriaceae, Trichococcus, Thauera and Klebsiella genera. And, Yersinia frederiksenii and Thauera were regarded as partial denitrication, which have the capacity to reduce nitrate to nitrite, but not further to nitrogen gas. 15,37 Also, Yersinia frederiksenii and Thauera were the dominant groups with a percentage of 32.1% and 14.3% in the embedding pellets by the phylogenetic classication according to Table 2 and Fig. S4, † which might be responsible for the high nitrite accumulation in this study. Hydrogenophaga, Trichococcus, Klebsiella and Dechloromonas genera were also considered denitrication biomass. It was determined that the denitrifying biomass accounted for about 75% of the whole biomass community. In addition, the anammox biomass of Candidatus Kuenenia sp. was the fraction group (2/28) aer continuous incubation in the embedding pellets.
Recently, different research directions of anammox bacteria were followed with interests by scientists, such as resuscitation aer dormancy, 38 identication of a new species and its characteristics, 39 analysis of the cellular structure 40 and the engineering application of sidestream and mainstream. 41 Some researchers' work could bend to increase the nitrite productivity rate by ammonia-oxidizing bacteria. 42 Researchers found that partial denitrication also could produce nitrite, but it was not constant and unstable. 37 Also, Wang et al. reported that a partial denitrication could couple with the immobilization of anammox in a continuous upow reactor with a nitrogen removal efficiency of 88.5%. 43 Our research focused on accelerating the nitrite accumulation of denitrication by RM addition, which provided another way of thinking to integrate the anammox process for nitrogen removal. With 1-AQ addition, the nitrite accumulation increased 1.55-fold with C/N ¼ 2 by denitrication compared with that of the control, and the NRRs of the integrating process reached 1801.2 g TN m À3 d À1 through 188 days incubation.
The mechanism that RMS could accelerate the bacteria activities was also controversial. Qiao et al. revealed that RMS could catalyze the enzyme activities instead of coenzyme Q. 28 The addition of RMs might expedite the electron transfer between the electron acceptors and donors. Van der Zee et al. recognized that part of these differences may be directly related to the E0 0 of the shuttling compounds investigated. 44 In addition, other physico-chemical properties of the shuttling compound may play a role in determining its feasibility and strength as a mediator for azo-dye reduction. However, it was regretful that the method for how to choose the optimal RMs for the special biomass to treat contaminants did not have a unied view. In our study, the purpose of the RM addition was to increase the nitrite accumulation. Thus, the chosen RMs could accelerate the NR enzyme activities even higher than the NIR activities. And, the much bigger gaps of NR and NIR activities by partial denitrication had, the better we would want.

Conclusion
In this study, it was demonstrated that the nitrite accumulation via partial denitrication with 1-AQ could be increased 86.1% at the optimal dosing concentration of 75 mM. The NRR of the reactor via immobilizing partial denitrication biomass and 1-AQ to integrate with anammox process reached 1788.4 g TN m À3 d À1 using only ammonia and nitrate as the inuent nitrogen resources in the long-term operation. The 16S rDNA sequencing results proved that the partial denitrication species was the primary group of the denitrifying biomass.

Conflicts of interest
We declare that we have no nancial and personal relationships with other people or organizations that can inappropriately inuence our work, there is no professional or other personal interest of any nature or kind in any product, service and/or company that could be construed as inuencing the position presented in, or the review of, the manuscript entitled.