Hydrogenophaga sp. H7 was isolated from a copper/iron mine soil in Daye City, Hubei, China, as previously reported (Fan et al., 2019). The basal medium (BM) was used to enrich strain H7. The nitrification medium (NM) and denitrification media (DM) were used to measure the abilities of strain H7 to remove NH₄⁺-N or NO₃⁻-N (DM-1) or NO₂⁻-N (DM-2), respectively. The simultaneous nitrification and denitrification mixed media (SNDM) were used to measure the abilities of strain H7 to remove NH₄⁺-N, NO₃⁻-N and oxidize As(III) (SNDM-1) simultaneously or remove NH₄⁺-N, NO₂⁻-N and oxidize As(III) simultaneously (SNDM-2) (see Supplementary material 1 for the detailed composition of these media).

Strain H7 was cultured in BM medium at 28°C with shaking at 150 rpm until the OD₆₀₀ was approximately 1.0. For the As(III) oxidation assay, 1% (v/v) strain H7 was inoculated in 100 ml of BM medium by adding 30 mg/l As(III). For the nitrogen removal assay, the cells were collected by centrifugation at 8,000 g for 5 min, washed three times with normal saline, and suspended in normal saline to the same OD₆₀₀. The above bacterial suspension was then inoculated (1%, v/v) in 100 ml of NM, DM-1, DM-2, SNDM-1, or SNDM-2 media. The culture samples were cultured at 28°C with shaking at 150 rpm, and samples were taken to measure the OD₆₀₀ and the concentrations of NH₄⁺-N, NO₃⁻-N, NO₂⁻-N, NH₂OH, total nitrogen (TN), As(III), and As(V) at designated times (see Supplementary material I for the measurement methods) (APHA, 1998; Yang, 1999; Liao et al., 2013; Wang et al., 2013). All of the above experiments were performed in triplicate. The removal or oxidation rate was calculated as (C₀-C_t)/t. The removal or oxidation efficiency was calculated as (C₀-C_t)/C₀*100%. C₀ is the initial concentration, and Ct is the final concentration at time t. All the results are shown in the form of the mean.

The effects of carbon source, C/N ratios, pH, temperature, and shaking speed on nitrogen removal and As(III) oxidation were investigated by single factor tests. Detalied parameters are as follows: (1) Carbon sources: glucose, 4-hydroxybenzoate (4-HBA), sodium acetate, and sodium citrate. (2) C/N ratios: 3, 5, 8, 10, and 12. (3) pH: 6.0, 7.0, 8.0, 9.0, and 10.0. (4) Temperature: 15°C, 20°C, 28°C, 37°C, and 40°C. (5) Shaking speed: 0 rpm, 50 rpm, 100 rpm, 150 rpm, and 200 rpm (see Supplementary material I for the detailed steps). The culture samples were taken to measure the OD₆₀₀ and the concentrations of NH₄⁺-N, NO₃⁻-N, NO₂⁻-N, TN, As(III), and As(V) at 20 h. All of the above experiments were performed in triplicate.

The wastewater was collected from a pig farm in Wuhan City, Hubei, China, after being treated with the Moving Bed Biofilm Reactor reaction (wastewater O1). Glucose (200 mg/l) and As(III) (5 mg/l) were added to the wastewater in this study. For the single cycle test, strain H7 with 10⁷ CFU (colony forming units) was inoculated in 100 ml wastewater to analyze the effect of strain H7 on nitrogen and arsenic removal. The treated wastewater was cultured at 28°C with shaking at 75 rpm for 16 h, and then 108 mg/l FeCl₃ was added at 16 h. For the batch-to-batch experiment, strain H7 with 10⁷ CFU was inoculated in 100 ml wastewater and incubated at 28°C with shaking at 75 rpm for 18 h. The bacterial cells were collected by centrifugation at 8000 g for 5 min, washed three times with normal saline, and suspended in normal saline. The suspension was then inoculated with 100 ml of fresh wastewater. A total of 108 mg/l FeCl₃ was added at the end of the cycle. The culture samples were taken to measure the concentrations of NH₄⁺-N, NO₃⁻-N, TN, As(III), and As(V) at designated times.

One experimental group was designed: NH₄⁺-N versus control (strain H7 cultured in NM medium vs. strain H7 cultured in NM medium, but 45 mg/l urea was the sole nitrogen source). The method of inoculation and culture for strain H7 was consistent with that of the NH₄⁺-N removal experiment. The cells were collected by centrifugation (8,000 g, 5 min) for 8 h and then freeze-dried, which were detected and analyzed by Wuhan Gene Create Ltd., Wuhan, China (see Supplementary material I for the detailed steps) (Bradford, 1976; Shilov et al., 2007).

qRT-PCR was used to detect the expression of the arsenite oxidase-encoding gene aioA in strain H7 (see Supplementary material I for the detailed steps). Gene expression was normalized by 2^−ΔΔCT analysis with an iQ5 Real-Time PCR Detection System (Bio-Rad, United States). The primers used in this experiment see Supplementary Table S1.

The NH₄⁺-N removal capacity of Hydrogenophaga sp. H7 is shown in Figure 1A. The OD₆₀₀ reached a maximum value of 0.502 ± 0.019 at 16 h when NH₄⁺-N was used as a sole nitrogen source. Meanwhile, the removal efficiencies of NH₄⁺-N and TN reached 99.50% with a maximum removal rate of 2.71 mg/l/h and 97.31% with a maximum removal rate of 2.34 mg/l/h, respectively (Supplementary Table S2). The concentration of hydroxylamine (NH₂OH), which is the product of nitrification, increased to the maximum value of 1.00 mg/l/h at 12 h (Figure 1A). In addition, strain H7 failed to grow and remove NH₄⁺-N without the addition of extra organic carbon or carbonate as the sole carbon source (data not shown). These results suggested that strain H7 removed NH₄⁺-N by heterotrophic nitrification, and its NH₄⁺-N removal efficiency was higher than that of previous heterotrophic nitrification strains, such as Bacillus sp. LY (0.43 mg/l/h) (Zhao et al., 2010), Pseudomonas sp. ADN-42 (1.38 mg/l/h) (Jin et al., 2015), Pannonibacter phragmitetus B1 (1.16 mg/l/h) (Bai et al., 2019) and Pseudomonas tolaasii Y-11 (2.04 mg/l/h) (He et al., 2016). Strain H7 could use NO₃⁻-N and NO₂⁻-N as a sole nitrogen source to support its growth and remove them (Figures 1B,C). The OD₆₀₀ of strain H7 reached a maximum value of 0.431 ± 0.012 when NO₃⁻-N was as a sole nitrogen source (Figure 1B). The removal efficiencies of NO₃⁻-N and TN by strain H7 were 99.82% with a maximum removal rate of 1.53 mg/l/h, and 97.71% with a maximum removal rate of 2.16 mg/l/h, respectively (Supplementary Table S3). Meanwhile, the concentration of NO₂⁻-N increased to the maximum value of 5.60 mg/l/h at 12 h with a gradual decrease in NO₃⁻-N and then decreased to zero at 16 h. These results suggested that strain H7 removed NO₃⁻-N by aerobic denitrification and that its NO₃⁻-N removal efficiency was also higher than that of some NO₃⁻-N removal strains, such as Pseudomonas putida P1 (0.68 mg/l/h) (Xiang et al., 2006), P. phragmitetus B1 (0.81 mg/l/h) (Bai et al., 2019) and Rhodococcus sp. CPZ24 (0.93 mg/l/h) (Chen et al., 2012). The OD₆₀₀ of strain H7 reached a maximum value of 0.369 ± 0.017 when NO₂⁻-N was used as a sole nitrogen source (Figure 1C). The removal efficiency of NO₂⁻-N was 100% with a maximum removal rate of 1.95 mg/l/h (Supplementary Table S4), which is higher than that of strains Pseudomonas sp. yy7 (0.76 mg/l/h) (Wan et al., 2011), P. phragmitetus B1 (0.77 mg/l/h) (Bai et al., 2019) and Acinetobacter sp. T1 (1.69 mg/l/h) (Yang et al., 2019). The maximum removal rate of NH₄⁺-N, NO₃⁻-N and NO₂⁻-N of strain H7 and above bacteria was shown in Supplementary Table S5. Meanwhile, the removal efficiency of TN was 97.26% with a maximum removal rate of 1.92 mg/l/h. Taken together, these results suggested that strain H7 removes NO₃⁻-N and NO₂⁻-N by denitrification.

The As(III) oxidation ability of strain H7 was then investigated. Strain H7 grew well in BM medium, and its OD₆₀₀ reached 2.0 (Figure 1D). Interestingly, strain H7 could oxidize 30 mg/l As(III) to As(V) within 4 h, while the OD₆₀₀ was less than 0.1 (Figure 1D). The As(III) oxidation rate of strain H7 was 7.5 mg/l/h, which was higher than that of some As(III)-oxidizers, such as Halomonas sp. HAL1 (0.31 mg/l/h, OD₆₀₀ > 1.0) (Chen et al., 2015), A. tumefaciens GW4 (3.75 mg/l/h, OD₆₀₀ > 0.5) (Wang et al., 2015) and Bosea sp. AS-1 (6.25 mg/l/h, OD₆₀₀ > 0.5) (Lu et al., 2018). Our previous study showed that the OD₆₀₀ of strain H7 was also <0.1 when it completely oxidized 30 mg/l As(III) in 1/10 ST medium (Fan et al., 2019). The H7 strain was able to efficiently oxidize As(III) at low biomass, which was beneficial in treating arsenic-contaminated wastewater.

The above results suggest that strain H7 was a nitrification and denitrification bacterium with As(III) oxidation ability. Therefore, NH₄⁺-N and NO₃⁻-N (or NH₄⁺-N and NO₂⁻-N) were used as mixed nitrogen sources to investigate the simultaneous nitrogen removal and As(III) oxidation capacity of strain H7. As shown in Figure 2A, the removal efficiency of NH₄⁺-N was 99.67% with a maximum removal rate of 1.43 mg/l/h, and the efficiency for NO₃⁻-N was 100% with a maximum removal rate of 0.88 mg/l/h (Supplementary Table S6). In this process, no hydroxylamine or NO₂⁻-N was detected. Furthermore, the removal efficiency of NH₄⁺-N or NO₂⁻-N was 99.53% or 100% with maximum removal rates of 1.62 mg/l/h and 2.33 mg/l/h, respectively (Figure 2B; Supplementary Table S7). Notably, the removal rates of TN in these two experiments reached over 97.90%. Moreover, strain H7 could completely oxidize As(III) to As(V) in 4 h under the above conditions (Figures 2C,D). Taken together, these findings suggest that strain H7 was an HNAD bacterium capable of removing nitrogen and oxidizing As(III) simultaneously.

For the nitrogen removal assays, most of the NH₄⁺-N-type nitrogen was removed in the first 8 h, while NO₃⁻-N and NO₂⁻-N were removed from 8 h to 16 h (Figures 2A,B). These results reveal that strain H7 may be removing NH₄⁺-N in a mixed nitrogen source before NO₃⁻-N and NO₂⁻-N, which might be due to the higher enzyme activity of NH₄⁺-N oxidization than that of NO₃⁻-N reduction and NO₂⁻-N reduction (Shi et al., 2013). In addition, strain H7 removed 11.32 mg/l NH₄⁺-N from the mixed nitrogen source within 12 h, while it took only 12 h to remove 22.63 mg/l NH₄⁺-N from the sole nitrogen source (Figures 1A, 2A,B), indicating that additional NO₃⁻-N or NO₂⁻-N may inhibit the removal of NH₄⁺-N by strain H7. These results were consistent with the conclusions from strains P. putida ZN1 (Zhang N. et al., 2019) and Paracoccus versutus LYM (Shi et al., 2013).

It is well known that NO₂⁻-N is toxic to humans (Bednarek et al., 2014). Several denitrified bacteria, such as P. putida ZN1 (71.57%) (Zhang N. et al., 2019), P. phragmitetus B1 (98.73%) (Bai et al., 2019), and Acinetobacter sp. T1 (57%) (Yang et al., 2019) could not remove NO₂⁻-N completely. In addition, some denitrified bacteria, such as Enterobacter sp. Z1 and Klebsiella sp. Z2, had difficulty completely removing NO₂⁻-N generated by the reduction of NO₃⁻-N during denitrification (Zhang Y. et al., 2019). However, strain H7 could completely remove NO₂⁻-N in either the sole nitrogen source or mixed nitrogen sources. Therefore, strain H7 has certain advantages in the removal of NO₂⁻-N.

Wastewater O1 was chosen to analyze the ability of strain H7 to remove nitrogen and oxidize As(III) simultaneously in industrial wastewater. The main characteristics of wastewater O1 are listed as follows: pH 7.82, 20.21 ± 0.21 mg/l NH₄⁺-N, 12.38 ± 0.33 mg/l NO₃⁻-N, 35.6 ± 0.55 mg/l TN, and 0 mg/l NO₂⁻-N. As shown in Figure 4, 98.11% of NH₄⁺-N, 98.12% of NO₃⁻-N, and 95.87% of TN were removed, and As(III) was completely oxidized to As(V) with the addition of strain H7 after 16 h (Supplementary Table S8). Fe³⁺ was added and led to the full removal of As(V) within 2 h (Figure 4D). The concentrations of NH₄⁺-N, NO₃⁻-N, TN, and As remained at 0.38 mg/l, 0.26 mg/l, 1.46 mg/l, and 0 mg/l, respectively. However, only 45.31% of NH₄⁺-N, 10.06% of NO₃⁻-N, and 31.92% of TN were removed, and As(III) was not oxidized to As(V) in the wastewater O1 control without adding strain H7 (Figure 4E), indicating the chemical oxidation will not be occur by dissolved oxygen. Adding Fe³⁺ removed 93.6% of As(III) in the wastewater O1 control, but the residual As(III) concentration (0.32 mg/l) did not reach the integrated wastewater discharge standard (GB3838-2002). These results indicate that strain H7 has great application prospects for remediating nitrogen and arsenic in co-contaminated wastewater.

The ability to remove nitrogen is one of the keys to remediating multiple nitrogen and arsenic contaminations. To date, several HNAD bacteria have been applied to real wastewater. For example, the removal efficiencies of NH₄⁺-N, NO₃⁻-N, and TN by Pseudomonas mendocina TJPU04 in industrial wastewater were 91% (127 mg/l), 52% (64 mg/l), and 75% (190 mg/l), respectively (He et al., 2018) and Klebsiella sp. Z2 removed 95.14% (980 mg/l) of NH₄⁺-N and 93.37% (1,000 mg/l) of TN (Zhang Y. et al., 2019). Relatively low concentrations of nitrogen remained even in these strains with a high denitrification efficiency, thus failing to meet the Surface Water Environmental Quality Standard of China (GB3838-2002). In this study, the wastewater O1 collected from a pig farm that had been treated by the MBBR reaction belonged to this lower nitrogen concentration type. Thus, strain H7 has great application potential to remediate wastewater with low nitrogen concentrations.

To further identify the ability of strain H7 to remediate nitrogen and As(III) co-contamination, five cycles of a batch-to-batch system were performed. As shown in Figure 5, the removal efficiency for NH₄⁺-N, NO₃⁻-N, and TN reached over 96.14, 99.08, and 94.68%, respectively, within 18 h in each cycle. As(III) was completely oxidized to As(V) in each cycle and was removed by the subsequent addition of Fe³⁺. After five cycles, the concentrations of NH₄⁺-N, NO₃⁻-N, TN, and As(III) remaining in wastewater O1 were 0.78 mg/l, 0.11 mg/l, 1.89 mg/l, and 0 mg/l, respectively, which still met the V level of Surface Water Environmental Quality Standard of China (GB3838-2002). These results suggested that strain H7 showed good continuity and effectiveness in the bioremediation of nitrogen and arsenic co-contaminated wastewater.

The genome of strain H7 had been reported in our previous study (Fan et al., 2019). Analyzing the genomic data combined with the Kyoto Encyclopedia of Genes and Genomes, it contained the nitrate reductase gene nasA and nitrite reductase gene nirBD, which participated in the assimilation of nitrate reduction (Figure 6A). This was consistent with the phenotype that strain H7 could use NO₃⁻-N or NO₂⁻-N as a sole nitrogen source to grow. In addition, a complete denitrification pathway was found, including the aerobic napA and anaerobic nitrate reductase gene narGHI, nitrite reductase gene nirS, nitric oxide reductase gene norBC, and nitrous oxide reductase gene nosZ (Figure 6A), which was consistent with the phenotype that strain H7 could remove NO₃⁻-N and NO₂⁻-N. Therefore, the NO₃⁻-N removal pathway was speculated, as shown in Figure 6B. In addition, an As(III) oxidation island was also found in the strain H7 genome (Figure 6A), and aioA was upregulated 2.2-fold by qRT-PCR (Supplementary Figure S2), which was consistent with the phenotype that strain H7 was able to oxidize As(III). The detailed gene information is shown in Supplementary Table S9.

To explore the NH₄⁺-N removal pathway of strain H7, iTRAQ was performed. NH₄⁺-N vs. control was designed. A total of 568 proteins showed differential expression; 66 proteins were upregulated, and 182 proteins were downregulated. Detailed information regarding the differentially expressed proteins related to nitrogen metabolism is shown in Supplementary Table S10.

Analysis of the proteins related to nitrogen removal showed that NirS, NorB, NorC, and NosZ were upregulated 3.1-, 5.3-, 2.5-, and 1.7-fold, respectively. However, the NO₃⁻-N reductases NapA, NarG, NarH, and NarI were downregulated 1.9-, 2.0-, 1.5-, and 1.7-fold, respectively. Usually, the NH₄⁺-N removal pathway by HNAD bacteria is as follows: (1) NH₄⁺-N was oxidized to NH₂OH by ammonia monooxygenase AMO; (2) NH₂OH was oxidized to NO₂⁻-N by hydroxylamine oxidase HAO; (3) NO₂⁻-N is NO₃⁻-N by NO₂⁻-N oxidase; (4) NO₃⁻-N was reduced to NO₂⁻-N by NapA or NarGHI; (5) NO₂⁻-N was reduced to NO by NirS; (6) NO was reduced to N₂O by NorBC; (7) N₂O was reduced to N₂ by NosZ (Yang et al., 2019). Nitrification occurred from the first step to the third step, and denitrification occurred from the fourth step to the seventh step. NH₄⁺-N removal could occur via the nitrate pathway (NH₄⁺ → NO₃⁻ → NO₂⁻ → NO→N₂O → N₂) or nitrite pathway (NH₄⁺ → NO₂⁻ → NO→N₂O → N₂) (Yang et al., 2019). In the NH₄⁺-N removal process of strain H7, no NO₂⁻-N or NO₃⁻-N accumulated. The nitrite-reducing NirS, nitric oxide reductase NorBC, and nitrous oxide reductase NosZ were all upregulated, while the nitrate reductases NapA and NarGHI were downregulated. Thus, the NH₄⁺-N removal pathway of strain H7 occurred via the nitrite pathway, which is called shortcut nitrification–denitrification (Figure 6B). Compared with the nitrate pathway, it was reported that shortcut nitrification–denitrification can reduce over 40% of the carbon source addition and 25% of the oxygen supply and greatly improve the denitrification rate (Mavinic and Turk, 1987; Kornaros et al., 2010).