The conventional ammonium nitrification pathway is coordinated by two types of bacteria: (i) ammonia-oxidizing bacteria, which are responsible for oxidizing ammonium to nitrite; and (ii) nitrite-oxidizing bacteria, which are responsible for oxidizing nitrite to nitrate. Nitrate is finally converted to gaseous nitrogen by heterotrophic denitrifying bacteria via the denitrification pathway under anaerobic conditions (Jaroszynski and Oleszkiewicz, 2011). The conversion efficiency of HNADs is better than that of the conventional process of ammonium removal because there are differences in intermediate products and key catalytic enzymes.

In the 1990s, a classical coupled model of ammonium nitrification in HNAD bacteria was reported (Wehrfritz et al., 1993). Ammonium is oxidized to hydroxylamine in the periplasm by ammonia monooxygenase (AMO), followed by oxidation to nitrite by hydroxylamine oxidoreductase (HAO), and finally to nitrate in the cytoplasm (Song et al., 2021). In HNADs, nitrate is reduced via denitrification by a series of oxidoreductases under aerobic conditions, and stable isotopes and enzyme inhibitors are used for more-accurate detection of intermediate metabolites in the study of nitrogen conversion pathways in HNADs.

Alcaligenes faecalis NR can convert ammonium to N₂O and N₂ with ammonium as the sole nitrogen source, and intermediate nitrate and nitrite are not detected under aerobic conditions (Zhao et al., 2012). When hydroxylamine is the sole nitrogen source, gaseous nitrogen is still detected, but only hydroxylamine oxidase activity is detected, and nitrate and nitrite reductase activity is undetected (Zhao et al., 2012). Hydroxylamine is converted directly to nitrous oxide replacing nitrite, and a similar finding has been reported for Acinetobacter calcoaceticus HNR (Zhao et al., 2010). It has been speculated that the ammonium-removal pathway of A. calcoaceticus HNR and A. faecalis NR is NH₄ ⁺-N→NH₂OH→N₂O→N₂ (a shortcut ammonium metabolic pathway), and experiments have confirmed that Photobacterium sp. NNA4 can transform hydroxylamine directly to N₂O when hydroxylamine is the sole nitrogen source (Liu et al., 2019).

Enzyme activation inhibitors can block metabolic pathways, resulting in the accumulation of intermediate products or the disappearance of end products, and they are often used to study the metabolic pathways of nitrogen conversion. In Alcaligenes sp. TB, when nitrite reductase inhibitor (pb²⁺) is present, the accumulation of nitrate and nitrite is detected, but gaseous nitrogen is not detected, whereas when nitrate reductase inhibitor (Na₂WO₄) is present, the accumulation of nitrate and gaseous nitrogen is detected simultaneously; these results show that the strain TB carries out ammonium removal via the pathway NH₄ ⁺-N→NH₂OH→NO₂ ⁻-N→N₂O→N₂ (an incomplete ammonium metabolic pathway) (Chen et al., 2016). The production of gaseous nitrogen is found to be accompanied by the interconversion of nitrite and nitrate, which has similarly been reported for Agrobacterium sp. LAD9 (Chen and Ni, 2012). In addition, Alcaligenes sp. TB can convert approximately 35.7% of the ammonium to gaseous nitrogen without enzyme inhibitors via another pathway, that is, NH₄ ⁺-N→NH₂OH→NO₂ ⁻-N→NO₃ ⁻-N→NO₂ ⁻-N→N₂O→N₂ (a complete ammonium metabolic pathway) (Chen et al., 2016). The same nitrogen-removal pathway has been reported for B. subtilis A1 (Yang et al., 2011), K. pneumoniae CF-S9 (Padhi et al., 2013), B. cereus GS-5 (Rout et al., 2017), and Rhodococcus sp. CPZ24 (Chen et al., 2012).

Different nitrogen-source combinations also affect the denitrification pathway of ammonium. When ammonium and nitrite are present simultaneously, the ammonium is transformed directly to gaseous nitrogen via a shortcut ammonium metabolic pathway, but when ammonium and nitrate are present simultaneously, the ammonium is first transformed to nitrate and then reduced to gaseous nitrogen via a complete ammonium metabolic pathway (Wei et al., 2021). Furthermore, compared with the traditional autotrophic nitrification pathway, the hydroxylamine pathway is simpler, which makes it more advantageous to use HNADs in treating high-concentration nitrogenous wastewater (Peng and Zhu, 2006).

The reduction of nitrate to gaseous nitrogen by HNADs is accomplished via four reductases under aerobic conditions in four processes, that is, NO₃ ⁻-N→NO₂ ⁻-N→NO→N₂O→N₂ (Song et al., 2021). Nitrite reductase is also a bottleneck in the denitrification pathway of HNADs because it is as sensitive to oxygen as are traditional anaerobic denitrifying bacteria. The ability of P. stutzeri T13 to convert nitrate efficiently with nitrate as the sole nitrogen source has been investigated, finding that the conversion efficiency of intermediate nitrite metabolites is significantly higher at low levels of dissolved oxygen (DO) than at high levels (Sun et al., 2015). This sensitivity to oxygen inhibits the reduction of nitrite, and it accumulates during denitrification with high DO levels. Some HNAD bacteria have different properties and are effective in converting nitrite to gaseous nitrogen at high DO levels, such as P. stutzeri ZF31, P. mendocina X49, and Vibrio sp. AD2 (Huang et al., 2015; Ren et al., 2021; Xie et al., 2021).

Nitrite is an intermediate product of the denitrification metabolic pathway, and because of its cytotoxicity, few HNADs can use nitrite directly for denitrification conversion. It has been reported that Pseudomonas sp. yy7 can use nitrite via assimilation and denitrification (Wan et al., 2011), and isotope-tracking experiments have shown that P. stutzeri converts nitrous oxide (N₂O)—a precursor of denitrification—to nitrogen gas (N₂) under aerobic conditions (Desloover et al., 2014). The denitrification pathway with nitrite or nitrous oxide as the initial nitrogen source is an incomplete metabolic pathway.

The reduction of nitrate to nitrite is the start of the denitrification reaction and is carried out by NR. Outside the cell is periplasmic nitrate reductase (Nap), and bound to the cell membrane is nitrate reductase (Nar) (Figure 3) (Vivián et al., 1999).

Nar is a Mo-containing enzyme that has three catalytic subunits: NarG, NarH, and NarI. NarG is the largest catalytic subunit, comprising a [4Fe-4S] center (FS0) and a Moco active site. NarH binds to NarG and is located in the cytoplasm, containing three [4Fe-4S] centers (FS1, FS2, and FS3) and one structurally distinct [3Fe-4S] center (FS4); it is mainly responsible for electron transfer. NarI is the smallest subunit; it is anchored to the cell membrane and is responsible for the transmembrane movement of protons (Coelho and Romao, 2015). The expression of Nar has been detected in B. cereus GS-5 under aerobic conditions and was encoded by 561 bp nar (Rout et al., 2017).

Nap is an extracellular enzyme that reduces nitrate in the periplasm under aerobic conditions and plays a critical role in HNADs; it was first purified from Paracoccus pantotrophus (Sears et al., 1995; Yang J. et al., 2020). Nap may be a key factor for nitrogen conversion by HNAD bacteria under aerobic conditions. The first crystal structures of periplasmic nitrate reductases were reported in Desulfovibrio desulfuricans at 1.9-Å resolution (Dias et al., 1999). Nap contains two subunits, that is, NapA and NapB, and the genes encoding these are napA and napB, respectively; napA has been successfully amplified in P. stutzeri XL-2 (Zhao et al., 2018).

Nitrite is converted to nitric oxide catalyzed via NIR, which contains two types of catalase: cd1Nir and CuNir (Figure 3). These two enzymes are nonhomologous isozymes and usually cannot coexist in the same microorganism (Zumft et al., 1994). However, the coexistence of two enzymes has been reported in the same microorganism, such as in Bradyrhizobium oligotrophicum S58 (Sanchez and Minamisawa, 2018) and Rhodothermus marinus (Graf et al., 2014).

Cd1Nir contains two identical subunits with cofactors heme d1 and heme c; it is a homologous dimer protein and is encoded by nirS (Nojiri et al., 2009). In Agrobacterium sp. LAD9, 799 bp of the nirS gene fragment have been amplified under aerobic conditions (Chen and Ni, 2012).

CuNir is a homologous trimeric protein containing three identical subunits with two Cu-binding sites on each (Godden et al., 1991). CuNir is usually encoded by the nirK gene and sometimes by the nirV gene (Yang J. et al., 2020). The nirK gene of nitrite reductases in S. pararoseus Y1 and in Alcaligenes faecalis NR has been amplified (Huang et al., 2018; Zeng et al., 2020).

NOR is embedded in the cell membrane with the catalytic site in the periplasm and is responsible for the reduction of nitric oxide to nitrous oxide (Figure 3). NOR is divided into three types according to the differences in proton and electron transfer centers: cNor, qNor, and Cu_ANor (Hino et al., 2010; Matsumoto et al., 2012).

Two structurally distinct subunits (NorB and NorC) constitute cNor. NorC is a smaller subunit that is on the periplasmic side and has one c-type heme as its electron acceptor. NorB is a larger subunit containing three metal catalytic centers, that is, heme b, heme b ₃ , and nonheme Fe _B . A segment of NOR-encoding gene (2,904 bp norB) that can be translated into a 759 amino acid sequence has been amplified successfully in Alcaligenes faecalis NR (Huang et al., 2018). qNor contains only one subunit without heme, and Cu_ANor has two subunits and has been reported in Bacillus (Suharti and De Vries, 2005).

NOS reduces nitrous oxide (N₂O) to nitrogen (N₂), with its site of action located in the periplasm (Figure 3). Two copper-containing enzymatic catalytic sites, that is, Cu_A and Cu_Z, are located on the same monomer (Cu_A and Cu_Z) (Brown et al., 2000a; Brown et al., 2000b). The intermediate product (N₂O) is a more harmful greenhouse gas than carbon dioxide (CO₂), and emissions of nitrous oxide must be reduced (Lee et al., 2019), but some microorganisms also emit nitrous oxide into the atmosphere as a terminal gaseous nitrogen (Gaimster et al., 2018). Therefore, reducing nitrogen oxide emissions is a very important issue in protecting the atmosphere. NOS is encoded by the nosZ gene and the accessory gene nosRDFYL (Honisch and Zumft, 2003), and the expression of the nosZ gene has been detected in B. californica K1 (Fang et al., 2021).

The conversion of ammonium into hydroxylamine via AMO is the first step in the HNAD nitrification process, and hydroxylamine is an important intermediate metabolite (Hooper et al., 1997). AMO-containing quinone oxidase has two subunits, which are inhibited by light and chelating agents during oxidation, but copper ions can increase the enzyme activity (Moir et al., 1996). The encoding gene of AMO is amoA, which was amplified and 397 bp in S. pararoseus Y1. Strain Y1 can convert ammonia into hydroxylamine under aerobic conditions (Table 2) (Zeng et al., 2020).

Oxidation of hydroxylamine to nitrite by HNADs occurs in the periplasm via HAO. The HAO encoding gene has been expressed successfully in B. cereus GS-5 (Rout et al., 2017). Recent reports suggest that HAO catalyzes hydroxylamine to produce nitric oxide rather than nitrite as found in previous studies, and nitric oxide is re-oxidized to nitrite (Caranto and Lancaster, 2017). In addition, hydroxylamine has been found to be oxidized to N₂O by P460 (Caranto et al., 2016; White and Lehnert, 2016) under anaerobic conditions and HAO enzymes (Zhao et al., 2019) under aerobic conditions in both ammonia oxidizing bacteria and heterotrophic denitrifying bacteria, respectively.

Different carbon sources—such as glucose, acetic acid, succinic acid, citric acid, methanol, and sucrose—have different effects on the growth and nitrogen removal of HNADs, as given in Table 3. The molecular weight and chemical structure of the carbon sources vary greatly and have a significant impact on nitrogen removal, and HNAD bacteria usually prefer carboxylates such as acetate to carbohydrates such as glucose (Jia et al., 2019). The reason for this may be that carboxylate has a simple chemical structure and a low molecular weight, making it more conducive to absorption and utilization through the tricarboxylic acid cycle. When performing nitrogen removal under aerobic conditions, sodium acetate is the optimal carbon source for B. subtilis A1 and V. diabolicus SF16 (Yang et al., 2011; Duan et al., 2015), and sodium pyruvate and succinate are the best recommended carbon sources for Klebsiella sp. TN-10 and B. methylotrophicus L7 (Zhang et al., 2012; Li et al., 2019). However, other strains such as P. aeruginosa P-1, A. contaminans HA, and E. cloacae HNR prefer glucose to carboxylate (Wei et al., 2021; Chen et al., 2015; Guo et al., 2016).

As well as conventional carbon sources, some solid ones have been tried in denitrification applications. As a by-product of tea extraction, tea residue has been used as a solid carbon source to cultivate the fungus F. solani RADF-77 and can be used as a biofilm attachment carrier in denitrification systems (Cheng et al., 2020). Wood-chip leachate has also been used as a carbon source in the nitrogen removal of P. tropicum IS0293 under aerobic conditions (Yao et al., 2020).

The C/N ratio affects the growth metabolism, energy conversion, and denitrification efficiency of HNADs, and it is a parameter that reflects the electron donor and acceptor requirements (Huang and Tseng, 2001; Rajta et al., 2020). A low carbon-source concentration affects not only the growth requirements of the strain but also the uptake and conversion efficiency of the microorganisms regarding the nitrogen source, and the optimal C/N ratio for most HNADs is between 8 and 15 (Table 4).

In Acinetobacter sp. FYF8, it has been reported that the nitrogen-gas conversion ratio is 39.88, 68.85, and 78.79% at a C/N ratio of 2.0, 2.5, and 3.0, respectively (Fan et al., 2021); these results show that the C/N ratio has a significant effect on nitrogen conversion by a given strain under particular incubation conditions. In addition, different nitrogen conversion pathways also require different C/N ratios; for B. methylotrophicus L7, the optimal C/N ratio is six for heterotrophic nitrification but 20 for aerobic denitrification (Zhang et al., 2012).

Within certain ranges of carbon-source concentration and C/N ratio, the higher the C/N ratio, the faster the bacterial growth and the higher the nitrogen-removal efficiency (Patureau et al., 2000). It has been found that the denitrification rate of A. junii YB increases with increasing C/N ratio in the range of 2–15, and nitrogen-balance analysis has shown that more ammonium is converted to biological nitrogen for cell growth at high C/N ratio (Ren et al., 2014). However, although it has been reported that a higher C/N ratio is more favorable for microbial growth and ammonium removal (Chen and Ni 2012), the latter can be inhibited if the C/N ratio is too high (Wang et al., 2016; Zhao et al., 2020). Most HNADs prefer a high C/N ratio, but some, such as Acinetobacter sp. Y16, can have high nitrogen-removal capacity at low C/N ratio (i.e., 2), which is a great advantage when treating wastewater with a low C/N ratio (Huang et al., 2013).

The inorganic ions Cu²⁺, Mg²⁺, Ni²⁺, Pb²⁺, Mn²⁺, Zn²⁺, Fe³⁺, and Cr⁶⁺, along with other heavy metal ions, are ubiquitous in wastewater and are either toxic or potent and affect the nitrogen-removal efficiency of HNADs (He T. et al., 2019). The nitrogen-removal efficiency of Arthrobacter arilaitensis Y-10 is significantly higher—by 10.88%—when compared with the control in the presence of copper at 0.1 mg/L but is strongly inhibited in the presence of copper at 0.25 mg/L (He T. et al., 2019). These results show that a low concentration of copper ions helps in ammonia–nitrogen conversion, but a high concentration reduces the removal efficiency.

Metal ions affect not only the growth of the strain but also its ability to remove nitrogen, and the type and concentration of ions have significantly different effects. When metal-ion addition experiments were conducted with A. aeruginosa HN-02, a low concentration of Cu²⁺ (0.5 mg/L) barely affected its ammonium-removal efficiency, but it decreased by ∼83% at 1.0 mg/L and became almost zero at 1.5 mg/L; compared to Cu²⁺, the ammonium-removal efficiency with the addition of Zn²⁺ (two to eight mg/L) was essentially the same as that of the control, and the toxicity of Cu²⁺ and Zn²⁺ in co-existence was much higher than that of single-addition Cu²⁺ (Chen et al., 2014). However, certain metal ions—such as Fe³⁺—can improve the functionality of the microorganism by stimulating its metabolism. For P. stutzeri T13, it has been reported that the nitrogen-assimilation capacity and maximum nitrate reduction rate increase significantly upon adding Fe³⁺ (Feng et al., 2021).

Compared with other nitrifying bacteria, Cupriavidus sp. S1 shows high nitrogen-removal capacity and excellent resistance to metal ions when various highly concentrated metal ions are added (Sun et al., 2016). In the presence of a high concentration of copper ions, Pseudomonas aeruginosa ZN1 has a copper-resistance gene and protein that allow the strain to grow and remove nitrogen at a copper–ion concentration of 800 mg/L (Zhang N. et al., 2018).

Salinity affects the denitrification capacity and cell growth of HNADs. High osmotic pressure may lead to strain disintegration, resulting in the loss of either cellular or enzymatic activity; therefore, high concentrations of salt inhibit cell growth and nitrogen removal. Previous studies have shown that most HNADs tolerate salinity below 40 g/L NaCl (Table 5), and halotolerant HNADs (Ventosa et al., 1998) may be more advantageous for treating wastewater with high salinity.

Although a low salt concentration is more conducive to cell growth and reproduction, some bacteria isolated from specific environments such as seawater and high-saline effluents have higher nitrogen-removal efficiency with appropriate salinity. In B. litoralis N31 isolated from the marine environment, it has been shown that an appropriate salinity (30–40 g/L) is beneficial for removing ammonium, probably because this is more suitable for enzyme activity (Huang F. et al., 2017). Similar phenomena have also been reported for Marinobacter NNA5 (Liu et al., 2016).

It has been found that pH and salinity also interact in nitrogen removal. In salinity tests, S. marcescens W5 showed a significant difference in ammonium removal between pH seven and pH 10 (Wang et al., 2016). However, studies on the mechanisms that produce this phenomenon are very limited.

Most HNADs are more adapted to neutral-to-alkaline environments (pH 6.0–9.0), and the optimal pH range for nitrogen removal is 7.0–8.0 (Table 6) (Chen et al., 2014; Liu et al., 2016; Hou et al., 2021). Some bacteria such as Acinetobacter sp. JR1 and S. marcescens W5 have a broad spectrum of pH adaptability (pH 4.5–10.0) (Wang et al., 2016; Yang et al., 2019). For S. marcescens W5 and P. stutzeri PCN-1, high ammonium-removal efficiency has been reported under both neutral and alkaline conditions (Zheng et al., 2014; Wang et al., 2016). Few strains can grow at extreme pH values, but strain HN-02 shows excellent acid adaptability and can grow cells and remove nitrogen at a pH value as low as 2.3 (Chen et al., 2014). Fungi are more advantageous than bacteria in converting organic nitrogen sources because of their better acid tolerance and denitrification ability (Cheng et al., 2020). HNADs have a broad spectrum of pH adaptability and can be used in various wastewater treatments.

The best growth state and highest nitrogen-removal efficiency of HNADs are usually found in the optimum pH range, such as for Acinetobacter sp. T1, Klebsiella sp. TN-10, and Marinobacter NNA5 (Liu et al., 2016; Chen et al., 2019; Li et al., 2019). However, the optimal cell-proliferation pH and the highest nitrogen-removal pH of Photobacterium sp. NNA4 are inconsistent, at 5.0–6.0 and 7.0, respectively (Liu et al., 2019).

Temperature affects microbial growth and the catalytic efficiency of enzymatic activity, and the optimal temperature range for HNADs is 30–37°C (Table 7) (Li et al., 2019; Liu et al., 2019; Xia et al., 2020). The higher the temperature in the enzyme-activity temperature range, the higher the enzyme reaction rate; the nitrification efficiency doubles for each 10 °C increase, and the denitrification efficiency doubles for each 4 °C increase (Zaitsev et al., 2008; Wu et al., 2019). The growth of most HNADs is inhibited at low temperature, but at 2 °C, Acinetobacter sp. Y16 can still achieve half of its nitrogen-removal efficiency at the optimal temperature (20 °C), with good low-temperature adaptability (Huang et al., 2013). The optimal temperature for B. simplex H-b is 10 °C, which is lower than that for other HNADs (Yang Q. et al., 2021). The growth trend of microorganisms is usually consistent with nitrogen removal, but at lower temperatures, microbial growth requires a longer lag period. Low-temperature nitrogen-removal HNADs can be used in wastewater treatment in cold areas.

In contrast to denitrifying bacteria, DO affects HNAD microbial colonization and nitrogen-removal efficiency and is a control parameter of the denitrification pathway (Hocaoglu et al., 2011). Conventional anaerobic denitrifying bacteria use nitrate as an electron acceptor to obtain energy, but the aerobic HNADs require the participation of oxygen to act as an electron acceptor for the denitrification process (Yang J. et al., 2020). In nitrogen removal by the denitrification pathway, nitrite reductase is sensitive to DO and can be inhibited by high DO concentration (Ka et al., 1997). It has been found that nitrite removal by P. stutzeri T13 is increased significantly by reducing the DO concentration and adjusting the rotational speed: Low rotational speed with low DO concentration is more useful for removing nitrite (Sun et al., 2015). HNADs have adapted to high-DO conditions for nitrogen removal and have higher efficiency than anaerobic denitrifying bacteria, with the optimal speeds given in Table 8.

The DO level is usually achieved by adjusting the rotation rate, but the DO concentration differs under different experimental conditions. The optimal shaking speed for both Acinetobacter sp. T1 and Photobacterium sp. NNA4 is 160 rpm, but their respective DO values are 5.1 mg/L and 5.89 mg/L (Chen et al., 2019; Liu et al., 2019).

Acinetobacter sp. ND7 requires different levels of DO for ammonium removal via the nitrification pathway and for nitrate removal via the denitrification pathway, with high DO favoring ammonium removal and low DO favoring nitrate removal (Xia et al., 2020).