The sludge used to isolate the denitrifying bacteria was taken from an aeration tank in the LIW treatment plant located in Busan, Korea. The sludge sample was agitated to obtain homogeneous suspension in sterile 0.2% NaCl. One mL of the suspended liquid was pipetted into a 10-mL tube that contained ‘PYK medium’: 5 g peptone, 3 g yeast extract and 2 g KNO₃ in 1 L tap water (pH 7.0). Another 1 mL of the suspended liquid was also pipetted into a 10-mL tube that contained ‘Succinate medium’: 4 g succinate, 1 g (NH₄)₂SO₄, 0.3 g KH₂PO₄, 0.1 g K₂HPO₄, 0.2 g Mg SO₄·7H₂O and 2 g KNO₃ in 1 L tap water (pH 7.0). After 3 days incubation at 30°C and 180 rpm, the liquid culture was spread with a platinum loop onto the same media solidified with 1.5% nutrient agar. The separated colonies formed on the agar plates were picked up serially, and a purified isolate was obtained by repeated streaking onto fresh agar plates. Each isolate was maintained on the agar plate at 4°C, and transferred to a fresh agar plate every 2 weeks until use.

The ability of aerobic denitrification for each isolate was tested by the use of the syringe technique under an aerobic condition [21]. First, 1 mL of each isolate was inoculated into a 50-mL glass syringe containing a 20-mL sterile PYK medium. At the beginning, 10 mL of pure O₂ was supplied using a Hamilton gastight syringe, after which the syringe reactor was incubated at 30°C and 180 rpm for 72 hr. Oxygen was supplied into the syringe reactor twice more before depleted. To measure the moles of gas produced by denitrification, 20 μL of gas sample was taken periodically using a 50 μL gas-tight syringe and analyzed by gas chromatography (GC). At the same time, a syringe reactor under an anaerobic condition was also tested in parallel. The number of moles of gas produced in the syringe reactor was calculated by the ideal gas law.

The isolate that produced the largest amount of N₂ gas in the syringe test was primarily identified on the basis of colony morphology, Gram reaction, and microscopic examination.

The specific identification of the isolate was performed using 16S-rDNA sequence analysis. Chromosomal DNA of the isolate was extracted from cells grown in the given medium with AccuPrep^® Genomic DNA extraction kit (Bioneer, Korea) as the manufacturer’s instructions. PCR amplification of the DNA using the 27F (5″-AGAGTTTGATCCTGGCTCAG-3″) and 1492R (5″-GGTTACCTTGTTACGACTT-3″) primers was performed with a DICE model TP600 PCR thermal cycler (Takara, Japan) as described by Kim et al. [21], and the 16S rDNA genes were determined by the Macrogen Company (Korea). These sequences were compared with GenBank (National Center for Biotechnology Information, USA) entries using the Advanced BLAST similarity search option [22], which is accessible from the homepage of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/). BioEdit Sequence Alignment Editor version 5.0.9 was used to verify the alignment and remove all positions with gaps before calculating distances with DNAdist programme in PHYLIP (version 3.5c). Phylogenetic tree was gained on the ground of the given sequences and their close relatives using the neighbor joining method with 1,000 bootstrap replicates.

The characteristics of aerobic denitrification by the isolates were investigated in 1-L five-neck flasks. The flask was comprised of five necks, and at each neck, a thermometric sensor, a pH probe, tubing for sampling, oxygen-inlet tubing connected to a membrane filter (0.2-μm pore size), and oxygen-outlet tubing were installed. To avoid contamination from foreign microorganisms, two consecutively connected 1-L flasks containing 10 N NaOH were placed for the discharging gas. This flask set-up was placed in a hot stirring bath system (Eyela, Japan) and maintained at 30 ± 0.2°C. Stirring the medium inside the flask was accomplished with the Variomag Telesystem (H+P Labortechnik AG, Germany) at 500 rpm. Oxygen (1.5 kg_f/cm²) was supplied continuously into the flask from an oxygen tank (80% purity), and 10-fold diluted Antifoam 204 was used when the foam occurred severely.

For the preparation of inoculums, each isolate was separately cultivated in a sterile PYK medium until a late-log growth phase. Then, equal amounts (5 mL) harvested from each of eight isolates were combined together and inoculated to the flask containing the PYK medium under an aseptic condition. To examine the effect of one potent isolate on the aerobic denitrification, a flask inoculated with the combined cells at the ratio of 10:1 (one potent isolate to other seven isolates) was cultivated in parallel. During the experiments, the culture broth was sampled periodically from the flask by a peristaltic pump using Tygon tubing, and the changes in the reaction parameters were measured. From these measurements, the percentage of nitrogen converted to N₂ gas by the isolated aerobic denitrifiers was calculated as the following formula:

where the concentration of final TN was the total concentrations of NH₄⁺–N, NO₂⁻–N, NO₃⁻ –N, organic N and N in biomass at the final stage of aerobic denitrification. The biomass composition was assumed to be C₅H₇O₂N.

To investigate the potential of the isolated aerobic denitrifiers in plant-scale treatment, they were applied to a LIW treatment plant (Busan, Korea) in which effluent quality was difficult to meet the N regulatory standard. The process flow diagram of the LIW treatment plant is shown in Fig. 1. The LIW treatment plant was comprised of collection tank (180 m³), equalization tank (10,472 m³), four relay tanks (758 m³), three clarifiers (6,666 m³), fermentation and synthesis tank (1,769 m³), eight-stage aeration tank (18,572 m³), two sludge thickeners (2,773 m³), two bio-contact tanks for treatment of nitrogen by microalgae (1,144 m³), silica tank for coagulation of colloids (450 m³), carbon tank for adsorption of organic compounds (336 m³), mineral tank for deterioration of organic compounds (453 m³) and industrial water tank (593 m³). The maximal capacity of this plant for LIW treatment was 9,000 m³ per day. The existing treatment for this LIW had been conducted by the use of microbial consortium (Korean patent: 10-1231977) consisting of Bacillus, Paenibacillus, Leuconostoc, Kurthia, Sphingobacteirum, etc. The microbial consortium was seeded daily into equalization tank (0.2%, v/v), fermentation and synthesis tank (0.04%, v/v), aeration tank (0.02%, v/v) and two sludge thickeners (each 0.04%, v/v), respectively. This treatment was set as a control experiment in this study.

Against the control, the experiments using additional input of isolated strains were performed for better N removal of this LIW to meet the N regulatory standard. For the preparation of seed culture, each isolated strain was actively cultivated at 30°C and 180 rpm in a 1-L flask containing the PYK medium, and then cells were collected after centrifugation. The collected cells were combined at a 10:1 ratio of one potent isolate to other seven isolates, based on the result of lab-scale experiment. After 1 day acclimation in the PYK medium at 30°C and 180 rpm under an aerobic condition, the actively grown cells were transferred to a 150 L reactor under an aseptic condition. With addition of 50-mg lactose/L, the seed culture for the plant-scale experiment was further cultivated in a 3-ton tank for 3 days under an aerobic condition. Finally, the proliferated cells prepared in this way were seeded into the aeration tank in two different ways: 1) the proliferated cells of 12 m³ were seeded at the first day and then the cells of 3.6 m³ were seeded daily from the second day to the end (designated as ‘treatment 1’), and 2) The cells were seeded as the above strategy and moreover 0.6-m³ lactose as a carbon source was additionally supplied every 3 days (designated as ‘treatment 2’). The aeration tank was divided into eight stages, and the hydraulic residence time was approximately 42.2 hr (with maximal flow rate of 440 m³/hr). Samples were periodically taken from the aeration tank, and the changes of the reaction parameters were measured according to standard methods for the water pollution. Experiments were carried out for eight weeks.

The dry-cell weight (DCW) was determined by weighing the cell pellet after being dried in an oven at 105°C for 12 hr. The cell pellet was prepared by centrifuging a 5 mL sample of broth culture at 5,000 rpm for 10 min and then by decanting the supernatant after washing twice with distilled water. To measure the nitrogen and carbon dioxide gases produced by aerobic denitrifiers in the syringe experiments, 20 μL samples (injection volume) were taken for GC/TCD (Perkin Elmer Instruments) analysis. The columns used were a ‘molecular sieve 13X’ and ‘carboxen 1000’ for nitrogen and carbon dioxide, respectively. In the analyses of both gases, the following conditions were equally applied: the carrier gas was helium at a flow rate of 30 mL/min while the injector and detector temperatures were 100 and 200°C, respectively. However, the oven temperature for nitrogen gas was 40°C, whereas that for carbon dioxide gas was initially 40°C for 3 min and then increased to 170°C at a rate of 30°C/min.

In the aerobic denitrification experiments, the following parameters were analyzed: 5-day biological oxygen demand (BOD₅), chemical oxygen demand-dichromate (COD_Cr), total phosphorus (TP), suspended solids (SS), and total nitrogen (TN), total Kjeldahl nitrogen (TKN), ammonium, nitrite and nitrate. These analyses were carried out according to standard methods for the water pollution [23]. The concentration of dissolved oxygen (DO) and the value of oxidation-reduction potential (ORP) were monitored with an YSI DO probe (Model 58; USA) and Isteck ORP probe (Model 730P; Korea), respectively.

By repeated streaking on agar plates, eight strains were purified. The eight isolates were given the names KH1 to KH8, and their characteristics of colonies and morphology were tabulated in Table 1. Each isolate exhibited its own characteristics of colony and morphology distinctively. Among these strains, the strain KH8 exhibited the most prominent ability of N₂ gas production in the syringe experiment. For the efficient treatment of nitrous concentration contained in LIW, N₂ gas as an end metabolite must be produced from the nitrous compounds at a higher conversion rate. Hence, the denitrification ability of the strain KH8 under an aerobic condition was further examined against that under an anaerobic condition. As shown in Table 2, 72.1-μmole N₂ gas was produced anaerobically for 72 hr of cultivation with 28.3-μmole CO₂ gas. Under the aerobic condition, N₂ production was not active at initial stage. After then, however, N₂ was steadily produced, and 64.2-μ mole was produced after 72 hr with a much higher amount of 411.9-μmole CO₂. This indicates that the strain KH8 needed some time to adapt to an aerobic condition but its denitrification ability under an aerobic condition was almost comparable to the anaerobic denitrification ability. For this reason, the strain KH8 was identified and characterized in later experiments.

The species-specific identification of the isolate was performed using 16S rDNA sequence analysis. The 1,418 bp-sized fragment of the 16S rDNA gene of the isolate was amplified and sequenced. Homology searches revealed that the isolate KH8 was closely related to Pseudomonas aeruginosa R12 (DQ073454; 100% similarity) and P. aeruginosa LCD12 (FJ194519; 99% similarity). The phylogenetic tree based on the partial 16S rDNA gene showed the relation of the isolate KH8 and other related strains (Fig. 2). From these distinctive features, the isolate KH8 was designated as P. aeruginosa R12.

The changes in reaction parameters were examined during the aerobic denitrification using different compositions of isolated strains as seed culture. The characteristics of aerobic denitrification starting with equal amounts of isolated strains under an aerobic condition are shown in Fig. 3(a). The initial DO was measured to be 5.8 mg/L under the continuous supply of oxygen. As the aerobic denitrification started, DO levels decreased to 0.3 mg/L within 2 hr, then fluctuated, and maintained over 0.9 mg/L after 32 hr. This implies that the supplied oxygen did not match the amount of DO consumed by isolates during the active reaction, probably due to the low solubility of oxygen [24]. As DO levels dropped, ORP decreased from 56.7 to −44.0 mV within 1 hr, decreased further to −148.4 mV after 4 hr, then increased and maintained at positive values after 10.5 hr. This indicates that the decrease in ORP value was related to the decrease in DO. The pH was 7.24 at the beginning, and decreased to 7.09 within 1 hr. After this, the pH increased to 8.01 until 5 hr due to alkalinity production by denitrification [25]. After 5 hr, the pH decreased steadily to a final value of 6.20. During the experiment for 48 hr, the concentration of COD_Cr reduced from 10,695 to 1,632 mg/L, the concentration of TN reduced from 2,355 to 1,767 mg/L, and the concentration of TKN reduced from 1,333 to 588 mg/L. From these data, the removal efficiencies of COD_Cr, TN and TKN were estimated to be 84.7%, 25.0% and 55.9%, respectively. The resulting COD_Cr/TN ratio decreased from 8.6 (at the beginning) to 2.0 (at the end). It is known that the C/N ratio may influence the metabolic pathway of organic matter utilization [26]. Accordingly, the decrease in the C/N ratio in a later stage of aerobic denitrification implies that external carbon may be necessary for high efficiency of N removal. Along with these reaction parameters, profiles of NH₄⁺–N, NO₂⁻–N and NO₃⁻–N concentrations are also shown in Fig. 3(a). The concentration of NH₄⁺–N increased from 73.7 to 354.0 mg/L for first 12 hr, and then decreased slowly to the end. On the other hand, the concentration of NO₃⁻–N decreased from 270.7 to 104.6 mg/L for first 12 hr, and then increased slowly to the end. The concentration of NO₂⁻–N as an intermediate metabolite in aerobic denitrification maintained low for first 12 hr, then increased to 202.7 mg/L after 24 hr, but decreased to almost zero at the end. From the analyses of all parameters, it was found that active aerobic denitrification took place in the first 12 hr when equal amounts of eight isolates were used as inoculums.

At a cell-combination ratio of 10:1 (the isolate KH8 to the other seven isolates), the characteristics of the aerobic denitrification are shown in Fig. 3(b). From the beginning of the experiment, DO levels maintained below 1 mg/L for the first 5.5 hr even under the continuous supply of oxygen. After then, the DO levels increased and maintained over 2.5 mg/L until the end. Due to active denitrification, the ORP values decreased as DO levels dropped. The ORP decreased to −182 mV after 3 hr, then increased steadily, and maintained at positive values after 11 hr. This was followed by a slight increase to 113.3 mV after 48 hr. The initial pH was 7.24, and decreased slightly to 7.20 within 1 hr. After then, the pH increased steadily to 8.34 until 24 hr. This duration of increase in pH was much longer than that exhibited during the experiment using equal amounts of isolated strains (shown in Fig. 3(a)). This indicates that higher alkalinity production resulted from stronger denitrification by the dominant action of the isolate KH8. After 34 hr, the pH somewhat decreased to a final value of 7.75. In particular, severe foams were generated between 1 hr and 6 hr. As a result, the changes in DO, ORP and pH appeared mostly in the first 12 hr. During the 48-hr reaction, the concentration of COD_Cr reduced from 7,381 to 1,099 mg/L, the concentration of TN reduced from 3,513 to 1,464 mg/L, and the concentration of TKN reduced 1,471 to 728 mg/L, respectively. From these data, the removal efficiencies of COD_Cr, TN and TKN were estimated to be 85.1%, 58.3% and 50.5%, respectively. Although there was no significant difference in the COD_Cr removal, the TN removal efficiency doubled in comparison with that (25.0%) obtained from the experiment using equal amounts of isolated strains (shown in Fig. 3(a)). Thus, this result indicates that the increase in the composition of isolate KH8 in the seeded cells resulted in improvement of N removal efficiency. The initial ratio of COD_Cr/TN was 5.5, but it decreased to 3.3 after 12 hr. After then, the C/N ratio continued to decrease, and was calculated to be 2.2 at the end. This result indicates that the aerobic denitrification efficiency was influenced by C/N ratio [27]. Along with these reaction parameters, changes in NH₄⁺–N, NO₂⁻–N and NO₃⁻–N concentrations were also examined. The concentration of NH₄⁺–N increased from 101.1 to 524.0 mg/L for first 12 hr, and then decreased to 144.9 mg/L at the end. On the other hand, the concentrations of NO₃⁻–N and NO₂⁻–N decreased from 202.3 to 18.0 mg/L and from 24.5 to 0 mg/L for first 12 hr, respectively. After then those concentrations increased slowly to 82.9 and 477.0 mg/L at the end, respectively. Therefore, the final concentration of NO₃⁻–N was almost nine times lower, compared with the result shown in Fig. 3(a), indicating that aerobic denitrification was adequately taken place by the dominant isolate KH8.

To examine the characteristics of denitrifying isolates under an aerobic condition, nitrogen balances for their aerobic denitrifications were investigated at different compositions of seeded cells (Table 3). At the cell-combination ratio of 1:7, N loss was estimated to be approximately 32.8% during the 12-hr aerobic denitrification, with the cell concentration of 9 mg/L. This result was presumed that 32.8% of the initial N in the culture medium was converted to N₂ gas by the denitrifying activities of isolated strains under an aerobic condition. However, this aerobic denitrification slowed down during the next 36 hr. The N loss taking place in the aerobic denitrification was somewhat improved when the composition of isolate KH8 was increased within the seeded cells (at the cell-combination ratio of 10:1). The percentage of N loss was 42.5 during the first 12 hr, and it was 58.4 during an entire period of this experiment. From these results, the potential of isolate KH8 was found to be obvious in aerobic denitrification. The similar results were found in the previous studies using potential aerobic denitrifiers [10, 13, 20].

To examine the potential ability of the isolates for better N removal, plant-scale aerobic denitrification using the isolates was performed in the aeration tank under three different types of treatments (control, treatment 1 and treatment 2). Fig. 4 shows the concentration profiles in the influent and effluent during the entire course of the experimental study. The LIW flowed into this treatment plant had the following characters: pH of 8.5, DO of 1.0 mg/L, COD_Cr of 3,000 mg/L, BOD₅ of 2,500 mg/L, TN of 450 mg/L, NH₄⁺–N of 400 mg/L, TP of 50 mg/L, SS of 3,000 mg/L, and undetectable NO₂⁻–N and NO₃⁻–N. After the LIW passed through equalization tank, it flowed into fermentation and synthesis tank, and then the eight-stage aeration tank, as shown in Fig. 1. The influent and effluent water quality of the aeration tank under each treatment was tabulated in Table 4. From the control experiment, the influent and effluent concentrations of DO, BOD₅, TP and SS were found to be 1.0 and 2.4 mg/L (DO), 1,248 and 71 mg/L (BOD₅), 33 and 20 (TP), and 5,174 and 2,976 mg/L (SS), respectively. The increase in SS concentration was due to the return sludge, and this outcoming SS settled down in tertiary clarifier. When the method of treatment 1 (additional input of isolates, as described in the section of 2.4) or treatment 2 (additional input of isolates and C-source, as described in the section of 2.4) was applied, DO levels increased slightly and the removal efficiency of BOD₅ also increased slightly. However, the removal efficiencies of TP and SS by these treatments were least different. During the experimental period, the variances in pH, COD_Cr and TN were obvious among the different types of treatments. The best removal efficiencies of COD_Cr (97.0%) and TN (89.8%) were achieved by the treatment 2. This TN removal efficiency was much higher than that (51%) obtained from the control experiment. Along with higher TN reduction, a higher pH value (7.34) revealed in the effluent at day 56 under the treatment 2, because more active denitrification occurred under an aerobic condition. In all the treatments, NO₂⁻–N was not detected in the effluent at day 56, implying that it converted to N₂ by denitrification. The lowest concentrations of NH₄⁺–N (20mg/L) and NO₃⁻–N (8.9 mg/L) in the effluent at day 56 were achieved by the treatment 2, in proportion to the TN removal efficiency.

To understand the characteristics of the isolated aerobic denitrifiers in the aeration tank during the plant-scale operation, the major parameters were monitored. The profiles of pH, COD_Cr and TN values under three different types of treatments are shown in Fig. 4, and those of NH₄⁺–N, NO₂⁻–N and NO₃⁻–N are shown in Fig. 5. As shown in Fig. 4(a) (in case of control experiment), the pH was 7.20 at the beginning and it ended at 7.03 without significant variance. This result indicates that poor denitrification resulted in pH decrease at the end [28]. The influent concentration of COD_Cr fluctuated because it was dependent upon daily LIW content. The influent TN was in a range of 290–300 mg/L, and was removed steadily. As it turned out, the removal efficiencies of COD_Cr and TN were 92.8% and 51%, respectively. The influent C/N ratios were in a range of 3.7–4.0. Hence, the composition of carbons and nitrogens in the influent was not significantly fluctuated. These removal efficiencies were improved when the isolated aerobic denitrifiers were additionally seeded into the aeration tank (in case of treatment 1, as shown in Fig. 4(b)). Those efficiencies were estimated to be 95.6% and 84.2%, respectively. The intense reductions in COD_Cr and TN started after 14 day, with the increase of pH. The influent pH at 7.18 decreased to 6.78 after 14 day due to strong nitrification. After then, the pH increased steadily to 7.19 due to the alkalinity production by denitrification. These results indicate that the isolated aerobic denitrifiers needed at least 14 day to be established in the aeration tank. The influence of the isolated aerobic denitrifiers on denitrification in the aeration tank was more obvious when these cells were seeded with lactose as carbon source (in case of the treatment 2, as shown in Fig. 4(c)). With addition of a carbon source, the influent C/N ratios (3.9 to 4.2) increased slightly. In this treatment, the pH decreased during the first 14 day, and then it increased by the strong denitrification. The effluent pH revealed 7.34, which was higher by 0.3 than that (7.03) obtained from the control experiment. The similar results could be found in the previous report of heterotrophic nitrification and aerobic denitrification for treatment of high-strength ammonium [29]. By this improved denitrification, the removal efficiencies of COD_Cr (97.0%) and TN (89.8%) increased further.

As shown in Fig. 5(a), the concentrations of NH₄⁺–N, NO₂⁻–N and NO₃⁻–N were changed in the control experiment. The influent NH₄⁺–N concentration in a range of 240–255 mg/L converted to NO₂⁻–N and NO₃⁻–N by nitrification, and final concentration of effluent NH₄⁺–N was 40 mg/L. Although the influent did not contain NO₂⁻–N, its concentration in the effluent at day 0 was 10 mg/L due to the effect of nitrification. However, NO₂⁻–N was not detected after 40 day. The NO₃⁻–N was detected first in the fermentation and synthesis tank. As a result, the initial concentration of NO₃⁻–N flowing into the aeration tank was 53 mg/L. The initial NO₃⁻–N of 50 mg/L in the effluent increased rapidly to 165 mg/L during the first 14 day, but it did not decrease adequately after 56 day, implying that active denitrification did not occur to offset the nitrification. This reduction of NO₃⁻–N after 14 day was obvious in the result of treatment 1 (Fig. 5(b)). The effluent NO₃⁻–N of 165 mg/L at 14 day decreased steadily to 20.1 mg/L until 56 day, implying that active denitrification occurred in the aeration tank. Along with NO₃⁻–N, the concentration in the effluent NH₄⁺–N decreased from 251 to 35 mg/L in parallel. Similarly, the effluent NO₂⁻–N also decreased, and was not detected after 45 day. This trend was more obvious in the result of the treatment 2 (Fig. 5(c)). The concentration of effluent NO₃⁻–N increased from 49 to 184 mg/L during the first 14 day, and then decreased rapidly to 8.9 mg/L after 56 day. At the same time the effluent NH₄⁺–N decreased from 247 to 20 mg/L for 56 day. The effluent NO₂⁻–N of 9.6 mg/L at the beginning decreased slightly, and was not detected after 47 day. Therefore, more active denitrification by the treatment 2 resulted in the best N removal efficiency among the three different types of treatments. Besides, variances in the values of parameters were not significant after 42 day operation. This result implies that the seeded aerobic denitrifiers for the removal of excess N existing in LIW were stabilized in the aeration tank for six weeks. It is important that operation under a stabilized system is a key to reliable N treatment [30]. Suneethi and Joseph [31] reported that a large-scale anaerobic ammonium oxidation (ANAMMOX) process was stabilized in 130 day after ANAMMOX bacteria were seeded. Compared with this result, the stabilization of the seeded isolates in this study was completed in a shorter time although the LIW contained refractory organic N compounds to some extent [32]. Under the treatment 2, the final water quality in the effluent leaving the treatment plant was as follows: pH 7.3, DO of 3.0 mg/L, COD_Cr of 30 mg/L, BOD₅ of 24 mg/L, TN of 18 mg/L, NH₄⁺–N of 12 mg/L, NO₂⁻–N of 0 mg/L, NO₃⁻–N of 6 mg/L, TP of 5 mg/L, and SS of 15 mg/L, all of which were below the water-quality standard concentrations. As a result, efficient LIW treatment was achieved by additional input of isolated aerobic denitrifiers with a carbon source. Therefore, the isolated aerobic denitrifiers could be an attractive candidate for application to wastewater treatment plants in which the N removal is difficult after nitrification due to poor denitrification or advanced treatment is required for N under the existing aeration-tank capacity.