Ammonium-Nitrogen (NH 4 + -N) Removal from Groundwater by a Dropping Nitriﬁcation Reactor: Characterization of NH 4 + -N Transformation and Bacterial Community in the Reactor

: A dropping nitriﬁcation reactor was proposed as a low-cost and energy-saving option for the removal of NH 4 + -N from contaminated groundwater. The objectives of this study were to investigate NH 4 + -N removal performance and the nitrogen removal pathway and to characterize the microbial communities in the reactor. Polyoleﬁn sponge cubes (10 mm × 10 mm × 10 mm) were connected diagonally in a nylon thread to produce 1 m long dropping nitriﬁcation units. Synthetic groundwater containing 50 mg L − 1 NH 4 + -N was added from the top of the hanging units at a ﬂow rate of 4.32 L day − 1 for 56 days. Nitrogen-oxidizing microorganisms in the reactor removed 50.8–68.7% of the NH 4 + -N in the groundwater, which was aerated with atmospheric oxygen as it ﬂowed downwards through the sponge units. Nitrogen transformation and the functional bacteria contributing to it were stratiﬁed in the sponge units. Nitrosomonadales -like AOB predominated and transformed NH 4 + -N to NO 2 − -N in the upper part of the reactor. Nitrospirales -like NOB predominated and transformed NO 2 − -N to NO 3 − -N in the lower part of the reactor. The dropping nitriﬁcation reactor could be a promising technology for oxidizing NH 4 + -N in groundwater and other similar contaminated wastewaters. ,

Organization (WHO) guideline for drinking water of 1.5 mg L −1 [10] is substantially less than the levels for these different locations. High NH 4 + -N levels in water supplies lead to unpleasant odors and taste. These high levels are a major cause for reducing the disinfection efficacy of chlorine and other halogens, as well as increasing the risk of pathogen contamination during water treatment and distribution; hence, the levels of NH 4 + -N in groundwater must be decreased before consumption.
The in situ permeable reactive barrier (PRB) is a promising groundwater NH 4 + -N removal technology. Several laboratory-and full-scale PRBs have effectively removed NH 4 + -N from groundwater [6,8,[11][12][13]. Although a PRB is cost-effective in the long term, it, nonetheless, requires large-scale construction and incurs a high initial cost [14]. Therefore, low-cost, energy-sparing, simple, and compact alternative technologies are needed to remove groundwater NH 4 + -N, particularly in developing countries and small rural communities.
In the present study, a dropping nitrification reactor was proposed for the removal of NH 4 + -N from contaminated groundwater ( Figure S1). This configuration simulated a down-flow hanging sponge (DHS) system. The DHS system is a low-cost, energy-saving option that is simple in construction and easy in operation and maintenance compared to the PRB. The DHS reactor was originally developed for the treatment of domestic sewage. DHS systems efficiently reduce organic compounds, chemical oxygen demand (COD), biochemical oxygen demand (BOD), and nitrogen in sewage [15][16][17][18][19][20]; they have also been used for the treatment of several kinds of industrial wastewater and for various other purposes [21]. The DHS reactor is a trickling filter using polyurethane sponge as the microbial carrier. Wastewater trickles down from the top of the sponge under the influence of gravity. In the down-flow process, wastewater is exposed to oxygen in the air; thus, this reactor does not require any external aeration [17,22,23]. The organic compounds and NH 4 + -N in wastewater can be oxidized by the microorganisms on the surface and in the interior of the sponge. During DHS wastewater treatment and other biological nitrogen removal processes, NH 4 + -N is subjected to ammonia-oxidizing bacteria (AOB) that generate nitrite-nitrogen (NO 2 − -N). The latter, in turn, is oxidized to nitrate-nitrogen (NO 3 − -N) by nitrite-oxidizing bacteria (NOB) [22]. To the best of our knowledge, however, no prior study has investigated the application of DHS for the treatment of NH 4 + -N-contaminated groundwater containing little or no organic carbon. Here, we hypothesized that a dropping nitrification reactor can effectively remove NH 4 + -N from contaminated groundwater. This reactor will be a low-cost, low-energy consuming, easily operable, and environmentally friendly option for NH 4 + -N removal from groundwater.
The aims of this study were to test the potential of the dropping nitrification reactor for the removal of NH 4 + -N from contaminated groundwater to elucidate the single axis, top-to-bottom nitrogen removal pathway and to identify the characteristics of the microbial community, including AOB and NOB in the reactor.

Setup of the Laboratory-Scale Dropping Nitrification Reactor
Polyolefin sponges (10 mm × 10 mm × 10 mm; Sekisui Aqua Systems Company, Osaka, Japan) served as the medium for the reactor. Polyolefin sponges are superior to polyurethane sponges in DHS (more durable, firmer shape, and higher load bearing capacity). Approximately 82 sponge cubes were connected diagonally in a nylon thread to form a hanging sponge unit with an effective length of 1 m ( Figure S1) and a dry weight of approximately 2.94 g. The sponge units were incubated with 10 L of a synthetic NH 4 + -N-contaminated groundwater mixed with the activated sludge (10:1, v/v).
The activated sludge from a municipal wastewater treatment plant in Kofu, Yamanashi, Japan, served as the bacterial inoculum for the sponge units. The suspension was aerated with a pump at a flow rate of 1 L min −1 for 10 days before setting up the reactor to establish microbial communities in the sponge media. A laboratory-scale reactor was prepared using a stainless-steel frame (1.2 m width, 0.45 m depth, 1.5 m height) and set up in a greenhouse at the University of Yamanashi, Kofu, Yamanashi, Japan. The reactor consisted of four separate hanging sponge units ( Figure S1). The reactor was covered with a sun-shielding sheet to prevent microalgal growth.

Operating Conditions of the Laboratory-Scale Dropping Nitrification Reactor
The groundwater NH 4 + -N removal reactor was operated for 56 days.

Physicochemical Parameters and Nitrogen Concentration in Water Samples
Water temperature, pH, ORP, and DO of the influent and effluent samples were measured on-site with a pH/Temp meter (AS600, AS ONE Corporation, Osaka, Japan), a waterproof ORP meter (ORP-6041, Custom Corp., Chiyoda-ku, Japan), and a DO meter (PDO-520, Fuso Inc., Chuo-ku, Japan), respectively.
Before determining the nitrogen concentration, water samples were filtered through a membrane filter (polypropylene, 0.45 µm pore size; Membrane Solutions Co. Ltd., Minato-ku, Japan). NH 4 + -N, NO 2 − -N, and NO 3 − -N concentrations were measured in accordance with the Standard Methods for the Examination of Water and Wastewater [25]. The indophenol method was used to determine NH 4 + -N concentration, and the N-(1-naphthyl) ethylenediamine and UV adsorption (220 and 275 nm) methods were used to determine NO 2 − -N and NO 3 − -N concentrations, respectively.

Kinetics of NH 4 + -N Removal Along the Single Axis of the Dropping Nitrification Reactor
The NH 4 + -N profiles along the single axis of the reactor from the top (influent, 0 m) to the bottom (effluent, 1 m) were evaluated according to zero-and first-order kinetic reactions as shown in Equations (1) and (2), respectively: where C is the NH 4 + -N concentration, dC dt is the rate of change in NH 4 + -N concentration per unit time, and k 0 and k 1 are the zero-and the first-order kinetic rate constants, respectively.

DNA Extraction from Sponge Samples and Quantification of Functional Microbial Genes
The sponge samples (approximately 100 mg wet weight = approximately 7.6 mg dry weight) were cut from the corners of the sponges and used for the microbial community analyses. Microbial DNA was extracted from the sponge samples with NucleoSpin Tissue (MACHEREY-NAGEL GmbH, Duren, Germany) according to the manufacturer's protocol.
Bacterial 16S rRNA and the ammonia monooxygenase (amoA), nitrite oxidoreductase (nxrA), and nitrite reductase (nirK and nirS) genes were quantified by real-time quantitative polymerase chain reaction (RT-qPCR) based on functional gene-targeted primer sets (Table S1) in a Thermal Cycler Dice Real-Time System II (TaKaRa Bio Inc., Shiga, Japan). Each qPCR assay was conducted in a 25 µL reaction mixture including 12.5 µL SYBR Premix Ex Taq (TaKaRa Bio, Shiga, Japan), 0.5 µM of each forward and reverse primer (Table S1), 2 µL template DNA, and 9.5 µL deionized H 2 O. The reaction conditions were as follows: initial denaturation by preheating at 95 • C for 30 s, 40 cycles of 98 • C for 5 s, annealing at the specified temperatures (which varied with primer type; Table S1) for 50 s, and an extension at 72 • C for 1 min, followed by a dissociation stage (95 • C for 15 s, 60 • C for 30 s, and 95 • C for 15 s). A standard curve was plotted for each gene using a synthetic plasmid carrying the target sequence. All qPCRs were conducted in duplicate. The gene abundances in the nitrification unit (copies per unit) were calculated using the gene abundances in the sponge material (copies per mg) and the total weight of sponge materials in the unit (mg per unit). 8.0.1623_i86linux64 (https://www.drive5.com/usearch/). In these analyses, contigs were formed, and error sequences and chimeras were removed. All operational taxonomic units (OTUs) were clustered at a cutoff of 0.03 (97% similarity). OTUs with < 1% relative abundance among all sequences in all samples were summed as "others". Sequencing and sequence-read analyses were conducted at FASMAC (Kanagawa, Japan).

Statistical Analysis
The mean and standard deviation of the physicochemical parameters and nitrogen concentrations were calculated from the data of four replicates of 1 m long dropping nitrification units. Gene abundances (± SD) were calculated using duplicate experiments for each sample. A t-test was used to compare the pairs of groups for significant differences (P < 0.05). The data were processed in SPSS v. 20 (IBM Corp., Armonk, NY, USA).

Changes in Physicochemical Parameters
Changes in pH, ORP, and DO in the influent and effluent samples over 56 days of experimentation are shown in Figure 1. Water temperature in the influent and effluent samples ranged from 22 to 32 • C. The pH of the influent was 8.0 ± 0.1 and that of the effluent ranged from 5.7 to 7.4. The pH of the effluent was significantly lower (P < 0.05) than that of the influent. The pH decrement from the top to the bottom of the reactor indicated that H + ion was released (acidification) as a result of microbial nitrification [26,27].
The ORP measurements in the influent and effluent samples were in the ranges of 109-155 and 127-174, respectively. The DO measurements in the influent and effluent samples were in the ranges of 6.0-8.3 and 6.5-8.5, respectively. The ORP and DO after 1 week of operation were significantly higher (P < 0.05) in the effluent than in the influent. Groundwater was fed from the top to the bottom of the reactor. During downward flow, oxygen diffusion into the groundwater from the air caused an increase in the DO concentration in the groundwater. Similar results have been obtained in the DHS reactors treating anaerobic wastewater [19,28,29]. Elevated DO levels are favorable for microbial nitrification [28]. nitrification in which NH 4 + -N was oxidized to NO 2 − -N and NO 3 − -N. Ammonia volatilization was not significant in the reactor, because pH value was < 9.3 in the synthetic groundwater [30]. For up to 35 days, the NO 2 − -N concentration was higher than the NO 3 − -N concentration in the effluent. Notably, after 35 days, the opposite trend was observed; therefore, the oxidation of NH 4 + -N to NO 2 − -N was effective and fast after starting operation of the reactor when compared to the oxidation of NO 2 − -N to Water 2019, 11, x FOR PEER REVIEW 6 of 13 steady state within a short period of operation. The total inorganic nitrogen (NH4 + -N, NO2 --N, and NO3 --N) in the effluent samples was approximately the same as the influent NH4 + -N concentration after 3 days of operation. The decrease in NH4 + -N concentration from the influent to the effluent was equivalent to the increase in the NO2 --N and NO3 --N concentrations in the effluent where the latter two compounds accumulated; thus, NH4 + -N removal was caused exclusively by biological nitrification in which NH4 + -N was oxidized to NO2 --N and NO3 --N. Ammonia volatilization was not significant in the reactor, because pH value was < 9.3 in the synthetic groundwater [30]. For up to 35 days, the NO2 --N concentration was higher than the NO3 --N concentration in the effluent. Notably, after 35 days, the opposite trend was observed; therefore, the oxidation of NH4 + -N to NO2 --N was effective and fast after starting operation of the reactor when compared to the oxidation of NO2 --N to NO3 --N.

Variations in the Abundances of the Functional Microbial Genes Involved in the Nitrogen Transformations
The abundances of bacterial 16S rRNA, amoA, nxrA, nirK, and nirS in the reactor over 56 days of operation are shown in Figure 3. The bacterial 16S rRNA gene abundance ranged from 8.2 × 10 12 to 4.5 × 10 13 copies per unit. The abundance of amoA significantly increased (P < 0.05) from 1.5 × 10 10 to 3.3 × 10 11 copies per unit within the initial 14 days and then reached and maintained a peak at 5.5 × 10 11 copies per unit. The abundance of nxrA gradually and steadily increased from 5.2 × 10 7 to 1.3 × 10 12 copies per unit over 56 days; thus, the AOB population and activity increased faster in the reactor than those of NOB. These changes in amoA and nxrA abundance corresponded to the changes in NH4 + -N, NO2 --N, and NO3 --N concentration in the effluent (Figure 2). The AOB population should be larger than that of the NOB in balanced nitrifying systems [31,32]. The NOB population was smaller than the AOB population in wastewater treatment systems, such as in sequencing batch reactors [33] and in combined activated sludge-rotating biological contactor and anaerobic-anoxicoxide (A2O) processes [34]. NOB growth is relatively slow in the absence of nitrite [35]. Similar to these previous reports, the AOB population increased more quickly in this reactor after the onset of the operation compared to the NOB population.

Variations in the Abundances of the Functional Microbial Genes Involved in the Nitrogen Transformations
The abundances of bacterial 16S rRNA, amoA, nxrA, nirK, and nirS in the reactor over 56 days of operation are shown in Figure 3. The bacterial 16S rRNA gene abundance ranged from 8.2 × 10 12 to 4.5 × 10 13 copies per unit. The abundance of amoA significantly increased (P < 0.05) from 1.5 × 10 10 to 3.3 × 10 11 copies per unit within the initial 14 days and then reached and maintained a peak at 5.5 × 10 11 copies per unit. The abundance of nxrA gradually and steadily increased from 5.2 × 10 7 to 1.3 × 10 12 copies per unit over 56 days; thus, the AOB population and activity increased faster in the reactor than those of NOB. These changes in amoA and nxrA abundance corresponded to the changes in NH 4 + -N, NO 2 − -N, and NO 3 − -N concentration in the effluent (Figure 2). The AOB population should be larger than that of the NOB in balanced nitrifying systems [31,32]. The NOB population was smaller than the AOB population in wastewater treatment systems, such as in sequencing batch reactors [33] and in combined activated sludge-rotating biological contactor and anaerobic-anoxic-oxide (A2O) processes [34]. NOB growth is relatively slow in the absence of nitrite [35]. Similar to these previous reports, the AOB population increased more quickly in this reactor after the onset of the operation compared to the NOB population. In contrast, the abundances of nirK and nirS were 4.3 × 10 9 -2.6 × 10 10 and 1.2 × 10 11 -3.1 × 10 11 copies per unit, respectively, and did not change significantly over 56 days. Functional denitrification genes (nirK and nirS) were detected in the sponge media. Nevertheless, no denitrification was observed in this experiment as the reactor was highly aerobic (Figure 1c) and the organic carbons (i.e., electron donors) for heterotrophic denitrification were not included in the synthetic groundwater.   In contrast, the abundances of nirK and nirS were 4.3 × 10 9 -2.6 × 10 10 and 1.2 × 10 11 -3.1 × 10 11 copies per unit, respectively, and did not change significantly over 56 days. Functional denitrification genes (nirK and nirS) were detected in the sponge media. Nevertheless, no denitrification was observed in this experiment as the reactor was highly aerobic (Figure 1c) and the organic carbons (i.e., electron donors) for heterotrophic denitrification were not included in the synthetic groundwater.   The decrease in NH 4 + -N concentration along the single axis of the nitrification units more nearly approximated a first-order than a zero-order kinetic reaction ( Figure S2). The first-order kinetic constant (k 1 ) significantly increased (P < 0.05) from 0.35 to 1.55 m −1 within the first 14 days and then ranged from 1.01 to 1.28 m −1 ( Figure 5). The first-order kinetic constant for the increase in NH 4 + -N removal within the initial 14 days corresponded to the increase in amoA gene abundance (Figure 3).
Water 2019, 11, x FOR PEER REVIEW 9 of 13 The decrease in NH4 + -N concentration along the single axis of the nitrification units more nearly approximated a first-order than a zero-order kinetic reaction ( Figure S2). The first-order kinetic constant (k1) significantly increased (P < 0.05) from 0.35 to 1.55 m −1 within the first 14 days and then ranged from 1.01 to 1.28 m −1 ( Figure 5). The first-order kinetic constant for the increase in NH4 + -N removal within the initial 14 days corresponded to the increase in amoA gene abundance (Figure 3).

Bacterial Community Structure Profiles along the Single Axis of the Dropping Nitrification Reactor
Bacterial community compositions at the class and order levels along the single axis of the reactor on the 56th day are shown in Figure 6. At the class level, Betaproteobacteria (37.9% of all classes), Alphaproteobacteria (24.2%), and Gammaproteobacteria (17.2%) predominated at the top of the nitrification reactor, whereas Alphaproteobacteria (16.2%), Betaproteobacteria (15.1%), and Nitrospira (NOB class; 25.6%) predominated at the bottom.
At the order level, Nitrosomonadales (AOB order; 34.5%) and Xanthomonadales (16.6%) predominated at the top of the nitrification unit; however, their abundances decreased to 11.4% and 7.3%, respectively, at the bottom part. On the other hand, the abundance of Nitrospirales (NOB order) increased along with the downward flow starting from the middle and reached 25.7% at the bottom of the unit. Nitrosomonas spp. (Nitrosomonadales) and Nitrospira spp. (Nitrospirales) were the dominant AOB and NOB species, respectively, in aerobic biological wastewater treatment plants [36] and DHS systems [37]. Nitrosomonadales and Nitrospirales members were also detected in the nitrification units of the present study. Moreover, the abundance of Nitrosomonadales-like AOB was relatively higher in the upper part, while the abundance of Nitrospirales-like NOB was higher in the lower part of the nitrification units. These spatial differences in AOB and NOB along the single axis of the nitrification unit were consistent with the profiles for the NH4 + -N, NO2 --N, and NO3 --N concentrations and the abundances of amoA and nxrA in the same system ( Figure 4). Thus, NH4 + -N is efficiently oxidized and transformed to NO2 --N by Nitrosomonadales-like AOB in the upper part, and NO2 --N is oxidized and transformed to NO3 --N by Nitrospirales-like NOB in the middle to lower part. In addition, Nitrobacterlike NOB (Order; Rhizobiales) might be responsible for NO2 -oxidation in the reactor, because Rhizobiales (2.8%-6.3% of all orders) was detected in the reactor. The coexistence of AOB and NOB has been reported for various environments [38,39], wastewater treatment plants [40][41][42], and drinking water treatment plants [43]. AOB and NOB function independently, but their synergistic relationships mutually benefit their growth and activity [44]. AOB and NOB are in close proximity within biofilms, but the active NH4 + -oxidizing zone is separate from the NO2 − -oxidizing zone at the micrometer colony diameter scale [45]. In the present study, we demonstrated that NH4 + and NO2 - At the order level, Nitrosomonadales (AOB order; 34.5%) and Xanthomonadales (16.6%) predominated at the top of the nitrification unit; however, their abundances decreased to 11.4% and 7.3%, respectively, at the bottom part. On the other hand, the abundance of Nitrospirales (NOB order) increased along with the downward flow starting from the middle and reached 25.7% at the bottom of the unit. Nitrosomonas spp. (Nitrosomonadales) and Nitrospira spp. (Nitrospirales) were the dominant AOB and NOB species, respectively, in aerobic biological wastewater treatment plants [36] and DHS systems [37]. Nitrosomonadales and Nitrospirales members were also detected in the nitrification units of the present study. Moreover, the abundance of Nitrosomonadales-like AOB was relatively higher in the upper part, while the abundance of Nitrospirales-like NOB was higher in the lower part of the nitrification units. These spatial differences in AOB and NOB along the single axis of the nitrification unit were consistent with the profiles for the NH 4 + -N, NO 2 − -N, and NO 3 − -N concentrations and the abundances of amoA and nxrA in the same system ( Figure 4). Thus, NH 4 + -N is efficiently oxidized and transformed to NO 2 − -N by Nitrosomonadales-like AOB in the upper part, and NO 2 − -N is oxidized and transformed to NO 3 − -N by Nitrospirales-like NOB in the middle to lower part. In addition, Nitrobacter-like NOB (Order; Rhizobiales) might be responsible for NO 2 − oxidation in the reactor, because Rhizobiales (2.8-6.3% of all orders) was detected in the reactor.
The coexistence of AOB and NOB has been reported for various environments [38,39], wastewater treatment plants [40][41][42], and drinking water treatment plants [43]. AOB and NOB function independently, but their synergistic relationships mutually benefit their growth and activity [44]. AOB and NOB are in close proximity within biofilms, but the active NH 4 + -oxidizing zone is separate from the NO 2 − -oxidizing zone at the micrometer colony diameter scale [45]. In the present study, we demonstrated that NH 4 + and NO 2 − oxidation in the nitrification unit treating contaminated groundwater was distinctly and widely separated from the top to the bottom of the flow.
Water 2019, 11, x FOR PEER REVIEW 10 of 13 oxidation in the nitrification unit treating contaminated groundwater was distinctly and widely separated from the top to the bottom of the flow. This study demonstrated the highly efficient NH4 + -N removal from groundwater by using the dropping nitrification reactor and revealed the characteristics of the reactor for the first time. In future studies, we will examine the NH4 + -N oxidation efficiency of the reactors in series and the operational lifetime of the sponge media. However, NO2 --N and NO3 --N were accumulated in the effluent of the reactor. Excess NO3 --N (>11 mg L −1 for drinking water [10]) has a negative impact on human health; thus, the effluent is still not recommended for drinking purposes. Therefore, we plan to conduct another study of NH4 + -N removal from groundwater by combining nitrification with a denitrification or anammox system for complete nitrogen removal. This study demonstrated the highly efficient NH 4 + -N removal from groundwater by using the dropping nitrification reactor and revealed the characteristics of the reactor for the first time. In future studies, we will examine the NH 4 + -N oxidation efficiency of the reactors in series and the operational lifetime of the sponge media. However, NO 2 − -N and NO 3 − -N were accumulated in the effluent of the reactor. Excess NO 3 − -N (>11 mg L −1 for drinking water [10]) has a negative impact on human health; thus, the effluent is still not recommended for drinking purposes. Therefore, we plan to conduct another study of NH 4 + -N removal from groundwater by combining nitrification with a denitrification or anammox system for complete nitrogen removal.

Conclusions
The  Figure S1: Schematic diagram of a dropping nitrification reactor, Figure S2: Zero-order (a) and first-order (b) kinetics for the decrease in NH 4 + -N along the single axis of the dropping nitrification units. NH 4 + -N concentrations represent the mean of four replicate reactors at different sampling depths. Comparison of the coefficients of determination (R 2 ) between the zero-and first-order kinetic reactions (c), Table S1: Target genes for qPCR analysis, primers and sequences, annealing temperatures, and amplification sizes.