This study used six air-cathode single-chamber MFCs consisting of two identical chambers (a Plexiglas cylinder, 2 cm in length and 3 cm in diameter) with a total working volume of 30 mL. The anode consisted of a circular carbon cloth (W0s1009, Phychemi Co. Ltd., China) with a diameter of 3.0 cm. A commercial carbon cloth (W1s1009, Phychemi Co. Ltd., China) laminating a blocking layer of active carbon on one side was used to prepare the air-cathode (Gao et al., 2021). Four poly(dimethylsiloxane) (PDMS, Sylgard 184, Dow Corning, Auburn, MI) layers brushed on the filled side were considered diffusion layers exposed to the air, while Pt/C of 0.5 mg/cm² coated on the other side was regarded as a catalyst layer exposed to the solution (Zhang et al., 2010). The anode and cathode were placed on opposite sides of each chamber and connected to an external resistor of 1kΩ using a titanium wire.

The startup of DNMFCs was shortened by hanging the anodes on a mature anode in the long-term, stable, large-scale MFC beforehand for more than 1 month. As the inoculum, the composition of the anodic microbial community in the anode chamber of the dual-chambered MFC is shown in Supplementary Figure S1. The pre-enrichment stage lasted 35 days (∼5 batches). The synthetic wastewater containing 1.0 g/L CH₃COONa (∼700 mg/L COD), 10.13 g/L Na₂HPO₄, 6.08 g/L NaH₂PO₄, 0.13 g/L KCl, 0.71 g/L NaNO₃ (96 mg/L NO₃ ⁻-N), and 12.5 mL/L Wolfe’s mineral solution was fed. The COD/NO₃ ⁻-N ratio of wastewater was approximately 7, which was purged with N₂ for 30 min s before injecting into the reactor. The reactors were operated in a fed-batch mode at an ambient temperature of 30°C ± 1°C.

After stable operation, a series of tests were conducted (such as organics removal, electron recovery, and nitrogen detection) to evaluate electricity generation and nitrogen conversion. For further analysis of special microorganisms, the distribution of DNB, and their contribution to DNMFCs, the original single chamber was divided into two identical chambers after 50 days of stable operation, which were separated by a proton exchange membrane (Nafion 117, Dupont) (Figure 1). The DNMFCs with single-chamber and dual-chamber were termed DNMFCs and DC-DNMFCs, respectively. Other parameters remained unchanged without special instructions, and various parameters affecting nitrate reduction were investigated. All tests were conducted in triplicate, and parallelization of more than three cycles was tested under identical conditions in each test to ensure the accuracy of the results.

During the MFC batch operation stage, effluent samples were withdrawn from each batch using a syringe and passed through a syringe filter (0.45 μm pore diameter) before analysis. The concentrations of chemical oxygen demand (COD), NH₄ ⁺-N, NO₂ ⁻-N, and NO₃ ⁻-N were measured according to the APHA standard methods (APHA, 2001). The output voltages were recorded at 5-min intervals using a computer with a data acquisition system, as described previously (Jin et al., 2022). Polarization and power density curves were calculated by varying the external resistor from 10 to 0.01㏀. The generated electric charge (Ee, Q), theoretical consumed electric charge of nitrate conversion (E_N, Q), theoretical generated electric charge of acetate (E_A, Q), Coulombic efficiency (CE, %), and electron transfer efficiency (ETE, %) were calculated using Eqs 1–5: Ee = ∫ 0 t I d t , (1) E N = F × V × ∆ C N × 5 14 , (2) E A = F × V × ∆ C O D × 4 32 , (3) C E = E e E A × 100 , (4) ETE = Ee + E N E A × 100 , (5) where I is the current, t is the time, F is the Faraday’s constant (96,485 Q/mol), V is the volume (30 mL), ΔC_N is the amount of NO₃ ⁻-N reduced, and ΔCOD is the amount of COD consumed.

Throughout the process, a commercial DO microelectrode (Unisense, Denmark) was separately inserted in the middle of the anode and cathode chambers to measure the DO value. Before each measurement, the microelectrode was calibrated following the manufacturer’s instructions. Independent triplicate experiments were conducted for each stability test.

The micromorphology of the electrode biofilms was observed using a scanning electron microscope (SEM) (JSM-6510LV, Japan) and a confocal laser scanning microscope (CLSM) (inVia Reflex, UK). Before SEM testing, the samples were pretreated as follows: (1) the anode samples were immersed for 4 h in the fixative containing 2.5% glutaraldehyde at 4 °C; (2) the samples were rinsed with 0.1 mol/L PBS solution; (3) they were then dehydrated with 50, 70, 80, and 90% ethanol, respectively; and (4) they were further dehydrated with tert-butyl alcohol; and (5) finally, they were tested after being coated with a 10 nm layer of gold to reduce charging during SEM analysis. Before CLSM testing, the samples were pretreated as follows: (1) the anode samples were rinsed with 0.1 M PBS solution to eliminate the original medium; (2) stained using a LIVE/DEAD BacLight Viability Kit (L13152, Life Technologies, USA) for 20 min; and (3) rinsed with 0.1 M PBS solution to eliminate excess dye. A CLSM with ×20 objective was used to visualize the spatial live/dead topography of the biofilms.

Total genomic DNA was extracted from an entire piece of each electrode, and the bacterial 16S rDNA was PCR-amplified using primers 338F and 806R targeting the variable V3–V4 region (forward primer: 5′-CCT​ACG​GGA​GGC​AGC​AG-3′; reverse primer: 5′-GGACTACHVGGGTWTCTAAT-3′) (Jin et al., 2022). The PCR products were sequenced using the MiSeq Illumina platform (Majorbio Bio-pharm Technology, Shanghai, China). These data analyzed were used to express the microbial community distribution at the class and genus levels.

Ammonium was used instead of nitrate to compare the electricity performance and nitrogen-removal abilities of DNMFCs treating NH₄ ⁺- and NO₃ ⁻-containing wastewater (Figures 2A–C). First, NH₄ ⁺ loss in DNMFCs was only 6.2%, and null concentration of NO₂ ⁻ and NO₃ ⁻ was detected in the bulk solution throughout the cycle operation, indicating that ammonium oxidation did not occur on the air-cathode surface (Figure 2A). It means that the nitrifying process does not contribute to organic oxidation and electron flux and that the nitrifying bacteria do not participate in the oxygen and nitrate reduction reactions. Both the pre-enrichment of anode biofilm and the high activity of air-cathode with Pt/C maintained a low DO concentration in the solution even during startup, which finally inhibited the occurrence of nitrification in this system.

When DNMFCs were treated with nitrate-containing wastewater, NO₃ ⁻ was rapidly reduced with no noticeable accumulation of intermediates during the entire process, demonstrating complete biological denitrification by DNB (Yang et al., 2019). Second, the cycle time of MFCs in the presence of NH₄ ⁺ was approximately two times longer than that of NO₃ ⁻. Therefore, the relative electron recovery of DNMFCs sharply decreased from 106.8 ± 0.3 to 46.7 ± 0.2 μmol/e, respectively (Figures 2A, F). Furthermore, similar waves of power density and polarization curves of DNMFCs with ammonia- and nitrate-containing wastewater treatment indicated that the appearance of stable and mature biofilms on the electrodes was not affected by nitrogen composition.

According to polarization curves, the anode potentials in MFCs were highly consistent with the increase in current densities, attributed to the pre-enriched biofilm of the anodes before the reactor startup. The cathode potential also tended to be stable, suggesting that the microbial community in the electrode biofilms of DNMFCs strongly tolerated nitrogen stress. Therefore, the method for DNMFC startup in this study is successful and efficient. The cathode biofilm in the DNMFCs displayed an increased EDNB abundance, which partly compensated for the active site reduction of Pt/C. The oxygen reduction reaction (ORR) by the chemical catalyst of Pt/C can be performed in a series of reactions with the cooperation of Pt/C, DNB, and EAB. Compared to the metal-free carbon cloth used as the cathode, the metal-coated cathode in SCMFCs has a shorter startup time, higher bioelectrochemical activity, and higher nitrogen removal efficiency (Xin et al., 2020; Yan et al., 2012; Yang et al., 2019).

In DNMFCs, the electrochemical activity and performance of electricity generation are closely related to the ratios of COD/NO₃ ⁻-N in wastewater. As shown in Figures 2D–F, the electron fluxes were further investigated at COD/NO₃ ⁻-N ratios from 7 to 3, and high efficiencies of electrochemical activity and nitrogen removal were obtained in DNMFCs. NO₃ ⁻ was majorly removed when the COD/NO₃ ⁻-N ratio was above 4, which is close to the value of 3.5 for nitrate reduction by heterotrophic denitrifying bacteria (Knowles, 1982) and much lower than that of 10 in practical wastewater treatment for nitrogen removal (Sun et al., 2010). Furthermore, compared to general heterotrophic denitrification, electrochemical techniques for nitrate removal have the advantages of high efficiency, active sludge decrement, and lack of additional chemical reagent addition. However, CEs sharply declined with the decrease in the COD/NO₃ ⁻-N ratio, indicating that the nitrate reduction reaction is very important and should be unignored in DNMFCs. As shown in the results, the rate of nitrate reduction is much quicker than that of other reduction reactors. Therefore, electron transfer efficiency seems much more reasonable than characterizing the performance of DNMFCs. In DNMFCs, ETEs increased from 52.5% ± 0.5% to 71.1% ± 0.8% and then decreased to 44.2% ± 0.5% at a COD/NO₃ ⁻-N ratio of 3. Under the condition of insufficient organic matter, the intermediate of NH₄ ⁺ considerably accumulated, suggesting the presence of electrochemical nitrate reduction to ammonia in DNMFCs.

Considering the voltage waves, electrode potentials, DO, and COD, DNMFCs with a single chamber and two chambers exhibited similar behaviors under the same operational conditions, which could be due to the minimal effects exerted by protons and oxygen on the electricity performance (Supplementary Figure S2). Furthermore, the biofilm growing on the wall also slightly affected the denitrification because there was no significant difference in the nitrate removal rate after removing the biofilms from the DNMFCs (Figure 3A). Additionally, a high concentration of NO₂ ⁻-N was observed before the biofilm removal, implying that the intermediate accumulation was mainly attributed to the process of traditional biological denitrification (Sun et al., 2010). As shown in Figure 3B, different trends of nitrogen removal rate were observed in the anode and cathode chambers of DC-DNMFCs. Considering NO₃ ⁻-N removal rates, the mean value in the anode chamber (22.6 mg/L/h) was considerably higher than that in the cathode chamber (13.4 mg/L/h), demonstrating that the high abundance of electroactive biofilm promotes anode heterotrophic denitrification by extracellular electron transfer, as previously described (Tong and He, 2013; Sotres et al., 2016). NO₂ ⁻-N accumulation in the cathode chamber was notably greater than that in the anode chamber, verifying that DNB-mediated biological denitrification produced more intermediates than that produced by EDNB (Tong and He, 2013). Nitrous oxide (N₂O) as an intermediate product of biological denitrification is significant for its strong greenhouse gas effect and has been detected in our previous studies (Jin et al., 2022). However, the percentage of residual gaseous N₂O was less than 0.0001%, which can be ignored during the mass conversion of nitrate reduction in DNMFCs by heterotrophic denitrification.

The distribution of acetate, nitrate, and electrons in DNMFCs changed with the COD/NO₃ ⁻-N ratios. In this system, we designed a series of experiments for calculating the proportion of each reaction at a COD/NO₃ ⁻-N ratio of 7. For example, the hollow slab at the air side of the cathode was replaced with a solid plate to block the air, which can calculate the generated electrons accepted by nitrate. The results showed that the voltages of DNMFCs without air supply sharply decreased, and the time for a complete cycle lasted less than 10 h (Figure 3C). As shown in Supplementary Table S1, NO₃ ⁻-N removal remained unaffected in both conditions of air supply and no air supply. However, COD removal sharply declined, and only approximately 7.8 μmol/e of electrons were recovered by DNMFCs without air supply. The value of E_N should be 7.8 μmol/e; therefore, the concentration of NO₃ ⁻-N reduced by electrochemical denitrification was 7.5 mg/L using Eq. 2, which accounted for 7.7% of the original concentration. The residual nitrate level of 92.3% was reduced by heterotrophic denitrification. Conversely, the voltage of DNMFCs with air supply maintained a stable output of 400 mV in 15 h, and the entire cycle lasted for more than 35 h, implying that ORR was still dominant in the air-cathode of DNMFCs. Approximately 46.7 μmol/e of electrons were recovered in DNMFCs with air supply. Therefore, in DNMFCs, nitrate as an electron acceptor only accounted for 16.7% of all collected electrons, and the other electrons were accepted by oxygen. Similarly, the amount of COD from the number of collected electrons could be calculated as 107.2 mg/L using Eq. 3. The theoretical consumed COD for electricity generation can be calculated, assuming that a CE of 38.0% is constant in DNMFCs. The results showed that 43.4% of COD was used for power recovery, and the rest was used for heterotrophic denitrification. Above all, the distributions of acetate, nitrate, and electron in DNMFCs at the COD/NO₃ ⁻-N ratio of 7 are shown in Figure 3D.

Nitrate stimulates the growth of nitrogen-metabolizing bacteria, resulting in DNB enrichment, with a remarkable ecological impact on DNMFCs (Huang et al., 2018). Azonexus and Pseudomonas, which are known as DNB and EDNB, prefer to grow in nitrogen-containing solutions, becoming the dominant species in DNMFCs (Jangir et al., 2016; Yun et al., 2018; Yang et al., 2019). DC-DNMFCs can be considered the transition state of the dynamic development in DNMFCs. The biomass density of electrode biofilms of DNMFCs considerably differed when viewed using an SEM and a CLSM (Supplementary Figure S3). During long-term operations, copious amounts of bacteria in the inner layer of the electrode biofilms died, demonstrating that the inner layer of the thick biofilm cannot satisfy the nutrient demand in the oligotrophic micro-environment (Yuan et al., 2021).

The microbial community structure at the class level (abundance >1%) and the heatmap of their bacterial distribution at the genus level (abundance >1%) between DNMFCs and DC-DNMFCs were analyzed, as shown in Figures 4A–C. At the class level, Alphaproteobacteria and Gammaproteobacteria accounted for 51.6% and 17.8%, respectively, in the anode biofilm of DNMFCs, which decreased to 25.0% and 7.9% in the DC-DNMFCs. In contrast, Flavobacteriia and Ignavibacteria remarkably increased in the anode of DC-DNMFCs. Compared to the anode, the gap between microbial communities was closer in the cathode between DNMFCs and DC-DNMFCs, and the relative abundances of Proteobacteria (including Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria) reached 84.5% and 83.5%, respectively. Moreover, Flavobacteriia dominated the cathode of DNMFCs, while Bacteroidia dominated the DC-DNMFCs. At the genus level, the dominant genera of anode biofilm in DNMFCs were in the order of Azonexus, Pseudomonas, Comamonas, Chryseobacterium, and Aminiphilus, which are considered DNB and EAB (Figure 4C). However, biofilms attached to the anode of DC-DNMFCs noticeably enriched the sequences related to Chryseobacterium, Ignavibacterium, Rhodocyclus, and Pseudomonas. Pseudomonas accounted for 49.6% of the DNMFCs and 40.4% of the DC-DNMFCs in cathode biofilms, respectively. The high abundance absolutely dominated the cathode microbial community, indicating that this special genus remarkably reduced nitrate by heterotrophic/electrochemical denitrification (Li et al., 2019). The genus Aquamicrobium exhibited a remarkable similarity between DNMFCs and DC-DNMFCs, which might relate to the original inoculum. Moreover, the genus Pannonibacter (Sun et al., 2016), a new type of denitrifier, was also enriched in DNMFCs, which dramatically increased in DC-DNMFCs.

Additionally, the microbial community compositions of biofilms growing on the walls of the entire DCMFC, the DC-DNMFC anode chamber, and the DC-DNMFC cathode chamber were also investigated to understand the dynamic rule of bacterial change. Meanwhile, the cathode biofilms, divided into the surface and inner layers, were also analyzed at the genus level. The former can be easily washed away by running water, and the latter tightly adheres to the cathode. Furthermore, the proportions of special genera significantly differed on the anode, wall, cathode, cathode surface layer, and cathode inner layer. For example, the proportions of the dominant genus Pseudomonas distributed in the DNMFCs were 17.6% (anode), 27.8% (wall), 49.6% (cathode), 30.6% (surface layer), and 49.8% (inner layer). This genus grew majorly on both the electrodes and wall, which were maintained at high levels in DNMFCs. In the DC-DNMFCs, the percentages of Pseudomonas changed to 7.3% (anode), 6.5% (anode wall), 40.4% (cathode), 35.9% (cathode wall), 27.6% (surface layer), and 25.3% (inner layer). These results indicate that Pseudomonas prefers a micro-aerobic environment of cathode for growth, although it presents high activity for nitrate reduction under a strictly anaerobic environment of anode. Few aerobic bacteria (including Aquamicrobium and Azoarcus) exhibited remarkably high relative abundance in the cathode chamber, consuming O₂ permeated from the air-cathode, thereby making the cathode a hypoxic environment suitable for the growth of denitrifiers. In contrast, the genus Azonexus dominated the anode of DNMFCs with a high relative abundance of 33.1%, which practically disappeared in the cathode biofilm, implying that Azonexus as an alternative DNB needs anaerobic conditions for growth. The proportions of Azonexus in DC-DNMFCs at the anode and the wall of the anode chamber accounted for 5.8% and 32.8%, respectively, which is a critical proof to verify the above speculations. Furthermore, functional genera consisted of Proteiniphilum, Comamonas, Rhodocyclus, Chryseobacterium, and Aminobacter. Moreover, other genera with high percentages in the anode biofilm and low proportions in the wall could be considered EAB, which included Rhodocyclus, Desulfomonile, and Ignavibacterium. Additionally, Methanosarcina, methane-producing bacteria, also increased to 11.1% at the anode of DC-DNMFCs, compared to 7.2% at the anode of DNMFCs. These results were consistent with the fact that DNB and EAB could be dominant in the MFC-based wastewater treatment process (Yang et al., 2020; Yang et al., 2021).

In this study, functional bacteria can be grouped into three categories: DNB, EAB, and EDNB. First, DNB were rapidly accumulated in the solution and then adhered to the reactor walls or adsorbed onto the electrodes (anaerobes on the anode and facultative anaerobes on the cathode). DNB with low electron transfer capacity, such as Azonexus, Azoarcus, and Comamonas, possessed low relative abundances in the electrode biofilm but high abundances in the wall biofilm. Second, EAB that could adapt to the nitrate-containing aqueous environment were either retained or gradually eliminated during the system operation. The relative abundances of EAB, such as Ignavibacterium, Rhodocyclus, and Desulfomonile, except for Geobacter, notably increased in DC-DNMFCs, indicating that these genera were likely to have a positive response to nitrate exposure (Yang et al., 2021). Finally, the relative abundance of Pseudomonas increased considerably at both the cathode and anode biofilms, which identified its capacity to metabolize nitrogen and transfer electrons. Previous studies have confirmed that Pseudomonas is widespread in the electrode, which contains nitrate-reducing species (Ilamathi et al., 2019). In this system, Pseudomonas may be related to simultaneous nitrate reduction and electron transfer, which was consistent with the previous study (Jin et al., 2022).

According to the taxonomy information, the evolution process of the dominant functional bacteria from anode to cathode, especially the distribution of the dominant genera in DNMFCs, is illustrated in Figure 5. Azonexus and EAB were noticeably enriched in the anode chamber, which gradually weakened until they completely disappeared at the cathode, indicating that the bacteria are strictly anaerobic. However, other DNB, mostly known as facultative anaerobes, grew throughout the reactor with increased relative abundances. Furthermore, DNMFC systems are believed to favor the growth of EAB with nitrate-reducing ability, such as Pseudomonas, which was considered the most remarkable contributor, absolutely dominating nitrate-containing wastewater treatment (Sun et al., 2016; Yang et al., 2020). Therefore, nitrate exposure not only influences the microbial community structure but also enriches the relative functional bacteria, resulting in the increase in an cathode biofilm (Li et al., 2019; Guo and Liu, 2020).