Long-term effect of polyethylene microplastics on the bioelectrochemical nitrogen removal process

Nitrogen pollution in wastewater has been considered a worldwide risk to ecosystems. Bioelectrochemical technologies have been developed recently to remove nitrogen. In addition to nitrogen, microplastics, as emergency pollutants in wastewater, could potentially affect nitrogen removal efficiency due to their toxic effects on the activity of microorganisms. This study explored how microplastics would influence nitrogen removal in the bioelectrochemical process. It was found that nitrogen removal declined following the addition of micro-plastic during a long-term exposure experiment. With exposure to microplastics, the biofilm viability and the content of EPS declined significantly. Microbial community structure shifted significantly following the addition of microplastics. As the most abundant genera and denitrifier, Thiobacillus shrank largely with the addition of microplastics. Moreover, the reduction of the total abundance of the denitrification bacteria and the denitrification-related functional genes was also observed. The results unveil the mechanisms of how micro-plastics inhibit nitrogen removal and offer insights into the application of bioelectrochemical in nitrogen removal from wastewater rich in microplastic.


Introduction
Nitrogen pollution in the aquatic environment has drawn worldwide concern due to its deteriorating effect on water ecosystems [1].In the biological wastewater treatment processes, ammonium is mainly oxidized to nitrate through nitrification, followed by denitrification, which reduces nitrate into nitrogen gas.Organic carbon sources are necessary to complete nitrogen removal as the electron donor.The denitrification process would be restrained for wastewater or groundwater with a low C/N ratio because of the shortage of electron donors [1,2].Although external carbon addition could remedy the insufficient electron donor, it would increase the cost of wastewater treatment and might cause excessive organic effluent.Numerous efforts have been put into treating low C/N wastewater through autotrophic denitrification processes, among which hydrogen, iron, and sulfur were widely used as electron donors [3,4].In addition, significant progress has been made in developing autotrophic denitrification processes through other pathways.
Recently, microbial electrochemical systems (MES), which involve ammonium oxidation in bioanode and denitrification at biocathode, have been developed as an alternative approach to remove nitrogen from wastewater [5].Zheng et al. have demonstrated that ammonia oxidation and denitrification could be achieved in single microbial electrolysis cells (MEC) at a suitable anodic potential [1].In the study of Zhu et al. [6], the ammonium anodic oxidation process was promoted in a single chamber MEC with an anodic potential control.Besides, it was verified that cathode potential in a denitrifying MEC could shift the nitrate removal rate, and the highest denitrification rate was achieved at − 123 mV vs SHE [7].MES biocathode can reduce nitrate to nitrogen gas without adding organic carbon sources [8,9].The denitrifiers in the biofilm could gain electrons from the cathode electrode directly or indirectly from the hydrogen gas produced on the cathode to reduce nitrate to nitrogen gas [9][10][11].Therefore, MES could overcome carbon scarcity when treating nitrogen-polluted wastewater with a low C/N ratio.However, the performance of MES could be turbulent due to the changes in the wastewater characteristics, such as pH [12], copper [13] and nickel ions [14].
Microplastics, tiny plastic particles with a size <5 mm, have attracted increasing attention in recent years due to their potential toxic effects and widespread distribution.Although the ocean is regarded as the largest sink for microplastics, around 80% of that comes from the terrestrial environment [15].Microplastics have been detected in almost all aquatic terrestrial environments, including municipal wastewater [16,17], rivers [18], lakes [15], rainwater [19], reservoirs [20], and even groundwater [21].Previous studies have demonstrated that the occurrence of microplastics could alter microbial community structure and the performance of the microbial-based system.Sedimentary nitrification and denitrification processes could be inhibited by polyvinyl chloride microplastics by affecting the microbial community structure and functional gene abundance [22].It was reported that the microplastics at 1000 μg/L decreased ammonium removal efficiency and accumulated the nitrate and nitrite in sequence batch reactors (SBR) [23].Microplastics could influence nitrogen transformation through shifting microbial community structure, nitrogen cycling genes, and related enzyme activities [24].It has even been found that microplastics can affect the performance of exoelectrogenic biofilm treating organic wastewater [25].However, the role of microplastics in MES-based nitrogen removal processes remains an enigma.Deciphering how microplastics could affect the nitrogen removal ability in MES would improve our understanding of the ability to withstand external stress of MES, which is necessary for the engineering application of MES to treat nitrogen-polluted wastewater.
In this study, polyethylene microplastic (PE-MP) was selected to investigate its effect on bioelectrochemical nitrogen removal.The electrochemical and microbiological response of the MEC systems to PE-MP was evaluated.Electrode biofilm viability and extracellular polymer substrate (EPS) distribution on the electrode biofilm matrix were quantified.Microbial communities and functional genes were analyzed to underline the bacterial composition changing and the genes involved in the denitrification process.This study provides new insight into the role of microplastic in the bioelectrochemical nitrogen removal process.

Reactor configuration
Single-chamber rectangular reactors with an effective working volume of 200 ml (5 × 5 × 8 cm) were used to perform the experiment.The reactors were assembled with stainless bolts, nuts, and rubber gaskets described previously [26].Three-electrode system was configured for the bioelectrochemical nitrogen removal.A couple of carbon felt were used as the anode and cathode to enrich the biofilm with the ability to nitrify and denitrify, respectively.The surface area of the two electrodes was 16 cm 2 (4 × 4 cm), and they were placed at a distance of 3 cm.The reference electrode (Ag/AgCl, 0.197 V vs. SHE) was positioned close to the cathode at a distance of 1 cm.The schematic of the reactor was shown in Fig. S1.Cathode potential was controlled by a potentiostat (Ivium-n-Stat, Ivium Technologies, the Netherlands), with the cathode as the working electrode, anode as the counter electrode, and Ag/AgCl as the reference electrode.Cathode potentials were set to enhance the denitrification process at low C/N [7,27].

Inoculation and operation
All the reactors were initially inoculated with active sludge collected from the wastewater treatment plant (Lundtofte, Lyngby, Denmark) for nitrogen removal [28].The experiment was conducted continuously in overflow mode.The inoculated bacteria would form biofilm on the surface of the electrode.[1].Vitamin solution and mineral solution were supplemented with 12.5 ml/L and 2.5 ml/L, respectively [29].The low C/N ratio in the influent was controlled at 0.5, and the influent tank was stored in a 4 • C fridge with an agitator stirred continuously.Few dissolved oxygen was supplied in the influent to support the nitrification process with the assistance of the anode biofilm.Influent flew into the anode, and the ammonium could be oxidized to nitrate with organic carbon and oxygen consumption.Then the wastewater flew to the cathode side to perform the denitrification process and flew out of the reactors with an overflow model.The continuous experiment was operated for 96 days, which was divided into three stages.In stage 1, all the reactors were run as the baseline stage without microplastics as the control.In stage 2 and stage 3, the experiment was conducted at low concentration (10 µg/L) and high concentration (1000 µg/L) PE-MP levels [23].PE-MP (Cat no.434272.Sigma-Aldrich) with a size of 40-48 µm was selected as the model microplastic because it was regarded as the most ubiquitous synthetic polymer type in the aquatic environment [18].To investigate the effect of microplastics on the performance of the bioelectrochemical denitrification, different cathode potentials were set for each reactor at open, − 123 mV, − 303 mV, and − 703 mV vs SHE, named R1, R2, R3, and R4, respectively.Every two days, the samples were collected from the effluent of each reactor for physical-chemical analysis.Triplicate biofilm samples were collected at the end of stage 1 and stage 3 from the electrode randomly.The experiment was conducted at room temperature.

Biofilm viability and EPS characterization
Electrode biofilm collected at the end of stage 1 and stage 3 from all the reactors were visualized using Confocal Laser Scanning Microscopy (CLSM) (Leica TCS SP5, Germany) to investigate the electrode biofilm viability and EPS characterization under the control and influence of microplastics.Biofilm viability was observed under the z-stack mode, and all the samples were pretreated following the previous study [25].In short, the collected biofilm samples were rinsed in 50 mM PBS three times directly, followed by staining live/dead cells for 15 min using a Bacterial Viability kit (LIVE/DEAD BacLight, Invitrogen, Carlsbad, U.S. A.). SYTO-9 is a green fluorescent nucleic acid stain, which could enter both live and dead cells.While red fluorescent nucleic acid stain propidium iodide could only enter dead cells with broken membranes.The live and dead cells emitted green and red fluorescence after being stained by STYO-9 and propidium iodide, respectively.The rest of the dye was washed using 50 mM PBS.Excitation and emission wavelength for these two dyes were summarized in Table S1.Under the z-stack model, the stained biofilm was visualized layer by layer.Biofilm threedimensional structures were constructed using Imaris (version 7.4.2) to quantify the coverage of both live and dead cells separately [30].The ratio of the dead cells was calculated based on the mean fluorescence intensity value picked from three random places from the electrode.
The characterization of EPS for the biofilm in the reactors without (control) and with PE-MP was analyzed according to the previous studies [31,32].The main part of EPS, including proteins, polysaccharides, and lipids, were stained by specific dyes separately and visualized through CLSM.The biofilm was first rinsed using 50 mM PBS three times to remove the adsorbed impurities.Before staining, sodium bicarbonate buffer (0.1 M) was used to maintain the amine group with a non-protonated form.Afterward, Fluorescein isothiocyanate (FITC) (1.0 g/L) was added to mark the proteins.Subsequently, the biofilm was incubated into concanavalin A (CA) (250 mg/L) and Nile red (NR) (10 mg/L) for 30 min and 10 min, respectively, to label polysaccharides and lipids.Lastly, the extra stain was removed by rinsing the biofilm three times with 50 mM PBS.The wavelengths of excitation and emission and the target for these three dyes were summed up in Table S1.CLSM visualized different stained EPS sections under the z-stack model, and the three-dimension structure was constructed by Imaris software.The content of each fraction was calculated based on the fluorescence intensity value of each.

Microbiological analysis
The biofilm samples collected at the end of stage 1 and stage 3 were S. Wang et al. used to analyze the microbial community structure variation using highthroughput sequencing.Only cathode biofilm was collected because cathode denitrification was the process promoted by the applied cathode potential.Hence, to evaluate the effect of the microplastic in the bioelectrochemical nitrogen removal, the impact on the cathode denitrification performance is critical.The total genomic DNA was extracted using DNeasy Powerlyzer PowerSoil Kit (Qiagen, Germany) following the manufacturer's standard.Then the PCR amplification was performed using the universal primers 388F (ACTCCTACGGGAGGC AGCAG) and 806R (GGAC-TACHVGGGTWTCTAAT) targeting the V3-V4 region of the bacterial 16S rRNA gene [33].Subsequently, the products were purified and sequenced through Illumina MiSeq PE 250 platform by Majorbio Co. Ltd. (Shanghai, China).The treatment process of the obtained sequence data followed our previous study [34,35].High-quality sequences were filtered out and clustered into Operational Taxonomic Units (OTUs) at a 97% similarity level by UPARSE (version 7.0.1090).Taxonomic classification of the clustered OTUs was performed against to 16S rRNA Silva database using the RDP classifier.The rarefaction curve and alpha diversity were constructed using Mothur (version v.1.30.2).Raw data were deposited in the NCBI database with the accession number PRJNA903200.Representative functional genes involved in the denitrification process were quantified with real-time qPCR to evaluate which denitrification pathway could be influenced under the stress of microplastics [36].The primers used for qPCR were showed in Table S2.

Analytical methods and visualization
The concentration of ammonium-N, nitrate-N, and nitrite-N was quantified using SKALAR (San++, Netherlands).Cyclic voltammetry (CV) of both anode and cathode electrodes was examined at the end of stage 1 and stage 3 to evaluate the electrochemical characteristics of the biofilm.The CV was performed from − 0.8 to 0.8 V with a scanning rate of 2.5 mV/s.To estimate the beta diversity of each group of samples, principal coordinates analysis (PCoA) was carried out based on the Bray Curtis distance.Circos plot was used to show the bacterial composition at the phylum level using Circos (0.67-7).The phylogenetic tree of the abundant genus was generated through R with "ggtree" package and then visualized by iTOL.The significant differences in biofilm viability, EPS content, and microbial community diversity between stage 1 and stage 3 were identified by Student t-tests (p < 0.05).

Long-term effects of PE-MP on nitrogen remove
The nitrogen removal efficiency was first investigated in all the reactors without PE-MP during stage 1 under different cathode potentials.The influent flew into the bottom of the anode electrode and overflew from the cathode.The ammonia oxidation occurred on the anode biofilm, releasing electrons to the anode (Eqs.( 1)-( 2)) [1].Meanwhile, the low oxygen content from the feed solution could further induce ammonium oxidation (Eqs.( 3)-( 4)) [23].The oxygen was almost consumed around the anode electrode.While the nitrate accumulated in this low C/N ratio wastewater because of the low content of electron donor.Previous research demonstrated that the nitrate removal rate would increase with the decrease of cathode potential from + 597 mV to − 403 mV [7].In the MES systems, the electrotrophic bacteria could capture electrons from the cathode.This could drive heterotrophic denitrification (Eqs.( 5)-( 6)) [37,38].It has also been reported that ammonia oxidation and denitrification could be achieved in the singlechambered MEC [1].
Nitrogen removal performance was evaluated during 96 days of the operation targeting NH 4 + -N, NO 3 --N, and NO 2 --N.In stage 1 (0-38 d), these four reactors were operated without PE-MP to evaluate the nitrogen removal efficiency under different cathode potentials (Fig. 1).It was noteworthy that the nitrogen removal was enhanced with the cathode potential control.The ammonium concentration in the effluent decreased from around 82 mg/L in R1 to 33 mg/L in R2.Especially for the denitrification process, it was clear that the nitrate removal was significantly increased in R2-R4 compared to R1.An enhancement of denitrification was reported in cathode bioelectrochemical systems using autotrophic denitrifiers [9].The denitrifier could accept electrons from the cathode to reduce nitrate.The highest nitrogen removal was achieved in R2, indicating that at potential − 123 mV, both nitrification and denitrification processes reached the highest rate.Previous experiments also demonstrated that the nitrate removal rate could be strongly influenced by cathode potential [7].
At stage 2 (38-60 d), when a low concentration (10 µg/L) of PE-MP was added, it was observed that the remaining ammonium in the effluent of R2, R3, and R4 increased a bit.However, it was not significantly influenced in R1.Previous research has proved that exposure to microplastics in riverbanks could reduce ammonium removal [39].When it comes to the content of nitrate and nitrite, the behaviour was similar.The increasing amount of nitrate and nitrite represented that the denitrification process, which was enhanced by the cathode potential control, was blocked with the addition of PE-MP.When the content of PE-MP was reinforced (1000 µg/L) in stage3 (60-96 d), more profound influences on nitrogen removal were found in ammonium, nitrate, and nitrite (Fig. 1).Even in reactor R1, the ammonium removal declined.The same was true in R2-R4.The nitrification process was further restricted.The high content of microplastic would be toxic to the microbial communities, which could result in low microbial activities.One example is that high-density PE-MP has been reported to be toxic to some metabolic pathways and enzyme activities of microorganisms [40,41].Therefore, the activity of both nitrification and denitrification bacteria might be inhibited under high content of microplastic.Although the rapid hormonal response has been reported to promote the growth of some cells at the beginning under low-dose exposure, the high-dose microplastic could inhibit that [41].In the previous report, the presence of polyvinyl chloride microplastics could inhibit the electrochemical activities of the electroactive bacteria [42].Thus this could also hinder the biofilm bacteria from accepting the electron from the electrode.CV test was conducted at the end of stage 1 and stage 3 to evaluate the influence of microplastics on the electrochemical activity of the potential-controlled biofilm (Fig. S2).Although no clear reduction peaks were observed, a low current was found with the exposure of PE-MP for both the anode and cathode.The reduced electroactivity in the anode could decrease the removal of ammonium and, on the cathode, could shrink the denitrification process.It has been verified in MFC that the presence of microplastics could retard the electroactivity of Geobacter [42].

Biofilm viability and EPS matrix architecture
Biofilm viability was further investigated using CLSM at the end of stage 1 and stage 3 to evaluate the PE-MP effect.As shown in Fig. S3a, during stage 1, the cell viability was different under potential control.It was significant that with the stress of microplastic, the biofilm viability was enhanced in the anode of all the reactors.In our previous study, it has also been found that the low content of microplastics could stimulate the growth of anode biofilm [43].The ratio of dead cells to total cells was calculated further to assess the influence of PE-MP on biofilm viability.A significant increase in the dead/live cell ratio was observed in both anode and cathode at the end of stage 3, indicating the considerable inhibition of the microplastics to electrode biofilm viability (Fig. S3b, S3c).In the anode electrode biofilm, the dead/live cell ratio increased from 28.95-52.36%to 29.73-58.43%from stage 1 to stage 3.A similar and higher content of dead/live cell ratio could be found in the cathode of stage 3.The presence of the microplastic might kill some of the bacteria and would be harmful to microbiology activities.Previous studies also unravelled the toxicity effect of microplastics on microbial viability [16,44].There additive leached from the microplastic could be toxic to the microorganisms.The adverse effects of microplastics on marine microorganisms have also been reported [45].
EPS is the fundamental component of the biofilm and shrouds the cells.Therefore the electron transfer between the cells and the electrode has to flow through the EPS.Different properties of the EPS, including the structure and components, could result in various redox abilities [46].Thus the EPS of both the anode and cathode of the biofilm was visualized using CLSM at the end of stage 1 and stage 3.The total content of EPS from the anode electrode showed a decreased trend with the exposure of microplastics at the end of stage 3 compared to stage 1 (Fig. S4).The applied electricity could stimulate the secretion of the biofilm EPS when comparing the total content between R1 and other reactors.It could be found that the presence of PE-MP could suppress the content of the biofilm EPS on cathode biofilm (Fig. 2).Meanwhile, similar results could also be observed on the anode electrode biofilm (Fig. S4b).When we compared the EPS content of the anode and cathode, we found a higher EPS on the cathode biofilm (Fig. S4c).Total EPS intensity showed that the cathode potential control could stimulate the secretion of the biofilm EPS.It increased from 2.28 × 10 8 in R1 to 8.10 × 10 8 in R2, which was higher in R3 and R4 than in R1.EPS mainly include proteins, polysaccharides, and lipids, and they all have been reported to possess semi-conductive properties acting as ionic semiconductors [46].Besides, EPS could also serve as the carbon source supplying energy to the biofilm.Hence the high-content EPS biofilm possesses higher energy and a high electron transfer rate, leading to a high nitrogen removal efficiency.At the end of stage 3, a significant decline in EPS was observed in each reactor, indicating the suppression of microplastics in EPS secretion (Fig. 2b).Especially in R2, total EPS sharply shrank from 8.10 × 10 8 to 4.70 × 10 8 .Comparing the individual ingredients content of EPS, we found that each section would significantly decrease (Fig. 2c).Therefore, the biofilm EPS content of electroactive biofilm could also be inhibited by microplastic, leading to a low nitrogen removal efficiency.This result is consistent with the previous studies, which reported that exposure to microplastics could suppress the secretion of PES and inhibit the COD removal efficiency in the anaerobic digestion process [16].

Response of microbial community dynamics to PE-MP exposure
High throughput sequencing was applied to analyze the dynamics of the microbial community from the cathode biofilm under the effect of microplastics.Cathode biofilm was chosen to do the high throughput sequencing due to the cathode biofilm is the main research object under the cathode potential.Sobs index showed that the microbial richness from the samples of stage 3 was increased compared to that at stage 1 (Fig. 3a), indicating that the presence of PE-MP could escalate the community richness.Similar results could also be observed through rarefaction curves (Fig. S5).According to the Simpson index, microbial diversity also increased at stage 3, as microorganisms were more S. Wang et al. sensitive to the toxicant released from the microplastics.The increased microbial diversity has also been confirmed in SBR nitrogen removal reactors at a certain concentration of nano plastics [23].The high (more than 0.97) coverage of stage 1 and stage 3 suggested that the sequencing depth was enough to capture the bacteria diversity in this system.PCoA was executed based on OUT information to show differences in the microbial community between the two groups of stages (Fig. 3b).Results showed that the microbial communities were significantly (p < 0.05) separated between stage 1 and stage 3, demonstrating that the microplastics could shift the microbial community structure.Besides, the significant (p < 0.05) difference between stage 1 and stage 3 could also be observed through ANOSIM, and the difference inside stage 3 was higher than that in stage 1, indicating the shift of microplastics to the microbial community (Fig. S6).
To compare the variation of microbial community dynamics, the taxonomic composition of the microbial community at both phylum and genus level were further explored.At the phylum level, more than 90% of the OTUs were assigned to Proteobacteria (average relative abundance: 74.53%), Bacteroidota (11.97%),Chloroflexi (4.06%), Acidobacteriota (2.09%), and Desulfobacterota (1.70%) (Fig. S7).As the most predominant phyla, Proteobacteria lessen from 79.51 ± 6.03% at stage 1 to 69.56 ± 10.97% at stage 3 based on average relative abundance.Moreover, the average relative abundance of Bacteroidota increased from 8.51 ± 1.21% to 15.44 ± 4.93% with exposure to PE-MP at stage 3. Proteobacteria was widely found in the MEC hydrogenotrophic denitrifying biofilm [47].Bacteroidota was the most abundant phyla in bioelectrochemical autotrophic denitrifying bacteria at a low C/N ratio [48].
The distribution of the abundant genus was further estimated to show the variation of the abundant genus and their function (Fig. 4).Same with the composition at the phylum level, most of the abundant genera belong to the abundant phylum: Proteobacteria, Bacteroidota, Chloroflexi, Acidobacteriota, and Desulfobacterota.A total of three functions of the bacteria were identified in this bioelectrochemical nitrogen removal system.Specifically, functional genera involved in nitrogen cycling, including nitrification, denitrification, and nitrogen fixation, were identified and marked.It was noticeable that most of the abundant genera were denitrification bacteria.The primary reason     might be that the cathode potential control could enrich denitrification bacteria.Cathode could also supply electrons for the bioelectrochemical denitrification process [47].As the most abundant genera, Thiobacillus is well-known as a denitrification bacteria and could act as autotrophic denitrification bacteria [49].It has been reported that Thiobacillus could remove nitrate with high efficiency and complete nitrate reduction [50].At the same time, Thiobacillus is abundantly distributed in the bioelectrochemical denitrification process, and it could reduce nitrate through DIET with Geobacter [51].Other abundant genera, such as Acidovorax, Limnobacter, Denitratisoma, and Ignavibacterium were also reported as denitrification bacteria.To elucidate the genus abundance variation between stage 1 and stage 3 in the cathode biofilm, significant analyzes based on the relative abundance of different genera were performed (Fig. S8).Compared to stage 3, a distinct decrease was observed for the abundance of the genus Thiobacillus and f_Blastocatellaceae in stage 1.In addition, few nitrification and nitrogen fixation bacteria were observed in this cathode biofilm.Regarding the variation of total relative abundance of each separate function genus, the range of the denitrification bacteria was 67.22-90.42%(Fig. 4c).However, it is 0.27%-3.51%for nitrification bacteria and only 0.11%-0.76%for nitrogen fixation bacteria (Fig. 4b, 4d).The enrichment of the denitrification bacteria could be observed in the cathode potential controlled samples from stage 1.The abundance of that enriched from 77.79% in R1 to 83.03%, 90.42%, and 79.25% in R2, R3, and R4 with the cathode potential control.It is evident that with the addition of the PE-MP, the total abundance of the denitrification bacteria decreased in all reactors.It dropped from 77.79% to 67.22% in R1 and from 90.42% to 84.94% in R3.In this case, the presence of the PE-MP could inhibit nitrate removal in the bioelectrochemical nitrogen removal process by altering the microbial community composition.This could also respond to total nitrogen removal because of the suppression of the cathode denitrification process.It has been reported that microplastics could reduce nitrification and denitrification by changing the microbial-based nitrogen removal process [52].The ammonium conversion efficiency could also be inhibited under exposure to microplastics and nanoplastic [53].Microplastics could inhibit microbial community activity by releasing additives, producing ROS, and killing some microorganisms.This could also lead to the inhibition of microbial electrochemical nitrogen removal.

Functional genes variation
Functional genes involved in the denitrification process were quantified through real-time qPCR to further clarify the inhibition metabolism of nitrogen removal in this bioelectrochemical system.Here we only focus on the denitrification process because it has been enhanced with the cathode potential control.In the primary process of the biological denitrification process, nitrate is first converted to nitrite, followed by nitrogen monoxide, nitrous oxide, and nitrogen gas.Therefore, the representative genes for each pathway were retrieved, including narG, nirS, norB, and nosZ [47,49] (Fig. 5).The denitrification bacteria could capture electrons from the low content of organic compounds and the cathode electrode to perform nitrate reduction [47].At the same time, the genes-encoded enzymes could participate in the conversion process from nitrate to nitrogen gas.The nirS gene appeared abundantly in this system, which could be caused by the fact that nirS encoding denitrifiers were always the most well-known to be the complete denitrifiers [22].
The gene copies of narG ranged from 805.521 ± 134.600 to 3278.760 ± 418.620 copies/ng DNA in stage 1 and significantly higher than that under the stress of microplastics at stage 3. Regarding the nitrite reductases related gene, the abundance of the nirS gene sharply decreased at stage 3, showing the apparent inhibition of the microplastics to the nitrite reduction.Meanwhile, the copies of norB ranged from 658.764 ± 102.724 to 6700.345 ± 2640.654 copies/ng DNA at stage 1, and it was 110.256 ± 13.592 to 434.250 ± 82.660 copies/ng DNA at stage 3.The abundance of the nosZ gene was also higher at stage 1 than that in stage 3. Totally, the downregulated denitrification involved genes under the stress of the PE-MP could also clarify the inhibitory of the microplastics to nitrogen removal at the gene level.This decline might result from the decreased denitrification bacteria abundance (Fig. 4).In the previous research, the decreased nitrification genes and denitrification genes were observed in the sedimentary nitrogen cycling with the addition of polylactic acid microplastics [22].Besides, it has also been reported that nanoplastic at 1000 µg/L could reduce the denitrification-related genes in the SBR nitrogen removal process [23].The declined genes could respond to lower nitrogen metabolism activities inducing a low nitrogen removal efficiency.

Implications
The influence of PE-MP on the bioelectrochemical nitrogen removal process was demonstrated for the first time in this study.PE-MP could inhibit the nitrogen removal ability by affecting the EPS content, biofilm viability, microbial community structure, and functional genes.
Although we have revealed the potential mechanism underlying the impact of microplastics on bioelectrochemical nitrogen removal, several limitations and implications still need to be considered in the future.The observation in this study was based on a single type and exact size of microplastics.In the natural environment, various kinds of microplastics with a wide size range may have different effects on the performance of the bioelectrochemical systems.It has been reported that different types of microplastics could influence the nitrification and denitrification processes differently.For instance, polyurethane foam and polylactic acid microplastics could promote nitrification and denitrification processes, but polyvinyl chloride microplastics could inhibit both [22].Meanwhile, evaluating the concentration range that could influence the bioelectrochemical nitrogen removal is essential.Moreover, the additives leached from the microplastic degradation must also be assessed because they may disrupt bacterial growth and energy metabolism [54].Finally, to better evaluate the effect of microplastic on bioelectrochemical systems, the environmental risk and biological toxicity of microplastics should be considered.Thus, more research efforts need to be made in this field to provide more guidance for future practical applications.

Conclusions
This study reveals the inhibition effect of PE-MP on the bioelectrochemical nitrogen removal process.The microplastics at low (10 µg/L) and high (1000 µg/L) levels could inhibit the nitrogen removal capacity to a different extent.Adding PE-MP could significantly decrease the cell viability resulting in a higher dead cell ratio.Besides, an apparent decline of EPS content was also observed under the stress of microplastics at the end of stage 3 compared to stage 1.Microbial analysis based on PCoA showed that exposure to microplastics could significantly (p < 0.05) shift the microbial community structure.As the most abundant genera and the denitrification bacteria, Thiobacillus significantly shrank with a high concentration of microplastic addition.Simultaneously, the total relative abundance of denitrification bacteria also decreased at the end of stage 3 with PE-MP.Finally, the functional genes involved in denitrification were downgraded in all reactors with exposure to microplastics.This study profoundly explains how PE-MP could inhibit nitrogen removal in bioelectrochemical systems.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 2 .
Fig. 2. Distribution character of the cathode EPS.(a) Z-Stacked CLSM images of different sections of EPS; (b) total intensity of the EPS; (c) the distribution of individual EPS between stage 1 and stage 3.

S
.Wang et al.

Fig. 4 .
Fig. 4. The distribution of the functional genus.(a) Heatmap of the abundance of the individual abundant genus; (b) total relative of the nitrification bacteria; (c) total relative abundance of denitrification bacteria; (c) total relative abundance of nitrogen fixation bacteria.

S
.Wang et al.

Fig. 5 .
Fig. 5. Functional genes involved in the denitrification process and their abundance change at stage 1 and stage 3.

S
.Wang et al.