Chemical Engineering Journal 420 (2021) 127576

Available online 31 October 20201385-8947/© 2020 Elsevier B.V. All rights reserved.

Microbial and genetic responses of anammox process to the successive exposure of different antibiotics

Quan Zhang, Jing Wu, Ye-Ying Yu, Yi-Jun He, Yong Huang, Nian-Si Fan *, Bao-Cheng Huang, Ren-Cun Jin *

Laboratory of Water Pollution Remediation, School of Life and Environmental Sciences, Hangzhou Normal University, Hangzhou 311121, China

A R T I C L E I N F O

Keywords: Anammox Spiramycin Streptomycin ARGs Microbial community

A B S T R A C T

The issues of antibiotics and corresponding resistance genes (ARGs) have attracted public attention due to their environmental destructiveness. Herein, the response of the anammox process to the typical antibiotics strepto-mycin (STM) and spiramycin (SPM) was systematically explored. In terms of process performance, the anammox system was highly resistant to STM (50 mg L− 1), while 3 mg L− 1 SPM led to a significant inhibition. Exposure to STM resulted in the selective cultivation of Candidatus Kuenenia, while SPM significantly reduced its abundance. In addition, the relative abundance of bacteria belonging to Chloroflexi increased significantly during the SPM feeding phase, implying its potential resistance. The intI1 and corresponding ARGs (STM resistance genes: aadA, aadB, and aph3ib; SPM resistance genes: ereA, ermF and mphA) generally increased under stress from antibiotics. Co-occurrence network analysis showed that the collaboration among multiple ARGs might be the strategy used to relieve STM stress in the anammox system. This study systematically revealed the response of the anammox process to successive exposure to STM and SPM, and provides guidance for the treatment of wastewater multiple antibiotics containing.

1. Introduction

Environmental pollution related to antibiotics has received extensive attention for decades, especially for wastewater treatment. Antibiotics have been detected in many wastewaters, including wastewaters from municipalities [1], livestock [2], the pharmaceuticals industry and hospitals [3]. The concentrations of sulfonamides and macrolides in sewage sludge from wastewater treatment plants were observed to be 22.7 and 85.1 μg kg− 1 dry weight (dw), respectively [1]. Antibiotic concentrations in hospital and pharmaceutical wastewater have reached 103 and 104 μg L− 1, respectively [3]. The antibiotics in wastewater treatment systems could suppress the growth and activity of microor-ganisms, which poses challenges to the biological treatment perfor-mance. Moreover, as emerging pollutants, antibiotic-resistant bacteria (ARB) and antibiotic resistance genes (ARGs) derived from antibiotics cause harm to the environment that also cannot be ignored [4].

Compared with the traditional nitrification–denitrification process, the anaerobic ammonium oxidation (anammox) process has over-whelming advantages due to its high efficiency and economical nature, e.g., no organic carbon requirement, low sludge production, and low

oxygen consumption [5,6]. However, the sensitivity of anammox bac-teria (AnAOB) to environmental changes has restricted the application of anammox technology [7]. The inhibition of anammox by antibiotics has been widely reported. Regarding the short-term effects, Sguanci et al. (2017) showed that 50 mg L− 1 doxycycline reduced anammox activity by 22% within 48 h [8]. In the long-term, tetracycline (1.0 mg L− 1) could significantly deteriorate the process performance of anam-mox [2]. However, actual wastewater is often a complex matrix in which multiple antibiotics coexist. Thus, the establishment of an anammox process resistant to compound antibiotics is of great significance for expanding the application of this process. A previous study showed that the anammox process developed multiple resistances when exposed to oxytetracycline and sulfamethoxazole at the same time [9]. In addition, exploring the response of the anammox process to exposure to multiple antibiotics is helpful to optimize the process of switching operating modes for different types of antibiotic-containing wastewater in prac-tical applications.

In this study, two upflow anaerobic sludge blanket reactors were operated to investigate the response of the anammox process to alter-nating exposure to multiple antibiotics. Streptomycin (STM) and

* Corresponding authors. E-mail addresses: fns1907@163.com (N.-S. Fan), jrczju@aliyun.com (R.-C. Jin).

Contents lists available at ScienceDirect

Chemical Engineering Journal

journal homepage: www.elsevier.com/locate/cej

https://doi.org/10.1016/j.cej.2020.127576 Received 7 September 2020; Received in revised form 17 October 2020; Accepted 26 October 2020

Chemical Engineering Journal 420 (2021) 127576

2

spiramycin (SPM) were selected because they are widely used for disease treatment and exist in actual wastewater [10,11]. A previous study re-ported that STM had a strong inhibitory effect on the nitrifying microbes [10]. SPM could also deteriorate the performance of the biological ni-trogen removal system, while the increase of ARB and ARGs will further raise environmental risks [11]. However, few studies have explored their effects in anammox process. With the focus on process perfor-mance, the dynamic changes in the microbial community and ARGs were also thoroughly explored in this study. The primary objective of this study was to reveal the response of the anammox process to sequential exposure to multiple antibiotics and provide guidance for the treatment of antibiotic-containing wastewaters.

2. Materials and methods

2.1. Construction and operation of bioreactors

Two UASB reactors named R0 and R1 (effective volume of 1.5 L) were constructed (Fig. S1) and inoculated with 0.7 L mature anammox granules. R1 was used as the experimental group to explore the influence of antibiotics, while R0 was the control reactor without antibiotic addition. The inoculum was obtained from an anammox bioreactor that had been in stable operation for more than one year in our laboratory. The nitrogen loading rate (NLR) and nitrogen removal rate (NRR) of the inoculation reactor were 5.68 ± 0.4 and 5.38 ± 0.4 kg N m− 3 d− 1, respectively. The dominant bacteria was Candidatus Kuenenia [11]. The two reactors in this study were operated in thermostatic (35 ± 1 ◦C) and dark conditions and were continuously fed with synthetic wastewater using (NH4)2SO4 and NaNO2 as substrates. Besides, as shown in Table S1, trace elements and minerals were included in the synthetic wastewater. The hydraulic retention time of the reactors was set at 2.25 h, and the NLR was 5.97 kg N m− 3 d− 1, as previously described [11].

R0 was set as the control reactor, and R1 was the experimental reactor fed with STM (0–50 mg L− 1) and SPM (0–5 mg L− 1) successively; values were set according to a previous study and actual concentrations in wastewater [11,12]. The specific operating strategies of the reactors are shown in Table 1.

2.2. Collection and characterization of samples

The effluent and influent of the reactors were collected for routine analyses. The NH4

+-N, NO2–-N, NO3

–-N and pH of water samples were measured daily according to standard methods [13]. RS (consumption of NO2

–-N/NH4+-N) and RP (production of NO3

–-N/NH4+-N) were calculated

to evaluate the anammox reactor ratio. On the days 40, 65, 105, 125, 170, 195, 235 and 270, the sludge samples were withdrawn from each reactor to determine the sludge characteristics and related molecular biological indicators. Batch assays were used to determine the specific anammox activity (SAA) of anammox granules as previously described

[14]. Extracellular polymeric substances (EPS) were extracted by the ‘heating’ method, and the determination of polysaccharides (PS) and protein (PN) was performed using the modified Lowry and anthrone methods, respectively [15].

2.3. DNA extraction and gene quantification

The Power Soil DNA Kit (MoBio Laboratories, Carlsbad, CA) was used for DNA extraction from the collected samples. The quality and quantity of DNA were verified by NanoDrop-1000 (Thermo Scientific, USA) and stored at − 20 ◦C.

The resistance genes for STM (aadA, aadB, and aph3ib) and SPM (ermF, ereA, and mphA), intI1, and functional genes (hzsA and hdh) were quantified using quantitative polymerase chain reaction (qPCR). The corresponding primer pairs and annealing temperatures are listed in Table S2. Multiple 10-fold serial diluted plasmids containing the target genes were used to prepare the standard curve. Sterile water was used as a negative control, and each set of experiments was repeated in tripli-cate. To ensure a linear relationship over the concentration ranges used in this study, the r2 value of each standard curve exceeded 0.99.

2.4. 16S rRNA gene sequencing

High-throughput sequencing of 16S rRNA genes was performed for microbial community analysis as previously reported [16]. Briefly, the V3-V4 region of the 16S rRNA genes was amplified by the bacterial primers 338F and 806R [11]. Library construction and sequencing of amplicons for different samples were performed by Majorbio Bio-pharm Technology (Shanghai, China). QIIME software was used to analyze the sequencing results, including quality filtration and taxonomic classifi-cation [10]. A value of 97% was selected as the threshold for defining the operational taxonomic units (OTUs).

2.5. Statistical analysis

A t-test was performed to assess the significant differences in the results employing the Statistical Package for the Social Sciences 18.0 (SPSS Inc., USA). Statistical correlation analysis was implemented using R software (www.r-project.org) with the Psych package. According to a previous study [17], network analysis was performed to visualize the correlations among ARGs, functional genes, and bacterial taxa. Network visualization was conducted in the software Gephi (version 0.9.1). CANOCO 4.5 software (Microcomputer Power Co., USA) was used for principal component analysis (PCA) and the corresponding visualiza-tion. The visualization of other results was completed by OriginPro 9.1 (OriginLab, USA).

3. Results and discussion

3.1. Process performance of the UASB reactors

As Fig. 1 shows, R0 exhibited a relatively stable operation for 286 days with an average nitrogen removal rate (NRE) of 92.3 ± 0.8%, and the NO2

–-N concentration in the effluent remained at approximately 10 mg L− 1. After 40 days, 1 mg L− 1 STM was added to R1, which had no significant effect on process performance. Subsequently, the concen-tration of STM was increased to 50 mg L− 1, and R1 was repeatedly exposed to this high concentration of STM. Although high-concentration antibiotics have been reported to be toxic to anammox bacteria, the nitrogen removal performance of R1 remained stable. In phase 5, the NLR, NRR and NRE of R1 were 6.2 ± 0.9 and 5.6 ± 0.6 kg N m− 3 d− 1 and 91.6 ± 2.9%, respectively. These findings indicated that the anammox granules had a relatively high tolerance to STM stress, which also fills the gaps in our knowledge of the response of the anammox system to STM stress. The resistance of the anammox process to antibiotics has been widely reported in previous studies. In a study by Zhang et al.

Table 1 The overall operational strategy of R1 in this study.

Phase Duration (d) Concentration of antibiotics in R1

STM (mg L− 1) SPM (mg L− 1)

0 1–40 0 0 1 41–82 1 0 2 83–115 50 0 3 116–125 0 0 4 127–150 50 0 5 151–170 50 0 6 171–195 0 1 7 196–210 0 0 8 211–235 0 3 9 236–255 0 0 10 256–270 0 5 11 277–286 0 0

Q. Zhang et al.

Chemical Engineering Journal 420 (2021) 127576

3

(2018b) [18], the anammox process maintained a stable nitrogen removal performance under the long-term stress of gradually increasing the norfloxacin concentration to 50 mg L− 1. Another study also observed the long-term resistance of anammox process to erythromycin at con-centrations below 1 mg L− 1 [19]. Such tolerance benefits from the stable structure of granular sludge [20], and the physiological responses of AnAOB, such as adjusting the secretion of EPS and increasing the abundance of ARGs (as discussed below). In addition, the basic prop-erties of antibiotics may also contribute to the difference in response. STM and SPM belong to the aminoglycoside and macrolide antibiotics, respectively, with different molecular structure. SPM is poorly soluble in water, but soluble in methanol. Conversely, STM is freely soluble in water, and remains stable under weakly acid or neutral condition. The influent pH for anammox process is generally weekly alkaline [21], which could reduce the titer of STM and corresponding effects.

On day 170, the antibiotic in the influent was changed to SPM (1 mg L− 1). Although the dosage of SPM was much lower than that of STM, the performance of the anammox process degraded significantly. After 12 days of operation, the concentrations of NO2

–-N and NH4+-N in the

effluent of R1 reached 23.6 ± 6.6 and 50.1 ± 9.7 mg L− 1, respectively. Correspondingly, the NRE decreased to 72.6 ± 1.4%, indicating that anammox bacteria were inhibited by the SPM. The variations in the ratio of effluent NO2

–-N and NH4+-N reflected that other functional bacteria

involved in the nitrogen cycle within the mixed cultures. Interestingly, the process performance was restored under the same operating condi-tions, indicating the ability of the anammox process to adjust in response to antibiotic stress. Nevertheless, at the end of phase 6, the NRE of 87.9 ± 5.5% and NRR of 4.9 ± 0.9 kg N m− 3 d− 1 were still lower than those of the previous phase (NRE: 91.6 ± 3.0%; NRR: 5.8 ± 0.6 kg N m− 3 d− 1). This result suggested that the inhibitory effect of SPM on anammox might be higher than that of STM.

After a recovery period of 14 days in phase 7, the reactor was re- exposed to SPM (3 mg L− 1) on day 211. At the end of phase 8, the concentrations of NO2

–-N and NH4+-N in the effluent of R1 reached 102.1

± 15.8 and 96.5 ± 12.3 mg L− 1, respectively, and the NRE was 74.6 ±8.3%. This result showed that 3 mg L− 1 SPM had a significant impact on the process performance, which was consistent with a previous study. Research by Wu et al. (2020) indicated that 3 mg L− 1 SPM was the threshold for effects on the anammox process under long-term operation [11]. Subsequently, the NRE recovered to 90.2% after stopping the addition of SPM.

Increasing the concentration of SPM to 5 mg L− 1 resulted in a sig-nificant deterioration in anammox performance. On 270 day, the NRR and NRE of the reactor decreased to the lowest levels of 3.25 kg N m− 3

d− 1 and 53.08%, respectively. To avoid further deterioration, SPM dosing was interrupted at this time. It is worth noting that the process performance quickly recovered to the initial level within 14 days, accompanied by a rising NRE to 94.5 ± 1.3% at the end of phase 11. A previous study revealed that after inhibition by tetracycline, the anammox performance could rapidly recover [2]. Zhang et al. (2018b) reported that anammox performance could also be restored within 30 days after being stressed by norfloxacin [18]. The variations in the tolerance of AnAOB to different types of antibiotics may be a significant factor in determining the recovery time. Additionally, the rapid recovery of process performance in this study might be attributed to the multi- antibiotic resistance caused by the pre-exposure to STM in phases 1–5. The plasmid-borne multi-antibiotic resistance was mainly attributed to the horizontal transfer of ARGs, and the high-concentration antibiotic has been recognized as a main driving force [22]. Therefore, the multi- antibiotic resistance might be achieved under conditions of long-term exposure to different antibiotics. Li et al. (2010) have validated the multi-antibiotic resistance in bacteria in a wastewater treatment system

Fig. 1. Nitrogen removal performance of R0 (a) and R1 (b) during the whole period.

Q. Zhang et al.

Chemical Engineering Journal 420 (2021) 127576

4

suffering oxytetracycline [23]. The stoichiometric molar ratios RS (1.32) and RP (0.26) can indi-

rectly reflect the performance of anammox [24–26]. The RS and RP of R0 remained stable during 286 days of operation. Although the RS and RP of R1 fluctuated obviously, the fluctuation range was still close to the theoretical value. This result indicated that although the performance of the reactor was affected by antibiotic exposure, anammox was still the dominant nitrogen removal process.

3.2. Variation in granule characteristics

Changes in granular characteristics (EPS and SAA) were monitored during different phases. The relative SAA (rAA) was calculated accord-ing to Eq. (1).

rAA =SAA1

SAA0× 100% (1)

where SAA1 and SAA0 are the SAA of R1 and R0 in different phases. As the concentration of STM reached 50 mg L− 1, rAA dropped to 71.9%. Interestingly, rAA reached 347.9% after recovery in phase 3, and the SAA of R1 was significantly higher than that of R0 (t-test, p < 0.05), which indicated that exposure to antibiotics increased anammox activ-ity. Zhang et al. (2015b) observed that florfenicol exposure induced an increase in SAA in the UASB reactor [27]. The overcapacity of reactor could be achieved by previous exposure to severe conditions, which may contribute to relieving the side effects of following shock [28]. There-fore, the increase in SAA of R1 might be attributed to the overcapacity of microorganisms after the previous long-term inhibition. The rAA in phase 5 was 88.6%, implying tolerance in the anammox system. Under the subsequent addition of SPM, rAA remained relatively stable and reached 103.9% in phase 10. Under the same pressure of 5 mg L− 1 SPM, the SAA obtained in this study was 141.4 ± 8.6 mg N g− 1 VSS d− 1, which was much higher than that of 47.2 ± 14.1 mg N g− 1 VSS d− 1 reported by Wu et al. (2020) [11]. Therefore, the excellent SPM tolerance may be linked to the previous exposure to STM, which provides ideas for the establishment of a multi-antibiotic resistant anammox process.

In terms of EPS, the content of EPS in R0 was stable (Fig. S2). The first addition of SPM led to an obvious increase in EPS production in R1 (Fig. 2b). From phase 0 to phase 2, the content of PN and PS increased from 34.1 ± 1.3 and 6.3 ± 1.7 to 47.0 ± 5.9 and 12.4 ± 0.4 mg g VSS− 1, respectively. Regulating the secretion of EPS, the first barrier between microorganisms and the environment, is an important strategy of AnAOB to resist environmental stress [29]. On the one hand, increasing the secretion of EPS will delay the spread of poison to the cells. In an anammox system treating wastewater with a copper content of 50 mg L− 1, approximately 50% copper remained in the reactor, of which 8.97 mg L− 1 was trapped by EPS [30]. On the other hand, the secretion of EPS is conducive to the maintenance of the compact structure of granules [20]. The accumulation of bacteria in a specific area will trigger the operation of the quorum sensing system, and then the metabolic pathway of AnAOB will be adjusted to cope with environmental stress [29].

Subsequently, the content of EPS decreased when the STM concen-tration reached 50 mg L− 1 during phase 3–5. These results and the process performance together suggest that the anammox system adapted to the STM stress at this time, resulting in a decrease in EPS. AnAOB might be more inclined to alleviate STM inhibition by increasing the abundance of ARGs than by synthesizing EPS [19], because the synthesis and secretion of EPS require energy consumption. Moreover, excessive EPS is not conducive to mass transfer and substrate availability [31,32], which was probably the internal reason for the decrease in EPS content. Further exposure to SPM caused another increase in EPS content, with the highest value of 115.8 mg g− 1 VSS in phase 10 (Fig. 2b). This trend was also in line with the process performance, reflecting the adaptive process of the anammox system to a new environmental pressure of

SPM.

3.3. Dynamic changes in the microbial community

High-throughput sequencing of 16S rRNA revealed the dynamic changes in the microbial community in the anammox process under STM and SPM stress. Alpha diversity analysis showed that the presence of antibiotics reduced the diversity and richness of microorganisms. While the microbial diversity in R0 remained stable, the Chao1 and ACE index of R1 decreased from 420.2 and 423.0 to 319.6 and 340.7, respectively (Table S3).

At the phylum level, Planctomycetes, Chloroflexi, and Proteobacteria were the dominant phyla in the two reactors (Fig. 3a). Actinobacteria, Patescibacteria, Bacteroidetes and Acidobacteria also possessed a certain abundance. Notably, compared with phase 0 (38.29%), the relative abundance of Chloroflexi decreased to 13.61% in phase 5 and returned to 38.14% in phase 10, suggesting that Chloroflexi might be resistant to SPM. A recent study also observed the stability and high abundance of Chloroflexi under SPM pressure in the anammox system [11]. As shown in Fig. 3b, the relative abundance of Planctomycetes fluctuated during the STM loading phase (phases 0–5) but stabilized when exposed to SPM (phases 6–11). This phenomenon preliminarily indicated that anammox bacteria might have acquired multiple resis-tance under the conditions of alternating STM and SPM exposure.

Ca. Kuenenia was the dominant genus in all systems (Fig. 3b), and was also the only type of anammox bacteria detected in this study. First exposure to 50 mg L− 1 STM led to an increase in the relative abundance of Ca. Kuenenia from initial 23.40 to 46.60% at end of phase 2. After the

Fig. 2. Characteristics of anammox granules at different phases, including the relative specific anammox activity (rAA) (a) and the extracellular polymeric substances (EPS) (b). PN: protein; PS: polysaccharide.

Q. Zhang et al.

Chemical Engineering Journal 420 (2021) 127576

5

recovery, the relative abundance of Ca. Kuenenia remained at 36.26% when re-exposed to 50 mg L− 1 STM at phase 5, which was still signifi-cantly higher than the initial value and that of the control (p < 0.05). This phenomenon also suggested that Ca. Kuenenia might be more resistant than other bacteria in the system to STM. Other types of AnAOB exhibited a relatively high sensitivity to antibiotics. For example, Can-didatus Brocadia decreased from initially 0.45% to 0.33, 0.16, 0.01, 0.01, 0 and 0.01%, with the increasing concentration of erythromycin [33]. Subsequently, the abundance of Ca. Kuenenia decreased signifi-cantly due to the addition of SPM. This result was also consistent with the process performance under SPM exposure. Notably, the relative abundance of OLB13 increased from 1.98% in phase 5 to 29.60% in phase 8 with the addition of SPM. The change in OLB13 was the dominant factor causing the abundance increase in Chloroflexi. In addition to providing potential SPM resistance, the appearance of a heterotrophic environment also provided conditions for the growth of Chloroflexi. On the one hand, antibiotic stress could lead to increased secretion of EPS and the release of organic matter by the lysis of dead bacteria, which provides electron donors for reactions catalyzed by heterotrophic bacteria [34]. On the other hand, some enzymes in EPS may have decomposition effects on antibiotics [35], and the decom-posed products (organic compounds) may also promote the establish-ment of heterogeneous systems. Bacteria associated with AnAOB, norank_o_SBR1031, norank_f_A4B and SM1A02, have also been detected, but their function is still unclear.

PCA was used to further elucidate the influence of environmental

factors in R1 on changes in microbial communities. At the phylum level (Fig. 3c), the addition of STM and SPM led to clear separation along the PC2 vector (30.89%), which was mainly due to the proliferation of Chloroflexi, Bacteroidetes and Patescibacteria. Similar separation was also observed at the genus level (Fig. 3d), which represented 92.6% of the total microbial variability. Notably, Ca. Kuenenia showed a positive correlation with STM but a negative correlation with SPM. This result was also in line with the process performance. Additionally, PCA anal-ysis at the phylum and genus levels jointly indicated that the influence of SPM on the microbial community was greater than that of STM.

3.4. Changes in functional genes, ARGs, and intI1

The functional genes hzsA and hdh encode hydrazine synthase and hydrazine dehydrogenase, respectively, which have regulatory signifi-cance for the anammox pathway [36]. As shown in Fig. S3, the changes in two functional genes showed a similar trend during the experimental period. At phase 2, the abundances of hzsA and hdh increased, which was inconsistent with the process performance and the changes in AnAOB abundance. Slight reductions occurred at phase 6 and phase 8, mainly caused by the inhibition of SPM. Afterwards, the abundance of both genes gradually increased during the last three phases, which was different with the abundance of AnAOB. The possible reason was the expression level of functional genes decreased, meaning that the tran-scription process was affected by the high-concentration SPM. Gener-ally, neither STM nor SPM exposure led to significant changes in the

Fig. 3. Microbial community structure of anammox granules at the phylum level (a) and the genus level (b); Principal component analysis (PCA) on the composition of the microbial communities at the phylum (c) and genus (d) level.

Q. Zhang et al.

Chemical Engineering Journal 420 (2021) 127576

6

abundance of functional genes with variations over an order of magni-tude. To some extent, this result indicated that the metabolic process of AnAOB was not affected by the long-term effect of SPM or STM, which also provided a basis for the relatively stable process performance.

In terms of the ARGs, as shown in Fig. 4, exposure to 50 mg L− 1 STM resulted in an increase in the corresponding ARGs (aadA, aadB, and aph3ib) in R1. aadA had the most obvious change, increasing from 5.06 × 103 to 1.14 × 104 copies ng− 1 DNA from phase 0 to phase 2, indicating that resistant bacteria might preferentially express aadA in response to STM stress. When the addition of STM was stopped, the relative abun-dance of aadA decreased gradually. During the entire exposure to anti-biotics (STM and SPM), magnitude. Under the same operating conditions, differences in ARG abundance may be caused by different resistance mechanisms. Regulating the synthesis of nucleotidyl-transferases is the resistance mechanism of aadA and aadB, while aph3ib

is responsible for phosphotransferase [10]. Notably, although SPM was not added before phase 5, there was a slight increase in the abundance of ARGs related to SPM, which provided the basis for the hypothesis of multiple resistances proposed previously.

In phase 6, under the stress of 1 mg L− 1 SPM, the abundance of ereA, ermF, and mphA) tended to increase obviously (Fig. 5). The abundance of ereA, and mphA increased from 1.38 × 102 and 3.60 × 104 to 1.38 × 105

and 5.19 × 106 copies ng− 1 DNA, respectively. Notably, the abundance of mphA decreased significantly while that of ereA remained stable in phase 7 (without SPM addition), which suggested that mphA might be more susceptible to antibiotics in the environment due to a special resistance mechanism. By regulating the synthesis of oxidoreductase, mphA can chemically modify antibiotics to degrade them or replace the active groups [18,37]. The subsequent increase in SPM exposure con-centration did not cause significant changes in the abundance of related ARGs, which may be responsible for the inhibition of process perfor-mance at the end of the experiment.

As a gene encoding mobile genetic element, the presence of intI1 can induce the horizontal transfer of resistant genes, thereby increasing multiple resistance [38,39]. In this study, the variation in intI1 was similar to that in ARGs. The abundance of intI1 increased from 1.01 ×104 to 1.21 × 105 due to STM exposure (phases 0–2). Subsequently, exposure to 1 mg L− 1 SPM led to a significant increase in the abundance of intI1 (3.07 × 105), but the value returned to its initial level at the end of the experiment. Li et al. (2010) detected three types of ARGs in intI1, including aadA, aadB and aacA4, in isolated bacteria from an antibiotic- contaminated waterbody [23]. Therefore, integrators carry multiple resistance genes and can be transmitted between bacteria, which is the leading factor in the development of multiple resistance in bacteria.

3.5. Co-occurrence of ARGs and bacteria in anammox system

Co-occurrence network analysis was performed to explore the com-plex interactions among ARGs, functional genes, intI1 and bacteria. As shown in Fig. 6a, only Denitratisoma had an obvious interaction with AnAOB at the genus level, which was consistent with a previous study. As Denitratisoma is denitrifying bacteria, the interaction between Deni-tratisoma and AnAOB was understandable. On the one hand, the over- secretion of EPS from AnAOB under antibiotic stress has a potential feeding effect on denitrifying bacteria [9]. On the other hand, some secretory products of denitrifying bacteria can stimulate anammox ac-tivity [40]. Notably, the bacteria belonging to Chloroflexi (norank_f_A4b, norank_o_SBR1031 and OLB13) almost have abundance interactions, which may benefit filamentous biomass formation [41]. This is also conducive to shaping a stable granular sludge structure and maintaining an anaerobic environment, thereby optimizing the habitat of AnAOB [42].

In terms of ARGs and intI1, there were abundant interactions among the ARGs of STM (aadB, aadA and aph3ib), which suggested that the collaboration of multiple ARGs might be an important strategy for the anammox system to relieve STM stress. Fan et al. (2019) found a sig-nificant positive correlation between tetracycline resistance genes in anammox process treating tetracycline containing wastewater [2]. In contrast, there was no interaction between SPM resistance genes. However, it was noteworthy that there were the most abundant in-teractions between ereA and bacteria. This phenomenon suggested that the mechanism of esterification enzyme regulated by ereA might be the main protection strategy of anammox system in response to SPM. Be-sides, there was also a strong interaction between intI1 and ereA, and intI1 was associated with many bacteria including Planctomycetes, Patescibacteria and Chloroflexi. In fact, the effect of intI1 on prolifera-tion and transport of ereA has been widely reported. In Vibrio cholera, ereA was shown to be the gene cassette of intI1 [43]. The significant positive correlation between intI1 and ereA was also found in the anaerobic system for spiramycin wastewater treatment [44].

In addition to being related to ereA transport, intI1 plays a potential Fig. 4. Quantification results of streptomycin (STM) ARGs (aadA (a), aadB (b), and ahp3ib (c)) in R0 and R1.

Q. Zhang et al.

Chemical Engineering Journal 420 (2021) 127576

7

Fig. 5. Quantification results of spiramycin (SPM) ARGs (ereA (a), ermF (b), and mphA (c) and intI1 (d)) in R0 and R1.

Fig. 6. Co-occurrence network revealing the complex interaction between ARGs, functional genes, intI1 and dominant bacteria (a). The positive interactions between bacteria are indicated by thin gray curves; positive interactions between genes are indicated by thick red curves; positive interactions between genes and bacteria are indicated by thick blue curves. The size of each node is proportional to the number of connections (For an interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article). Correlation analysis between genes and bacteria (b).

Q. Zhang et al.

Chemical Engineering Journal 420 (2021) 127576

8

role in the transport of other ARGs. As shown in Fig. 6b, intI1 was positively correlated with almost all detected ARGs and bacteria, which again highlights the importance of intI1 in the transport of ARGs. Therefore, the tolerance of the anammox system to STM and SPM could be interpreted as follows. In terms of the physiological behavior of bacteria, excessive EPS secretion under antibiotic stress could protect bacteria (as discussed in section 3.2). Moreover, considering the mi-crobial community, potential resistant microorganisms also play a key role. The increases in the relative abundances of AnAOB and some Chloroflexi under STM and SPM exposure indicated potential resistance (section 3.3). Due to the high aggregation of microorganisms in granular sludge, the presence of resistant bacteria makes it possible to optimize the anammox system [29]. Finally, the interaction of multiple resistance mechanisms and the transfer of ARGs by intI1 provide the basis for the existence of multiple resistance in the anammox system.

3.6. Implications of this work

The establishment of antibiotic-resistant wastewater treatment sys-tems is imperative because of antibiotic contamination. In practical applications, it is possible to switch the working conditions of the pro-cess in response to different types of antibiotics. This is the first attempt to study the response of the anammox process to successive STM and SPM exposure. First, early STM exposure resulted in an increase in the abundance of SPM-related ARGS in the anammox system (section 2.4), which provided the possibility for active switching of process condi-tions. In addition, this study might help popularize the anammox pro-cess for the treatment of wastewater containing multiple antibiotics, at least the wastewater containing aminoglycoside and macrolide antibi-otics. However, the outcomes of this study indicate that multi-antibiotic resistance builds up during the long-term exposure to antibiotics. How to relief the side effects or even elimination of multi-antibiotic resistance and multi-resistant bacteria in this system urgently need to be addressed.

4. Conclusions

The anammox process had higher tolerance to STM than SPM. An SPM concentration of 3 mg L− 1 led to a significant inhibition of process performance, while the process performance was relatively stable under 50 mg L− 1 STM. The presence of antibiotics reduced the diversity and richness of microorganisms. Exposure to STM resulted in the accumu-lation of Ca. Kuenenia, while the abundance of Ca. Kuenenia decreased significantly due to the addition of SPM. Exposure to antibiotics, whether STM or SPM, induced an increase in the abundance of corre-sponding ARGs. Furthermore, co-occurrence network analysis showed that collaboration among multiple ARGs (aadA, aadB, and aph3ib) might be the strategy for the anammox system to relieve STM stress. All detected ARGs and most bacteria were positively correlated with intI1, highlighting the role of horizontal transfer in proliferation and implying multiple resistance of the anammox system.

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.

Acknowledgments

The authors are grateful for financial support from the Natural Sci-ence Foundation of Zhejiang Province (LQ20E080014), National Natu-ral Science Foundation of China (51878231), and the Science and Technology Development Program of Hangzhou (20191203B11).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi. org/10.1016/j.cej.2020.127576.

References

[1] W. Li, Y. Shi, L. Gao, J. Liu, Y. Cai, Occurrence, distribution and potential affecting factors of antibiotics in sewage sludge of wastewater treatment plants in China, Sci. Total Environ. 445–446 (2013) 306–313.

[2] N.-S. Fan, X.-L. Zhu, J. Wu, Z. Tian, Y.-H. Bai, B.-C. Huang, R.-C. Jin, Deciphering the microbial and genetic responses of anammox biogranules to the single and joint stress of zinc and tetracycline, Environ. Int. 132 (2019), 105097.

[3] M. De Cazes, R. Abejon, M.P. Belleville, J. Sanchez-Marcano, Membrane bioprocesses for pharmaceutical micropollutant removal from waters, Membranes (Basel) 4 (2014) 692–729.

[4] Q. Zhang, G. Tian, R. Jin, The occurrence, maintenance, and proliferation of antibiotic resistance genes (ARGs) in the environment: influencing factors, mechanisms, and elimination strategies, Appl. Microbiol. Biotechnol. 102 (2018) 8261–8274.

[5] B. Kartal, L. Van Niftrik, J. Rattray, J.L. Van De Vossenberg, M.C. Schmid, J. Sinninghe Damste, M.S. Jetten, M. Strous, Candidatu s ‘Brocadia fulgida’: an autofluorescent anaerobic ammonium oxidizing bacterium, FEMS Microbiol. Ecol. 63 (2008) 46–55.

[6] J.G. Kuenen, Anammox bacteria: from discovery to application, Nat. Rev. Microbiol. 6 (2008) 320.

[7] R.-C. Jin, G.-F. Yang, J.-J. Yu, P. Zheng, The inhibition of the Anammox process: a review, Chem. Eng. J. 197 (2012) 67–79.

[8] S. Sguanci, T. Lotti, C. Caretti, S. Caffaz, T. Dockhorn, C. Lubello, Inhibitory effects of veterinary antibiotics on anammox activity: short- and long-term tests, Environ. Technol. 38 (2017) 2661–2667.

[9] Q.-Q. Zhang, Y.-H. Bai, J. Wu, L.-Z.-J. Xu, W.-Q. Zhu, G.-M. Tian, P. Zheng, X.- Y. Xu, R.-C. Jin, Microbial community evolution and fate of antibiotic resistance genes in anammox process under oxytetracycline and sulfamethoxazole stresses, Bioresour. Technol. 293 (2019), 122096.

[10] Z. Tian, A. Palomo, H. Zhang, X. Luan, R. Liu, M. Awad, B.F. Smets, Y. Zhang, M. Yang, Minimum influent concentrations of oxytetracycline, streptomycin and spiramycin in selecting antibiotic resistance in biofilm type wastewater treatment systems, Sci. Total Environ. 720 (2020), 137531.

[11] J. Wu, N.-S. Fan, Y.-Y. Yu, Y.-J. He, Y.-H. Zhao, Q. Zhang, B.-C. Huang, R.-C. Jin, Insight into the microbial and genetic responses of anammox granules to spiramycin: comparison between two different dosing strategies, J. Cleaner Prod. 258 (2020), 120993.

[12] X. Luan, H. Zhang, Z. Tian, M. Yang, X. Wen, Y. Zhang, Microbial community functional structure in an aerobic biofilm reactor: impact of streptomycin and recovery, Chemosphere 255 (2020), 127032.

[13] W.E. Federation, A.P.H. Association, Standard methods for the examination of water and wastewater, American Public Health Association (APHA): Washington, DC, USA (2005).

[14] Z.-Z. Zhang, Q.-Q. Zhang, J.-J. Xu, R. Deng, Z.-Q. Ji, Y.-H. Wu, R.-C. Jin, Evaluation of the inhibitory effects of heavy metals on anammox activity: a batch test study, Bioresour. Technol. 200 (2016) 208–216.

[15] J. Wu, H.-M. Zhou, H.-Z. Li, P.-C. Zhang, J. Jiang, Impacts of hydrodynamic shear force on nucleation of flocculent sludge in anaerobic reactor, Water Res. 43 (2009) 3029–3036.

[16] Z.-Z. Zhang, H.-Y. Hu, J.-J. Xu, Z.-J. Shi, Y.-Y. Shen, M.-L. Shi, R.-C. Jin, Susceptibility, resistance and resilience of anammox biomass to nanoscale copper stress, Bioresour. Technol. 241 (2017) 35–43.

[17] X. Tang, Y. Guo, B. Jiang, S. Liu, Metagenomic approaches to understanding bacterial communication during the anammox reactor start-up, Water Res. 136 (2018) 95–103.

[18] X. Zhang, T. Chen, J. Zhang, H. Zhang, S. Zheng, Z. Chen, Y. Ma, Performance of the nitrogen removal, bioactivity and microbial community responded to elevated norfloxacin antibiotic in an Anammox biofilm system, Chemosphere 210 (2018) 1185–1192.

[19] Q.-Q. Zhang, Y.-H. Zhao, C.-J. Wang, Y.-H. Bai, D. Wu, J. Wu, G.-M. Tian, M.-L. Shi, Q. Mahmood, R.-C. Jin, Expression of the nirS, hzsA, and hdh genes and antibiotic resistance genes in response to recovery of anammox process inhibited by oxytetracycline, Sci. Total Environ. 681 (2019) 56–65.

[20] D. Kang, L. Guo, Q. Hu, D. Xu, T. Yu, Y. Li, Z. Zeng, W. Li, X. Shen, P. Zheng, Surface convexity of anammox granular sludge: Digital characterization, state indication and formation mechanism, Environ. Int. 131 (2019), 105017.

[21] Q. Zhang, X. Zhang, Y.-H. Bai, W.-J. Xia, S.-K. Ni, Q.-Y. Wu, N.-S. Fan, B.-C. Huang, R.-C. Jin, Exogenous extracellular polymeric substances as protective agents for the preservation of anammox granules, Sci. Total Environ. 747 (2020), 141464.

[22] Y. Wang, J. Lu, L. Mao, J. Li, Z. Yuan, P.L. Bond, J. Guo, Antiepileptic drug carbamazepine promotes horizontal transfer of plasmid-borne multi-antibiotic resistance genes within and across bacterial genera, ISME J. 13 (2018) 509–522.

[23] D. Li, T. Yu, Y. Zhang, M. Yang, Z. Li, M. Liu, R. Qi, Antibiotic resistance characteristics of environmental bacteria from an oxytetracycline production wastewater treatment plant and the receiving river, Appl. Environ. Microbiol. 76 (2010) 3444–3451.

Q. Zhang et al.

Chemical Engineering Journal 420 (2021) 127576

9

[24] R. Jin, B. Xing, J. Yu, T. Qin, S. Chen, The importance of the substrate ratio in the operation of the Anammox process in upflow biofilter, Ecol. Eng. 53 (2013) 130–137.

[25] C.-J. Tang, P. Zheng, C.-H. Wang, Q. Mahmood, J.-Q. Zhang, X.-G. Chen, L. Zhang, J.-W. Chen, Performance of high-loaded ANAMMOX UASB reactors containing granular sludge, Water Res. 45 (2011) 135–144.

[26] A.A. Van de Graaf, A. Mulder, P. de Bruijn, M.S. Jetten, L.A. Robertson, J. G. Kuenen, Anaerobic oxidation of ammonium is a biologically mediated process, Appl. Environ. Microbiol. 61 (1995) 1246.

[27] Z.-Z. Zhang, Q.-Q. Zhang, Q. Guo, Q.-Q. Chen, X.-Y. Jiang, R.-C. Jin, Anaerobic ammonium-oxidizing bacteria gain antibiotic resistance during long-term acclimatization, Bioresour. Technol. 192 (2015) 756–764.

[28] C. Reino, M.E. Suarez-Ojeda, J. Perez, J. Carrera, Stable long-term operation of an upflow anammox sludge bed reactor at mainstream conditions, Water Res. 128 (2018) 331–340.

[29] Q. Zhang, N.-S. Fan, J.-J. Fu, B.-C. Huang, R.-C. Jin, Role and application of quorum sensing in anaerobic ammonium oxidation (anammox) process: a review, Crit. Rev. Environ. Sci. Technol. 1–23 (2020).

[30] Y. Ma, D. Wei, X. Zhang, H. Fu, T. Chen, J. Jia, An innovative strategy for inducing Anammox from partial nitrification process in a membrane bioreactor, J. Hazard. Mater. 379 (2019), 120809.

[31] J. Li, P. Qi, Z. Qiang, H. Dong, D. Gao, D. Wang, Is anammox a promising treatment process for nitrogen removal from nitrogen-rich saline wastewater? Bioresour. Technol. 270 (2018) 722–731.

[32] Z.-Z. Zhang, Y.-F. Cheng, Y.-H. Zhou, X. Buayi, R.-C. Jin, A novel strategy for accelerating the recovery of an anammox reactor inhibited by copper(II): EDTA washing combined with biostimulation via low-intensity ultrasound, Chem. Eng. J. 279 (2015) 912–920.

[33] X. Zhang, Z. Chen, Y. Ma, T. Chen, J. Zhang, H. Zhang, S. Zheng, J. Jia, Impacts of erythromycin antibiotic on Anammox process: performance and microbial community structure, Biochem. Eng. J. 143 (2019) 1–8.

[34] Y. Sun, Y. Guan, H. Wang, G. Wu, Autotrophic nitrogen removal in combined nitritation and Anammox systems through intermittent aeration and possible microbial interactions by quorum sensing analysis, Bioresour. Technol. 272 (2019) 146–155.

[35] H.-Q. Yu, Molecular insights into extracellular polymeric substances in activated sludge, Environ. Sci. Technol. (2020).

[36] Y. Wang, X. Ma, S. Zhou, X. Lin, B. Ma, H.-D. Park, Y. Yan, Expression of the nirS, hzsA, and hdh genes in response to nitrite shock and recovery in Candidatus Kuenenia stuttgartiensis, Environ. Sci. Technol. 50 (2016) 6940–6947.

[37] Y. Zhu, T.A. Johnson, J. Su, M. Qiao, G.X. Guo, R.D. Stedtfeld, S.A. Hashsham, J. M. Tiedje, Diverse and abundant antibiotic resistance genes in Chinese swine farms, Proc. Natl. Acad. Sci. U.S.A. 110 (2013) 3435–3440.

[38] J.P. Brooks, A. Adeli, M.R. Mclaughlin, Microbial ecology, bacterial pathogens, and antibiotic resistant genes in swine manure wastewater as influenced by three swine management systems, Water Res. 57 (2014) 96–103.

[39] M.R. Gillings, W.H. Gaze, A. Pruden, K. Smalla, J.M. Tiedje, Y. Zhu, Using the class 1 integron-integrase gene as a proxy for anthropogenic pollution, ISME J. 9 (2015) 1269–1279.

[40] J. Guo, S. Wang, J. Lian, H.H. Ngo, W. Guo, Y. Liu, Y. Song, Rapid start-up of the anammox process: effects of five different sludge extracellular polymeric substances on the activity of anammox bacteria, Bioresour. Technol. 220 (2016) 641–646.

[41] A.D. Pereira, C.D. Leal, M.F. Dias, C. Etchebehere, C.A.L. Chernicharo, J.C. de Araújo, Effect of phenol on the nitrogen removal performance and microbial community structure and composition of an anammox reactor, Bioresour. Technol. 166 (2014) 103–111.

[42] L.-Z.-J. Xu, Q. Zhang, J.-J. Fu, J.-T. Zhang, Y.-H. Zhao, L.-Y. Jin, N.-S. Fan, B.- C. Huang, R.-C. Jin, Deciphering the microbial community and functional genes response of anammox sludge to sulfide stress, Bioresour. Technol. 302 (2020), 122885.

[43] M. Thungapathra, Amita, K.K. Sinha, S.R. Chaudhuri, P. Garg, T. Ramamurthy, G. B. Nair, A. Ghosh, Occurrence of antibiotic resistance gene cassettes aac(6′)-Ib, dfrA5, dfrA12, and ereA2 in Class I integrons in Non-O1, Non-O139 Vibrio cholerae strains in India, Antimicrob. Agents Chemother. 46 (2002) 2948–2955.

[44] M. Liu, R. Ding, Y. Zhang, Y. Gao, Z. Tian, T. Zhang, M. Yang, Abundance and distribution of Macrolide-Lincosamide-Streptogramin resistance genes in an anaerobic-aerobic system treating spiramycin production wastewater, Water Res. 63 (2014) 33–41.

Q. Zhang et al.