Does the biological treatment or membrane separation reduce the antibiotic resistance genes from swine wastewater through a sequencing-batch membrane bioreactor treatment process

Environment International 118 (2018) 274–281

Contents lists available at ScienceDirect

Environment International journal homepage: www.elsevier.com/locate/envint

Does the biological treatment or membrane separation reduce the antibiotic resistance genes from swine wastewater through a sequencing-batch membrane bioreactor treatment process

T

⁎

Qianwen Suia,b, Chao Jianga,b, Junya Zhanga,b, Dawei Yua,b, Meixue Chena,b, , Yawei Wanga,b, ⁎⁎ Yuansong Weia,b,c, a State Key Joint Laboratory of Environmental Simulation and Pollution Control, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China b Department of Water Pollution Control Technology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China c University of Chinese Academy of Sciences, Beijing 100049, China

A R T I C LE I N FO

A B S T R A C T

Keywords: Antibiotic resistance gene Mobile genetic element Sequencing-batch membrane bioreactor Solid retention time Swine wastewater

Swine wastes are the reservoir of antibiotic resistance genes (ARGs), which can potentially spread from swine farms to the environment. This study establishes a sequencing-batch membrane bioreactor (SMBR) for ARG removal from swine wastewater, and analyzes the effect of biological treatment and membrane separation on the ARG removal at different solid retention times (SRTs). The SMBR removed 2.91 logs (copy number) of ARGs at a short SRT (12 days). Raising the SRT reduced the removal rates of the detected genes by the biological treatment. Under the relative long SRT (30 days), ARGs and mobile genetic elements (MGEs) were maximized within the reactor and were well removed by membrane separation, with the average genes removal rate of 2.95 (copy number) and 1.18 logs (abundance). At the relatively low SRT, the biological treatment showed the dominant ARG removal effect, while the membrane separation took the advantages of ARG removal especially at the relatively long SRT. The ARG profile was related to the shift of the microbial community structure. The ARGs coexisted with the functional bacteria (ammonia oxidizing bacteria, nitrite oxidizing bacteria and denitrifiers), suggesting they are hosted by the functional bacteria.

1. Introduction According to a recent survey of antibiotics usage in China, animal production consumed 52% of China's total antibiotic consumption in 2013 (162,000 tons) (Zhang et al., 2015). Antibiotics are widely used in the livestock industry, as they not only prevent and cure disease, but also promote animal growth. However, antibiotic resistance is an emerging concern, and antibiotic resistance genes (ARGs) are recognized environmental pollutant (Pruden et al., 2006), originating from both hospitals and wider environments (Martínez, 2008). Swine wastewater is an important reservoir of ARGs. The ARGs in swine wastewater are disseminated to the adjacent environment through discharge and land application of the swine wastewater. The effluent quality can be improved by treating the swine wastewater in a membrane bioreactor (MBR). Previously, we established a

sequencing-batch MBR (SMBR) with high TN, COD and total bacteria removal efficiencies (Sui et al., 2017). Munir et al. (2011) reported a 1–3 logs higher reduction of effluent ARGs in an MBR than in a conventional treatment process. Membrane separation is another effective technique for reducing bacteria number (Harb and Hong, 2017). A membrane module augmented with dense membrane foulants facilitated the reduction of ARGs (Zhu et al., 2018). But owing to the relatively high density of bacterial cells, biofilm, and ARGs within the MBR, a high frequency of horizontal gene transfer (HGT) is observed in the mixed liquor of MBR (Yang et al., 2013). To facilitate ARGs removed by biological process and/or membrane separation, we must comprehensively analyze the ARGs occurrence in the mixed liquor, foulants and effluent. Regulating the solid retention time (SRT) is a common strategy to control the properties of mixed liquor and microbial community.

⁎ Correspondence to: M. Chen, Department of Water Pollution Control Technology, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China. ⁎⁎ Correspondence to: State Key Joint Laboratory of Environmental Simulation and Pollution Control, Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing 100085, China. E-mail addresses: [email protected] (M. Chen), [email protected] (Y. Wei).

https://doi.org/10.1016/j.envint.2018.06.008 Received 28 January 2018; Received in revised form 24 May 2018; Accepted 7 June 2018 0160-4120/ © 2018 Published by Elsevier Ltd.

Environment International 118 (2018) 274–281

Q. Sui et al.

centrifuged at 10000 rpm for 5 min.

Relatively short SRT (10–12 days) was applied to increase ammonia oxidizing bacteria (AOB) population while inhibiting nitrite oxidizing bacteria (NOB) population in order to enhance the TN removal rate in wastewater with low C/N ratio (Fitzgerald et al., 2015). Ma et al. (2011) reported that increasing SRT from 10 to 20 days significantly reduces the copy number of several genes (sul1, sul2, tetC, tetG and tetX) in mesophilic anaerobic digestion. Both the microbial community and environmental factors influence ARG occurrence (Zhang et al., 2015b). Nonetheless, how the SRT influences ARG occurrence in typical anoxic/ aerobic treatments used in swine farms for biological nutrient removal has not (to our knowledge) been clarified. In this study, a lab-scale SMBR was used to treat swine wastewater for the removal of ARGs. To separate the effects of ARG reduction by biological process and membrane separation, the ARG quantities in the mixed liquor, foulant and effluent were analyzed. The ARG occurrences in different regions of the SMBR were evaluated for various SRTs (12, 15 and 30 days). Especially, the impacts of the environmental factors, microbial community and HGT on the ARG occurrence were explored during the treatment of swine wastewater in the SMBR.

2.2. DNA extraction The liquid samples of influent (8 mL), supernatant (20 mL) and effluent (200 mL) were first filtered through 0.22-μm filters, and the remaining sludge samples (ca. 0.04 g-dry weight) were subjected to DNA extraction using the FAST DNA extraction Kit (MP Biomedicals, USA) according to the manufacturer's instructions. The extracted genomic DNA was detected and quantified by 1% agarose gel electrophoresis and NanoDrop 2000 (Thermo Scientific, USA), respectively, and then stored at −80 °C until further analysis. 2.3. Quantitative PCR (qPCR) ARGs of tetM, tetG, tetX (tetracycline resistance), ermB, ermF, mefA, ereA (macrolide resistance), sul1 and sul2 (sulfonamide resistance) and blaTEM (β-lactam resistance), as well as mobile genetic elements (MGEs) of intI1 (class 1 integron gene) Tn916/1545 (conjugative transposon Tn916/1545), ISCR1 (insertion sequence common region I gene), and total bacteria (16S rRNA) were quantified by qPCR. The primers and annealing temperature of the determined genes are listed in Table S2. The plasmids containing these specific genes were manufactured by Zhejiang Tianke Biotechnology Company (Zhejiang, China). The standard samples were diluted to yield a series of 10-fold concentrations and subsequently used for qPCR standard curves. The 25 μL PCR reaction mixtures contained 12.5 μL SYBR Green qPCR Super-Mix-UDG with Rox (Invitrogen, USA), 0.5 μL each of 10 μM forward and reverse primers, 10.5 μL DNA-free water, and 1.0 μL standard plasmid or DNA extract. The thermo cycling steps for qPCR amplification were as follows: (Ahmed et al., 2007) 50 °C, 2 min; (Berendonk et al., 2015) 95 °C, 5 min; (Bonfante & Anca, 2009) 95 °C, 20 s; (Caporaso et al., 2011) annealing temperature, 30 s; (Edgar et al., 2011) 72 °C, 31 s; (Fitzgerald et al., 2015) plate read, go to (Bonfante & Anca, 2009)~(Edgar et al., 2011), 39 more times; (Fu et al., 2017) Melt-curve analysis: 60 °C to 95 °C, 0.2 °C/read. The reaction was conducted using an ABI Real-time PCR system 7500 (ABI, USA). The specificity was assured by the melting curves and gel electrophoresis. Each gene was quantified in triplicate with a standard curve and negative control.

2. Materials and methods 2.1. Experimental design and sampling methods The SMBR (length × width × height = 260 mm × 260 mm × 450 mm; effective volume 30 L) was operated at ambient temperature (20–25 °C). The framework of the reactor and its operational methods are described in the Supplementary Information and our previous study (Sui et al., 2017). The raw swine wastewater was frequently taken from a confined swine farm (Beijing, China) with 5000 head capacity and stored in a cooler room at 4–6 °C. Each collected sample of raw swine wastewater could feed the reactor for 2 weeks to one month. The SMBR was implemented at three SRTs (30, 15 and 12 days) by daily wasting sludge from the reactor. The operational parameters, water quality of the influent and effluent are listed in Table 1, and antibiotic concentrations are shown in Table S1. The whole experiment continued for 262 days, in which the period of these three SRT treatments was operated for 123, 82 and 57 days, respectively. During each treatment period, triplicate samples were collected on days 44, 83, 110, 148, 175, 203, 223, 241 and 259. Samples were taken from the influent (Inf), the mixed liquor suspended solid (ML), the supernatant (Supn), the foulant attached on the membrane (Mem), and the effluent (Eff) of the SMBR. The seed sludge from a municipal wastewater treatment plant was named as S0. After the end of oxic phase, the mixed liquor sample was collected. The Supn was the supernatant after the mixed liquor sample settled for 30 min. The Mem sample was obtained by washing the membrane surface with deionized water, and then

2.4. High-throughput sequencing and bioinformation analysis The V4 region of the bacterial 16S ribosomal RNA gene was amplified by PCR (95 °C for 2 min, followed by 25 cycles of 95 °C for 30 s, 55 °C for 30 s, and 72 °C for 30 s and a final extension at 72 °C for 5 min) using the primers 515F/806R (Caporaso et al., 2011). Barcode at the reverse primer is an eight-base sequence unique to each sample. PCR reactions were performed in triplicate in 20 μL containing 4 μL of

Table 1 The operational parameters of the SMBR at three SRTs. Item

Parameter

SRT 30

SRT 15

SRT 12

Operational parameters

HRT (d) SRT (d) Duration time (d) MLSS (mg/L) COD load (kgCOD/kgVSS d) TN load (kgTN/kgVSS d) COD (mg/L) TN (mg/L) NH4+-N (mg/L) COD (mg/L) TN (mg/L) NH4+-N (mg/L) NO2−-N (mg/L) NO3−-N (mg/L)

5–6 30 123 11,745 ± 690 0.15 ± 0.01 0.02 ± 0.00 7737 ± 899 1116 ± 91 822.3 ± 38.1 392 ± 102 88 ± 25 13.7 ± 6.3 16.9 ± 0.6 3.5 ± 0.6

5–6 15 82 12,508 ± 753 0.13 ± 0.01 0.02 ± 0.00 7118 ± 313 1027 ± 110 831.9 ± 99.0 317 ± 42 95.6 ± 26 22.9 ± 10.3 12.4 ± 5.7 12.2 ± 11.25

4–5 12 57 10,592 ± 1438 0.14 ± 0.02 0.02 ± 0.00 5453 ± 682 874 ± 125 760.7 ± 102.4 209 ± 49 45 ± 9 3.5 ± 0.4 13.4 ± 1.6 2.3 ± 0.7

Influent

Effluent

275

Environment International 118 (2018) 274–281

Q. Sui et al.

5 × FastPfu Buffer, 2 μL of 2.5 mM dNTPs, 0.8 μL of each primer (5 μM), 0.4 μL of FastPfu Polymerase, and 10 ng of template DNA. Amplicons were extracted from 2% agarose gels and purified using the AxyPrep DNA Gel Extraction Kit (Axygen, USA) according to the manufacturer's instructions and quantified using QuantiFluor™ -ST (Promega, U.S.). Purified amplicons were pooled in equimolar and paired-end sequenced (2 × 250) on an Illumina MiSeq platform (Illumina, USA) according to the standard protocols of Sangon Co., Ltd., China. Sequencing reads were assigned to each sample according to the unique barcode of each sample. Pairs of reads from the original DNA fragments were firstly merged using FLASH (Magoc and Salzberg, 2011), and then PRINSEQ was used for the quality control of the merged reads (Schmieder and Edwards, 2011). The barcode and primers then were removed. All reads were further uploaded to MG-RAST (http://metagenomics.anl.gov/linkin.cgi?project=19384). PCR chimeras were filtered out using UCHIME (Edgar et al., 2011). After the above filtration, the average sequencing depth was ca. 30,000 clean reads. The taxonomic classification of the sequences in each sample was conducted using the Ribosomal Database Project (RDP) Classifier as previously suggested (Zhang et al., 2015a). The sequences were assigned to different taxonomy levels at the bootstrap cutoff of 50% suggested by the RDP (Wang et al., 2007). The putative pathogenic bacteria and denitrifiers were estimated according to our previous study (Sui et al., 2017).

Meanwhile, frequent HGT may occur in the reactor under the SRT of 30 days. As shown in Fig. 1(b), the quantities of MGEs (intI1, ISCR1 and Tn916/1545) were the higher at the SRT of 30 days than that at the other two SRTs. Yang et al. (2013) demonstrated that the possible occurrence of horizontal ARG transfer in an MBR, which combined a high density of cells, biofilms, and ARGs. Comparing the copy numbers between the Inf and Supn samples, and between the Supn and Eff samples, we can visualize the ARGs and MGEs removed by biological treatment and membrane separation, respectively. The results are shown in Fig. 2. According to influent versus effluent comparison, the SMBR severely reduced the all copy numbers of all detected genes, but did not obvious negate the abundance. Through the biological treatment, comparing the influent and supernatant, we find that raising the SRT linearly decreased the removal rate of ARGs in terms of both copy number and abundance (see Fig. S1). As the SRT increased, the copy number reduced more rapidly than the abundance. The ARG copy numbers reduced by biological process were on the average of 1.68 ± 2.27, 1.09 ± 2.16 and −1.16 ± 1.29 logs at the SRT of 12, 15 and 30, respectively. And the average abundances removed by biological process were in corresponding to 0.91 ± 2.03, 0.82 ± 2.10 and −0.45 ± 0.87 logs, respectively. Membrane separation achieved relatively high ARG and MGE removal rates with average copy number reduction of 1.83 ± 1.00 and 1.33 ± 1.15 logs, respectively, and no obvious dependence on SRT treatment. However, the membrane separation performed excellently on 30-SRT supernatant, despite the remarkable gene increase in this sample. All genes in the 30-day SRT supernatant were greatly reduced by membrane separation, with the average copy-number and abundance removal rates of 2.95 ± 1.84 and 1.18 ± 0.76 logs, respectively. The supernatant of the 12-day SRT was relatively abundant in tetG and ereA. This result is inconsistent with our previous study of on- site investigation, in which the tetG and ereA abundance were evaluated after the biological wastewater treatment process (anaerobic digestion, lagoon, and SBR) (Sui et al., 2015). Interestingly, ermB, ermF and blaTEM abundances were significantly correlated (p < 0.05), oppositely to the other genes, the copy number of these genes decreased dramatically in the supernatant and then increased in the effluent, especially at SRT of 15 and 12 days. It indicated that the host bacteria of ermB, ermF and blaTEM were probably at relatively long doubling times, and the host bacteria and the extracellular genes of ermB, ermF and blaTEM were greatly diminished and discharged in the waste sludge under the relatively short SRT. The predominance of ermB, ermF and blaTEM in waste sludge was confirmed by Zhang et al. (2015b). The ARG composition in the MBR effluent largely differed from those in the influent and supernatant, probably reflecting the shift in water quality from eutropher to oligotropher. The microbial community further changes after membrane separation, as analyzed in Sections 3.3 and 3.4. Harb and Hong (2017) examined the performance of two MBR systems treating municipal wastewater. They reported a wide range of bacterial reduction (< 2 to > 5 logs) in both system. In our previous study, the total bacteria were reduced by 2.77 logs between the influent and effluent of the SMBR. The membrane separation largely contributed to this reduction, with a removal amount of 1.83 logs (Sui et al., 2017). Despite the small pore size of the membrane (0.1 μm) in the present study, the ARGs were not completely reduced in the effluent, probably because the extracellular DNA could pass through the filter. Another possible reason is reproduction of the bacteria outside of the membrane in the relatively low nutrient conditions of the effluent.

2.5. Statistical analysis The abundances of ARGs and mobile genetic elements (MGEs) were reported as the relative concentrations normalized to the copy number of 16S rRNA. Statistical calculations and data analysis were performed in the SPSS 20 statistical software package (IBM, USA), and the figures were plotted by OriginPro 9.0 (OriginLab, USA). The associations between the operating variables and the microbial and biochemical parameters were assessed by Spearman rank correlations. Statistical significance was defined as p < 0.05. Principal component analysis (PCA), redundancy analysis (RDA) and partial RDA were performed in Canoco 5.0 (Microcomputer Power, USA). Based on the Spearman results, the potential bacterial interactions in the SMBR were evaluated in a network analysis implemented in the Gephi platform (Bonfante and Anca, 2009). 3. Results and discussion 3.1. ARG and MGE occurrences in the liquid samples The copy number and abundance of ARGs and MGEs in the influent (Inf), supernatant (Supn) and effluent (Eff) are shown in Fig. 1. The highest removal rate of ARG copy numbers was occurred at the relatively short SRT (12 days). The ARG copy numbers were reduced by1.56, 0.96 and 2.91 logs, and the abundances were reduced by 0.40, −0.40 and −0.04 logs, at SRT of 30, 15 and 12 days, respectively. Although the ARG concentration fluctuated in the influent of the raw swine wastewater, the SMBR largely reduced the absolute copy numbers of ARGs and MGEs. During the experimental periods, the copy numbers of ARGs and MGEs in supernatant were highly increased compared to those of the influent and effluent at the SRT of 30 days. Reducing the SRT to 15 and 12 days successfully reduced the ARG and MGE copy numbers in both supernatant and effluent. The relatively long SRT (30 days) yield a relatively high ARG copy numbers in supernatant. Bacteria may be retained and/or proliferated in the reactor, as indicated by the copy numbers of 16S rRNA in the Supn and ML samples of 30-daySRT (Fig.1(b)). Though the total bacteria number was relatively low in the influent, it was remarkably increased by 0.72 logs in the reactor probably because the relative long biomass retention time allows the bacteria to proliferate to high levels.

3.2. ARG and MGE occurrences in the sludge samples The copy numbers and abundances of ARGs and MGEs in sludge samples, including mixed liquor (ML) and the foulant attached to the membrane (Mem) are presented in Fig. 3. The copy numbers and 276

Environment International 118 (2018) 274–281

Q. Sui et al. 1E+10

ARGs (copies/mL)

1E+09

0.09

0.77

1E+08

0.4

0.60

2.32

0.05

ereA mefA ermF ermB sul2 sul1 tetX tetG tetM blaTEM

0.11

0.87

ARGs/16S rRNA

a

2.31

1E+07 1E+06

0.3

0.15 0.35

0.2

0.22 0.18

1E+05 0.1 1E+04 1E+03

0.0

Inf(30) Supn(30) Eff(30) Inf(15) Supn(15) Eff(15) Inf(12) Supn(12) Eff(12)

1E+13

0.38

1E+07

1E+07

1.30

0.35 0.61

2.62

1E+04

0.77

1E+06

1E+01

1E+05

1E-02

1E+04

0.14

1E+10

1E+08

Tn916/1545 ISCR1 intI1 16S rRNA

0.16

MGEs/16S rRNA

MGEs (copies/mL)

1E+09

16S rRNA (copies/mL)

b

Inf(30) Supn(30) Eff(30) Inf(15) Supn(15) Eff(15) Inf(12) Supn(12) Eff(12)

0.12 0.10 0.08 0.60

0.06 0.04

Inf(30) Supn(30) Eff(30) Inf(15) Supn(15) Eff(15) Inf(12) Supn(12) Eff(12)

0.37

0.56 0.48

0.02

1E-05

0.00

0.65

0.11

Inf(30) Supn(30) Eff(30) Inf(15) Supn(15) Eff(15) Inf(12) Supn(12) Eff(12)

Fig. 1. The copy number and abundance of ARGs and MGEs in influent, supernatant and effluent (a ARGs, b MGEs; Inf influent, Supn supernatant, Eff effluent; the arrow indicates the log10 changes of ARGs and MGEs).

a Inf-Eff(12)

Inf-Supn(12)

Supn-Eff(12)

Inf-Eff(15)

Inf-Supn(15)

Supn-Eff(15)

Inf-Eff(30)

Inf-Supn(30)

Supn-Eff(30)

-2

0

2

4

-4

-2

Absolute reduction (copies/mL)

0

2

4

6

16S rRNA Tn916/1545 ISCR1 intI1 ereA mefA ermF ermB sul2 sul1 tetX tetG tetM blaTEM -6

8

Absolute reduction (copies/mL)

-4

-2

0

2

4

6

Absolute reduction (copies/mL)

b Inf-Eff(12)

Inf-Supn(12)

Supn-Eff(12)

Inf-Eff(15)

Inf-Supn(15)

Supn-Eff(15)

Inf-Eff(30)

Inf-Supn(30)

Supn-Eff(30)

-2

-1

0

1

2

3

Relative reduction (genes/16S rRNA)

-2

0

2

4

6

Relative reduction (genes/16S rRNA)

-8

Tn916/1545 ISCR1 intI1 ereA mefA ermF ermB sul2 sul1 tetX tetG tetM blaTEM -6

-4

-2

0

2

4

Relative reduction (genes/16S rRNA)

Fig. 2. The reduction of genes absolute copy number and relative abundance between influent, supernatant and effluent (a absolute copy number reduction; b relative abundance reduction; Inf influent; Supn supernatant; Eff effluent).

trend was influenced by the property of ML samples. However, the copy numbers of ARGs and MGEs were lower in Mem samples than in the ML at the SRT of 30 days. During filtration, a cake layer formed on the surface of the membrane, and the extracellular polymeric substance (EPS) was in relation to the formation of cake layer (Ahmed et al., 2007). The cake layer and the attached EPS on the membrane surface influenced the property of the ARGs in Mem. Wu et al. (2011) reported that shortening SRT from 60 to 20 days increased the EPS production.

abundances of the ARGs were of the same magnitude in the ML and Mem, but were relatively high at the SRT of 30 days. In the ML samples, both the ARG copy numbers and abundances were higher at SRT of 30 days than at the SRT of 12 and 15 days. The MGEs were also highest at SRT of 30 days and exhibited the same property as the detected ARGs. Meanwhile the quantity of 16S rRNA in the ML was slightly higher at SRT of 30 days than at the other SRTs(see Figs. 1(b) & 3(b)). Shortening the SRT decreased the abundance of MGEs in Mem. This 277

Environment International 118 (2018) 274–281

Q. Sui et al.

a

0.5

4.0E+11

ARGs/16S rRNA

9E+10 8E+10 7E+10 6E+10 5E+10 4E+10 3E+10 2E+10 1E+10

0.3

0.2

0.1

ML(30)

ML(15)

ML(12)

Mem(30) Mem(15) Mem(12)

0.0

S0

b

1E+06 1E+03

1E+08

MGEs/16S rRNA

1E+09

ML(15)

ML(12)

Mem(30) Mem(15) Mem(12)

S0

Tn916/1545 ISCR1 intI1

0.012

1E+12 1E+09

ML(30)

0.014

1E+15

1E+10

MGE (copies/g-TS)

ereA mefA ermF ermB sul2 sul1 tetX tetG tetM blaTEM

0.4

3.0E+11

16S rRNA (copies/g-TS)

ARGs (copies/g-TS)

3.5E+11

1E+00

16S rRNA

0.010 0.008 0.006 0.004 0.002

1E+07 ML(30)

ML(15)

ML(12)

Mem(30) Mem(15) Mem(12)

S0

1E-03

0.000

ML(30)

ML(15)

ML(12)

Mem(30) Mem(15) Mem(12)

S0

Fig. 3. The copy number and abundance of ARGs and MGEs in mixed liquid and foulant attached to the membrane (a ARGs, b MGEs; ML mixed liquid, Mem foulant attached to membrane surface).

hydrolysis and biodegradation process in anaerobic digestion, the SMBR alters between anoxic and oxic environment, facilitating the grow and proliferate of facultative bacteria and aerobic bacteria. A long SRT enables proliferation of bacteria with a wide range of doubling time, increasing the richness and diversity of the microbial community. Additionally, the membrane played an important role in solid-liquid separation, increasing concentration of biomass (planktonic and biofilm states) within the reactor. In turn, the dense biomass could increase the frequency of HGT. The copy number and abundance of ARGs and MGEs were higher in the sludge of the SMBR (ML samples) than those in seed sludge (S0). Because the swine wastewater is an ARG reservoir (Berendonk et al., 2015), with long-term operation of the SMBR, antibiotic resistance bacteria were retained and proliferated in the reactor, meanwhile the HGT incorporated intracellular and extracellular ARGs could transfer through different species.

Meanwhile, Fu et al. (2017) reported that lengthening the SRT increased the MLSS concentration, but lowed the cake layer resistance, and the higher MLSS did not significantly contribute to the faster formation of cake layer. As shown in Fig. 3, the copy numbers of ARGs and MGEs were relatively low in the Mem samples at SRTs of 30 days, possibly reflecting the lower quantity of cake layer formed on membrane surface. The ARG abundances in Mem sample at SRT of 30 days were relatively high, possibly because the high MGE abundance increased the HGT. The foulant attached to the membrane surface may assist the emergence and dissemination of ARGs through HGT. The HGT would be promoted by polymicrobial nature of biofilms and close proximity between species (Olsen, 2015). The highly hydrate matrix favors the transfer and natural transformation of extracellular DNA (Madsen et al., 2012). At the SRT of 30 days, the Mem sample was relatively rich in both ARGs and MGEs, indicating frequent occurrences of HGT. Among the detected MGEs, the class 1 integrase gene (intI1) and conjugative transposon gene (Tn916–1545) were more prevailing than insertion sequencing (ISCR1). Integron and insertion sequencing are known to capture ARGs, and conjugative transposon and plasmid could transfer genes between microbes through cell-to-cell contact (Zhang et al., 2015b). The ARG abundances were higher in ML and Supn than in Mem. Specifically, the ARG abundances of the ML, Supn and Mem samples were 44.0%, 32.3% and 30.0% respectivly at SRT of 30 days, 9.3%, 12.4% and 5.7% respectively at SRT of 15 days, and 20.6%, 31.3% and 12.0% respectively at SRT of 12 days. The mixed liquor was the hot spot of ARGs in the MBR. For a given SRT, the ARG properties were similar in the ML and Supn, and were higher in suspended activated sludge than in the foulant, which may be caused by the high MLSS concentration in the SMBR. The SRT greatly influenced the contents of the Supn, ML and Mem within the reactor. At SRT of 30 days, the Supn, ML and Mem samples were relatively abundant in ARGs and MGEs. Ma et al. (2011) indicated that lengthening the SRT enhances the removal of the absolute ARG copies during anaerobic digestion. However, the results of the oxic SMBR in the present study contradict these findings. Unlike the

3.3. Effect of SRT on the microbial community The shift of microbial community composition on phylum level of different SRTs is shown in Fig. S2. The influent was relatively rich in Proteobacteria, Bacteroidetes and Fimicutes, while the effluent was dominated by Proteobacteria. The ML, Mem and Supn samples showed similar properties, and were mainly composed of Proteobacteria, Bacteroidetes, Candidatus Saccharibacteria, and Actinobacteria. Bacteroidetes and Firmicutes are highly abound in animal intestines and feces (Niu et al., 2015). The microbial community of the activated sludge was influenced by the swine wastewater and various controlling parameters. The microbes retained and survived in the SMBR, and the bacterial community shift was likely the result of adaptation to the controlling parameters. The shift of microbial community on genus level was investigated in a PCA. As shown in Fig. 4, samples from different SRT treatments clustered in separate groups, and the effluent samples clustered apart from the others. The Supn, ML, and Mem with similar properties clustered into same SRT groups and were influenced by treatment process and controlling parameters. The profiles of the effluent microbial 278

Environment International 118 (2018) 274–281

Q. Sui et al.

Fig. 4. Principal component analysis (PCA) of microbial community based on the top 10 abundant genera of different SRTs.

The network is divided into two parts. Two groups of ARGs,tetX, sul2, ereA, and ermF, blaTEM and ermB, are distributed in the lower part of the diagram are closely correlated with Nitrosomonas. The MGEs are also distributed around the Nitrosomonas. In the upper part of the diagram, two ARGs (tetG and sul1) are positively and closely correlated with NOB (Nitrobacter and Nitrospira). Some of the denitrifiers (Thauera, Azoacus, and Pseudochrobactrum) are also closely correlated with AOB, and most of denitrifiers distributed in the upper part of the network. The putative bacteria are evenly distributed around the functional bacteria. Under the SRT control, it can be observed that co-occurrence of functional bacteria and ARGs, suggesting that when the SRT was regulated, the host bacteria of ARGs were responded to the shifting composition of functional bacteria. The potential of ARGs carried by functional bacteria, and/or metabolism regulation during the biological treatment should be further analyzed.

communities differed from other samples, and were also altered by the SRT. The effluents at SRT of 30 days distributed far from those at the 12-day and 15-day SRT. Considering microbial community shift, it was suggested that the effluent characteristics totally differ from the characteristics of the samples within the reactor, and were less influenced by the SRTs. The effluent was physical filtrated by membrane module and was subjected to the adverse environment factors (such as low carbon and nutrient concentration, and low bacteria population). The clustering of the samples inside and outside the reactor under the three SRT conditions suggests that ARG profiles were supported by the shifts in microbial community compositions. 3.4. The impacting factors on ARG evolution The impacting factors on ARG abundance evolution are shown in both Figs. S3 and S4. The ARG evolution is influenced by the water quality parameters, controlling parameters, microbial community and the MGEs. Through redundancy analysis (RDA), the analyzed impacting factors collectedly accounted for 82.0% of ARG variation. In the partial RDA, the environmental factors (SRT, SS, Temp, COD, NH4+-N, NO2−N, and NO3−-N) explained 61.4% of ARG variation, the microbial community structure (Proteobacteria, Bacteroidetes, Actinobacteria, Firmicutes, candidate division WPS-2, Candidatus Saccharibacteria, Chloroflexi, Verrucomicrobia) explained 51.7% of the variation, and MGEs (intI1, ISCR1, Tn916–1545) explained 31.8% of the variation. SRT was the most important factor, contributing 11.6% to the ARG variation. The feed explained 1.5% of ARGs variances in the Mem, ML, Supn and Eff samples. The selective pressures of antibiotics on ARGs were determined in a Spearman correlation analysis (see Table S3). The strong antibiotic-antibiotic correlations indicated their synergic removal during the wastewater treatment. However, the antibiotics and ARGs were weakly correlated, with one exception: mefA was significantly correlated with new spiramycin (p < 0.01). Fig. 5 is a network of interactions between ARGs and their potential host bacteria, determined by Spearman correlation analysis. The ARGs were significantly (p < 0.05) and positively correlated with the functional bacteria (AOB, NOB and denitrifiers) and putative pathogenic bacteria in the ML, Mem, Supn samples within the SMBR. The correlation between the ARGs and functional bacteria is probably due to the controlled SRT.

4. Conclusions An SMBR was established for ARG removal from swine wastewater. To better understand the effects of biological treatment and membrane separation on ARG removal, the mixed liquor, membrane foulant and effluent were analyzed in the samples from the SMBR. The SRT was varied as 30, 15 and 12 days. The 12-day SRT condition maximized the ARG removal, with 2.91 logs removal on ARG copies. Through the biological treatment, the removal rates of the detected genes were decreased with the increase of the SRT in both copy number and abundance. The membrane separation showed good removal performance especially at SRT of 30 days. The copy number and abundance of ARGs and MGEs of the samples inside the reactor were higher at the 30-day SRT than at the 15- and 12day SRT conditions, indicating the proliferation and/or horizontal gene transfer of ARGs in the SMBR. The ARG profiles were highly associated with the microbial community structure. The ARGs co-occurred with functional bacteria (AOB, NOB and denitrifiers) under the SRT control, indicating that functional bacteria can host ARGs. The relationship required further analysis in the future work. Author contributions The manuscript was written through contributions of all authors. All 279

Environment International 118 (2018) 274–281

Q. Sui et al.

Fig. 5. Network analysis showing the significant (p < 0.05) and positive correlation through Spearman correlation between ARGs and involved bacteria including AOB, NOB, denitrifiers and putative pathogenic bacteria. The bigger of the node, the greater the number of correlations.

authors have given approval to the final version of the manuscript.

References Ahmed, Z., Cho, J., Lim, B.-R., Song, K.-G., Ahn, K.-H., 2007. Effects of sludge retention time on membrane fouling and microbial community structure in a membrane bioreactor. J. Membr. Sci. 287, 211–218. http://dx.doi.org/10.1016/j.memsci.2006. 10.036. Berendonk, T.U., Manaia, C.M., Merlin, C., Fatta-Kassinos, D., Cytryn, E., Walsh, F., Bürgmann, H., Sørum, H., Norström, M., Pons, M.-N., Kreuzinger, N., Huovinen, P., Stefani, S., Schwartz, T., Kisand, V., Baquero, F., Martinez, J.L., 2015. Tackling antibiotic resistance: the environmental framework. Nat. Rev. Microbiol. 1–8. http:// dx.doi.org/10.1038/nrmicro3439. Bonfante, P., Anca, I.-A., 2009. Plants, mycorrhizal fungi, and bacteria: a network of interactions. Annu. Rev. Microbiol. 63, 363–383. http://dx.doi.org/10.1146/ annurev.micro.091208.073504. Caporaso, J.G., Lauber, C.L., Walters, W.A., Berg-Lyons, D., Lozupone, C.A., Turnbaugh, P.J., Fierer, N., Knight, R., 2011. Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proc. Natl. Acad. Sci. U. S. A. 108 (Suppl), 4516–4522. http://dx.doi.org/10.1073/pnas.1000080107. Edgar, R.C., Haas, B.J., Clemente, J.C., Quince, C., Knight, R., 2011. UCHIME improves sensitivity and speed of chimera detection. Bioinformatics 27, 2194–2200. http://dx. doi.org/10.1093/bioinformatics/btr381. Fitzgerald, C.M., Camejo, P., Oshlag, J.Z., Noguera, D.R., 2015. Ammonia-oxidizing microbial communities in reactors with efficient nitrification at low-dissolved oxygen. Water Res. 70, 38–51. http://dx.doi.org/10.1016/j.watres.2014.11.041.

Acknowledgements This study was funded by National Key Research and Development Program [grant number 2016YFD0501405], the National Natural Science Foundation of China [grant numbers 41501513 and 21577161], Special Fund for Agro-scientific Research in the Public Interest [grant number 201303091], and National Major Science & Technology Projects for Water Pollution Control and Management [grant number 2015ZX07203-007].

Appendix A. Supplementary data Supplementary data to this article can be found online at https:// doi.org/10.1016/j.envint.2018.06.008.

280

Environment International 118 (2018) 274–281

Q. Sui et al.

btr026. Sui, Q., Zhang, J., Tong, J., Chen, M., Wei, Y., 2015. Seasonal variation and removal efficiency of antibiotic resistance genes during wastewater treatment of swine farms. Environ. Sci. Pollut. Res. 1–10. http://dx.doi.org/10.1007/s11356-015-5891-7. Sui, Q., Jiang, C., Yu, D., Chen, M., Zhang, J., Wang, Y., Wei, Y., 2017. Performance of a sequencing-batch membrane bioreactor (SMBR) with an automatic control strategy treating high-strength swine wastewater. J. Hazard. Mater. http://dx.doi.org/10. 1016/j.jhazmat.2017.05.010. Wang, Q., Garrity, G.M., Tiedje, J.M., Cole, J.R., 2007. Naïve Bayesian classifier for rapid assignment of rRNA sequences into the new bacterial taxonomy. Appl. Environ. Microbiol. 73, 5261–5267. http://dx.doi.org/10.1128/AEM.00062-07. Wu, B., Yi, S., Fane, A.G., 2011. Microbial behaviors involved in cake fouling in membrane bioreactors under different solids retention times. Bioresour. Technol. 102, 2511–2516. http://dx.doi.org/10.1016/j.biortech.2010.11.045. Yang, D., Wang, J., Qiu, Z., Jin, M., Shen, Z., Chen, Z., Wang, X., Zhang, B., Li, J.W., 2013. Horizontal transfer of antibiotic resistance genes in a membrane bioreactor. J. Biotechnol. 167, 441–447. http://dx.doi.org/10.1016/j.jbiotec.2013.08.004. Zhang, Q.-Q., Ying, G.-G., Pan, C.-G., Liu, Y.-S., Zhao, J.-L., 2015. A comprehensive evaluation of antibiotics emission and fate in the river basins of China: source analysis, multimedia modelling, and linkage to bacterial resistance. Environ. Sci. Technol. 49, 6772–6782. http://dx.doi.org/10.1021/acs.est.5b00729. Zhang, J., Cai, X., Qi, L., Shao, C., Lin, Y., Zhang, J., Zhang, Y., Shen, P., Wei, Y., 2015a. Effects of aeration strategy on the evolution of dissolved organic matter (DOM) and microbial community structure during sludge bio-drying. Appl. Microbiol. Biotechnol. http://dx.doi.org/10.1007/s00253-015-6640-z. Zhang, J., Chen, M., Sui, Q., Tong, J., Jiang, C., Lu, X., Zhang, Y., Wei, Y., 2015b. Impacts of addition of natural zeolite or a nitrification inhibitor on antibiotic resistance genes during sludge composting. Water Res. 91, 1–5. http://dx.doi.org/10.1007/s13398014-0173-7.2. Zhu, Y., Wang, Y., Zhou, S., Jiang, X., Ma, X., Liu, C., 2018. Robust performance of a membrane bioreactor for removing antibiotic resistance genes exposed to antibiotics: role of membrane foulants. Water Res. 130, 139–150. http://dx.doi.org/10.1016/j. watres.2017.11.067.

Fu, C., Yue, X., Shi, X., Ng, K.K., Ng, H.Y., 2017. Membrane fouling between a membrane bioreactor and a moving bed membrane bioreactor: effects of solids retention time. Chem. Eng. J. 309, 397–408. http://dx.doi.org/10.1016/j.cej.2016.10.076. Harb, M., Hong, P.-Y., 2017. Molecular-based detection of potentially pathogenic bacteria in membrane bioreactor (MBR) systems treating municipal wastewater: a case study. Environ. Sci. Pollut. Res. 24, 5370–5380. http://dx.doi.org/10.1007/s11356-0168211-y. Ma, Y., Wilson, C.A., Novak, J.T., Riffat, R., Aynur, S., Murthy, S., Pruden, A., 2011. Effect of various sludge digestion conditions on sulfonamide, macrolide, and tetracycline resistance genes and class i integrons. Environ. Sci. Technol. 45, 7855–7861. http:// dx.doi.org/10.1021/es200827t. Madsen, J.S., Burmølle, M., Hansen, L.H., Sørensen, S.J., 2012. The interconnection between biofilm formation and horizontal gene transfer. FEMS Immunol. Med. Microbiol. 65, 183–195. http://dx.doi.org/10.1111/j.1574-695X.2012.00960.x. Magoc, T., Salzberg, S.L., 2011. FLASH: Fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957–2963. http://dx.doi.org/10.1093/ bioinformatics/btr507. Martínez, J.L., 2008. Antibiotics and antibiotic resistance genes in natural environments. Science 321, 365–367. http://dx.doi.org/10.1126/science.1159483. Munir, M., Wong, K., Xagoraraki, I., 2011. Release of antibiotic resistant bacteria and genes in the effluent and biosolids of five wastewater utilities in Michigan. Water Res. 45, 681–693. http://dx.doi.org/10.1016/j.watres.2010.08.033. Niu, Q., Li, P., Hao, S., Zhang, Y., Kim, S.W., Li, H., Ma, X., Gao, S., He, L., Wu, W., Huang, X., Hua, J., Zhou, B., Huang, R., 2015. Dynamic distribution of the gut microbiota and the relationship with apparent crude fiber digestibility and growth stages in pigs. Sci. Rep. 5 (9938). http://dx.doi.org/10.1038/srep09938. Olsen, I., 2015. Biofilm-specific antibiotic tolerance and resistance. Eur. J. Clin. Microbiol. Infect. Dis. 34, 877–886. http://dx.doi.org/10.1007/s10096-015-2323-z. Pruden, A., Pei, R., Storteboom, H., Carlson, K.H., 2006. Antibiotic resistance genes as emerging contaminants: studies in northern Colorado. Environ. Sci. Technol. 40, 7445–7450. http://dx.doi.org/10.1021/es060413l. Schmieder, R., Edwards, R., 2011. Quality control and preprocessing of metagenomic datasets. Bioinformatics 27, 863–864. http://dx.doi.org/10.1093/bioinformatics/

281