Antibiotic resistance genes removal and membrane fouling in secondary effluents by combined processes of PAC/BPAC–UF

Antibiotics have been widely used to treat infectious diseases and to protect the health of humans and animals (Zhang et al. 2009; Di Cesare et al. 2016). The overuse and misuse of various antibiotics lead to the emergence and spread of antibiotic resistance genes (ARGs) in the environment (Li et al. 2017). Ever since the report of ARGs as an emerging type of pollutant in the environment (Pruden et al. 2006), the harmfulness of ARGs has attracted great attention worldwide. According to the World Health Organization (WHO), approximately 700,000 deaths worldwide each year have been attributed to antibiotic resistance. These circumstances have created a sense of urgency worldwide, with antibiotic resistance cited as one of the most critical human health risks (Zarei-Baygi et al. 2019).

Wastewater treatment plants (WWTPs), which are major sources of antibiotic-resistant bacteria (ARB) and ARGs (Storteboom et al. 2010), receive various antibiotic pollutants from domestic, hospitals and animal production sewage. The oxidation ditch process and the Anaerobic–Anoxic–Oxic (A/A/O) process are common in WWTPs, and significant ARG removals (the overall log removal of ARGs ranged from 1.2 to 3.8) has been observed in these WWTPs (Luo et al. 2014). Nonetheless, the secondary effluents from WWTPs still contain high levels of ARGs (Munir & Xagoraraki 2011). Among the ARGs, studies have demonstrated that three tetracycline resistance (tet) genes (tetA, tetW and tetO) and two sulfonamide resistance (sul) genes (sulI and sulII) are detectable using high frequencies (Chen & Zhang 2013). In addition, there is an increasing shortage of water resources, and the secondary effluent from WWTPs was chosen as a promising source of water reuse due to its large quantity and relatively good quality (Hu et al. 2014). However, a large number of ARGs are safety issues that cannot be ignored in the secondary effluent reuse process. Therefore, promising technologies should be developed to deal with ARGs.

Given the wide application of ultrafiltration (UF) in recent years, some studies also evaluated the significance of the UF membrane on removing ARGs. It was found that UF could reduce ARGs content by 0.6–1.5 orders of magnitude (Breazeal et al. 2013). However, the molecular weight of ARGs can range from hundreds to millions of daltons (Cai et al. 2007). Thus, UF can only remove a portion of ARGs, while a considerable amount of ARGs would remain in the effluent. Therefore, it is an important direction for the development of UF technology to combine it with other purification units to establish a combination process and method for different pollutants. Powdered activated carbon (PAC) can effectively adsorb contaminants in water. Besides, Duan (2014) found that amorphous granular activated carbon could significantly reduce the concentration of tetracycline and sulfonamide ARGs in the secondary effluents of WWTPs through adsorption. Furthermore, PAC would gradually convert to biological PAC (BPAC) during long-term use, thus enhancing the function of the microbial degradation of pollutants. However, few studies focused on ARG removal using PAC/BPAC–UF combined processes.

Therefore, this study aims to evaluate the potential of the PAC/BPAC–UF combined process for ARGs removal from secondary effluents of the WWTPs and provide recommendations for potential engineering applications. Specifically, we investigate the presence of two sul genes (sulI and sulII), two tet genes (tetA and tetW) and class I integron intI1 in a WWTP. We study the removal of ARGs, intI1 and dissolved organic carbon (DOC) in secondary effluents by PAC/BPAC–UF combined processes. In addition, we examine the mitigation of membrane fouling by the two combined processes.

In the current study, the experimental raw water was collected from the secondary sedimentation tank of a WWTP in Beijing, China. All samples were aseptically collected in sterile containers and transported to the laboratory on ice for immediate processing. The source water qualities are shown in Table 1.

The experimental PAC was shell-activated carbon purchased from Shanghai (Heatton) Environmental Sci-Tech Co., Ltd. The PAC was washed with pure water to the value of UV₂₅₄ being close to zero, dried at 105 °C for 3 h and then placed in a vacuum drying pan. During the cultivation of BPAC, the concentration of PAC was 2 g/L. The raw water was manually added once per 24 h, and the reactor was controlled to have an effective volume of 2 L. An aeration head was installed at the bottom of the organic glass column reactor with a diameter of 15 cm and a height of 25 cm. An aeration rate of 3 L/min was used to provide oxygen for microorganisms while avoiding the deposition of activated carbon at the bottom. Through this process, BPAC was cultivated for use in the experiment. Polyethersulfone (PES) flat-sheet UF membranes purchased from Millipore (USA) with molecular weight cutoff (MWCO) of 100 and 1 kDa were used in this study. The effective membrane area was 41.8 cm².

In this experiment, different doses of PAC and BPAC (0, 20, 40, 60 and 80 mg/L) were added to raw water, respectively. After 24 h of adsorption and degradation, the mixture was filtered through UF membranes with MWCO of 100 kDa. A schematic representation of the experimental setup is illustrated in Figure 1. UF experiments were conducted in a filtration cell (Amicon 8400, Millipore, USA) in a dead-end mode at a constant trans-membrane pressure (TMP) of 100 kPa, which was maintained by a nitrogen gas bottle connected to the secondary water reservoir. The volume of Millipore UF filter bowl was 400 mL. During the experiment, the filtration process was set to three cycles, and each cycle took 10 min to filter out 400 mL of the mixture and backwash for 1 min. The backwashing method was carried out as follows: the UF membrane was washed for 1 min using 200 mL ultrapure water in a filter bowl with magnetic mixers revolving at the speed of 250 rpm. To determine the recovery degree of membrane flux, the cleaned membrane was used to filter pure water at a pressure of 0.10 MPa. The permeate flux was monitored by an electronic balance, and the output data were automatically recorded via a computer using a data acquisition system. The membrane effluent after each cycle was taken and concentrated by passing through a 1 kDa PES UF membrane to concentrate microbial biomass for the quantification of gene copies. The membrane effluent volume of each process is 1.2 L. After filtration, the membranes were stored at −20 °C prior to DNA extraction and subsequent quantification. All the experiments were performed at room temperature (23 ± 1 °C).

DNA was extracted using the FastDNA Spin Kit for soil (MP Biomedicals, Santa Ana, CA, USA) following the protocol of the manufacturer. The concentration and quality of the extracted DNA were measured by spectrophotometry (NanoDrop ND-2000, Thermo, USA). The extracted DNA samples were stored at − 20 °C until the quantification analysis.

• DNA length (tetA) = 2,902 bp.

• DNA length (tetW) = 2,860 bp.

• DNA length (sulI) = 2,854 bp.

• DNA length (sulII) = 2,882 bp.

• DNA length (intI1) = 2,972 bp.

• DNA length (16Sr) = 2,834 bp.

Figure 2 summarizes the abundances of the four ARGs, intI1 and 16SrDNA detected in raw water and different process effluents.

As shown in Figure 2, two tet genes (tetA and tetW) and two sul genes (sul I and sul II) were detected in secondary effluents from the WWTP. The concentrations of tetA, tetW, sulI and sulII were 10^5.40–10^6.10, 10^3.32–10^3.49, 10^6.73–10^7.19 and 10^6.79–10^7.21 copies/mL, respectively, and the concentrations of intI1 and 16SrDNA were 10^5.58–10^5.76 and 10^7.49–10^7.71 copies/mL, respectively. This was similar to the detection of ARGs in secondary effluents of WWTPs in the study by Zheng et al. (2017). Raw water through direct UF resulted in the removals of tetA, tetW, sulI and sulII by 1.56-log, 0.86-log, 1.48-log and 1.23-log, respectively. These indicate that UF can remove some ARGs in the secondary effluents. When PAC and BPAC were combined with UF, the reduction of ARGs was greater than that of direct UF. For the PAC–UF, increased PAC dosage enhanced the removal effect of ARGs. When the dosage of PAC was 60 mg/L, the removal effect of tetA, tetW, sulI and sulII was optimal, where the removals were 3.10-log, 1.35-log, 3.18-log and 3.35-log, respectively. When the dosage of PAC was raised to 80 mg/L, the removal effect was reduced. The adsorption of activated carbon and the interception of the UF membrane effectively reduced ARGs and intI1 in secondary effluents. When the adsorption was saturated, the removal effect was dropped. For the BPAC–UF, when the dosage of BPAC was 80 mg/L, the removal effect of tetA, tetW, sulI and sulII and intI1 was the best. The removal amounts were 3.73-log, 1.81-log, 3.72-log, 3.62-log and 2.73-log, respectively. This is because PAC has a large specific surface area and has a good adsorption capacity for contaminants in solution. The UF membrane combined with PAC can effectively retain ARGs. The BPAC–UF combined process merged adsorption, biodegradation and membrane separation (Sun et al. 2018); hence, its removal efficiency of ARGs was more effective compared to PAC–UF. In addition, the microbial community structure on BPAC may be beneficial for the removal of ARGs in the secondary effluent.

In this study, DOC removal from the effluents was also investigated as important indicators of the treatment efficiency of PAC/BPAC–UF. DOC was detected using a TOC-VCPH total organic carbon analyzer (Shimadzu Co., Japan).

As shown in Table 3, the removal efficiency of DOC in the secondary effluents by the PAC/BPAC–UF combined processes was better than that by direct UF (only 23.5%). When PAC was combined with UF, the removal efficiency of DOC increased with the increase of PAC dose. The treatment effect was the best when 60 mg/L PAC was added, where the removal rate of DOC was 65.2%. For BPAC–UF, the removal of DOC was different. Although the removal efficiency of DOC was the highest (58.3%) with 60 mg/L BPAC, it dropped to 52.9% with 80 mg/L BPAC. This does not mean that PAC is better than BPAC for the removal of organic matter. This is because PAC is fresh activated carbon with high adsorption efficiency. BPAC is a mature biological activated carbon, and its adsorption efficiency has been reduced. The BPAC removal process for DOC can be divided into two phases. In the first phase, the removal of DOC relies primarily on physical adsorption. In the second stage, the adsorption site of BPAC is gradually saturated, and the removal of DOC by BPAC is mainly accomplished by microbial degradation. In the whole process, the degradation of microorganisms and the physical adsorption play a synergistic relationship. The adsorption of organic matter by GAC increases the concentration of the substrate, which is beneficial to the growth and reproduction of microorganisms. The degradation of organic matter by microorganisms releases the adsorption site of BPAC and restores the partial adsorption capacity (Wang 2017).

The normalized membrane flux (J/J₀) change and the membrane fouling resistance in the three filtration cycles are shown in Figure 3.

The results show that the dosages of PAC and BPAC significantly influenced the J/J₀. Figure 6(a) demonstrates that the J/J₀ of the direct UF decreased significantly, which was 0.32 at the end of the filtration. When the PAC dosages were 20, 40, 60 and 80 mg/L, the J/J₀ at the end of the filtration decreased to 0.40, 0.37, 0.33 and 0.25, respectively. The experimental results indicate that the appropriate amount of PAC could increase the flux of the membrane. As illustrated in Figure 6(b), when the BPAC dosages were 20, 40, 60 and 80 mg/L, the values of J/J₀ at the end of the filtration were 0.43, 0.40, 0.34 and 0.22, respectively. The results reflect that the J/J₀ at the dosages of BPAC at 20 and 40 mg/L were obviously higher than those of direct UF, and the J/J₀ would decrease when the dosage was too high. At an appropriate dosage, PAC and BPAC would form a relatively loose and high porosity cake deposited on the membrane (Sun et al. 2018). In addition, the BPAC carried an extracellular polymeric substance and microorganisms. The microorganisms on the surface of the BPAC would continue to consume EPSs to reduce the contaminant adsorption or filling on the membrane surface to improve the membrane flux and reduce the membrane pore blockage. Therefore, BPAC–UF was better than PAC–UF in relieving membrane fouling.

Significant DOC, intI1, 16SrDNA and ARGs removal was achieved in this study. Based on the linear regression method, the relationships between DOC, intI1, 16SrDNA and tet genes (tetA and tetW), sul genes (sulI and sul II) removal in water samples were investigated. Figure 4 displays the fitting outcomes (p < 0.05 indicates statistically significant).

As illustrated in Figure 4, intI1, 16SrDNA and DOC showed significant positive correlations with tet genes and sul genes (p < 0.05). These suggest that the concentration of the four ARGs decreased significantly with the removal of intI1, 16SrDNA and DOC. ARGs also belong to the organic substance. The ARGs can interact with clay minerals and various organic colloidal particles. In this way, the ARGs could adsorb onto organic matter and get removed from wastewater together with organic colloidal particles (Breazeal et al. 2013). PAC has a good adsorption capacity for organic and colloidal particles in solution. BPAC combines physical adsorption and biodegradation. Therefore, the removal of some colloidal substances in the effluents contributed toward the reduction of ARGs. Note that integrons are important pathways for the transmission of ARGs between different species in the environment. Most of the ARGs in the environment exist on genetic transfer elements, which help to transfer ARGs within and between bacteria (Gaze et al. 2011). There were significant positive correlations between tet genes (tetA and tetW), sul genes (sulI and sulII) and intI1 gene, implying that tetA, tetW, sulI and sulII might be bound to intI1 in water samples. Therefore, the removal of intI1 contributed to the reduction of ARGs.

16SrDNA gene copy numbers reflect the number of microbes in samples. The replication and spread of ARGs in the environment can be affected by the concentration of 16SrDNA, but different genes would be affected differently (Wang et al. 2016). Activated carbon has a good adsorption effect on microorganisms in water (including bacteria, fungi and viruses). PAC/BPAC–UF combined processes removed most of the microorganisms in water by the dual action of adsorption and retention, which significantly reduced the concentration of ARGs. These results indicate that the decrease of microorganisms was the main cause of the removal of ARGs in PAC/BPAC–UF. In summary, the removals of intI1, 16SrDNA and DOC in secondary effluents were beneficial to the removal of ARGs.

To analyze the mechanism of the removal of ARGs by the microbial community structure on BPAC, the sample DNA sequences before and after the BPAC treatment were sequenced using the Illumina HiseqV4 PE250 (Illumina, USA) high-throughput sequencing platform. The RDP classifier was implemented to classify the species. After classification, the microbial species and relative abundance of each microorganism were determined and expressed using the statistical analysis. The community structure at the two levels of phylum and class was observed and compared (see Figure 5).

The microbial species were widely distributed on the surface of BPAC. Up to 32 species of microorganisms were detected at the phylum level. Nine species with higher relative abundance were selected in the figure, and the rest of the sum were others. As illustrated in Figure 5(a), the proportion of Proteobacteria on BPAC was 78%. At the class level (Figure 5(b)), Alphaproteobacteria accounted for 35%, and the Betaproteobacteria accounted for 39.0%. The bacteria in the Proteobacteria are Gram-negative bacteria (Juncker et al. 2010), including a variety of pathogens, such as Escherichia coli, Vibrio cholerae and Helicobacter pylori, and tetracycline resistance (tet) genes (tetO and tetW) are positively correlated (Garner et al. 2017), while sulfonamide resistance (sul) genes are common in human pathogens. The potential host strains of sulI and intI1 are extensive, and most of them are Proteobacteria (Duan 2017). The proportion of Firmicutes on BPAC was 5%. The bacterium includes a large class of bacteria, mostly Gram-positive. Tetracycline resistance (tet) genes are present in Gram-positive bacteria (Luna & Roberts 1998). At the class level, Bacilli and Clostridia form the main body of the Firmicutes. As illustrated in Figure 5(b), the proportion of Bacilli on BPAC was 6%. The previous study suggested that Bacillus was the host of a variety of ARGs (Zhang 2013). In addition, tetW and sulII are significantly associated with other potential host bacteria (Lysinibacillus, Streptococcus and unidentified_Xanthomonadales) (Gul et al. 2015).

The above colonies proliferated on BPAC and became dominant populations, which were beneficial to the transfer of ARGs in secondary effluents. The UF membrane traps the BPAC particles in the water, which simultaneously removes the bacteria carried on the particles. Therefore, the concentration of ARGs in secondary effluents reduced significantly. Most of the ARGs in the environment exist on genetic transfer elements, such as plasmids and integrons (Gaze et al. 2011). Microorganisms on the BPAC filter cake layer secreted enzymes that degrade the plasmid (Hernandezraquet et al. 2013) and thus reduced ARGs in secondary effluents. In addition, Proteobacteria can fully hydrolyze proteins, starches and other substances in the water to grow and proliferate (Guerrero-Feijóo et al. 2017). Therefore, the presence of Proteobacteria will promote the removal of organic pollutants in water. The previous chapter has already discussed that organic matter removal is closely related to resistance gene reduction, so ARGs will be removed in cooperation with organic matter.

The dominant reactions and mechanisms involved in the two different systems of PAC/BPAC–UF processes are shown in Figure 6.

Mechanical screening is the primary mechanism for UF membrane filtration. Macromolecular contaminants are trapped by UF membranes, and contaminants accumulate on the membrane surface. Contaminants with smaller molecular size than membrane pore size may be adsorbed on the inner wall of the membrane pores. The pores of the membrane are narrowed or blocked (Qu et al. 2012), causing membrane fouling. ARGs are a special type of contaminant with a wide range of molecular weights (Cai et al. 2007). Small molecular pollutants can be adsorbed by the appropriate amount of PAC to reduce the narrowing or clogging of the pores; the activated carbon and some substances in the mixture will form a filter cake layer with high porosity on the surface of the membrane and the pollutants in the water will continue to be adsorbed and retained. At the same time, the ARGs in the water are effectively removed. Therefore, membrane fouling is effectively alleviated.

As the mixture is filtered through the UF membrane, a small amount of organic matter can be adsorbed by an appropriate amount of BPAC, and BPAC and some substances form a cake layer on the surface of the membrane. Since BPAC is surrounded by microorganisms, the microflocs with increased particle size make the filter cake layer formed on the surface of the membrane more porous than PAC. Due to the action of the filter cake layer formed, the pollutants are trapped and adsorbed, and the organic matter is continuously biodegraded. The synergistic effect of the two causes the ARGs to be effectively removed, and the degree of contamination of the membrane surface is significantly reduced. Therefore, the membrane flux is increased, and membrane fouling is more mitigated.

This study found that the combined processes of PAC/BPAC–UF could effectively reduce ARGs in secondary effluents and mitigate membrane fouling. The conclusions are:

1. At different dosages, PAC/BPAC–UF had different removal efficiencies on the four ARGs and DOC. In practical applications, we recommend 40 mg/L PAC and BPAC as appropriate dosages. At this dosage, PAC/BPAC–UF could remove 1.26–2.69-log and 1.55–2.97-log ARGs, respectively; the DOC removal rates would be 60.7% and 54.1%, respectively. In addition, membrane fluxes would increase by 15.6% and 25.1%, respectively.

2. DOC, 16SrDNA and intI1 are significantly correlated with ARGs. The removals of intI1, 16SrDNA and DOC in secondary effluents were beneficial to the removal of ARGs.

3. After the water sample underwent the BPAC treatment, Proteobacteria and Bacilli became dominant populations on BPAC. In addition, BPAC was trapped by the membrane while promoting the removal of ARGs in the water. Relative to PAC–UF, the removal efficiency of ARGs in secondary effluents by BPAC–UF was better, and it was more efficient in relieving membrane fouling.

The research was supported by the National Natural Science Foundation of China (grant no. 51678027).

The Supplementary Material for this paper is available online at http://dx.doi.org/10.2166/wh.2019.160.