Fate of antibiotics and antibiotic resistance genes during conventional and additional treatment technologies in wastewater treatment plants

• Good removal (79-88%) of total antibiotics was observed at all WWTPs. • NEREDA® (2.3 log) and conventional WWTP (2.0 log) are more effective for log reduction of ARG than 1-STEP®filter

Information on the removal of antibiotics and ARGs in full-scale WWTPs (with or without additional treatment technology) is limited. However, it is important to understand the efficiency of full-scale treatment technologies in removing antibiotics and ARGs under a variety of conditions relevant for practice to reduce their environmental spreading. Therefore, this study was performed to evaluate the removal of antibiotics and ARGs in a conventional wastewater treatment plant (WWTP A) and two full-scale combined with additional treatment technologies. WWTP B, a conventional activated sludge treatment followed by an activated carbon filtration step (1-STEP® filter) as a final treatment step. WWTP C, a treatment plant using aerobic granular sludge (NEREDA®) as an alternative to activated sludge treatment. Water and sludge were collected and analysed for 52 antibiotics from four target antibiotic groups (macrolides, sulfonamides, quinolones, tetracyclines) and four target ARGs (ermB, sul 1, sul 2 and tetW) and integrase gene class 1 (intI1). Despite the high removal percentages (79-88%) of the total load of antibiotics in all WWTPs, some antibiotics were detected in the various effluents. Additional treatment technology (WWTP C) showed antibiotics removal up to 99% (tetracyclines). For ARGs, WWTP C reduced 2.3 log followed by WWTP A with 2.0 log, and WWTP B with 1.3 log. This shows that fullscale WWTP with an additional treatment technology are promising solutions for reducing emissions of antibiotics and ARGs from wastewater treatment plants. However, total removal of the antibiotics and ARGS cannot be achieved for all types of antibiotics and ARGs. In addition, the ARGs were more abundant in the sludge compared to the wastewater effluent suggesting that sludge is an important reservoir representing a source for later

Introduction
Antibiotics inhibit the growth of microorganisms and have been used widely since the 1930s to combat infectious diseases (Gallo et al., 1995;Chee-Sanford et al., 2009). Only a fraction of these antibiotics are completely metabolized within bodies of humans and animals, and 20% to 90% is generally excreted as parent compound or metabolite through urine and feces (Jelic et al., 2015).
As a result, antibiotics are widely present in our domestic sewage waters and enter WWTPs that have only limited capacities to remove these compounds, thus they end up in the environment. Even though antibiotics are detected at low concentrations (ng/L to μg/L scale) in the environment, these antibiotics can persist there for a long time (Kolář et al., 2001). This persistence depends not only on the characteristic of the respective antibiotics, but also on environmental conditions such as oxygen and other electron acceptors and donors, and light. As a result, some antibiotics are readily degraded, while others are not. Their presence in the environment can lead to the emergence and prevalence of antibiotic resistant bacteria (ARB) and antibiotic resistance genes (ARGs) (Andersson and Hughes, 2012). Antibiotic resistance develops naturally. However, use, misuse and inappropriate antibiotic prescriptions has accelerated the occurrence of ARB and ARGs in the environment (Kraemer et al., 2019), especially in wastewater (Sharma et al., 2016;Karkman et al., 2018;Pazda et al., 2019). WHO has listed antibiotic resistance as one of the world threats since 2014 (WHO, 2014).
Conventional WWTPs are designed to remove high concentrations of total organic carbon, and nutrients such as nitrates, and phosphates (Pronk et al., 2015;de Kreuk et al., 2010) but not specifically designed to remove micropollutants, including antibiotics and ARGs (Novo et al., 2013;Pal et al., 2015). In a conventional WWTP, activated sludge is the most common treatment technology for the biological treatment of wastewater (Samer, 2015). Two tanks are needed for activated sludge; one for aeration of biological reactions and for settling of the sludge.
The removal of antibiotics and ARGs varies (1-2 log) in conventional WWTPs (Pallares-Vega et al., 2019). Some studies showed an increased relative concentration of antibiotics and some ARGs after WWTP treatment (Pärnänen et al., 2019). Others reported a decrease in the prevalence of antibiotics and relative concentrations of ARGs after WWTP treatment (Liao et al., 2016;Pärnänen et al., 2019;Gao et al., 2012;Czekalski et al., 2012;Kulkarni et al., 2017). Even when a significant removal in the wastewater effluent was found, antibiotics were still detected in the receiving (surface) water body, ranging from 2 ng/L to 25 mg/L (Singh et al., 2019).
Therefore, additional treatment technologies may provide a promising contribution to reduce antibiotics and ARGs before effluent is discharged to the environment. Various additional treatment technologies have shown to remove antibiotics and/or ARGs from wastewater, e.g., physical treatment processes , disinfection (Shen et al., 2020;Khorsandi et al., 2019), advanced oxidation processes (Collivignarelli et al., 2018;Zhuang et al., 2015), as well as aerobic granular sludge (AGS) (Mihciokur and Oguz, 2016).
Activated carbon (AC) is a treatment technology that is based on adsorption and filter material, and can act as a carrier matrix for biomass. It is known to remove dissolved compounds, suspended matters, nitrogen, and phosphates (Hung et al., 2005). AC is a form of an amorphous carbonaceous material showing a high specific surface area (1000 m 2 / g) (Tadda et al., 2016) microporous structure, and large pore volume (Choi et al., 2005). Therefore, due to its high adsorptive capacities, AC has shown potential in removing color/odor/taste (Matsui et al., 2015;Huang et al., 2019b), disinfection by-products (Gopal et al., 2007), micropollutants (Kårelid et al., 2017;Choi et al., 2005), antibiotics (Yu et al., 2016;Pachauri et al., 2009) and ARB (Ravasi et al., 2019) and ARGs . AC was also applied as a post-treatment to solar photo fenton treatment to remove antibiotics from wastewater .
AGS has been introduced to overcome the drawbacks in activated sludge, (Wilén et al., 2018). AGS is known for the excellent settling ability, simultaneous removal of organic matter and nitrogen, high biomass concentration, and a good ability to withstand the high organic load. AGS contain granules comprised of self-immobilized cells and does not depend on material support for biofilm growth. It retains a large number of microorganisms, at the same time permitting rapid bioconversion of many compounds and improving the performance and stability of the reactor (Bassin, 2018). In addition, AGS is a compact (using only one tank for settling and aeration) and cost-effective wastewater treatment (Nancharaiah and Reddy, 2017). Due to these characteristics, AGS gain more attention to be implemented in the wastewater treatment. Some full-scale AGS have recently been implemented to treat municipal and industrial wastewater (Pronk et al., 2015;van der Roest et al., 2011). As this is a relatively new technology, limited data are available on the removal of antibiotics in full-scale plants (Wang et al., 2019) and ARB&Gs.
We hypothesize that additional treatment technologies have the potential to improve the removal of antibiotics and ARGs at current WWTPs. Most studies that compare different treatment technologies have been carried out at lab-scale (Sousa et al., 2017;Zheng et al., 2017;Karaolia et al., 2018;Guo et al., 2017). However, limited research on antibiotics and ARGs removal has been performed in full-scale WWTPs, and studies so far focused on the comparison between conventional and additional treatment technology, e.g. membrane nanofiltration and reverse osmosis in China (Lan et al., 2019), parallel membrane bioreactors in China  and four separate treatments based on UV irradiation in Spain (Rodríguez-Chueca et al., 2019).
Therefore, this study evaluates the removal of antibiotics and ARGs in full-scale WWTPs, with and without additional treatment technologies. A conventional WWTP operated with activated sludge is chosen as a control, receiving domestic wastewater for comparison purposes. In addition, two full-scale WWTPs with different additional treatment technologies were studies, a 1-STEP® filter (based on AC), and a NEREDA® technology (based on AGS). Grab samples were collected for two months and analysed for 52 antibiotics from 4 groups (macrolides, sulfonamides, quinolones, tetracyclines), 4 ARGs (ermB, sul 1, sul 2 and tetW) and integrase gene class 1 (intI1).

WWTPs
Three Dutch WWTPs were selected to study the removal of antibiotics, and ARGs; one conventional WWTP (WWTP A) and two WWTPs with additional treatment technologies (WWTP B and WWTP C). WWTP A employs conventional steps, such as a primary treatment (grit removal), a reactor with activated sludge, a sludge settling tank, and finally sand filtration.
WWTP B consists of a conventional activated sludge treatment followed by an AC filtration step (1-STEP® filter). The 1-STEP® filter is in operation since August 2012 and consists of a vertical, compact fixed bed activated carbon filter, combining filtration, denitrification, coagulation, flocculation, and adsorption in one single treatment unit (Bechger et al., 2009;Bechger et al., 2013). It is in use as final treatment step (as a tertiary treatment) to improve the nitrogen and phosphorus removal and the final effluent quality meets the requirements of the European Water Framework Directive 2000/60/EC (Council Directive, 1991).
WWTP C employs a NEREDA® technology, which is based on AGS as an alternative to activated sludge (as a secondary treatment). The NEREDA® technology is in operation since 2011 and developed in the Netherlands (van der Roest et al., 2011). The granules have a robust structure of aerobic granular biomass, with dense, compact, large particles (0.2-2 mm) with a high specific gravity (van der Roest et al., 2011). The granules consist of different microorganisms, including phosphate accumulating organisms, nitrifiers, denitrifiers, and glycogen accumulating organisms, which allows several processes simultaneously (Giesen et al., 2013). Due to the growth in dense grains, there are different oxygen levels within the grain, allowing different organisms to be active.
Process flow diagrams and sampling points of each WWTP are shown in Fig. 1. WWTP A has the capacity to treat 22,000 person equivalents (p.e.) with an average treated volume of 2860 m 3 /d, WWTP B treats 200,000 p.e. with an average treated volume of 26,000 m 3 /d and WWTP C is designed for 59,000 p.e and an average treated volume of 8000 m 3 /d. Details for the three WWTPs are shown in Table S1.

Sample collection and pre-treatment
Collection of grab samples (water and sludge samples) of the WWTPs was carried out during the winter season (February and March 2017) at 3 sampling points (WWTP A) and 4 sampling points (WWTP B and WWTP C), which are shown in Fig. 1. The sampling was performed between 9:00-12:00 (after the daily morning peak hour), two times each month in duplicate at all sampling points. Average air temperatures during sampling were between 8°C and 11°C (data from the WWTPs), respectively.
Water samples (n = 12 per WWTP A, n = 16 per WWTP B and C) were taken at the outfall of the tank. A bucket was dipped into a tank for water sample collection, and the samples were stored in 1 L sterile glass bottles. On the sampling day, the water was measured for pH, water temperature and dissolved oxygen (DO) by using an HQ40D portable meter (Hach, Germany). The water samples were divided into two parts at different storage: for physicochemical and DNA filtration, the samples were stored at 4°C and processed within 48 h. For antibiotics analyses, the samples were stored at −20°C until processed. All sludge samples (n = 8 per WWTP) were collected in duplicate. For sludge collection, a bucket was dipped at the influent and at the sludge return flow (WWTP A) or via a sludge tap (WWTP B and C). The sludge samples stored in a 50 mL tube at −20°C before DNA extraction.

Physicochemical parameters
During sampling, temperature, pH, and DO were measured using a probe (Hach, USA). Conductivity (in μS/cm) was measured in the laboratory by a conductivity probe (Hach, USA). Other analyses, total phosphate (TP), nitrogen (ammonium (NH 4 -N), nitrate (NO 3 -N) nitrite (NO 2 -N)), and COD, were determined using Hach kits in laboratory (USA; LCK 349, LCK 304, LCK 339, LCK 342, and LCK 1414 respectively). Total suspended solids (TSS) analyses were performed by filtering 200 mL of raw samples through a glass filter (Whatman, England) and placed at 105°C overnight. After weighing, the same samples were incubated at 550°C for 2 h for volatile suspended solids (VSS). All procedures of analyzing nutrients and COD were performed according to the international standards of the American Public Health Association (APHA, 2005).

Solid phase extraction (SPE) and LC-MS/MS
SPE purification and concentration, and LC-MS/MS procedures were performed as previously described by Sabri et al. (2020). This analysis was performed for only water samples and not for sludge samples. The target antibiotic groups (macrolides, sulfonamides, quinolones, tetracyclines) and their chemical characteristics are shown in Table S3. No significant difference was found after some samples at influent and effluent were tested for triplicate (data not shown). Therefore, the analyses were performed in a single measurement. In short, 10 mL of water sample of each sampling point was processed by using a Strata X Polymeric Reversed Phase SPE column (Phenomenex, USA). First, the SPE column was equilibrated by washing with 5 mL MeOH, followed by 5 mL EDTA-McIlvain buffer 0.1 M; pH 4.0. Then, the sample was loaded onto the SPE column and washed with 5 mL Milli-Q water. After this, the sample was eluted with 5 mL MeOH and dried under a nitrogen evaporator with a temperature of 40°C. Finally, the dried sample extracts were dissolved in 500 μl MeOH: Milli-Q water (20:80). The extracts were analysed by LC-MS/MS (Sciex, USA, QTRAP 6500), equipped with a BEH C18 Waters Acquity column (Waters Corporation, USA, 100* 2.1 mm), at a column temperature of 40°C. Data were analysed by using MultiQuant software (Sciex, version 3.0.2). The limit of detection ranged from 5 to N1000 ng/L (depending on the antibiotics). Details for each compound is presented in Table S4. For quality control, a known amount (100 μg/L) and internal standard of each compound were spiked to every sample. The recovery percentage of the spiked compound in the sample ranged from 70 to 120%.

DNA extraction and quantitative PCR (qPCR)
Molecular analyses were performed as previously described by Sabri et al. (2020). Briefly, the DNA extraction of 100 mL samples was filtered onto a membrane filter with a 0.2 μm pore size (Millipore, USA) and stored at −20°C until further use. DNA was extracted by using the PowerWater kit (MoBio Laboratories, USA) for water samples and PowerSoil kit (MoBio Laboratories, USA) for sludge samples, following the manufacturer's protocols. Four ARGs of interest were selected: ermB, sul1, sul2 and tetW, as well as 16S rRNA and intI1. These ARGs were chosen based on antibiotics that have been frequently used and detected in water (Ye et al., 2007). Each ARG corresponding to the respective antibiotic such as macrolides are corresponding to ermB, sulfonamides are corresponding to sul1 and sul2, and tetracyclines are corresponding to tetW. Meanwhile, intI1 was proposed by Gillings et al. (2015) as a good marker for environmental pollution since it is commonly linked to genes conferring resistance to antibiotics and rapid change in response to environmental pressures.
A PCR master mix was prepared, depending on the gene of interest. The master mix contained, as described in detail in Sabri et al. (2020), water, precision blue (Biorad, USA), a probe (if needed), the forward and reverse primer (Eurogentec, Belgium), and super mix (SYBR-Green or IQ (Bio-Rad, USA)). The primers are presented in Table S5. Negative control and a calibration curve were also included. The data were processed by Bio-Rad-CFX manager (Version 3.1).

Calculations and statistical analysis
Linear regression, t-test and correlation analyses were conducted to examine the influence of specific parameters on the performance of the WWTPs in removing antibiotics and ARGs. All statistical analyses were performed on the R platform (Version 3.5.2). The difference was considered statistically significant at p b 0.05.
The removal percentage of antibiotics and ARGs was calculated by comparing the total concentrations between influent and effluent, or before and after the respective additional treatment technology. The calculations for antibiotics (Eq. (1)) and ARGs (Eq. (2)) were used as below: Log removal ¼ Log ARGs before treatment technology−Log ARGs after treatment technology 3. Results and discussions

Performance of the studied WWTPs
The general performance of the studied WWTPs was identified by measuring removal of suspended solids, COD and nutrients. All measured parameters (DO, pH, water temperature, conductivity, TP, NH 4 -N, NO 3 -N, NO 2 -N, COD, TSS, and VSS) met the effluent regulatory targets of the European Water Framework Directive (Council Directive, 1991). The increase of NO 3 -N is the result of production through nitrification and associated with the decrease of NH 4 -N. The removal percentages are summarized in Table 1. Detailed data for each parameter is given in Table S2(a) and (b). More information is given in the Text S1.
Dissolved organic compounds, nutrients, and WWTP operating parameters can affect the concentration of antibiotics and ARGs in a WWTP and its effluent. In our study, we found a good correlation (Pearson correlation, r = 0.4 to 0.8, p b 0.05) between NH 4 -N and all antibiotics groups (only in February) and ARGs (in both sampling months) (Fig. S1). Huang et al. (2019c) observed that sul2 was positively correlated with COD and NH 4 -N, while other ARGs (ermB, intI1, sul1 and sul2) were positively correlated with pH. However, in this study we did not observe significant correlation between ARGs and water temperature, DO and COD. Literature also described that WWTP operating parameters such as hydraulic retention time could enhance the antibiotics removal ability, e.g. a better removal of antibiotics was observed at a higher longer hydraulic retention time with AGS (Liao et al., 2019). We did not study the correlation between dissolved organic compounds, nutrients, and WWTP operating parameters in-depth, due to the limited number of WWTPs involved. Furthermore, a full-scale WWTP is also affected by unpredictable conditions such as sudden heavy rain, unpredictable antibiotic consumption at certain days that affect the removal performance of a WWTP. Therefore, this warrants more research to study this further.

Occurrence and distribution of antibiotics in three WWTPs
The presence of antibiotics in the influent and effluent of the WWTPs in conventional and in additional treatment technologies WWTPs was investigated. In general, 8 out of 52 antibiotics were detected in the influent of all 3 WWTPs in our sampling campaigns of February and March 2017 (Fig. 2). These antibiotics are norfloxacin (NOR), ciprofloxacin (CIP), levofloxacin/ofloxacin (LEV), sulfamethoxazole (SMX), trimethoprim (TRI), sulfadiazine (SF), sulfapyridine (SP) and tetracycline (TET). The antibiotics chlortetracycline (CTC), doxycycline (DC), flumequine (FLU), oxytetracycline (OTC), and lincomycin (LIN) were found occasionally in the WWTPs. The other analysed antibiotics (Table S4) were all below the quantification limit.
The total antibiotics concentration ranged from 3000 ng/L to 6000 ng/L in the influent of all WWTPs for both months. The quinolones were the most abundant antibiotics found in all influents, with an average concentration of 2300 ng/L (WWTP A), 1900 ng/L (WWTP B) and 1600 ng/L (WWTP C). Two antibiotics showed a high variation between the sampling moments in two months; Norfloxacin was measured 18 times lower in February (80 ng/L) compared to March (1400 ng/L) in WWTP A, and doxycycline was measured 20 times higher in February (80 ng/L) compared to March (1500 ng/L) in WWTP C. Table 1 Removal (%) for selected parameters in three studied WWTPs. Data are mean value ± standard deviation (n = 12 per WWTP A, n = 16 per WWTP B and C). WWTP A 74 ± 31 97 ± 0 −511 ± 359 96 ± 0 100 ± 0 99 ± 3 WWTP B 73 ± 3 100 ± 0 −5 ± 5 99 ± 1 100 ± 0 99 ± 1 WWTP C 55 ± 27 98 ± 2 −525 ± 139 97 ± 1 97 ± 3 95 ± 4 An average of 3000 ng/L antibiotics was detected in the influent of all WWTPs, with the quinolones as most abundant antibiotics. This is in the same range as reported in winter time from other studies, e.g., 1000 ng/L (Huang et al., 2019a) to 3300 ng/L (Zheng et al., 2019) in Chinese wastewater influent, and can reach up to 6000 ng/L in Czech Republic (Golovko et al., 2014). It is known that more antibiotics are prescribed during winter periods, as a result of increased respiratory tract infections (Ferech et al., 2006;Werner et al., 2011). This is also in line with a study of Diwan et al. (2013) who showed that higher concentrations of quinolones were measured in the winter. Van Boeckel et al. (2014) reported that the consumption of antibiotics is highest between January and March in the northern hemisphere. For example, antibiotics prescription in the USA is 24.5% higher in the winter than in the summer (Suda et al., 2014), and 32% in Israel (Dagan et al., 2008). This high consumption of quinolones in winter also is shown in various countries in Europe (Adriaenssens et al., 2011). This elevated use in winter might explain the relatively high concentrations detected in the wastewater influent since our data were collected in the winter with maximum levels of antibiotics. As a consequence, this made it possible to study their removal in different treatments in a WWTP.
Along the WWTP treatments, a similar decreasing trend from influent to effluent was observed for all WWTPs in February and March. The only exception was the elevated concentration in the activated sludge tank in WWTP A and WWTP B, where higher concentrations of quinolones (LEV, CIP, and NOR) and of oxytetracycline were detected.
For the WWTP effluent concentrations, 5 of the 52 analysed antibiotics were detected in all WWTP effluents. These detected antibiotics are CIP, SMX, SP, SF, and TRI in which sulfonamides (SMX, SF, SP) and TRI were consistently detected in all WWTPs for both sampling months. Sulfonamides and quinolones were detected in the effluent of all studied WWTPs. Poor removal of sulfonamides in WWTPs was also observed by Marx et al. (2015) and Bengtsson-Palme and Larsson (2016). Sulfonamides group are not easy to degrade, have a low potential to volatilize, are very hydrophilic (K ow b1) and are highly mobile in the sand and groundwater infiltration systems (log sorption-distribution coefficients (K d ) b 2) (Wegst-Uhrich et al., 2014;Kolpin et al., 2002). Therefore, sulfonamides are easily transferred into the aquatic environment, which can explain their high reported occurrence in the water phase (Xu et al., 2007;Jiang et al., 2018).

Antibiotics removal within conventional and additional treatment technologies
The efficiency of the removal of antibiotics in the WWTPs with additional treatment technologies (WWTP B and C) and without (WWTP A) was evaluated by comparing the water before and after the individual treatment processes. All the treatments in the respective WWTPs showed removal in the total load of antibiotics, regardless of conventional or additional treatment technology, although differences within the treatment steps in WWTPs were observed.
First, two conventional treatment steps were evaluated; activated sludge in WWTP A and WWTP B were compared. Concentrations of some antibiotics were significantly increased in concentration in the water phase after the activated sludge process, while others decreased in concentration (p b 0.05). For example, WWTP A showed a 177% increase of tetracyclines and 66% of quinolones, whereas 55% of sulfonamides and 74% of macrolides were removed. In WWTP B, the tetracyclines increased with 1650%, quinolones with 200%, whereas 29% of sulfonamides were removed. No removal of macrolides was observed.
In the activated sludge treatment steps of WWTP A and WWTP B, the concentration of quinolones and tetracyclines increased. The increase of both antibiotics in water are most likely due to quinolones and tetracylines being released from hydrolyzed organic waste fractions that enter the activated sludge and water phase with the influent. This is followed by a redistribution over the water and sludge phase by sorption processes, as also shown by another study Jia et al. (2012). We observed that the TSS at this treatment step was higher than in the other sampling points within the WWTP (Table S2), indicating that the amount of antibiotics attached to the particles in the wastewater could have been higher (p b 0.05). This is supported by the high correlation between TSS content and antibiotic concentration (Pearson correlation 0.95 in February and 0.96 in March). This indicates that the higher the particle concentration in the wastewater, the higher concentration of the antibiotics. CIP and TET are multivalent zwitterions with strong dipole and exhibited significant sorption capacity onto suspended solids and sludge in previous research (Polesel et al., 2015). Quinolone sorption is high (log K d N 3) and it adsorbs to sludge surfaces through electrostatic interactions (Golet et al., 2003). Concentrations up to 18.4 mg/kg have been measured in sludge (Jia et al., 2012) and up to 2.4 mg/kg of dry weight (Golet et al., 2003).
Other treatment steps of conventional treatment technology, such as the settling tank and subsequent sand filtration, were present in WWTP A and WWTP C. In the sedimentation tank, average removal fractions were respectively 94% for tetracyclines, 5% for sulphonamides and 94% for quinolones. In the sand filtration, the average removal of tetracyclines was 100%, 20% for sulphonamides, and 16% for quinolones. Macrolide concentrations increased after both the sedimentation tank and sand filtration.
The main removal mechanism in the sedimentation tank is the sorption of antibiotics on the colloidal matter, followed by removal in the coagulation/flocculation/sedimentation process (Adams et al., 2002;Shah, 2008). Xing and Sun (2009) showed that this resulted in 87% antibiotics removal after sedimentation and suggested this as an effective removal step to treat wastewater of antibiotics and pharmaceutical manufacturers. Low removal percentage in the sand filtration in WWTP C, which only removed 0.3% (quinolones) to 0.4% (sulfonamides), showed that antibiotics are largely unaffected by sand filtration. This is consistent with Rooklidge (2004), who also indicated that sand filtration removed b4% of sulfonamide and demonstrated limited mobility of lincomycin, trimethoprim, and tylosin within the sand filter. This is indicating that sand filter possesses low sorption properties and has a high persistence of microorganisms not adapted to biodegradation of these specific antibiotics (Ternes et al., 2002).
For the additional treatment technologies studied, the 1-STEP® filter removed 19% of sulfonamides, and 65% of quinolones, and produced concentrations of macrolides increased with 113%. No tetracyclines were detected before 1-STEP®. The present activated carbon removes antibiotics by physico-chemical adsorption onto the activated carbon and by the biofilm on the activated carbon (Ahmed, 2017;Östman et al., 2019). Activated carbon removes effectively hydrophobic compounds with a log K ow N 4, for example tetracyclines, and quinolones (Grandclément et al., 2017;NCBI, 2018;Raevsky et al., 2009). The removal observed in this study is lower than reported in other studies, although those were lab-scale studies (Choi et al., 2008;Zhang et al., 2016b). In a full-scale WWTP, the lifetime of the AC, the saturation level of the AC with other organic compounds, the water flow, hydraulic retention times, and the concentration of antibiotics in the influent will influence the removal percentage.
Meanwhile, NEREDA® removed 100% of tetracyclines, 36% of sulfonamides, 84% of macrolides, and 74% of quinolones. No increase of antibiotics was observed within the treatment, indicating little accumulation and consequent desorption within the treatment. This can be explained by the relatively high microbial activity of aerobic granules (Wang et al., 2019). The bacteria produce compact granules compared to flocs in conventional activated sludge and these granules settle faster in the wastewater (Forster, 2019). The aerobic granules are formed by bacteria that produce extracellular polymeric substances (EPS) and are stabilized by slow growing microorganisms (Świątczak and Cydzik-Kwiatkowska, 2018). EPS influence the surface properties of biomass and increase the sorption of organic pollutants Schmidt et al., 2012). Xu et al. (2013) reported that protein in EPS interact and bind with sulfamethazine by hydrophobic interaction, contributed to the stability of the complex, and improve the efficient removal of sulfamethazine by harvesting the sludge EPS. This is also supported by Pi et al. (2019), and these authors reported that the chemisorption and hydrophobic interaction of tryptophan and tyrosine during the binding process to EPS and sulfonamides played an important role in adsorption capacity. The removal percentage in our NEREDA® technology is in the same range as other studies of AGS Wang et al., 2019;Wang et al., 2018). Not only AGS has a promising step to remove antibiotics from wastewater, but also anaerobic granular and flocculent sludge showed a high removal of micropollutants (Butkovskyi et al., 2017).
When we compare 1-STEP® filter and NEREDA®, 1-STEP® filter removed 30% -52%, meanwhile NEREDA® removed from 68% -82%. The location of the treatment step may contribute to the removal difference. However, we did not observe any significant difference between the locations (p N 0.05). The 1-STEP® filter is located after activated sludge and acts as a polishing step of the treatment. This contributes to a low load of antibiotics in the system, when compared to NEREDA® which is located after the primary treatment. This indicates that the location of the treatment technology (either in the secondary treatment or as a polishing treatment) did not affect the removal percentage of antibiotics. The difference of the antibiotic concentration can be the results of other factors, such as concentration of the influent and WWTP operating parameters.

Occurrence and distribution of ARGs over the treatment phases in the three WWTPs
In this study, the 16S rRNA gene, the intI1 gene, and four ARGs (ermB, sul1, sul2, and tetW) were detected at all sampling points in all WWTPs ( Fig. 3 and Table S7), in water and in sludge samples. This illustrates the prevalence of ARGs along the phases of WWTPs in water and sludge.
Generally, the absolute abundance of ARGs in the influent ranged from 4.6 × 10 5 to 1.1 × 10 8 copies/mL in WWTP A, 1.5 × 10 5 to 3.1 × 10 7 copies/mL in WWTP B and 4.3 × 10 5 to 1.4 × 10 7 copies/mL in WWTP C respectively. The most abundant ARG in the influent of all WWTPs was ermB with a range of 3.7 × 10 7 to 2.5 × 10 8 copies/mL in WWTP A, 2.1 × 10 5 to 7.9 × 10 7 copies/mL in WWTP B and 5.6 × 10 7 to 2.3 × 10 8 copies/mL in WWTP C. The second most abundant ARG was intI1, followed by tetW, sul1, and sul2. The relative abundance of the ARGs to the 16 s rRNA gene shows a decreasing trend for all ARGs, except for intI1, which is stable in the treatment transect from the influent to the effluent (Fig. S2).
In sludge samples, ARGs were detected before, and after dewatering at 3.5 × 10 6 to 8.3 × 10 10 copies/g dry weight in WWTP A, 3.1 × 10 6 to 1.0 × 10 12 copies/g dry weight in WWTP B, and 1.8 × 10 5 to 1.2 × 10 11 copies/dry weight in WWTP C. Sul1 and intI1 were the most abundant in each treatment unit of all WWTPs, followed by ermB, tetW and sul2. ARGs concentration increased in WWTP A and C when compared before and after the dewatering system.
Unlike the antibiotics, the total amount of ARGs did not accumulate in the water phase in the activated sludge tank in WWTP A. We observed an average removal of 0.44 log from the water phase and -in parallel-we observed a slight increase (0.34 log) in the sludge phase. This was expected, as the sludge or sediments are known as hot spots of high bacterial density, activities and biofilm formations (Heß et al., 2018). This accumulated ARGs from the water phase to attach to the sludge since the majority of bacteria are known to live in association with surfaces (Davey and O'Toole, 2000). It has been shown that reduction of microbial biomass might correlate with the reduction of ARGs in the water phase and lead to an equivalent accumulation in the sludge phase . The precise mechanism behind this is not fully clear. As a result, ARGs will accumulate in the sludge and also in the sediment and soil Peng et al., 2018).
Along the WWTP, all ARGs except sul1 and intI1 showed a decreasing trend. Sul1 and intI1 increased slightly but not significantly after some steps. Sul1 increased with 0.26 ± 0.20 log in WWTP A after grit removal and activated sludge treatment, and 0.12 ± 0.22 log after activated sludge in WWTP B. intI1 increased 0.20 ± 0.14 log at the activated sludge in WWTP A.
ARGs were present in the effluent ranging from 2.7 × 10 3 (sul2) to 6.3 × 10 5 (intI1). intI1 was the most abundant gene in the effluents of all WWTPs. This is followed by ermB, sul1, tetW, and sul2 in WWTP A, sul1, ermB, tetW, and sul2 in WWTP B and ermB, sul1, tetW and sul2 in WWTP C. Since a major amount of the ARGs ends up in the sludge phase, lower ARGs concentrations were detected at the effluent as compared to the influent. All WWTPs significantly (P b 0.05) reduced the total ARGs (copies/mL) from the influent to the effluent. A similar range of reduction about 1-3 log removal was observed in China (Lee et al., 2017;Chen and Zhang, 2013), 2.4 to 4.6 log removal in Michigan (Munir et al., 2011) and b2 log removal in Italy (Fiorentino et al., 2019). Our study showed that ermB was the most removed gene in all three WWTPs, as also reported by Rafraf et al. (2016). ErmB genes have been found mainly in gram-positive bacteria (Gupta et al., 2003), and it has been shown before that gram-positive bacteria were removed from influent to effluent (Forster et al., 2002).
In the effluent, intl1 was the most detected gene in all three WWTPs. This was also observed by Narciso-da-Rocha et al. (2014), who suggested that intI1 is stable in wastewater. Furthermore, ARGs and the intI1 gene were not efficiently reduced during wastewater treatment (Rafraf et al., 2016). We also observed low log removal of sul1 (0.60-1.63 log) in the three WWTPs, and a similar log removal (0.9-1.9 log) was observed by Chen and Zhang (2013). The limited removal of both sul1 and intI1 (r = 0.81) and sul2 and intI (r = 0.93) were strongly correlated (p b 0.05), as shown by others, as sul1 is one of the backbone genes of the 39-conserved segments in intI1 (Muziasari et al., 2014;Partridge et al., 2002).
Through the sorption of ARG-carrying bacteria, sludge has the potential to act as ARGs reservoir and mitigate the spread of antibiotic resistance in the environment through effluents (Munir et al., 2011). This situation can also increase the exposure risks, especially in countries applying WWTP sludge for agricultural purposes, or producing and using products made from WWTP sludge materials.
In our study, the antibiotics and ARGs showed different patterns of reduction in the investigated treatments. In WWTP A, the total amount of antibiotics increased after the activated sludge treatment, while the concentration of ARGs decreased. The ARGs were removed in all treatment steps in the WWTP. There are inconsistencies in the literature in determining the correlation between antibiotics and ARGs. Some studies reported there is a correlation between presence and removal of antibiotics and ARGs Wu et al., 2010), and some studies showed no or partial correlations (Gao et al., 2012;Xu et al., 2015). In this study, we did not find such correlations either. This is maybe due to the different environments and pollution levels associated to the three different full-scale wastewater treatment systems. Therefore, further and more extensive studies for a multitude of full-scale WWTPs should be performed in order to provide a better insight into the absence or presence of generic correlations between the removal from effluents and the accumulations into sludges of antibiotics, ARB, and ARGs.
Overall, all WWTPs reduced ARGs significantly (P b 0.05), with respectively 2.0, 1.3, and 2.3 log ARGs for WWTP A, B, and C ( Fig. 4 and Table S8). The highest removal was found for ermB, respectively 2.92, 2.22, and 3.11 log for WWTP A, B, and C. Finally, sul1 and intI1 were least removed in all WWTPs, with b1.4 log removal.

ARGs removal in conventional and additional treatment technologies
The efficiency of conventional and additional treatment technologies in removing ARGs was evaluated by comparing their presence in the water and sludge at different stages. First, two conventional treatments were evaluated, activated sludge in WWTP A and WWTP B.
Unlike antibiotics, ARGs did decrease significantly after the activated sludge process. In WWTP A, all ARGs decreased (except sul1 and intI1), Fig. 4. Log removal of ARGs in the water at different sampling points in the three studied WWTPs. Error bars indicate the standard error of the respective gene data set (duplicate). ranging from 0.26 log (sul2) to 1.29 log (ermB). Sul1 increased 0.26 log while intI1 increased 0.20 log. In WWTP B, conventional treatment removed 0.53 log sul2 to 2.45 ermB, whereas sul1 increased 0.12 log.
In the sedimentation tank, the average removal for ermB was 1.63 log, 1.45 log for sul1, 1.47 log for sul2, 1.66 log for tetW, and 1.26 log for intI1. For sand filtration, average removal for ermB was 1.75 log, 0.84 log for sul1, 0.71 log for sul2, 0.77 log for tetW, and 0.53 log for intI1. Interestingly, the sludge settling (sedimentation) tank of WWTP A showed the highest log removal in the water phase among the studied treatments. This implies that the sedimented sludge contains a large amount of the ARGs, as also observed by others (Nnadozie et al., 2017;Lee et al., 2017;Su et al., 2018). The final treatment in WWTP A, the sand filtration, also decreased the concentrations of ARGs. Similar removal in conventional WWTPs was shown by Hu et al. (2018), e.g. sul1, sul2, and intI1 were removed in the flocculation, sedimentation, and sand filtration tank.
The 1-STEP®-filter removed −0.03 log (tetW) to 0.55 log (ermB). Overall, 1-STEP® showed the least removal of ARGs. Its activated carbon filter is known for removing organic contaminants, natural organic matter, humic and fulvic acids, and biodegradable compounds. However, poor removal in this study indicated that the ARGs did not adsorb onto the activated carbon may be due to the majority of genes being present in viable bacterial and non-adhering cells, that were not removed from the water phase during filter passage.
The additional treatment technology NEREDA® removed 0.08 log (ermB) to 2.02 log (tetW) and showed the second highest log removal of ARGs. The granules in the NEREDA® retain organic waste fractions and bacteria in close proximity to each other, thus allowing interactions to occur, including cell-cell communication, and the formation of synergistic microbial consortia (Flemming and Wingender, 2010). Furthermore, the excellent settling properties result in high biomass concentrations (Liu et al., 2003). These properties result in the accumulation of the ARGs within the NEREDA® granules, those granules sink at the bottom and reduce the concentration ARGs in the water. The capture mechanism of ARGs either in solution, bound to suspended solids, or as present of in free bacterial cells by NEREDA granules, is yet to be defined.
The total ARGs in our study were increased in the sludge with an average of 1.26 log in WWTP C. The NEREDA® granules consist of EPSproducing bacteria such as bacteria belonging to the order Xanthomonadales, Sphingomonadales, and family of Rhizobiales (Hyphomicrobiaceae) (Świątczak and Cydzik-Kwiatkowska, 2018). EPS also has the potential to control the lateral transfer of ARGs. This may result in an accumulation of ARGs and indicates that sludge can represent a sink for resistant bacteria and might become an important reservoir for the ARGs (Zhang et al., 2016a).
This study, however, is subject to two critical points; the sampling method (grab sampling) and the limited duration of the sampling campaign (two months). Grab sampling may help in determining the presence of the compounds of interests; however, it captures the concentration of antibiotics and ARGs at a specific time. Furthermore, the data only represent two months in winter, and the result might differ in different seasons throughout the year. However, our approach provides data and insights on the performance of WWTPs with additional treatment technologies (in this study, 1-STEP® and NEREDA®) in removing antibiotics and ARGs. Future research could, for instance, perform studies during a longer time (e.g., 1 year) by using composite sampling. Such research could contribute to better understanding the performance of WWTPs with additional treatment technology over time. Improvement or upgrading of the treatment technology can then be more specifically proposed. Furthermore, only 1 WWTP with advanced treatment options per type of treatment was available for this study, basically because the number of full-scale installations with these additional treatments in the Netherlands is limited, due to their innovative character. As a result, the results might be affected by local sewage parameters. Results from additional WWTPs with similar treatments are therefore needed.

Implications on public health, water industry, and regulations
Clean water as a source for drinking water is increasingly becoming limited due to climate change, urbanization, and growing populations in the world. Therefore, wastewater reuse is considered as an alternative to tackle this problem (Angelakis et al., 2018). However, the increasing presence of antibiotics, ARB, and their associated ARGs in water are of concern (Hong et al., 2013). Water pollution has been listed as one of the top three concerns in water industry, together with climate change and political instability, from a survey conducted by American Water Works Association (AWWA, 2019). However, there are currently no legal regulations or guidelines that define the permitted levels of antibiotics or antibiotic resistance determinants that are allowed into the environment (Pazda et al., 2019).
This study shows that antibiotics and ARGs are present in WWTP effluent, even with additional treatment technologies. Such technologies can induce the mitigation of antibiotic and ARGs emissions to a limited extent. Here, we show that the removal efficiency of additional activated carbon and AGS differs. Therefore, techniques for advanced treatment should be chosen carefully, depending on the micropollutants targeted in a specific situation. For example, AC is not recommended when there is recreational water downstream and ARGs removal is needed. If limited human exposure to ARG is intended, AC only modestly increases ARG removal according to our results, and is thus insufficient. The wastewater macro-and micropollutants (antibiotics and ARGs included) have a high impact on public health if the removal is insufficient and discharged effluents are directly or indirectly reused water for irrigation, washing and drinking water preparation (Helmecke et al., 2020).

Conclusion
In this study, the removal of antibiotics and ARGs were studied in water and sludge of three WWTPs with (1-STEP® filter and NEREDA®) and without additional treatment technologies. Total concentrations of 3000 ng/L antibiotics were found in the influent and decreased over the different treatment steps to b1000 ng/L in the effluent. Tetracyclines and quinolones concentrations were elevated in the water after the activated sludge treatment step. This shows that these compounds were able to adsorb and desorb in the sludge, with the activated sludge acting as a reservoir for quinolones. Generally, good removal (79-88%) of total antibiotics were observed at all WWTPs. However, sulfonamides and quinolones were still present in the effluent in all three WWTPs, with or without additional treatment technologies.
All WWTPs showed 1-2 log removal for the analysed ARGs from influent to effluent. Of the measured ARGs, ermB was most abundant in the influent, and the most removed ARG in the three WWTPs. WWTPs with or without additional treatment technologies were able to reduce antibiotics with a similar efficiency, although ARGs were best removed in NEREDA®, followed by conventional treatment. This is the first study of removing antibiotics and ARGs in NEREDA®. The 1-STEP® filter decreased the concentrations of ARG with up to 0.5 log extra on top of the reduction in the conventional part of the plant. When looking at specific treatments, the sedimentation tank showed the highest log removal of ARGs. In the activated sludge, a relatively higher concentration of ARGs was detected compared to other treatment steps, suggesting that sludge is an important reservoir and transmission point for the ARGs. This study demonstrates that in most cases, WWTP with additional treatment technologies have the potential to provide a higher removal of both antibiotics and ARGs compared to conventional WWTP. Further research is needed to identify and optimize the most suitable treatment technology, and further reduce spreading of antibiotics and ARGs via WWTPs into the environment.

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.