Microbial community response to silver nanoparticles ...

Science of the Total Environment 616–617 (2018) 1014–1021

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Microbial community response to silver nanoparticles and Ag+ in nitrifying activated sludge revealed by ion semiconductor sequencing Carley A. Gwin a,b,c, Emilie Lefevre a, Christina L. Alito a,b, Claudia K. Gunsch a,b,⁎ a b c

Department of Civil and Environmental Engineering, Duke University, Box 90287, Durham, NC 27708, United States Center for Environmental Implications of NanoTechnology (CEINT), Duke University, Durham, NC 27708, United States Department of Civil and Environmental Engineering, Bucknell University, 1 Dent Drive, Lewisburg, PA 17837, United States

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• Gum Arabic-coated silver nanoparticles had long lasting impact on diversity and community structure. • High proportion of Zoogloea-related OTUs were detected. • Increase of silver nanoparticles in wastewater treatment plants has potential for increased sludge bulking. • Silver nanoparticles, not ionic silver, negatively affected Bacteroidetes.

a r t i c l e

i n f o

Article history: Received 28 August 2017 Received in revised form 20 October 2017 Accepted 21 October 2017 Available online 6 November 2017 Editor: D. Barcelo Keywords: Silver nanoparticles Wastewater Nutrient cycling Ion Torrent Next generation sequencing Metagenomics

a b s t r a c t Silver nanoparticles (AgNPs), which are known to act as biocides, are incorporated into medical and consumer products including athletic clothing, stuffed animals, liquid dietary supplements, and more. The increasing use of AgNPs in these products is likely to lead to their entry into both natural and engineered systems, which has the potential to disrupt bacterial processes including those involved in nutrient cycling in wastewater treatment. In the present study, sequencing batch reactors (SBR) mimicking secondary wastewater treatment were operated to determine the effects of AgNPs on the microbial communities contained within activated sludge of wastewater treatment plants (WWTP). SBRs were treated with 0.2 and 2 ppm of either gum Arabic (GA)-coated AgNPs, citrate (Ca)-coated AgNPs, or Ag+ as AgNO3. Cell samples were collected and DNA isolated periodically throughout SBR operation. DNA was used for Ion Torrent Next Gen Sequencing of the V3 region of the 16S rDNA gene. Subsequent analyses revealed that the microbial community both shifted and recovered quickly in response to Ag+. Both AgNP treatments resulted in slower initial community shifts than that observed with the Ag+ treatment. GA-AgNPs elicited the longest lasting effect. Additional examination of nitrogen removal bacteria suggested the possibility of an increase in sludge bulking species with increased concentrations of AgNPs in WWTPs. This study supports the hypothesis that Ag+ release from AgNPs is largely coating-dependent and thus a key driver in dictating AgNP toxicity. © 2017 Elsevier B.V. All rights reserved.

1. Introduction ⁎ Corresponding author at: Department of Civil and Environmental Engineering, Duke University, Box 90287, Durham, NC 27708, United States. E-mail address: [email protected] (C.K. Gunsch).

http://dx.doi.org/10.1016/j.scitotenv.2017.10.217 0048-9697/© 2017 Elsevier B.V. All rights reserved.

The entrance of silver nanoparticles (AgNPs) into the waste stream from manufacturing and use of consumer products containing AgNPs

C.A. Gwin et al. / Science of the Total Environment 616–617 (2018) 1014–1021

(e.g., textiles, dietary supplements, wound dressings (Silver, 2003) sippy cups (Tulve et al., 2015)) has become an issue of concern as both ionic silver (Ag+) and AgNPs exhibit strong antimicrobial properties (Fabrega et al., 2011; Mijnendonckx et al., 2013; Garner and Keller, 2014). Because wastewater treatment depends heavily on biological nutrient removal, with secondary treatment generally relying on the activated sludge process, the presence of silver in these systems is likely to have adverse effects on the activity and structure of the microbial communities at play. In particular, it is important to determine the impact of AgNPs on nitrogen removal as this process is a critical step in biological wastewater treatment and is carried out by highly sensitive microorganisms (Daims and Wagner, 2010; Arnaout and Gunsch, 2012; Alito and Gunsch, 2014). Nitrification relies on two distinct microbial processes: the sequential conversion of ammonia to nitrite and ultimately to nitrate under aerobic conditions mediated by ammonia-oxidizing bacteria (AOB) and nitrite-oxidizing bacteria (NOB), respectively (Daims and Wagner, 2010) and denitrification, catalyzed by facultative anaerobic heterotrophs using organic compounds as electron donors to reduce nitrite or nitrate to nitrogen gas (Daims and Wagner, 2010). Nitrification is more vulnerable than denitrification to AgNPs exposure (Jeong et al., 2012; Yuan et al., 2013). Nitrosomonas europaea, an AOB commonly found in nitrifying sludge, was reported to be particularly sensitive to Ag+ and AgNPs (Arnaout and Gunsch, 2012; Yuan et al., 2013), suggesting that ammonia oxidation is the biological process in wastewater treatment plants (WWTP) most likely to be affected by the entrance of silver. In these systems where microbial species are organized in complex communities, the effect of Ag+ and/or AgNPs on nitrification likely differs from that observed on monocultures maintained under laboratory-controlled conditions. Thus, it is critical to ascertain effects in complex systems. Over the last two decades, the direct application of molecular community fingerprinting techniques to complex microbial communities inhabiting WWT systems, has revealed that anthropogenic perturbations could have substantial impacts on the composition of microbial communities inhabiting these systems (Saikaly et al., 2005; Wang et al., 2008; Baek et al., 2010; Ye and Zhang, 2010; Wang and Gunsch, 2011; Wang et al., 2011; Yang et al., 2011; Sun et al., 2013; Zhang et al., 2014). Particularly, Alito and Gunsch (2014) who used T-RFLP to evaluate the effect of Ag+ and AgNPs on wastewater activated sludge microbial communities in bench-scale sequencing batch reactors (SBRs), showed that repeated exposure to silver resulted in profound changes in microbial community structure despite temporarily decreased ammonia removal efficiencies. However, due to the low resolution and lack of taxonomic information provided, a comprehensive characterization of the microbial community including taxonomic information of its members could not be undertaken. The relatively recent application of Next-Generation Sequencing (NGS) to waste treatment systems, including anaerobic piggery waste digesters (Whiteley et al., 2012), bench-scale digesters (Lefevre et al., 2016), nitrifying benchscale SBRs (Ma et al., 2015) and microcosms (Yang et al., 2014) as well as full scale WWTP (Shu et al., 2015) has expanded our understanding of microbial diversity, pathways, and the factors affecting these systems. In the present study, we applied a NGS method, namely ion semiconductor sequencing (i.e., Ion Torrent), to identify prokaryotic responders to AgNPs in a model wastewater treatment system with the goal of providing a more complete understanding of the potential ecological consequences of AgNPs to wastewater treatment microbiomes. 2. Materials and methods 2.1. Experimental setup The experimental setup used in this study is described in detail in Alito and Gunsch (2014). Briefly, eight 3 L SBRs were seeded with 0.5 L of nitrifying activated sludge from the North Durham Water Reclamation Facility (Durham, NC), covered in aluminum foil to prevent

1015

photoactivity, and continuously fed with synthetic wastewater (Zeng et al., 2003) (SWW). During operation the volume varied from 1 (post-decantation) to 2 L during the 8-h cycle, which included 30 min influent feeding, 6 h aeration and mixing, 1 h settling, and 30 min decantation. The hydraulic retention time (HRT) was 5.3 h and the solid retention time (SRT) was approximately 13 days. Sludge was wasted to maintain a concentration between 2000 and 2500 mg/L of mixed liquor suspended solids (MLSS). The dissolved oxygen (DO) was consistently between 6 and 6.5 mg/L. After the reactors had been operated at steady state (consistent ammonia and COD removal greater than or equal to 90%) for 30 days (120 days total), four treatments were applied in duplicate. The treatments were chosen based on preliminary experiments as well as the results of Arnaout and Gunsch (2012), where 0.2 ppm was the lowest concentration at which an effect was observed (AgNO3 at 0.2 ppm resulted in decreased membrane integrity of Nitrosomonas europaea), while 2 ppm total silver concentration resulted in nitrification inhibition. Current treatments consisted of: (1) citratecoated (Ca) AgNPs (15.5 ± 0.5 nm), (2) gum Arabic-coated (GA) AgNPs (32.3 ± 0.5 nm), (3) Ag+ as AgNO3, and (4) no silver control. Initial treatments consisted of 3 consecutive pulses of 0.2 ppm, each succeeded by a decreasing recovery period of 14, 6, and 2 days, and followed by a 0.2 ppm-continuous addition lasting three SRTs (approximately 39 days). Immediately after, the same regimen was repeated with 2 ppm additions of silver (Fig. 1). For pulse additions, silver was added directly to the reactors, but was added to the SWW influent for continuous addition. During the experiment, SBRs were monitored for treatment efficiency via COD and ammonia removal approximately every 3 days and following silver inputs, and samples were collected for silver characterization by ICP-MS and microbial community analysis. 2.2. Microbial community analyses Biomass samples were collected from the SBRs at 6 distinct sampling time points (Fig. 1). Samples were microcentrifuged for 1 min at 13000 ×g. The resulting cell pellets were then immediately stored at −20 °C until DNA extraction. DNA was extracted in duplicate then combined for each treatment using the UltraClean DNA Isolation Kit (MoBio Laboratories, Solana Beach, CA). Genomic DNA concentration was quantified using a Qubit ® 2.0 Fluorometer and the dsDNA BR Assay Kit (ThermoFisher Scientific; Waltham, MA, US). The hypervariable V3 region of the 16S rDNA gene (~150 bp) was amplified from a total of 24 samples using the 341F and 518R bacterial primers (Muyzer et al., 1993) modified with Ion Torrent adapter and Golay barcode sequences as described in Whiteley et al. (Whiteley et al., 2012). For each sample, 4 replicates of 50 μL PCR reactions were prepared, each consisting of 100 ng of genomic DNA, 200 nM of each primer, 200 μM of each dNTP, 2.5 units of Affymetrix FideliTaq DNA polymerase, 5 μL of 10× PCR reaction buffer provided with the Taq polymerase (Affymetrix, Cleveland, OH, US), and 0.5 mg/mL of bovine serum albumin (BSA). PCR amplifications were performed on a BioRad T100 thermal cycler using the following optimized conditions: an initial denaturation for 2 min at 95 °C, followed by 30 cycles of 30 s denaturation at 95 °C, annealing for 1 min at 60 °C, and elongation for 1 min at 68 °C, and a final elongation of 5 min at 68 °C. The PCR reactions were pooled, purified, and concentrated using the NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel; Bethlehem, PA, US). PCR products were eluted in 35 μL of molecular grade water and separated by electrophoresis on a 2.3% NuSieve 3:1 agarose (Lonza; Walkersville, MD, US) gel to verify the fragment size, ~ 220 bp (including the ~ 150 bp V3 region and the 69 bp Ion Torrent modified forward and reverse primers) and eliminate primer excess. Fragments were excised and gel-extracted using the Macherey-Nagel NucleoSpin Gel and PCR Clean-up kit. PCR products were eluted in 50 μL of molecular grade water and quantified using a Qubit 2.0 fluorometer. For each sample, equimolar amounts of PCR product were pooled and again gel purified. The sample was sequenced at the Duke IGSP Genome Sequencing and Analysis Core Facility (Durham, NC, US) on an Ion

1016


Fig. 1. Treatments and sampling regime. The silver concentrations applied as pulse and continuous additions are indicated above the timeline by vertical, and horizontal grey arrows, respectively. The 6 time points analyzed in this study using Ion Torrent sequencing are indicated below the timeline, in days, by back arrows. The total length of the experiment was 122 days.

Torrent PGM sequencing platform with the Ion 316 Chip v2 using the Ion PGM Template OT2 400 Kit (Life Technologies). The version 4.4.1 of the Torrent suite base-calling software was used to generate the read output file. 2.3. Ion Torrent data analyses QIIME (Caporaso et al., 2010) and PRIMER v6 (Clarke and Gorley, 2006) were used to analyze the reads generated by Ion Torrent sequencing (details of the QIIME scripts used can be found in Fig. S1). Reads were first filtered based on length and mean quality scores. Sequences were clustered together in operational taxonomic units (OTUs) using Uclust (Edgar, 2010) with a similarity cutoff of 97%. A total of 4 samples (day +1 Ca-AgNPs, day +7 control, day +15 control, day +76 Ca-AgNPs) were excluded from final analysis due to low read counts (470, 519, 622, and 776 reads, respectively) and non-asymptotic OTU rarefaction curves (Fig. S2). From a total of 501,717 raw reads obtained, 195,370 high quality sequences (9768.5 reads per sample on average) that clustered into 2082 OTUs remained after the removal of singletons. Community comparison analyses were performed using a CSS-transformed Bray-Curtis similarity matrix (Paulson et al., 2013). To visualize the degree of similarity in terms of community composition between samples, principal coordinate analyses (PCoA) were performed in QIIME. Additionally, analyses of similarity (ANOSIM) for a two-way layout structure design without replication (Clarke and Warwick, 1994) were performed in PRIMER v6 to test for the effect of treatment and time on the microbial community composition. 3. Results and discussion 3.1. SBR treatment efficiency Impacts on treatment performance were minimal even at the highest concentrations suggesting substantial functional redundancy allowed SBR communities to recover their functions relatively quickly (Fig. S3). COD removal efficiency for all SBRs never fell below 71% (GA-AgNP, first 0.2 ppm pulse), and the removal efficiencies did not differ statistically (p N 0.05) from the control SBRs'. Only a slight decrease, albeit not statistically significantly lower, in ammonia removal efficiency was measured in the Ca- and GA-AgNP treated SBRs after the first 0.2 ppm pulse addition (dropping from 98% pre-treatment to post-treatment values of 97% and 88%, respectively), but, in the Ag+ treated SBRs, the decrease was far greater, dropping from 98% down to 32%. However, within 24 h, AgNP-treated SBR function recovered to approximately 95% removal efficiency. The continuous 0.2 ppm feed did not affect ammonia removal except in the Ag+ treated SBRs; overall all SBRs were able to recover ammonia removal function. 3.2. Bacterial community α-diversity A total of 194,370 high-quality sequences that clustered into 2082 bacterial OTUs were obtained, which is comparable to previous studies using NGS on nitrifying sludge samples (Yang et al., 2014; Zhang et al., 2016). Chao1 individual-based rarefaction curves (Fig. S2) indicated that, with the exception of four samples that were excluded from the analyses due to low read counts, microbial communities were relatively

well sampled, allowing for diversity comparison analyses. Additionally, the Good's coverage value indicated that, on average, 96.5% of the diversity present in our samples was detected using Ion Torrent (Table 1). Shannon indices ranged from 1.90 to 6.55, which again are similar to what others have measured in nitrifying sludge samples (Yang et al., 2014; Zhang et al., 2016). Simpson‘s (1D) diversity indices were also calculated to compare the diversity obtained using NGS to the T-RFLPbased diversity previously reported in our lab (Alito and Gunsch, 2014). Although the number of samples analyzed here were fewer than those analyzed in our previous study, the species richness detected by NGS (i.e., # Observed OTUs and T-RFs in Table 1) was between 6 and 160 times higher than that detected by T-RFLP, confirming that T-RFLP largely underestimated the diversity present in the SBRs. The present study revealed an interesting trend in terms of changes in microbial diversity over time. Specifically, all treated SBRs, particularly the ones treated with GA- and Ca-AgNPs, experienced a decrease in microbial diversity during the recovery time between the first and the second 0.2 ppm-pulse addition (i.e., between Day + 1 and Day +13), likely suggesting that the first pulse addition of silver had a negative impact on microbial diversity. Others have observed similar trends via DGGE with the addition of AgNPs to an osmotic membrane bioreactor (Tan et al., 2015), as well as with the addition of carbon-coated AgNPs to activated sludge microcosms analyzed using pyrosequencing (Yang et al., 2014). At Day +15 (i.e., after the second 0.2 ppm-pulse addition) and Day + 76 (i.e., after microbial communities had been exposed to all 0.2 ppm treatments and two 2 ppm pulses), SBRs treated with Ca-AgNPs and Ag+ experienced an increase in microbial diversity, whereas the SBR treated with GA-AgNPs had not recovered from its first drop in diversity, suggesting that the impact of GA-AgNPs on the microbial diversity lasted longer compared to the other forms of silver. This may be a result of the difference in coatings; coating properties have been shown to play a role in the dissolution rate of Ag+, and may render a particular AgNP more stable or allow another to exhibit more toxicity (Yuan et al., 2013). Similarly, Choi and Hu (2008) found that AgNPs displayed the highest toxicity to nitrifying bacteria over silver chloride colloids or Ag+ additions, and Yang et al. (2014) saw a greater impact of carbon-coated AgNPs than with PEG-coated AgNPs or Ag+ as AgNO3. In terms of taxonomic composition of all OTUs found in all four SBRs, OTUs affiliated within 26 phyla, 57 classes, 104 orders, 169 families, and 234 genera. Similarly, the OTUs from Ma et al. (2015) corresponded to 23 phyla and 315 genera, and 13 phyla were identified by Zhang et al. (2016). Proteobacteria was by far the most dominant phylum, comprising on average 70.8% of the total reads generated (Fig. S4), which is much higher than what is usually found in nitrifying reactors (Zhang et al., 2012; Ma et al., 2015; Ramirez-Vargas et al., 2015), with the exception of 61.76% reported by Zhang et al. (2016). Proteobacteria were largely dominated by β-proteobacteria (70%, Fig. S4), but the αproteobacteria proportion was 7 times lower. β-proteobacteria are very common in activated sludge systems where their dominance over α-proteobacteria is generally indicative of an activated sludge rich in organic material (Ramirez-Vargas et al., 2015). Bacteroidetes, which comprised 22.6% of the reads obtained in this study, was the second most dominant phylum. Others have found similar community distributions in wastewater sludge samples Jeong et al., 2014, Ma et al., 2015, Ramirez-Vargas et al., 2015, Shu et al., 2015, Zhang et al., 2016), as Bacteroidetes have an important role in the degradation of

1017

C.A. Gwin et al. / Science of the Total Environment 616–617 (2018) 1014–1021 Table 1 Alpha diversity metrics. Samples Control: Day −18 Day +1 Day +7* Day +13 Day +15* Day +76 Gum Arabic AgNPs Day −18 Day +1 Day +7 Day +13 Day +15 Day +76 Citrate AgNPs Day −18 Day +1* Day +7 Day +13 Day +15 Day +76* Ionic silver Day −18 Day +1 Day +7 Day +13 Day +15 Day +76

Observed OTUs

687 178 88 105 96 122 487 241 111 103 174 161 433 101 191 205 187 93 874 173 882 119 274 484

Observed T-RFs (Alito and Gunsch, 2014)

4

8 5

3

21 19

6

19 9

27

36 3

Good's coverage (%)

Shannon diversity

Simpson diversity

98.5 94.4 90.6 95.1 91.3 96.1

4.12 3.06 5.00 3.06 5.12 3.41

0.814 0.652 0.946 0.692 0.951 0.732

99.2 94.8 95.9 95.7 97.2 96.7

4.67 4.87 2.84 2.87 2.84 2.76

0.904 0.868 0.645 0.664 0.720 0.526

99.0 89.5 96.0 98.0 94.7 93.7

2.15 5.56 3.45 1.90 4.31 4.46

0.384 0.956 0.695 0.446 0.818 0.912

99.1 95.3 99.6 93.7 92.8 97.3

4.08 3.67 3.42 3.91 6.55 6.01

0.767 0.836 0.729 0.737 0.973 0.949

Simpson diversity (Alito and Gunsch, 2014)

0.597

0.491 0.251

0.252

0.867 0.895

0.708

0.841 0.701

0.846

0.926 0.049

ime points marked by an asterisk correspond to time points analyzed in Alito and Gunsch (2014) using T-RFLP.

recalcitrant material such as cellulose and chitin, as well as in the uptake of high molecular weight organic compounds. Bacteroidetes have also been found to efficiently utilize extracellular polymeric substances and byproducts excreted by nitrifying bacteria (Ramirez-Vargas et al., 2015). The relative abundance of the other 24 detected phyla, including phyla frequently found in WWTPs such as Firmicutes, Chlorobi, Acidobacteria, Chloroflexi, and Actinobacteria, never exceeded 1% (Fig. S4). Another commonly observed feature of the microbial communities sampled in WWTPs, but particularly amplified in this study, is that only very few OTUs were highly represented; the large majority of the combined OTUs from all four SBRs were represented at very low abundance. Specifically, while four out of 2082 OTUs accounted for 58.8% of the total abundance, nearly 99.5% of the recovered OTUs (i.e., 2071 OTUs) each had a relative abundance below 1% (Fig. S5). The most abundant OTU in our reactors represented 25.5% of the reads generated overall, and affiliated to a β-proteobacteria genus, Zoogloea sp. (Fig. S5). Zoogloea ramigera is one of the most common and efficient denitrifiers found in WWTP, but is also well-known to form aggregates of cells embedded in extracellular gelatinous matrices, responsible for sludge bulking. Compared to other studies, an unusually high proportion of Zoogloea-related OTUs, represented by 296 total OTUs and accounting for 51.8% of the reads, was detected. However, this result is consistent with Ma et al. (2015) who reported that Zoogloea was the most resistant genus to high-load AgNP-treated SBRs. This high abundance of Zoogloea may have resulted as our SBRs were continuously fed with a synthetic wastewater medium for a long period of time (i.e. over four months by the time of our Day +1 time point), which could have allowed the selection and enrichment of Zoogloea spp. For example, between 19 and 61% of bacterial isolates were identified as members of the Zoogloea genus during an isolation experiment performed on activated sludge samples (Dias and Bhat, 1964). 3.3. Bacterial community β-diversity A PCoA analysis and an ANOSIM were performed on the whole bacterial community (Fig. 2) to determine if silver addition was the main

factor driving microbial community shifts in our SBRs. The ANOSIM revealed the addition of silver significantly affected the microbial community composition (i.e., R value of 0.667 (p b 0.05); Fig. 2). All samples collected 18 days prior to treatment application clustered together in

Fig. 2. Bray-Curtis matrix-based Principal Coordinate Analysis (PCoA), and results of the Analysis of Similarity (ANOSIM). The percentage of variation explained for the x and yaxis are indicated on the graph. C = Control; GA = Gum Arabic; Ca = Citrate; Ag = Ionic silver. Samples Ca +1, Ca +76, C +7, and C +15 were excluded from the analysis due to low read counts. The table below summarizes the results of the ANOSIM analysis performed. The null hypothesis (H0) states that there are no differences between groups in terms of community composition. The factors tested are treatment (i.e., C, GA, Ca, Ag) and time (Day +1, +7, +13, +15, +76). H0 is rejected if p ≥ 0.05. If H0 cannot be rejected, an R-value is calculated. An R-value close to 1 indicates an important difference between the groups tested, and an R-value close to 0 indicates a small difference between the groups tested in terms of community composition.

1018


Fig. 3. Relative taxonomic distribution of the reads obtained for each sample at the phylum level, and class level for Proteobacteria. The overall most dominant genus, Zoogloea (βproteobacteria) is also represented.

the PCoA (i.e., C −18, Ca −18, GA −18, and Ag −18; Fig. 2). The samples collected from the control SBRs at Day + 1, + 13, and + 76 clustered tightly together (Days + 7 and + 15 were excluded due to low read counts), but community composition varied from the Day − 18 samples (Fig. 2), suggesting that the control microbial community underwent an important shift in composition between Day − 18 and Day + 1. One possible explanation for this shift could be involuntary or uncontrolled changes in the laboratory environment or in the operation of the SBRs (e.g., influent flow, temperature), or perhaps overall differences in the laboratory operation of the SBRs as compared to the WWTP the microbes were sourced from, including feed composition and strategy. After Day +1, however, the microbial composition in the control SBR remained relatively unchanged until the end of the experiment. In the SBR treated with GA-AgNPs, one day following the first 0.2 ppm pulse addition, the community composition remained similar to that of the control (i.e., GA +1 is close to C + 1; Fig. 2), suggesting the absence of an immediate effect from GA-AgNPs. Similarly, the

second 0.2 ppm-pulse addition did not result in significant microbial compositional changes (between Day + 13 and Day + 15; Fig. 2). Seven days post the first pulse addition (Day +7), the microbial community composition shifted away from that of the corresponding control sample until Day +15, as indicated by the tight cluster formed by GA +7, +13, and +15 (Fig. 2). Therefore, although it was not immediate, the effect of GA-AgNPs on the microbial community composition resulted in relatively long-lasting compositional changes. In the SBR treated with Ca-AgNPs, at Day +7 the microbial community was still very similar to that of the corresponding control sample, and at Day + 13 the community composition greatly shifted away from that of the control and toward a community more similar to the corresponding GA-AgNPs sample. This suggests that the effect of Ca-AgNPs on community composition was similar to that of GA-AgNPs but delayed (i.e., the shift occurred at Day + 13 for Ca-AgNPs, while it occurred at Day + 7 for GA-AgNPs). In the SBR treated with Ag+, the effect of the first pulse addition was immediate and very apparent, as seen on the PCoA

Fig. 4. Overall distribution and relative abundance of 16S rDNA V3 regions reads affiliated to known nitrifiers (green) and denitrifiers (blue/grey).


1019

Fig. 5. Relative abundance (%) of Nitrifying OTUs detected using Ion Torrent. While 12 OTUs related to the Nitrosomonadaceae family were detected as the only ammonia-oxidizing bacteria, only one OTU related to Nitrospira sp. was detected as nitrite-oxidizing bacteria.

by the distance between Days +1 and +7, and the corresponding control samples. Alito and Gunsch (2014) observed a similar trend noting that Ag+ addition exhibited the most immediate effects on microbial communities, while AgNPs showed a more gradual shift. Although other mechanisms, such as reactive oxygen species generation, might also be involved in AgNPs' antimicrobial properties (Kim et al., 2007; Rai et al., 2009; Arnaout and Gunsch, 2012), it is likely that the progressive release of Ag+ from AgNPs as their coatings dissolve is the main mechanism underlying AgNP toxicity (Arnaout and Gunsch, 2012; Mijnendonckx et al., 2013; Yang et al., 2014). Therefore, this phenomenon may explain the delayed effect of GA- and Ca-AgNPs on microbial community composition compared to Ag+. At Day +13, however, the community exposed to Ag+ became similar to that of the corresponding control, suggesting that the community had recovered from the first Ag+ pulse addition, and the second Ag+ 0.2 ppm pulse addition did not affect the community composition as significantly as the first (e.g., Ag + 15 still relatively close to the control samples; Fig. 2), suggesting that the community had quickly built a certain resistance to the effects of Ag+. Graves et al. (2015) showed that Escherichia coli could evolve resistance to silver at the genomic level in only ~ 37.5 days (i.e., 225 generations) in monoculture when exposed to Ca-AgNPs and Ag+. Therefore, it is conceivable that in a complex community such as the SBRs studied here, resistance could develop even faster through a variety of methods including horizontal gene transfer of silver resistance genes (Li et al., 2015); the co-occurrence of arsenic, copper, and silver resistance genes on the same plasmids (Pal et al., 2015); or adaptive behavior such as the formation of cell aggregates (Khan et al., 2011; Yuan et al., 2013); or an increase in EPS production (Zhang et al., 2014). Regardless of the type of silver added, at Day +76, after the treated community had been exposed to silver multiple times, the community composition eventually returned to one similar to that of the corresponding control samples (i.e., GA-AgNP + 76 and Ag+ + 76 are closer to the control samples), demonstrating that the SBR microbial community composition is resilient (Allison and Martiny, 2008). Finally, it is important to note that despite the temporary compositional changes experienced by the microbial communities exposed to silver revealed in the present experiment, treatment performance was only transiently affected suggesting that the SBR microbial composition was likely functionally redundant. When comparing the taxonomic composition of the communities between the control and the treated SBRs, differences were noted (Fig. 3). While the proportion of β-proteobacteria slightly decreased (from 59.3 to 46.3%) in the control SBR between Day + 1 and Day + 13, the β-proteobacterial population substantially increased in the SBRs treated with both GA- and Ca-AgNPs (from 48.7 to 79.4%, and from 47.7 to 82.5%, respectively). The increase of the β-proteobacterial population

size in these two reactors was primarily due to an increase in Zoogloea-related OTUs, suggesting that a potential negative impact of an increase in AgNP concentration in WWTP influent could result in a rapid increase, over one or two weeks according to the present study, of species potentially responsible for sludge bulking. The SBR treated with Ag+ showed a decrease of nearly 50% of the β-proteobacterial population between Day +1 and +13, which may be explained by the fact that, in the Ag+-treated SBR, each pulse addition corresponded to a rapid increase in Ag+ concentration, while in the AgNP-treated SBRs, where progressive release of Ag+ occurs as the nanoparticle coating dissolves, each pulse addition only corresponded to a moderate increase in Ag+ concentration which may be better tolerated. The relative abundance of Bacteroidetes decreased in the reactors exposed to GA- and Ca-AgNPs, while it increased in the control and Ag+-treated SBRs between Day +1 and +13, suggesting that AgNPs, not Ag+, affected the Bacteroidetes population. However, it is difficult to determine if the effect of adding silver to the different microbial populations is direct, or due to the trophic interactions the microbial population establishes (e.g., competition for substrate, syntrophy). The members of the other phyla detected in this study did not display noticeable changes between treatments in terms of relative abundances. 3.4. Diversity of microorganisms carrying out nitrogen removal As the SBR seed was nitrifying sludge, the effect of silver treatments on the nitrogen removal community was examined; OTUs affiliating to taxa known as denitrifiers, ammonia-oxidizing bacteria (AOB), and nitrite-oxidizing bacteria (NOB) were retrieved from the dataset and separately analyzed. A total of 427 OTUs (representing 55.5% of the total reads obtained), including 414 denitrifiers, 12 AOB, and 1 NOB were identified (Table S1). The vast majority of denitrifiers belonged to the β-proteobacteria (98.4%), which were largely dominated by Zoogloea-related OTUs (Fig. 4). Nitrifiers were only represented by 13 OTUs, 12 of which affiliated with AOB but only one with NOB (Fig. 4). All AOB detected belonged to the Nitrosomonadaceae family, all of which cultivated so far catalyze ammonia oxidation (Prosser et al. 2014). Despite being frequently found in nitrifying WWTPs, Nitrosospira sp. (Baek et al., 2010) was not detected in our study. This can be explained as our SBRs were continuously fed with 40 mg NH+ 4 -N/L, which is in the higher range of NH+ 4 concentrations found in wastewater influent (Daims and Wagner, 2011), and unlike Nitrosomonadaceae spp., Nitrosospira spp. are less adapted to a high concentration of NH+ 4 (Daims and Wagner, 2011). While only a slight decrease in ammonia removal rate was reported in the Ca- and GA-AgNP treated SBRs after the first 0.2 ppm pulse addition (from 98% to 97%, and 88%, respectively), in Ag+ treated SBRs the decrease was far more significant, dropping from

1020


98% to 32%. Since Nitrosomonadaceae spp. were the only AOB identified in this study, one would have expected a decrease in their relative abundance proportional to the decrease in ammonia removal after the two silver pulse additions. However, Nitrosomonadaceae spp. abundances did not reflect these changes (Fig. 5). A possible explanation is silver could have only temporarily inhibited the nitrifying activity of Nitrosomonadaceae spp. without necessarily leading to cell death. Although higher concentrations of silver ultimately caused N. europaea cell membrane disruption in pure culture study, the main consequence of an exposure to concentrations closer to those used in this study was nitrification inhibition (Arnaout and Gunsch, 2012). An alternative explanation is that AOB other than the ones currently known might have been responsible for ammonia removal, but remained unidentified or undetected in this study. It is also possible that ammonia oxidizing Archaea, which were not analyzed for in this study, were performing most of the ammonia removal. Although Nitrospira and Nitrobacter are the most frequently encountered NOB in nitrifying sludge, only Nitrospira sp. was detected in this study. Yang et al. (2014) also found that Nitrospira were dominant over Nitrobacter in their microcosms, and Zhang et al. (2014) determined that Nitrospira were two orders of magnitude higher than Nitrobacter in a membrane bioreactor activated sludge system treated with low dose, continuous AgNPs. Conversely, high doses of both AgNPs and Ag+ impacted Nitrospira in nitrifying SBRs, but sequential loading did not (Ma et al., 2015). In nitrifying reactors, NOB are usually found at low relative abundance (0.001 to 1% in Baek et al., 2010; 0 to 7.8% in Ma et al., 2015; up to 4% in Vuono et al., 2015). In this study, the abundance of Nitrospira sp. was even lower, ranging from 0 to only 0.5% (Fig. 5), which suggests that nitrite oxidation may not have been a dominant process in our SBRs. Zeng et al. (2003), who worked on nitrifying SBRs fed with the same synthetic solution as that used in this study, verified that nitrification and denitrification could occur simultaneously. They also demonstrated that nitrogen removal was achieved by partial oxidation of ammonium to nitrite, which was then directly reduced to nitrogen gas by denitrifiers. Because of the low abundance of NOB and the predominance of denitrifiers in our SBRs, it is likely the same process occurred here. A strain of Zoogloea sp. recently isolated from an aerobic water reservoir was reported to be able to efficiently reduce nitrite when supplied as the sole source of nitrogen (Huang et al., 2015). Given that Zoogloea spp. represented nearly 94% of the denitrifiers detected in our SBRs, it is possible that Zoogloea spp. directly reduced nitrite to nitrogen gas, and that the role of Nitrospira sp. in oxidizing nitrite remained minor in our SBRs. An additional ANOSIM performed on the nitrogen removal community did not reveal any significant effect of silver treatments on the community composition (Fig. S6), suggesting that members other than the nitrogen removal community might be more sensitive to silver. This being said, the PCoA (Fig. S6) displays general patterns similar to what was observed for the whole community (Fig. 2), suggesting that taxa involved in nitrogen removal, rather than having been directly affected by silver addition, might have been affected through the impact silver had on other members of the community with which they may have tight interactions. 4. Conclusions The SBR microbial community composition shifted immediately upon exposure to Ag+ but recovered quickly, while the AgNP-treated communities shifted and recovered more slowly, with the longest lasting effect produced by GA-AgNPs. It is likely that these results can be explained by the progressive release of Ag+ as the nanoparticle coating dissolved following each pulse addition of AgNPs as opposed to a sudden increase in Ag+ in the Ag+ control. These findings support previous evidence of coating-dependent releases of Ag+ from AgNPs as the likely dominant mechanism of action of AgNPs against bacteria (Arnaout and Gunsch, 2012; Mijnendonckx et al., 2013; Yang et al., 2014). Our data also show that increases in AgNP addition to nitrifying WWTPs may

result in an increase of sludge bulking species. Additional work should monitor the frequency and abundance of these species to further determine the impact they may have on maintaining treatment efficiency in WWTPs more likely to be exposed to increasing concentrations of AgNPs in the future. Acknowledgments This material is based upon work supported by the National Science Foundation (NSF) and the Environmental Protection Agency (EPA) under NSF Cooperative Agreement EF-0830093 and DBI-1266252, Center for the Environmental Implications of NanoTechnology (CEINT). Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the NSF or the EPA. This work has not been subjected to EPA review and no official endorsement should be inferred. Appendix A. Supplementary data Detailed information regarding: 1) QIIME commands; 2) rarefaction curves; 3) reactor performance; 4) overall relative taxonomic distributions of OTUs; 5) ranked OTU relative abundance curve; 6) relative abundance and taxonomy of nitrogen-removal OTUs identified; 5) PCoA and ANOSIM for the nitrogen removal bacterial community. Supplementary data associated with this article can be found in the online version, at https://doi.org/10.1016/j.scitotenv.2017.10.217. References Alito, C.L., Gunsch, C.K., 2014. Assessing the effects of silver nanoparticles on biological nutrient removal in bench-scale activated sludge sequencing batch reactors. Environ. Sci. Technol. 48 (2), 970–976. Allison, S.D., Martiny, J.B., 2008. Resistance, resilience, and redundancy in microbial communities. Proc. Natl. Acad. Sci. 105 (Supplement 1), 11512–11519. Arnaout, C.L., Gunsch, C.K., 2012. Impacts of silver nanoparticle coating on the nitrification potential of Nitrosomonas europaea. Environ. Sci. Technol. 46 (10), 5387–5395. Baek, K.H., Park, C., Oh, H.-M., Yoon, B.-D., Kim, H.-S., 2010. Diversity and abundance of ammonia-oxidizing bacteria in activated sludge treating different types of wastewater. J. Microbiol. Biotechnol. 20 (7), 1128–1133. Caporaso, J.G., Kuczynski, J., Stombaugh, J., Bittinger, K., Bushman, F.D., Costello, E.K., Fierer, N., Pena, A.G., Goodrich, J.K., Gordon, J.I., 2010. QIIME allows analysis of highthroughput community sequencing data. Nat. Methods 7 (5), 335–336. Choi, O., Hu, Z., 2008. Size dependent and reactive oxygen species related nanosilver toxicity to nitrifying bacteria. Environ. Sci. Technol. 42 (12), 4583–4588. Clarke, K.R., Gorley, R.N., 2006. PRIMER V6: User Manual-Tutorial. Plymouth Marine Laboratory. Clarke, K., Warwick, R., 1994. Similarity-based testing for community pattern: the twoway layout with no replication. Mar. Biol. 118 (1), 167–176. Daims, H., Wagner, M., 2010. The microbiology of nitrogen removal. The Microbiology of Activated Sludge, pp. 259–280. Daims, H., Wagner, M., 2011. 8 in situ techniques and digital image analysis methods for quantifying spatial localization patterns of nitrifiers and other microorganisms in biofilm and flocs. Methods Enzymol. 496, 185. Dias, F., Bhat, J., 1964. Microbial ecology of activated sludge I. Dominant bacteria. Appl. Microbiol. 12 (5), 412–417. Edgar, R.C, 2010. Search and clustering orders of magnitude faster than BLAST. Bioinformatics https://doi.org/10.1093/bioinformatics/btq461. Fabrega, J., Luoma, S.N., Tyler, C.R., Galloway, T.S., Lead, J.R., 2011. Silver nanoparticles: behaviour and effects in the aquatic environment. Environ. Int. 37 (2), 517–531. Garner, K.L., Keller, A.A., 2014. Emerging patterns for engineered nanomaterials in the environment: a review of fate and toxicity studies. J. Nanopart. Res. 16 (8), 2503. Graves, J.L., Tajkarimi, M., Cunningham, Q., Campbell, A., Nonga, H., Harrison, S.H., Barrick, J.E., 2015. Rapid evolution of silver nanoparticle resistance in Escherichia coli. Front. Genet. 6 (42). Huang, T.-L., Zhou, S.-L., Zhang, H.-H., Bai, S.-Y., He, X.-X., Yang, X., 2015. Nitrogen removal characteristics of a newly isolated indigenous aerobic denitrifier from oligotrophic drinking water reservoir, Zoogloea sp. N299. Int. J. Mol. Sci. 16 (5), 10038–10060. Jeong, E., Chae, S., Kang, S., Shin, H., 2012. Effects of silver nanoparticles on biological nitrogen removal processes. Water Sci. Technol. 65 (7), 1298–1303. Jeong, E., Im, W.-T., Kim, D.-H., Kim, M.-S., Kang, S., Shin, H.-S., Chae, S.-R., 2014. Different susceptibilities of bacterial community to silver nanoparticles in wastewater treatment systems. J. Environ. Sci. Health A 49 (6), 685–693. Khan, S., Mukherjee, A., Chandrasekaran, N., 2011. Silver nanoparticles tolerant bacteria from sewage environment. J. Environ. Sci. 23 (2), 346–352. Kim, J.S., Kuk, E., Yu, K.N., Kim, J.-H., Park, S.J., Lee, H.J., Kim, S.H., Park, Y.K., Park, Y.H., Hwang, C.-Y., 2007. Antimicrobial effects of silver nanoparticles. Nanomedicine 3 (1), 95–101.

C.A. Gwin et al. / Science of the Total Environment 616–617 (2018) 1014–1021 Lefevre, E., Cooper, E., Stapleton, H.M., Gunsch, C.K., 2016. Characterization and adaptation of anaerobic sludge microbial communities exposed to Tetrabromobisphenol A. PLoS One 11 (7), e0157622. Li, A.-D., Li, L.-G., Zhang, T., 2015. Exploring antibiotic resistance genes and metal resistance genes in plasmid metagenomes from wastewater treatment plants. Front. Microbiol. 6. Ma, Y., Metch, J.W., Vejerano, E.P., Miller, I.J., Leon, E.C., Marr, L.C., Vikesland, P.J., Pruden, A., 2015. Microbial community response of nitrifying sequencing batch reactors to silver, zero-valent iron, titanium dioxide and cerium dioxide nanomaterials. Water Res. 68, 87–97. Mijnendonckx, K., Leys, N., Mahillon, J., Silver, S., Van Houdt, R., 2013. Antimicrobial silver: uses, toxicity and potential for resistance. Biometals 26 (4), 609–621. Muyzer, G., De Waal, E.C., Uitterlinden, A.G., 1993. Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplified genes coding for 16S rRNA. Appl. Environ. Microbiol. 59 (3), 695–700. Pal, C., Bengtsson-Palme, J., Kristiansson, E., Larsson, D.J., 2015. Co-occurrence of resistance genes to antibiotics, biocides and metals reveals novel insights into their coselection potential. BMC Genomics 16 (1), 1. Paulson, J.N., Stine, O.C., Bravo, H.C., Pop, M., 2013. Differential abundance analysis for microbial marker-gene surveys. Nat. Methods 10 (12), 1200–1202. Rai, M., Yadav, A., Gade, A., 2009. Silver nanoparticles as a new generation of antimicrobials. Biotechnol. Adv. 27 (1), 76–83. Prosser, J.I., Head, I.M., Stein, L.Y., Rosenberg, E., et al. (Eds.), 2014. The Prokaryotes – Alphaproteobacteria and Betaproteobacteria. Springer-Verlag, Berlin Heidelberg https://doi.org/10.1007/978-3-642-30197-1_372. Ramirez-Vargas, R., Serrano-Silva, N., Navarro-Noya, Y.E., Alcántara-Hernández, R.J., LunaGuido, M., Thalasso, F., Dendooven, L., 2015. 454 pyrosequencing-based characterization of the bacterial consortia in a well established nitrifying reactor. Water Sci. Technol. 72 (6), 990–997. Saikaly, P.E., Stroot, P.G., Oerther, D.B., 2005. Use of 16S rRNA gene terminal restriction fragment analysis to assess the impact of solids retention time on the bacterial diversity of activated sludge. Appl. Environ. Microbiol. 71 (10), 5814–5822. Shu, D., He, Y., Yue, H., Wang, Q., 2015. Microbial structures and community functions of anaerobic sludge in six full-scale wastewater treatment plants as revealed by 454 high-throughput pyrosequencing. Bioresour. Technol. 186, 163–172. Silver, S., 2003. Bacterial silver resistance: molecular biology and uses and misuses of silver compounds. FEMS Microbiol. Rev. 27 (2–3), 341–353. Sun, X., Sheng, Z., Liu, Y., 2013. Effects of silver nanoparticles on microbial community structure in activated sludge. Sci. Total Environ. 443, 828–835. Tan, J.-M., Qiu, G., Ting, Y.-P., 2015. Osmotic membrane bioreactor for municipal wastewater treatment and the effects of silver nanoparticles on system performance. J. Clean. Prod. 88, 146–151. Tulve, N.S., Stefaniak, A.B., Vance, M.E., Rogers, K., Mwilu, S., LeBouf, R.F., Schwegler-Berry, D., Willis, R., Thomas, T.A., Marr, L.C., 2015. Characterization of silver nanoparticles in

1021

selected consumer products and its relevance for predicting children's potential exposures. Int. J. Hyg. Environ. Health 218 (3), 345–357. Vuono, D.C., Benecke, J., Henkel, J., Navidi, W.C., Cath, T.Y., Munakata-Marr, J., Spear, J.R., Drewes, J.E., 2015. Disturbance and temporal partitioning of the activated sludge metacommunity. ISME J. 9 (2), 425–435. Wang, S., Gunsch, C.K., 2011. Effects of selected pharmaceutically active compounds on treatment performance in sequencing batch reactors mimicking wastewater treatment plants operations. Water Res. 45 (11), 3398–3406. Wang, S., Holzem, R.M., Gunsch, C.K., 2008. Effects of pharmaceutically active compounds on a mixed microbial community originating from a municipal wastewater treatment plant. Environ. Sci. Technol. 42 (4), 1091–1095. Wang, Z., Sheng, J., Tian, X., Wu, T., Liu, W., Shen, L., 2011. Optimization of the production of exopolysaccharides by Bacillus thuringiensis 27 in sand biological soil crusts and its bioflocculant activity. African Journal of Microbiology Research]–>Afr. J. Microbiol. Res. 5 (16), 2359–2366. Whiteley, A.S., Jenkins, S., Waite, I., Kresoje, N., Payne, H., Mullan, B., Allcock, R., O'Donnell, A., 2012. Microbial 16S rRNA Ion Tag and community metagenome sequencing using the Ion Torrent (PGM) Platform. J. Microbiol. Methods 91 (1), 80–88. Yang, C., Zhang, W., Liu, R., Li, Q., Li, B., Wang, S., Song, C., Qiao, C., Mulchandani, A., 2011. Phylogenetic diversity and metabolic potential of activated sludge microbial communities in full-scale wastewater treatment plants. Environ. Sci. Technol. 45 (17), 7408–7415. Yang, Y., Quensen, J., Mathieu, J., Wang, Q., Wang, J., Li, M., Tiedje, J.M., Alvarez, P.J., 2014. Pyrosequencing reveals higher impact of silver nanoparticles than ag+ on the microbial community structure of activated sludge. Water Res. 48, 317–325. Ye, L., Zhang, T., 2010. Estimation of nitrifier abundances in a partial nitrification reactor treating ammonium-rich saline wastewater using DGGE, T-RFLP and mathematical modeling. Appl. Microbiol. Biotechnol. 88 (6), 1403–1412. Yuan, Z., Li, J., Cui, L., Xu, B., Zhang, H., Yu, C.-P., 2013. Interaction of silver nanoparticles with pure nitrifying bacteria. Chemosphere 90 (4), 1404–1411. Zeng, R.J., Lemaire, R., Yuan, Z., Keller, J., 2003. Simultaneous nitrification, denitrification, and phosphorus removal in a lab-scale sequencing batch reactor. Biotechnol. Bioeng. 84 (2), 170–178. Zhang, T., Shao, M.-F., Ye, L., 2012. 454 pyrosequencing reveals bacterial diversity of activated sludge from 14 sewage treatment plants. ISME J. 6 (6), 1137–1147. Zhang, C., Liang, Z., Hu, Z., 2014. Bacterial response to a continuous long-term exposure of silver nanoparticles at sub-ppm silver concentrations in a membrane bioreactor activated sludge system. Water Res. 50, 350–358. Zhang, J., Jia, W., Wang, R., Ngo, H.H., Guo, W., Xie, H., Liang, S., 2016. Microbial community characteristics during simultaneous nitrification-denitrification process: effect of COD/TP ratio. Environ. Sci. Pollut. Res. 23 (3), 2557–2565.