Abstract With the growing application of high-throughput sequencing-based metagenomics for profiling antibiotic resistance genes (ARGs) in wastewater treatment plants (WWTPs), comparison of sample pretreatment and DNA extraction methods are needed to move toward standardized comparisons among laboratories. Three widely employed DNA extraction methods (FastDNA® Spin Kit for Soil, PowerSoil® DNA Isolation Kit and ZR Fecal DNA MiniPrep), with and without preservation in 50% ethanol and freezing, were applied to the influent, activated sludge and effluent of two WWTPs, in Hong Kong and in the USA. Annotated sequences obtained from the DNA extracted using the three kits shared similar taxonomy and ARG profiles. Overall, it was found that the DNA yield and purity, and diversity of ARGs captured were all highest when applying the FastDNA SPIN Kit for Soil for all three WWTP sample types investigated here (influent, activated sludge, effluent). Quantitative polymerase chain reaction of 16S rRNA genes confirmed the same trend as DNA extraction yields and similar recovery of a representative Gram-negative bacterium (Escherichia coli). Moreover, sample fixation in ethanol, deep-freezing and overseas shipment had no discernable effect on ARG profiles, as compared to fresh samples. This approach serves to inform future efforts toward global comparisons of ARG distributions in WWTPs. antibiotic resistance genes, DNA extraction, global comparison, wastewater treatment, metagenomics INTRODUCTION With increases in morbidity and mortality caused by antibiotic-resistant pathogens, antibiotic resistance genes (ARGs) are important markers for tracking the spread of antibiotic resistance and characterizing human health risk. Historically, ARG characterization has been primarily focused on clinical pathogens (Martinez 2008). However, pathogens can also acquire antibiotic resistance from non-pathogenic microbes in natural environments through horizontal gene transfer. Recent studies have examined the occurrence of ARGs in different environments (Schwartz et al.2003; Pei et al.2006; Segawa et al.2013), demonstrating that human inputs can alter and elevate the distribution of ARGs and contributing to a growing body of evidence that clinical pathogens can obtain ARGs from such reservoirs (e.g. homology of clinical ARGs and soil (Forsberg et al.2012) and ocean (Hatosy and Martiny 2015) system ARGs). Wastewater treatment plants (WWTPs) are critical for water sanitation and reclamation, but are also key nodes linking human and natural environments, creating the opportunity for horizontal gene transfer of ARGs between human pathogens and environmental bacteria (Finley et al.2013; Yang et al.2014). The dense and diverse microbial communities within WWTPs make sewage influent, activated sludge and the associated effluent favorable places for ARG transfer. Thus, there is a growing interest in understanding the occurrence and behavior of ARGs in WWTP systems as likely ‘hotspots’ for ARG proliferation (Renew and Huang 2004; Brown et al.2006; Watkinson et al.2009). The majority of research focused on ARGs in WWTPs has relied on PCR-based methods (Volkmann et al.2004; Auerbach, Seyfried and McMahon 2007; Szczepanowski et al.2009), which have limited capability to access the ‘resistome’ due to low throughput and their reliance on inherently limited and biased primers. With the emergence of high-throughput DNA sequencing, metagenomic approaches are rapidly expanding in application for profiling ARGs in WWTPs (Wang et al.2013; Yang et al.2013). Based on similarity-based annotation against ARG databases (Liu and Pop 2009; McArthur et al.2013; Yang et al.2013, 2016), such studies have enabled broad capture, discovery and profiling of ARGs associated with various stages of the WWTP treatment process. In recent years, a number of joint studies exploring ARGs as global pollutants have been launched (Pruden et al.2013; Li et al.2015). To compare results collected among different research groups, there is a need for standardized protocols, including DNA extraction, sequencing and data analysis. Although several studies have evaluated DNA extraction from activated sludge samples (Bushon et al.2010; Vanysacker et al.2010; Guo and Zhang 2013), until now, no study has evaluated the effect of sample pretreatment and DNA extraction method on metagenomic ARG profiles obtained from key WWTP compartments: activated sludge, influent and effluent, particularly on a global scale. In this study, samples were collected from two WWTPs (one in Hong Kong and the other in Virginia, USA) with six experienced researchers (three from The University of Hong Kong (HKU) and three from Virginia Tech (VT)) participating in sample pretreatment and DNA extraction. Microbial community structure and ARG profiles were explored using metagenomic approaches and the ARG types and taxa in each of the sequenced datasets were compared (Fig. 1). Real-time quantitative polymerase chain reaction (qPCR) aided in quantitative evaluation of the different sample preparation and extraction approaches to capture representative Gram-positive and negative bacteria. Ultimately, this study takes a step toward establishing standard approaches for global comparison of ARG profiles in WWTPs and other environmental systems. In particular, identifying consistent procedures suitable from various sample types is ideal to avoid bias of downstream analyses toward the specific extraction kit applied. Figure 1. View largeDownload slide Overview of the pipeline for sample pretreatment and DNA extraction (AS: activated sludge, EF: effluent, and IN: influent). Figure 1. View largeDownload slide Overview of the pipeline for sample pretreatment and DNA extraction (AS: activated sludge, EF: effluent, and IN: influent). MATERIALS AND METHODS Sample collection Influent, activated sludge and effluent samples were collected in March 2016 from the Shek Wu Hui WWTP located in Hong Kong (HK WWTP), China, and a WWTP located in southwest Virginia, USA (VA WWTP). The samples were subsequently transported either to a laboratory in HKU or VT within 1 h and immediately pretreated for DNA extraction. Pretreatment of samples Samples were pretreated accordingly upon receipt in the laboratory. For influent or effluent samples, after homogenization with stirring in a single container, nine independent replicates of equal volume were filter-concentrated using 0.22-μm mixed cellulose ester filters (Millipore, USA) until they clogged (Table S1, Supporting Information). After filtering, each filter was placed in a 2-mL sterile centrifuge tube and 1.5-mL 50% ethanol was added for sample fixation. Eleven replicate activated sludge samples were homogenized and 500 μL was transferred to 2 mL sterile centrifuge tubes, where 500 μL of 100% ethanol were added to 9 of the 11 replicates for fixation, with the other 2 replicates directly prepared for DNA extraction without fixation by centrifugation at 5000 g for 10 min. All of the fixed samples were immediately stored at –20°C. A total of 58 samples were prepared for DNA extraction, including nine ethanol-fixed replicates for influent, activated sludge and effluent samples and two fresh activated sludge samples from both WWTPs. DNA extraction DNA from each fresh activated sludge sample was extracted individually at the HKU and VT laboratories by different researchers using the FastDNA Spin Kit for Soil (MP Biomedicals, USA) (Enwall, Philippot and Hallin 2005; Zhao et al.2010; Zhang, Shao and Ye 2012) following the manufacturer's protocol (the optional 55°C step for 5 min was not used for the DNA elution in the present experiments). For the samples fixed in 50% ethanol, one third of the replicates of influent, activated sludge and effluent from the HK and VA WWTPs were immediately shipped to VT and HKU, respectively. After 3 weeks of storage at –20°C (for both the in-house and the shipped samples), DNA was extracted from the samples as described below. For influent and effluent samples, the filter that comprised each sample was removed from ethanol with sterilized tweezers, and was then torn into small pieces and transferred to an extraction tube. After centrifuging the remaining ethanol solution (5000 g for 10 min), the ethanol was carefully removed without disturbing the pellet. The pellet was then resuspended in the buffer employed for the first step in each corresponding DNA extraction kit and transferred to the extraction tube containing the filter. Activated sludge samples were pelleted and transferred to extraction tubes in the same manner. Three widely used commercial kits, all employing bead beating and chemical lysis buffer, were evaluated in this study: the FastDNA Spin Kit for Soil (MP Biomedicals) (Enwall, Philippot and Hallin 2005; Zhao et al.2010; Zhang, Shao and Ye 2012), the PowerSoil DNA Isolation Kit (MoBio Laboratories, Inc., USA) (Steinberg and Regan 2009; Viau and Peccia 2009; Zhang et al.2009) and the ZR Fecal DNA MiniPrep (Zymo Research Corporation, USA) (Ferrand et al.2014; Guzman et al.2015; Garcia-Mazcorro et al.2016). Three replicates of each sample (influent, activated sludge and effluent of two WWTPs) were extracted using each of the three kits following the manufacturer's instructions, with each extraction replicate conducted by a different researcher. For the final elution, the DNA was eluted with same volume DNase-free water (100 μL). The optional 55°C step for 5 min was not used for the DNA elution in the present experiments. The yield of the extracted DNA was measured using a Qubit 2.0 Fluorometer (Life Technologies, USA), and quality was determined using a Nanodrop® ND-1000 (Thermo, USA). For the high-throughput sequencing with short reads sequencing platform, the 260/280 value usually is more important than the 260/230 value. The residual protein (260/280) would influence the DNA quantification, the DNA fragmentation and the sequencing library construction, while the contamination from reagents (260/230), such as salt, would affect little to the processes for the Illumina sequencing platform (no 260/230 requirement are mentioned in the Illumina website; https://www.illumina.com). Therefore, only the 260/280 value was used to detect the DNA quality. The extracted DNA was examined on a 1% agarose gel in 1 × TAE buffer, run for 40 min under a voltage of 150 V and stained with RedSafeTM Nucleic Acid Staining Solution (iNtRON, South Korea), to assess integrity. Illumina sequencing Twenty-nine DNA samples extracted at VT were shipped to HKU and delivered within 5 days, without ice or other measures to control temperature during shipment. Together with the DNA extracted at HKU, a total of 58 DNA samples were used for library construction (250 bp insert). High-throughput sequencing was performed at the Beijing Genomics Institute (BGI, Shenzhen) on an Illumina HiSeq 4000 platform using the index PE150 sequencing strategy (true metagenomics approach). The raw data (Q20 ≥ 85% and Q30 ≥ 80%) was trimmed with the following parameters: (i) removal of reads containing the adapter, (ii) removal of reads containing > 5% ambiguous bases, (iii) removal of reads with single base quality scores < 20 in greater than 15% of the reads. At least 21 million clean metagenomic reads were generated for each of the 58 libraries constructed, resulting in a total of 315.4 Gb (gigabase) of data and 2.1 billion clean reads. The clean reads were then deposited to NCBI Sequence Read Archive database, and the accession number was listed in Table S2 (Supporting Information). ARG profiling and taxonomy analysis For each dataset, due to the limitation of the sequencing length (150 bp), sequenced reads aligned by searching against a database of constructed 16S rRNA gene V6 hypervariable regions using Usearch with a accel cutoff of 0.5, a E-value cutoff of 1e-5 and a maxaccepts cutoff with 10 (Yang et al.2016). Operational taxonomic units (OTUs) of these reads were performed using the lowest common ancestor algorithm for a taxonomic analysis of the samples and classified as their annotation. The taxonomy analyses were carried out at OTU level and other taxonomic classification levels (e.g. phylum level, family level) in the present study. The ARG-like reads among the 58 datasets were identified and classified into different types and subtypes from the metagenomic data using the integrated structured ARG database, which combines the ARDB (Liu and Pop 2009) and CARD (McArthur et al.2013) databases, but removes duplicated ARG sequences and classifies ARG sequences according to their target antibiotics (e.g. beta-lactams, tetracyclines, aminoglycosides) (Yang et al.2016). Specifically, metagenomic datasets were searched against the integrated structured ARG database using Usearch with an accel cutoff of 0.5 and E-value cutoff of 1e-5 (Yang et al.2016). Potential ARG-like reads were extracted using in-house Perl scripts and searched against the integrated structured ARG database using BLASTX for accurate annotation. A sequence was annotated as an ARG-like read if its best match in the database had an e-value cutoff of 1e-7, minimum alignment length of 25 amino acids and a minimum 80% amino acid similarity. The ARG profiles in different samples were compared at the type level, the subtype level and the gene level (the reference sequence level). Real-time qPCR Quantification of bacterial 16S rRNA genes, as well as of genes specific to Gram-positive and Gram-negative bacteria, was carried out as a means to further compare the extraction efficiencies and potential biases. The effects of different DNA extraction kits, researcher variability and sample fixation were assessed. Triplicate 10 μL qPCR reactions consisting of 400 nM of each primer, 1 × EVA Green Supermix (BioRad) and 1 μL of 1:50 diluted DNA extract were carried out for all of the samples extracted at VT. A dilution factor of 50 was chosen for optimum amplification and minimal inhibition after testing several dilutions on a representative subset of samples. The Escherichia coli-specific gadAB gene was targeted as a Gram-negative gene marker, while the Staphylococcus aureus-specific nuc gene was targeted as a Gram-positive gene marker. Thermal cycling consisted of an initial denaturation step of 98°C (16S rRNA and nuc genes) or 95°C (gadAB gene) for 2 min followed by 40 cycles of denaturing at 98 °C (16S rRNA and nuc genes) or 95°C (gadAB gene) for 5 s, and annealing-extension for 5 s. Primers and annealing temperatures applied are specified in Table S3 (Supporting Information). Six- to seven-point standard curves were analyzed in triplicate on each plate for each gene based on previously validated quantification standards. Positive detections were defined where at least two of three analytical replicates crossed the quantification cycle (Cq) line. Sample concentrations, in copies/mL, were calculated by multiplying the qPCR-measured starting concentrations by the dilution factor (i.e. 50), then multiplying by the volume of the DNA extract and then dividing by the sample volume. RESULTS AND DISCUSSION Selection of DNA extraction procedures Commercially available kits were considered ideal for the purpose of normalizing procedures for international and multiple-lab application. Unlike many other environmental samples, WWTP samples, especially influent and activated sludge, are primarily composed of microbial cells and their products (Frolund et al.1996; Liu and Fang 2003; Flemming, Neu and Wozniak 2007). These aggregated cells and associated exopolysaccharides pose a challenge to penetrating the cell wall through shearing or chemical reagents (Davies et al.1998). Previous studies have demonstrated that mechanical homogenization, especially bead beating, can improve DNA extraction from samples containing complex microbial communities (Lemarchand et al.2005; Guo and Zhang 2013). Therefore, three widely reported DNA extraction kits that employ bead beating for cell lysis were selected for this study. Assessment of DNA extract quality DNA extraction yields are summarized in Fig. 2a and Table S2 (Supporting Information). Each extracted DNA sample was eluted with 100 μL DNase-free water. The FastDNA Spin Kit for Soil provided the highest extraction yield (two tailed paired sample t-tests: P < 0.01) among the three kits for all six samples (10.3 ± 3.6 μg per sample, versus 6.5 ± 3.7 μg and 6.8 ± 3.2 μg for PowerSoil DNA Isolation Kit and the ZR Fecal DNA MiniPrep Kit, respectively). The DNA yield from fresh activated sludge samples using the FastDNA Spin Kit for Soil was 11.7 ± 5.0 μg, which was not significantly different from the yield of the ethanol-fixed activated sludge samples using the same kit (11.0 ± 2.2 μg). Not surprisingly, samples extracted by different researchers also showed little difference (one-way ANOVA test, P > 0.05). Comparison of 16S rRNA gene copy numbers recovered from each extraction using qPCR also indicated that the FastDNA Spin Kit for Soil provided superior yield, while the other two kits were comparable to each other (Fig. 3). All three kits provided sufficient DNA (>0.5 μg DNA, or ∼107 to 108 cells; Christensen, Olsen and Bakken 1995) for library construction, as required by the Illumina HiSeq 4000, the latest version currently available. Figure 2. View largeDownload slide DNA yields (a) and purity (b) of the activated sludge (AS), effluent (EF) and influent (IN) samples from VA and HK WWTPs using the three selected kits, as determined by spectrophotometric absorbance using Nanodrop. The marked area indicates the index of optimal DNA purity (1.8 to 2.0). Figure 2. View largeDownload slide DNA yields (a) and purity (b) of the activated sludge (AS), effluent (EF) and influent (IN) samples from VA and HK WWTPs using the three selected kits, as determined by spectrophotometric absorbance using Nanodrop. The marked area indicates the index of optimal DNA purity (1.8 to 2.0). Figure 3. View largeDownload slide qPCR-based total bacterial and Gram-specific DNA extraction yield comparisons of influent (IN), activated sludge (AS) and effluent (EF) extracted in the USA using the selected DNA extraction kit (ZY = ZR Fecal DNA Miniprep, MP = FastDNA Spin Kit for Soil, MO = PowerSoil DNA Isolation Kit). EG, VR and JM are the initials of the researcher carrying out the DNA extraction. Note: JM data not available for MO and ZY kits in person-to-person comparison graphs (second row). *Values below the limit of quantification, but above the limit of detection. Other values not shown were below detection. Figure 3. View largeDownload slide qPCR-based total bacterial and Gram-specific DNA extraction yield comparisons of influent (IN), activated sludge (AS) and effluent (EF) extracted in the USA using the selected DNA extraction kit (ZY = ZR Fecal DNA Miniprep, MP = FastDNA Spin Kit for Soil, MO = PowerSoil DNA Isolation Kit). EG, VR and JM are the initials of the researcher carrying out the DNA extraction. Note: JM data not available for MO and ZY kits in person-to-person comparison graphs (second row). *Values below the limit of quantification, but above the limit of detection. Other values not shown were below detection. Cell lysis and DNA recovery efficiency primarily affect the DNA yield (Martin-Laurent et al.2001). In the three selected commercial kits, the FastDNA Spin Kit for Soil contains glass beads of different sizes (0.1–4 mm in diameter), the PowerSoil DNA Isolation Kit uses irregularly shaped 0.7-mm garnet fragments and the ZR Fecal DNA MiniPrep uses 2-mm diameter glass beads. Additionally, the FastDNA Spin Kit for Soil employs a suspended matrix of DNA-binding material, which likely increases the available surface area and accessibility for DNA-binding relative to housing the binding material in a spin column, as applied in the other two kits. Thus, the combination of the variable diameter glass beads and the manner in which the binding matrix is employed likely both contributed to the superior DNA yield achieved by the FastDNA Spin Kit for Soil when applied to the wastewater samples. Figure 2b and Table S2 (Supporting Information) provide information regarding the purity of the extracted DNA. A 260 nm/280 nm optical density (OD) ratio between 1.8 and 2.0 is often cited for optimal purity (Chong 2001), as indicative of the wavelengths at which DNA and proteins optimally absorb light, respectively. The FastDNA Spin Kit for Soil (OD 260/280: 1.92 ± 0.03) and the PowerSoil DNA Isolation Kit (OD260/280: 1.84 ± 0.09) outperformed the ZR Fecal DNA MiniPrep (OD260/280: 1.63 ± 0.11) in terms of DNA purity. Meanwhile, the standard deviation of the OD260/280 ratio was consistently smaller for the FastDNA Spin Kit for Soil, suggesting that it provided more consistent DNA extract quality. Additionally, compared to the DNA extracted from fresh samples (OD260/280: 1.88 ± 0.05) using FastDNA Spin Kit for Soil, we found that biomass fixation using 50% ethanol had no measureable impact on the purity of the extracted DNA (P > 0.05), indicating that fixation in ethanol is appropriate for long distance sample shipping, even at ambient temperature. The size of recovered DNA fragments is also an important indicator of extracted DNA quality. As shown in Fig. S1 (Supporting Information), the DNA obtained using the selected commercial kits was generally characterized by relatively short segments (<20 kb), which is likely a result of the high shear force imposed by the bead-beating processing. (The bead-beating conditions were set as the manufactural protocol.) Although such DNA fragments may not be suitable for the construction of fosmid, cosmid or BAC libraries (Chong 2001), they were of sufficient length for the short-read sequencing as applied in this study (Guo and Zhang 2013). Additionally, we noted traces of fluorescence in the sample loading wells for lanes 9, 18, 26, 34, 35, 43 and 54, which is usually a result of protein contamination of the extracted DNA (Tan and Yiap 2009). Consistent with this observation, all seven of these samples were characterized by low OD260/280 and were extracted using the ZR Fecal DNA MiniPrep Kit. We noticed that the DNA yield of the same sample shown in Table S2 and Fig. S1 (Supporting Information) was not completely matched. As shown in Fig. S1 (Supporting Information), a few samples seemed to contain little DNA. This would possibly due to the protein contaminations or the operating error during the sample loading for the electrophoresis detection. Although the protein contaminations would affect the DNA quantification, the Illumina platforms only need a small quantity of DNA (<0.5 μg DNA) for sequencing library construction. Therefore, even the quantities of extracted DNA in the samples mentioned before were overestimated, the extracted DNA is still sufficient for the Illumina sequencing and the following analyses. ARG profiles To compare the ARG profiles of different samples, normalization by the 16S rRNA gene sequence length was conducted. ARG abundances were summarized in this study using the unit ‘copy of ARG per copy of 16S rRNA gene’. The three DNA extraction kits indicated similar trends when comparing the influent, activated sludge and effluent samples. Notably, influent samples, which primarily reflect human fecal input, were characterized by the highest abundance of ARGs, especially the HK influent samples (0.58 ± 0.045 copy of ARG per copy of 16S rRNA gene). Interestingly, ARG abundance of HK influent samples was significantly higher than that of VA influent samples (0.30 ± 0.024 copy of ARG per copy of 16S rRNA gene) (t-test P < 0.01), indicative of differences in ARG profiles in the gut bacteria of the residents of Hong Kong and US populations served by the respective WWTPs (Fig. S2, Supporting Information). Others have proposed that ARG abundance and diversity in the influent can be taken as an average of the ARG profile in the human gut in the gastrointestinal tracts of the residents of the WWTP catchment (Li et al.2015). The HK effluent samples contained the fewest ARGs (0.18 ± 0.012 copy of ARG per copy of 16S rRNA gene), indicating that, despite the high influent levels, the wastewater treatment processes effectively removed ARGs. In terms of the total ARG abundance (sum of all the relative ARG abundances), the FastDNA Spin Kit for Soil provided the most reproducible and consistent results, indicating an advantage for studies specifically focused on ARGs. Without any doubt, ARG profiles did not show significant difference in the same sample extracted by different researchers (one-way ANOVA test, P > 0.05). Figure 4 shows a summary of the ARG profiles of samples treated with the three kits and clustered based on the relative ARG abundances. For all three kits, the six individual sample locations (i.e. influent, activated sludge and effluent of each WWTP) each formed a separate cluster. This is an encouraging finding that, despite differences among the kits, the type of sample was still the main driver of the observed profiles (one-way ANOVA test, P < 0.001), and not the kit itself (one-way ANOVA test, P > 0.05). ARG abundance and diversity, derived from ARG copies per cell (the ARG abundances (ARG per copy of 16S rRNA gene) were corrected by the 16S rRNA gene copies per cell as a previous research (Yang et al.2016)), also indicated similar ARG profiles according to sample type (Fig. S3, Supporting Information). For all of the influent and effluent samples, the DNA extracted using the FastDNA Spin Kit yielded more similar ARG profiles relative to the ZR Fecal DNA MiniPrep, with some distinctions relative to that extracted with the PowerSoil DNA Isolation Kit. All three kits consistently indicated that multidrug resistance genes were the most abundant ARG type across all samples. Figure 4. View largeDownload slide ARG profiles captured by the three extraction kits (ZY = ZR Fecal DNA Miniprep, MP = FastDNA Spin Kit for Soil, MO = PowerSoil DNA Isolation Kit) applied to samples from activated sludge (AS), effluent (EF) and influent (IN) of the VA and HK WWTPs, compared by cluster analysis. The ARG profiles were determined as a function of the abundance (copy of ARG per copy of 16S rRNA gene) of different types. Figure 4. View largeDownload slide ARG profiles captured by the three extraction kits (ZY = ZR Fecal DNA Miniprep, MP = FastDNA Spin Kit for Soil, MO = PowerSoil DNA Isolation Kit) applied to samples from activated sludge (AS), effluent (EF) and influent (IN) of the VA and HK WWTPs, compared by cluster analysis. The ARG profiles were determined as a function of the abundance (copy of ARG per copy of 16S rRNA gene) of different types. Taxonomic comparison The microbial community structure, represented as the relative abundance of detected phyla, is summarized for each analyzed sample in Fig. 5. Across all samples, the most abundant phylum was Proteobacteria, which is consistent with previous studies (Zhang, Shao and Ye 2012). Notably, the ZR Fecal DNA MiniPrep kit might have selectively extracted significant more DNA from Gram-positive phyla, including Actinobacteria, Nitrospirae, Chloroflexi and Firmicutes, compared with the other two kits (two tailed paired sample t-tests: P < 0.01 compared to the PowerSoil DNA Isolation Kit, 0.01 < P < 0.05 compared to the FastDNA Spin Kit for Soil). Enhanced recovery of DNA from Gram-positive organisms by the ZR Fecal DNA MiniPrep kit was also indicated by qPCR (Fig. 3). Strikingly, VA effluent extracted using the PowerSoil DNA Isolation Kit clustered with the VA activated sludge samples, while all of the other samples from the same WWTP process clustered together. However, unlike the ARG profile, taxonomic profiles produced by the different kits did not cluster consistently. Additionally, taxonomic analyses at family level also showed similar profiles to that at the phylum level. (Figure S4 illustrated all the annotated families which belong to the phyla contributed more than 5% of the whole microbial community.) For VA activated sludge, VA influent and HK activated sludge samples, those extracted with the FastDNA Spin Kit for Soil clustered with those extracted using the PowerSoil DNA Isolation Kit. The effluent samples from the two WWTPs extracted using the FastDNA Spin Kit for Soil were more similar to the samples extracted using the ZR Fecal DNA MiniPrep. However, for the HK influent sample, the PowerSoil DNA Isolation Kit performed in a more similar manner to the ZR Fecal DNA MiniPrep. In sum, taxonomic clustering was driven by the sample type (one-way ANOVA test, P < 0.001), but relative similarities produced by the three kits varied among the influent, activated sludge and effluent samples (one-way ANOVA test, P > 0.05). Figure 5. View largeDownload slide Taxonomic profiles (P = phylum, U = unclassified) captured by the three extraction kits (ZY = ZR Fecal DNA Miniprep, MP = FastDNA Spin Kit for Soil, MO = PowerSoil DNA Isolation Kit) applied to samples of activated sludge (AS), effluent (EF) and influent (IN) of the VA and HK WWTPs, compared by cluster analysis. Distribution of taxa is represented as relative abundance at the phylum level (percentage). Figure 5. View largeDownload slide Taxonomic profiles (P = phylum, U = unclassified) captured by the three extraction kits (ZY = ZR Fecal DNA Miniprep, MP = FastDNA Spin Kit for Soil, MO = PowerSoil DNA Isolation Kit) applied to samples of activated sludge (AS), effluent (EF) and influent (IN) of the VA and HK WWTPs, compared by cluster analysis. Distribution of taxa is represented as relative abundance at the phylum level (percentage). PCoA of the taxonomic and ARG profiles Figure 6 shows the results of principal coordinate analysis (PCoA) based on the distribution profiles of the ARGs and the V6 region of the 16S rRNA genes across the 58 samples using Bray-Curtis ordination. The ARG profiles were calculated at the level of the ARG reference sequence (resistance gene level) in the target database, while the taxonomic profiles were calculated based on OTUs calculated using DNA sequence reads aligning with the V6 region of the 16S rRNA genes. Three-dimensional PCoA (Fig. 6) explained 66.6% and 68.8% of intersample variance for ARG and taxonomic profiles, respectively. All 58 samples clustered in accordance with the original source, except for the VA effluent sample extracted using the PowerSoil DNA Isolation Kit, which was significantly different from other VA effluent samples. An anomaly with this sample could be the reason for the unexpected result noted above with respect to the taxonomic analysis at the phylum level, i.e. that the DNA extracted from VA effluent using the PowerSoil DNA Isolation Kit did not cluster with the VA WWTP samples extracted with the other kits (Fig. 6). Meanwhile, the VA activated sludge cluster was the next nearest to the VA effluent cluster, in terms of both ARG and taxonomic profile, suggesting that bacteria from activated sludge escape to the effluent. Additionally, the clusters produced by PCoA confirmed no discernable difference between the fixed/transported/stored and freshly extracted samples in ARG reference sequence level (resistance gene level) and taxonomic OTU level. Figure 6. View largeDownload slide PCoA with the Bray-Curtis algorithm of the samples from activated sludge (AS), effluent (EF) and influent (IN) of the VA and HK WWTPs based on ARGs profiles at the level of reference sequences (a) and microbial community profile based on all referenced OTUs (b). *Only one sample of DNA extracted from VA effluent using the PowerSoil DNA Isolation Kit did not cluster with the VA WWTP samples extracted with the other kits. Figure 6. View largeDownload slide PCoA with the Bray-Curtis algorithm of the samples from activated sludge (AS), effluent (EF) and influent (IN) of the VA and HK WWTPs based on ARGs profiles at the level of reference sequences (a) and microbial community profile based on all referenced OTUs (b). *Only one sample of DNA extracted from VA effluent using the PowerSoil DNA Isolation Kit did not cluster with the VA WWTP samples extracted with the other kits. Procrustes analysis, which provides a descriptive summary and graphical comparison of both ARG and taxonomic profiles, was consistent with the above trends, indicating that both the two profiles of all 58 samples clustered according the six individual sampling locations and were consistently correlated (P < 10–4, M2 = 0.083) (Fig. S5, Supporting Information). This strong goodness-of-fit indicator (The M2 statistic (0 < M2 < 1.0) is the parameter of goodness of fit of the two profiles and lower M2 values indicate better fit (King and Jackson 1999)) for within-site comparison between the ARG profile and taxonomic composition is suggestive that the horizontal gene transfer of ARGs in the tested WWTP systems is not sufficiently frequent to obscure the association with their host genomes (Forsberg et al.2014). Diversity comparison It is noticed that due to the complexity of the target samples and the sequencing depth, ARGs’ exploration using high-throughput sequencing approaches would be difficult to exactly exhibit the ARGs with low abundance in the target samples in the present study (Yang et al.2016). The statistical analyses used in this study were based on the whole profiles of all the detected ARGs in reference sequence level (resistance gene level). It could be used to show the overall differentiated ARG profiles among all the tested samples. To further explore the potential extraction biases among the commercial kits, each dataset was normalized to 21 000 000 reads based on the minimum sequencing datasets obtained across the samples. The integrated, structured ARG database used in this study contained 4049 different reference ARG sequences and ARG-like sequences that were similar to 2530 sequences detected in this study across the 58 normalized datasets. Figure S6a (Supporting Information) shows that the HK influent samples contained the most diverse array of ARGs (1520 ± 34 different ARGs), followed by the VA influent samples (997 ± 44 different ARGs). Considering both of the ARG diversity and relative variance of the datasets, the FastDNA Spin Kit for Soil yielded superior performance for all three sample types at both WWTPs. The diversity analysis indicated a similar distribution of taxa detected in each sequencing dataset of each sample location. Based on the classification of the V6 region of the 16S rRNA genes, a total of 1367 taxa (at OTU level) were detected across all 58 datasets. HK_IN samples had the most diverse taxa (332 ± 17 different taxa), followed by the VA_IN samples (300 ± 16 different taxa). Figure S6b (Supporting Information) summarizes the diversity of the taxa in each dataset. Besides the VA effluent samples, for which the ZR Fecal DNA MiniPrep performed only slightly better than the FastDNA Spin Kit for Soil, the FastDNA Spin Kit for Soil had the best performance for all of the other five sample sets. Meanwhile, ARG and taxa rarefaction curves generated from representative samples (Figs S7 and S8, Supporting Information) indicated that the sequencing depths were all sufficient to characterize the ARG profiles at the subtype level and again demonstrated that the FastDNA Spin Kit for Soil performed best across all samples, except HK influent. However, we note that the HK influent samples contained the highest total ARG abundance (Fig. S2, Supporting Information) and that samples extracted using the PowerSoil DNA Isolation Kit contained the fewest taxa among all the HK influent samples. Additionally, 75.0% of the taxa were annotated as Proteobacteria in HK influent samples extracted using the PowerSoil DNA Isolation Kit, of which Gammaproteobacteria contributed 43.4% of the total. For the FastDNA Spin Kit for Soil and the ZR Fecal DNA MiniPrep treated HK influent samples, Proteobacteria contributed 70.2% and 63.2% and Gammaproteobacteria contributed 40.0% and 33.7%, respectively. Considering the fact that Gammaproteobacteria usually positively correlate with ARG abundance (Marti, Jofre and Balcazar 2013), it is possible that samples treated using PowerSoil DNA Isolation Kit contained higher abundance of ARGs than the samples treated using the other two kits would due to the DNA from Gammaproteobacteria were selective enriched during the DNA extraction. Therefore, considering the diversity of both the ARGs and the taxa, the FastDNA Spin Kit for Soil appears more efficient than the other two kits for recovery of DNA from minor groups of microorganisms. Sample fixation and transport To avoid shifts in microbial community composition and loss or damage of DNA during long-term transport or sample storage, sample fixation is often required, especially for international comparisons (Murphy et al.2002). We found that fixation of the activated sludge samples did not significantly (t-test P > 0.05) influence the diversity estimates of either the ARGs or taxa. Compared to DNA extracted from fresh VA and HK WWTP activated sludge samples (517 ± 10 and 550 ± 1 ARGs, respectively), fixed samples extracted using the FastDNA Spin Kit for Soil contained similar diversity of ARGs (541 ± 9 and 582 ± 16 ARGs in HK activated sludge and VA activated sludge, respectively). In terms of the diversity of taxa, the ethanol-fixed VA and HK activated sludge samples extracted using the FastDNA Spin Kit for Soil contained 208 ± 12 and 278 ± 7 taxa, while the fresh samples contained 210 ± 5 and 263 ± 8 taxa. qPCR analyses also indicated no apparent differences in gene yields, thereby supporting the conclusion that sample fixation with 50% ethanol is a viable technique for sample preservation prior to DNA extraction. The results of this study also demonstrated that even transport over a maximum distance (i.e. 12 time zones) and at ambient temperature had no impact on the taxonomic or ARG profiles for samples fixed in 50% ethanol (Fig. 6). Therefore, before DNA extraction, sample fixation using 50% ethanol (final concentration) is suitable for sample pretreatment in metagenomic analysis of ARGs, which is also commonly applied for fluorescence in situ hybridization (Xia, Kong and Nielsen 2007). CONCLUSION In summary, the three DNA extraction kits indicated similar trends when comparing the influent, activated sludge and effluent samples obtained from two WWTPs located on opposite sides of the globe. However, considering both yield and quality and the diversity of ARGs and taxa from the extracted DNA, plus the reproducibility of the extraction, the FastDNA Spin Kit for Soil exhibited superior performance for all sample locations. Ethanol fixation (50%) was also found to be effective for preserving samples, even when shipped at ambient temperature and maximal difference, when assessed both by qPCR and metagenomic analysis. This study takes an important step toward standardizing approaches for international comparison of ARG distributions in WWTPs, which will be key in assessing the impact of local policy, sanitation and treatment processes and informing effective strategies toward mitigating the spread of antibiotic resistance. SUPPLEMENTARY DATA Supplementary data are available at FEMSEC online. FUNDING The present study was funded by Hong Kong General Research Fund (GRF) 172099/14E and the U.S. National Science Foundation (NSF) Partnership for International Research and Education Award OISE:1545756. Ethical approval: This article does not contain any studies with human participants or animals performed by any of the authors. Conflict of interest. None declared. REFERENCES Auerbach EA, Seyfried EE, McMahon KD. Tetracycline resistance genes in activated sludge wastewater treatment plants. Water Res 2007; 41: 1143– 51. Google Scholar CrossRef Search ADS PubMed Brown KD, Kulis J, Thomson B et al. Occurrence of antibiotics in hospital, residential, and dairy, effluent, municipal wastewater, and the Rio Grande in New Mexico. Sci Total Environ 2006; 366: 772– 83. Google Scholar CrossRef Search ADS PubMed Bushon R, Kephart CM, Koltun GF et al. Statistical assessment of DNA extraction reagent lot variability in real-time quantitative PCR. Lett Appl Microb 2010; 50: 276– 82. Google Scholar CrossRef Search ADS Chong L. Molecular Cloning - A Laboratory Manual , 3rd edn, vol 292. New York, USA: Cold Spring Harbor Laboratory Press. 2001, 446. Christensen H, Olsen RA, Bakken LR. Flow cytometric measurements of cell volumes and DNA contents during culture of indigenous soil bacteria. Microbial Ecol 1995; 29: 49– 62. Google Scholar CrossRef Search ADS Davies DG, Parsek MR, Pearson JP et al. The involvement of cell-to-cell signals in the development of a bacterial biofilm. Science 1998; 280: 295– 8. Google Scholar CrossRef Search ADS PubMed Enwall K, Philippot L, Hallin S. Activity and composition of the denitrifying bacterial community respond differently to long-term fertilization. Appl Environ Microb 2005; 71: 8335– 43. Google Scholar CrossRef Search ADS Ferrand J, Patron K, Legrand-Frossi C et al. Comparison of seven methods for extraction of bacterial DNA from fecal and cecal samples of mice. J Microbiol Meth 2014; 105: 180– 5. Google Scholar CrossRef Search ADS Finley RL, Collignon P, Larsson DGJ et al. The scourge of antibiotic resistance: The important role of the environment. Clin Infect Dis 2013; 57: 704– 10. Google Scholar CrossRef Search ADS PubMed Flemming HC, Neu TR, Wozniak DJ. The EPS matrix: The “house of biofilm cells”. J Bacteriol 2007; 189: 7945– 7. Google Scholar CrossRef Search ADS PubMed Forsberg KJ, Patel S, Gibson MK et al. Bacterial phylogeny structures soil resistomes across habitats. Nature 2014; 509: 612– 6. Google Scholar CrossRef Search ADS PubMed Forsberg KJ, Reyes A, Wang B et al. The shared antibiotic resistome of soil bacteria and human pathogens. Science 2012; 337: 1107– 11. Google Scholar CrossRef Search ADS PubMed Frolund B, Palmgren R, Keiding K et al. Extraction of extracellular polymers from activated sludge using a cation exchange resin. Water Res 1996; 30: 1749– 58. Google Scholar CrossRef Search ADS Garcia-Mazcorro JF, Ivanov I, Mills DA et al. Influence of whole-wheat consumption on fecal microbial community structure of obese diabetic mice. Peerj 2016; 4: e1702. Google Scholar CrossRef Search ADS PubMed Guo F, Zhang T. Biases during DNA extraction of activated sludge samples revealed by high throughput sequencing. Appl Microbiol Biot 2013; 97: 4607– 16. Google Scholar CrossRef Search ADS Guzman CE, Bereza-Malcolm LT, De Groef B et al. Presence of selected methanogens, fibrolytic bacteria, and Proteobacteria in the gastrointestinal tract of neonatal dairy calves from birth to 72 Hours. PLoS One 2015; 10: e0133048. Google Scholar CrossRef Search ADS PubMed Hatosy SM, Martiny AC. The ocean as a global reservoir of antibiotic resistance genes. Appl Environ Microb 2015; 81: 7593– 9. Google Scholar CrossRef Search ADS King JR, Jackson DA. Variable selection in large environmental data sets using principal components analysis. Environmetrics 1999; 10: 67– 77. Google Scholar CrossRef Search ADS Lemarchand K, Berthiaume F, Maynard C et al. Optimization of microbial DNA extraction and purification from raw wastewater samples for downstream pathogen detection by microarrays. J Microbiol Meth 2005; 63: 115– 26. Google Scholar CrossRef Search ADS Li B, Yang Y, Ma L et al. Metagenomic and network analysis reveal wide distribution and co-occurrence of environmental antibiotic resistance genes. ISME J 2015; 9: 2490– 502. Google Scholar CrossRef Search ADS PubMed Liu B, Pop M. ARDB-antibiotic resistance genes database. Nucleic Acid Res 2009; 37: D443– 7. Google Scholar CrossRef Search ADS PubMed Liu Y, Fang HHP. Influences of extracellular polymeric substances (EPS) on flocculation, settling, and dewatering of activated sludge. Crit Rev Env Sci Tec 2003; 33: 237– 73. Google Scholar CrossRef Search ADS McArthur AG, Waglechner N, Nizam F et al. The comprehensive antibiotic resistance database. Antimicrob Agents Ch 2013; 57: 3348– 57. Google Scholar CrossRef Search ADS Marti E, Jofre J, Balcazar JL. Prevalence of antibiotic resistance genes and bacterial community composition in a river influenced by a wastewater treatment plant. PLoS One 2013; 8: e78906. Google Scholar CrossRef Search ADS PubMed Martin-Laurent F, Philippot L, Hallet S et al. DNA extraction from soils: Old bias for new microbial diversity analysis methods. Appl Environ Microb 2001; 67: 2354– 9. Google Scholar CrossRef Search ADS Martinez JL. Antibiotics and antibiotic resistance genes in natural environments. Science 2008; 321: 365– 7. Google Scholar CrossRef Search ADS PubMed Murphy MA, Waits LP, Kendall KC et al. An evaluation of long-term preservation methods for brown bear (Ursus arctos) faecal DNA samples. Conserv Genet 2002; 3: 435– 40. Google Scholar CrossRef Search ADS Pei RT, Kim SC, Carlson KH et al. Effect of river landscape on the sediment concentrations of antibiotics and corresponding antibiotic resistance genes (ARG). Water Res 2006; 40: 2427– 35. Google Scholar CrossRef Search ADS PubMed Pruden A, Larsson DGJ, Amezquita A et al. Management options for reducing the release of antibiotics and antibiotic resistance genes to the environment. Environ Health Persp 2013; 121: 878– 85. Google Scholar CrossRef Search ADS Renew JE, Huang CH. Simultaneous determination of fluoroquinolone, sulfonamide, and trimethoprim antibiotics in wastewater using tandem solid phase extraction and liquid chromatography-electrospray mass spectrometry. J Chromatogr A 2004; 1042: 113– 21. Google Scholar CrossRef Search ADS PubMed Schwartz T, Kohnen W, Jansen B et al. Detection of antibiotic-resistant bacteria and their resistance genes in wastewater, surface water, and drinking water biofilms. FEMS Microbiol Ecol 2003; 43: 325– 35. Google Scholar CrossRef Search ADS PubMed Segawa T, Takeuchi N, Rivera A et al. Distribution of antibiotic resistance genes in glacier environments. Env Microbiol Rep 2013; 5: 127– 34. Google Scholar CrossRef Search ADS Steinberg LM, Regan JM. mcrA-targeted real-time quantitative PCR method to examine methanogen communities. Appl Environ Microb 2009; 75: 4435– 42. Google Scholar CrossRef Search ADS Szczepanowski R, Linke B, Krahn I et al. Detection of 140 clinically relevant antibiotic-resistance genes in the plasmid metagenome of wastewater treatment plant bacteria showing reduced susceptibility to selected antibiotics. Microbiology-SGM 2009; 155: 2306– 19. Google Scholar CrossRef Search ADS Tan SC, Yiap BC. DNA, RNA, and protein extraction: The past and the present. J Biomed Biotechnol 2009; 3: 574398- Vanysacker L, Declerck SAJ, Hellemans B et al. Bacterial community analysis of activated sludge: an evaluation of four commonly used DNA extraction methods. Appl Microbiol Biot 2010; 88: 299– 307. Google Scholar CrossRef Search ADS Viau E, Peccia J. Survey of wastewater indicators and human pathogen genomes in biosolids produced by class A and class B stabilization treatments. Appl Environ Microb 2009; 75: 164– 74. Google Scholar CrossRef Search ADS Volkmann H, Schwartz T, Bischoff P et al. Detection of clinically relevant antibiotic-resistance genes in municipal wastewater using real-time PCR (TaqMan). J Microbiol Meth 2004; 56: 277– 86. Google Scholar CrossRef Search ADS Wang Z, Zhang XX, Huang KL et al. Metagenomic profiling of antibiotic resistance genes and mobile genetic elements in a tannery wastewater treatment plant. PLoS One 2013; 8: e76079. Google Scholar CrossRef Search ADS PubMed Watkinson AJ, Murby EJ, Kolpin DW et al. The occurrence of antibiotics in an urban watershed: From wastewater to drinking water. Sci Total Environ 2009; 407: 2711– 23. Google Scholar CrossRef Search ADS PubMed Xia Y, Kong Y, Nielsen PH. In situ detection of protein-hydrolysing microorganisms in activated sludge. FEMS Microbiol Ecol 2007; 60: 156– 65. Google Scholar CrossRef Search ADS PubMed Yang Y, Jiang X, Chai B et al. ARGs-OAP: online analysis pipeline for antibiotic resistance genes detection from metagenomic data using an integrated structured ARG-database. Bioinformatics 2016; 32: 2346. Google Scholar CrossRef Search ADS PubMed Yang Y, Li B, Ju F et al. Exploring variation of antibiotic resistance genes in activated sludge over a four-year period through a metagenomic approach. Environ Sci Technol 2013; 47: 10197– 205. Google Scholar CrossRef Search ADS PubMed Yang Y, Li B, Zou S et al. Fate of antibiotic resistance genes in sewage treatment plant revealed by metagenomic approach. Water Res 2014; 62: 97– 106. Google Scholar CrossRef Search ADS PubMed Zhang H, Banaszak JE, Parameswaran P et al. Focused-pulsed sludge pre-treatment increases the bacterial diversity and relative abundance of acetoclastic methanogens in a full-scale anaerobic digester. Water Res 2009; 43: 4517– 26. Google Scholar CrossRef Search ADS PubMed Zhang T, Shao MF, Ye L. 454 Pyrosequencing reveals bacterial diversity of activated sludge from 14 sewage treatment plants. ISME J 2012; 6: 1137– 47. Google Scholar CrossRef Search ADS PubMed Zhao LM, Wang YY, Yang J et al. Earthworm-microorganism interactions: A strategy to stabilize domestic wastewater sludge. Water Res 2010; 44: 2572– 82. Google Scholar CrossRef Search ADS PubMed © FEMS 2017. All rights reserved. For permissions, please e-mail: firstname.lastname@example.org
FEMS Microbiology Ecology – Oxford University Press
Published: Feb 1, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.
All for just $49/month
Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.
Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.
It’s easy to organize your research with our built-in tools.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera