- Email: [email protected]
Assessing the impact of water treatment on bacterial biofilms in drinking water distribution systems using high-throughput DNA sequencing
Chemosphere 117 (2014) 185–192
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Assessing the impact of water treatment on bacterial biofilms in drinking water distribution systems using high-throughput DNA sequencing Jennifer L.A. Shaw a,⇑, Paul Monis b, Rolando Fabris b, Lionel Ho b, Kalan Braun b, Mary Drikas b, Alan Cooper a a b
Australian Centre for Ancient DNA (ACAD), University of Adelaide, Darling Building, North Terrace, Adelaide, SA 5000, Australia Australian Water Quality Centre, SA Water Corporation, 250 Victoria Square, Adelaide, SA 5000, Australia
h i g h l i g h t s Four drinking water distribution systems were subjected to unique water treatments. Bacterial biofilm communities were characterised by sequencing the 16S rRNA gene. Initially, biofilm communities varied across different treatments. Biofilm biodiversity increased with distance from treatment application. A common community structure is eventually attained regardless of initial treatment.
a r t i c l e
i n f o
Article history: Received 11 February 2014 Received in revised form 20 June 2014 Accepted 20 June 2014
Handling Editor: O. Hao Keywords: Biofilm Metagenomics DNA sequencing Drinking water treatment Bacteria
a b s t r a c t Biofilm control in drinking water distribution systems (DWDSs) is crucial, as biofilms are known to reduce flow efficiency, impair taste and quality of drinking water and have been implicated in the transmission of harmful pathogens. Microorganisms within biofilm communities are more resistant to disinfection compared to planktonic microorganisms, making them difficult to manage in DWDSs. This study evaluates the impact of four unique drinking water treatments on biofilm community structure using metagenomic DNA sequencing. Four experimental DWDSs were subjected to the following treatments: (1) conventional coagulation, (2) magnetic ion exchange contact (MIEX) plus conventional coagulation, (3) MIEX plus conventional coagulation plus granular activated carbon, and (4) membrane filtration (MF). Bacterial biofilms located inside the pipes of each system were sampled under sterile conditions both (a) immediately after treatment application (‘inlet’) and (b) at a 1 km distance from the treatment application (‘outlet’). Bacterial 16S rRNA gene sequencing revealed that the outlet biofilms were more diverse than those sampled at the inlet for all treatments. The lowest number of unique operational taxonomic units (OTUs) and lowest diversity was observed in the MF inlet. However, the MF system revealed the greatest increase in diversity and OTU count from inlet to outlet. Further, the biofilm communities at the outlet of each system were more similar to one another than to their respective inlet, suggesting that biofilm communities converge towards a common established equilibrium as distance from treatment application increases. Based on the results, MF treatment is most effective at inhibiting biofilm growth, but a highly efficient post-treatment disinfection regime is also critical in order to prevent the high rates of post-treatment regrowth. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Biofilms are diverse microbial communities that grow on moisture-rich surfaces (Donlan, 2002). In oligotrophic environments, a biofilm habitat provides a selective advantage for ⇑ Corresponding author. Tel.: +61 8 8313 8245. E-mail address: [email protected] (J.L.A. Shaw). http://dx.doi.org/10.1016/j.chemosphere.2014.06.077 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
microbes compared to a planktonic lifestyle. For example, established biofilms secrete mucus that free-floating, organic particles can adhere to, creating a stationary food source for biofilm bacteria (Characklis and Marshall, 1990). Multiple trophic levels exist within the biofilm community; the waste of some members is utilised as food by others, creating symbiotic relationships between inhabitants (Characklis and Marshall, 1990). Further, flowing channels are created within the biofilm structure to allow
186
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
increased circulation and access to nutrients, promoting cell survival and growth (Donlan, 2002). Biofilms within drinking water distribution systems (DWDSs) pose financial and functional problems, as they can impair taste, odour and colour of drinking water, and increase the risk of pathogen transmission to consumers (Berry et al., 2006). Biofilms also reduce effective pipe diameters and increase frictional drag, which results in the need for increased pumping power and costs. Some species can also corrode pipes, creating rough surfaces upon which other biofilm species can more easily attach (Beech and Sunner, 2004). Furthermore, controlling biofilms in DWDSs is problematic as microbes within a biofilm matrix are more resistant to disinfectants than planktonic microbes (Lewis, 2001). Disinfectants, such as silver, UV radiation, chlorine and chloramine have all proven ineffective against DWDS biofilm formation (Lewis, 2001; Schwartz et al., 2003; Silvestry-Rodriguez et al., 2008). Given that biofilms cause significant disruptions to DWDSs and provide a survival advantage to harmful microorganisms, understanding how different treatment strategies impact biofilm development is financially and medically relevant. However, most previous DWDS biofilm studies have only focused on a small range of bacterial taxa, roughly estimating biodiversity using DGGE techniques (Emtiazi et al., 2004), or detecting specific species of interest by qPCR (Beumer et al., 2010) and laboratory culturing. More recently, high-throughput metagenomic sequencing has enabled rapid and inexpensive examination of entire bacterial communities from multiple samples in a single DNA sequencing reaction (Drury et al., 2013; Holinger et al., 2014; Douterelo et al., 2013). This new, high-resolution approach allows for the taxonomic identification of manifold species from the DNA present and allows a more detailed analysis as a result. Sample specific, unique ‘barcodes’ are incorporated into each sequence, enabling multiple samples to be processed in one sequencing reaction, greatly reducing the time and costs associated with processing large numbers of samples. This study used high-throughput metagenomic DNA sequencing to investigate the impact of four different water treatments on bacterial biofilms within DWDSs. Four experimental DWDSs received water subjected to one of the following treatments: (1) conventional coagulation, (2) magnetic ion exchange resin (MIEX) plus conventional coagulation, (3) MIEX plus conventional coagulation and granular activated carbon (GAC), and (4) membrane filtration (MF). Conventional coagulation treatment was used as a baseline control for the study, as it is currently the most widely employed water treatment in Australia. Coagulation, MIEX and GAC treatments all reduce organic matter within DWDSs, limiting the amount of nutrients available to bacteria, and therefore inhibiting bacterial growth. The MF treatment functions as a physical barrier, preventing bacterial cells and other organic materials from entering distribution pipelines, therefore averting biofilm formation. Biofilm communities downstream of these four water treatments were compared at two locations (immediately after treatment, and 1 km downstream). The metagenomic sequencing approach enabled taxonomic identification of specific bacterial biofilm taxa and provided information about the effect of different water treatments on biofilm community structure.
2. Materials and methods 2.1. Experimental facility and treatment processes Four pilot DWDSs were installed on site at the Mount Pleasant water treatment plant (WTP), near Adelaide, South Australia. Full details of these experimental water treatment systems have been previously described (Ho et al., 2012). The feed water was supplied
Fig. 1. Four treatment methods for the experimental distribution systems in the order of hypothetical increasing water quality: Stream 1 – Conventional treatment; Stream 2 – MIEX plus conventional treatment; Stream 3 – MIEX plus conventional treatment plus GAC; Stream 4 – Direct Microfiltration and Nanofiltration. As previously described by Ho et al. (2012).
from the Mt. Pleasant WTP inlet, which is sourced from the River Murray. Each DWDS pipe was 1 km in length and generated a constant feed flow rate of approximately 250 L h1. To decontaminate the pipes prior to the experiment, each DWDS pipe was flushed with product water from the main WTP with a high level of chlorine (sodium hypochlorite) dosing, until a consistent chlorine residual of 20–25 mg L1 was exiting the systems and turbidity was similar to the inlet. Following this, each system was sealed under pressure, and the chlorinated water held in contact with the pipes overnight, during which pressure was monitored for the integrity testing. After 24 h, WTP product water was flushed through the pipes until chlorine residual at the outlets again matched the inlets (1.5 mg L1). Once all four systems demonstrated that water quality was unaffected by passage through the pipes, the project began, and water from each of the treatments began flowing through the pipes (Fig. 1). During the experiment, the product water from each treatment was dosed with chlorine (sodium hypochlorite) prior to distribution, to achieve a C.t factor of 30 mg min L1. No chlorine residual was present on entry to the DWDS, with the aim of replicating the last 1 km of a DWDS where residual disinfectant has decayed. 2.1.1. Treatment 1: Conventional treatment (coagulation and filtration) The conventional treatment comprised of coagulation, flocculation and dual media (sand/anthracite) filtration, utilising an upflow clarifier and a gravity-fed Perspex filter column. The coagulant employed was aluminium sulphate as Al2(SO4)318H2O, with the dose ranging from 20 to 160 mg L1. A coagulation pH of between 6.0 and 6.5 was maintained through addition of sodium hydroxide or sodium bicarbonate buffering, depending on source water alkalinity. In addition, a coagulant aid, either anionic polyacrylamide (LT20, BASF Chemicals, Australia) or high molecular weight polyDADMAC (LT425, BASF Chemicals, Australia), was dosed downstream of the coagulant. The conventional treatment system was selected as a baseline control, as it represents the most widely applied drinking water treatment process currently employed in Australia.
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
2.1.2. Treatment 2: MIEX and conventional treatment The second treatment consisted of high rate magnetic ion exchange contact (MIEX DOC process, Orica, Australia) coupled with conventional coagulation, flocculation and dual media (sand/anthracite) filtration (Section 2.1.1). The average MIEX resin dose applied during the study was 15 L kL1. A continuous stirred tank reactor with a cone settler operating at 10% resin regeneration was employed for the MIEX DOC process. Due to the ability of the MIEX DOC process to efficiently remove absorbable organic materials, the subsequent coagulation treatment is primarily a clarification step, following the main organic carbon removal by the MIEX resin. As such, the coagulant demand is reduced, leading to a lower and less variable aluminium dose range (10–80 mg L1) compared with the conventional coagulation treatment system alone. 2.1.3. Treatment 3: MIEX, conventional treatment & GAC The third treatment system utilised product water from treatment two (Section 2.1.2) with the addition of two parallel pilotscale F400 GAC filters (Calgon, USA). F400 is a bituminous coalbased GAC with effective granule size 0.55–0.75 mm, which is commonly applied in water and wastewater applications to remove organic contaminants. Filtration was achieved using packed bed columns with gravity-fed empty bed contact times of approximately 14 min at 125 L h1 for each column. 2.1.4. Treatment 4: MF MF consisted of microfiltration using a single submerged hollow fibre membrane module (Memcor CMF-S system, USA) for particulate removal, followed by a single FILMTEC NF 270-4040 spiral wound nanofiltration membrane (DOW, USA). The microfiltration process was operated at 1000 L h1 with 75% permeate recovery. The nanofiltration process was operated in dead-end configuration with 43% permeate recovery, producing 325 L h1. Nominal pore size for the microfiltration was 0.2 m, and the molecular weight cut-off for the nanofiltration was 270 Da. 2.2. Bacterial cell enumeration by flow cytometry Bacterial cells present in the water streams were enumerated using flow cytometry (FCM) (Hammes et al., 2008). Duplicate 500 L aliquots were collected on 29 occasions, on a weekly to fortnightly basis, for twelve months leading up to biofilm sample collection. Bacterial cells were stained with SYTO-9 from the BacLight bacterial viability kit (Molecular Probes, USA), and FCM analyses were conducted using a FACS Calibur flow cytometer (Becton Dickinson, USA), as previously described (Hoefel, 2003). Data were analysed using CellQuest software (Becton Dickinson, USA) and a Mann–Whitney U statistical test was performed using SPSS statistical software (IBM v.21) to determine if there were differences between samples. 2.3. Sample collection and sequencing The systems ran for 28 months prior to biofilm sample collection, from March 2010 to July 2012. Bacterial biofilms located inside the pipes of each DWDS were sampled post-treatment, under sterile conditions, both (a) immediately before distribution (inlet) and (b) at a 1-km distance from treatment application (outlet). The outlet samples represent biofilm communities that would occur closer to consumer taps, whereas inlet samples represent biofilms occurring where the water has recently undergone treatment. Biofilm sampling was carried out using a method adapted from Storey and Kaucner (2009). Flow to the system was stopped and soil excavated from around the buried pipes, enabling easy access to the pipes and preventing cross-contamination of soil into the system. The surfaces of the pipes were cleaned and physically
187
decontaminated with 100-ppm sodium hypochlorite solution and then rinsed with ultrapure water to prevent chlorinated water from contacting the contents of the pipe. Pipe sections were then cut and removed with a sterilised pipe cutter and immediately transferred to a sterile container, sealed in a zip-lock bag and placed into an ice-chest for transport to the laboratory. Within 4 h, the pipe sections were delivered to the laboratory and biofilm samples were recovered from pipe surfaces using a sterile cell scraper before being transferred into 1/4 strength Ringers solution. DNA was then extracted from 250 mg of each biofilm sample using the Ultraclean soil DNA extraction kit (MoBio Laboratories, Solana Beach, CA) following manufacturer’s recommendations. Bacterial 16S ribosomal RNA (rRNA) amplicon libraries were prepared from genomic DNA using a two-step nested real-time PCR (RT-PCR). A nested approach was adopted to reduce barcoded-primer bias (Berry et al., 2011) and minimise contributions of degraded contaminant DNA. RT-PCR was used to analyse the DNA melt curves prior to NGS sequencing. Melt curve analyses enable the level of laboratory contamination present in the samples to be assessed by comparing the melt profiles of the samples to that of the non-template control. All PCR reactions were prepared in a dedicated sterile hood. Prior to adding GoTaq, the PCR mastermix was subjected to ultraviolet radiation for fifteen min to reduce laboratory contamination. DNA was added in a second sterile PCR hood in a separate room. In the first step of the nested RT-PCR, universal primers 27F (50 -AGAGTTTGATCCTGGCTCAG-30 ) and 1492R (50 -TACCTTGTTACGACTT-30 ) were used to amplify a 1500 basepair (bp) segment of the hypervariable V3 region within the 16S rRNA gene (Hoefel et al., 2005). Four replicate 25 lL PCR reactions per sample were prepared and pooled post-PCR to reduce PCR bias (Polz and Cavanaugh, 1998). Each reaction contained 5 lL of DNA extraction template, 5% dimethylsulfoxide (DMSO), 3.3 M SYTO-9 dye, 0.4 U lL1 Go Taq (Promega), 0.2 mM dNTPs, 0.5 lM of each primer, 1x PCR buffer, 2.5 mM of MgCl2 and 4.4 L of nuclease free water. The 1500 bp fragment was amplified using the following parameters: 95 °C for 3 min, 35 cycles at 94 °C for 30 s, 50 °C for 60 s and 72 °C for 120 s. Next, a smaller 170 bp region from within the 1500 bp PCR product was amplified using barcoded Ion torrent fusion primers 341F (50 -CCATCTCATCCCTGCGTGTCTCCGACTCAGnnnnnnnCCTACGGGAGGCAGCAG-30 ) and 518R (50 -CCTCTCTATGGGCAGTCGGTGATATTACCGCGGCTGCTGG-30 ). The primer sequence from Muyzer et al. (1993) is underlined, and ‘n’ represents a sample specific barcode sequence. Replicate reactions were again prepared in quadruplicate, and each 25 L reaction contained: 5 lL of DNA extraction template, 5% DMSO, 3.3 M SYTO-9 dye, 0.4 U lL1 Go Taq, 0.2 mM dNTPs, 0.5 lM of each primer, 1x PCR buffer, 2.5 mM of MgCl2 and 4.4 L of nuclease free water. The second round of RT-PCR amplification was performed with the following parameters: 95 °C for 3 min, 35 cycles at 95 °C for 20 s, 65 °C for 20 s and 72 °C for 30 s. Replicate PCR reactions were again pooled together prior to purification (Ampure, Agencourt Bioscience). The eight purified samples were then quantified and blended together to equimolar concentrations to make the DNA library. The amplicon library was sequenced using the Ion Torrent Personal Genome Machine (Life technologies, Carlsbad, CA, USA). A combined total of 279,671 raw sequence reads were obtained (Table SM-1 in supplementary material). Of these reads, 236,192 had zero mismatches within the barcode and proximal primer sequence and were therefore demultiplexed into their respective samples. Raw sample sequences are available from the NCBI Sequence Read Archive (SRA accession number: SRR1425418; BioProject ID: PRJNA253081). Reads with mismatches present were discarded from further analysis. Quality (Phred threshold of 20 for 90% of sequence bp) and length filtering (90 bp) were applied using FastX tookit (FASTX-toolkit v0.0.13; http://hannonlab.cshl.edu/
188
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
fastx_toolkit), and barcodes and primer sequences were trimmed off using CutAdapt v1.1 (Martin, 2011). Post quality filtering, the mean number of reads per sample was 5756 ± 2572 (Table SM-1). The subsequent fastq file containing the quality-filtered sequences was then converted to a QIIME software compatible .fna file (Available from: http://genomics.azcc.arizona.edu/help. php3). Within QIIME (Caporaso et al., 2010), sequences that were 97% similar were clustered to form operational taxonomic units (OTUs) using Uclust, and aligned using Pynast. Chimeric sequences were detected and removed using Chimera slayer (Haas et al., 2011). A standard (default) threshold (0.8) was applied to assign taxonomy using Greengenes database (version gg_12_8; http://greengenes.lbl.gov). All samples were rarefied at 2,590 sequences (i.e. the lowest number of reads present within a single sample) to ensure an even read coverage for each sample and minimise diversity estimates becoming skewed by read abundance. Further exploratory and statistical analysis of the data was carried out using QIIME software. To assess the diversity of a sample, Chao species richness and Shannon-Weiner diversity estimates were calculated. To test for significant differences across samples, a two-sample t-test within the QIIME software was conducted, whereby 2590 sequences per sample were repeatedly sub-sampled 999 times by Monte Carlo permutations and compared. Sample similarity was assessed within UPGMA trees with 1,940 jackknifed sequences per sample (e = 10 permutations) and viewed using FigTree V1.4 software (tree.bio.ed.ac.uk/software). Phylogenetic distance between taxa was calculated using weighted UniFrac analyses and visualised by hierarchal clustering in UPGMA trees (Lozupone and Knight, 2005). 3. Results
inlet to the outlet (782%), followed by the MIEX/conventional DWDS (98%). The cell count in the MIEX/GAC DWDS stayed relatively stable throughout the system, showing only a 4% decrease in mean bacterial cells from the inlet to the outlet. Oddly, the conventional DWDS cell count decreased by 45%. These results suggest that regardless of the initial treatment applied, the bacterial cell count within each system shifts towards a common, optimal number further downstream of the treatment application. 3.2. Biofilm species diversity and richness To assess the complexity of the microbial biofilm communities, the number of OTUs (Fig. 3a), Shannon–Weaver diversity (Fig. 3b) and Chao species richness (Table SM-3) were calculated for each sample. Post-rarefaction, the number of OTUs per sample ranged from 198 to 487. An increase in the number of OTUs and species diversity was observed from inlet to outlet for all DWDSs. The MF inlet sample had the lowest OTU count and diversity, yet the largest increase from inlet to outlet was observed in the MF system. In contrast, the conventional system inlet had the largest diversity and OTU count, and was most stable from inlet to outlet. Each inlet biofilm was significantly different to the other inlets (P = <0.05; two sample, non-parametric t-test; 999 Monte Carlo permutations; Table SM-4), demonstrating that each treatment initially causes significant variations in biodiversity and species richness. However, with the exception of the MF DWDS, the biofilms at the outlets were not significantly different to one another (Table SM-4), indicating that biofilms begin to shift towards a common, community equilibrium further downstream from treatment application. Interestingly, the diversity and species richness of the MF outlet biofilm were not significantly different from the inlet of the MIEX/GAC DWDS (P = 0.961; two sample, non-parametric
3.1. FCM FCM-based enumeration of the bacterial cells revealed large differences between the four DWDS inlets, with most system inlets displaying significantly different cell counts (Mann–Whitney U test; Table SM-2). The MF DWDS inlet had considerably lower bacterial cell counts compared to the other inlet DWDSs, whereas the conventional DWDS inlet had the highest cell count (Fig. 2). However, cell counts at the four DWDS outlets appear to be more uniform. The MIEX/Conventional and MIEX/GAC DWDS outlet samples were not significantly different from the MF DWDS outlet (Mann– Whitney U test; P-value = 0.380 and 0.176 respectively). The MF DWDS displayed the sharpest increase in bacterial cells from the
(a)
(b)
Fig. 2. Mean active bacterial cell count (cells ml1) at the inlet and outlet of each experimental DWDS measured by flow cytometry (FCM) on a fortnightly basis over a 12 month period prior to biofilm samples being collected (error bars are shown as one standard deviation about the mean).
Fig. 3. (a) Number of unique operational taxonomic units identified for each sample (OTUs; 3% nucleotide cut off); (b) Shannon-Wiener diversity index for each sample. All samples were rarefied to 2590 sequencing depth previously. Line depicts change from inlet to outlet.
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
t-test; 999 Monte Carlo permutations), suggesting that although the MF outlet biofilm is significantly different to the other outlet biofilms at a 1 km distance, it may still be evolving towards the same common established equilibrium but at a slower rate. In summary, the MF treatment appears to hinder biofilm formation to a greater extent than the other treatments, yet without a disinfectant residual, the diversity and species richness increases sharply further down the pipe, eventually producing a biofilm community more similar to that of other DWDS treatments.
3.3. Biofilm community structure Biofilm community structure was compared by calculating phylogenetic distances in a weighted UniFrac UPGMA tree. Once again, the inlet of the MF DWDS contained the most unique biofilm community (Fig. 4). The inlets of three of the DWDS (conventional treatment, MIEX/conventional and MIEX/conventional/GAC) clustered together, illustrating that these treatment types have similar effects on the biofilm species. Interestingly, the outlet biofilms for all of the treatments clustered together in the UPGMA analysis, rather than with their respective inlet, demonstrating that the biofilms shift to a common community structure as distance from treatment application increases, regardless of the initial treatment applied. Further, the inlet and outlet of the MF DWDS were the only samples that were significantly different to one another (UniFrac Monte Carlo significance; P = 0.03), indicating the large extent to which the community changed downstream of the MF system in the absence of a chlorine residual.
189
3.4. Taxa identified Specific bacterial OTUs within each sample were identified (Fig. 5; Table SM-5), and in all biofilms, except the inlet of the MF DWDS, Alphaproteobacteria were the dominant taxa. From the inlets to the outlets, Alphaproteobacteria significantly decreased (Conventional = 4%, MIEX/Conventional = 24%, MIEX/Conventional/GAC = 35%), yet conversely, a 47% increase was observed in the MF DWDS. Upon closer inspection, Rhizobiales and Sphingomonadales were the dominant taxa within the Alphaproteobacteria class. Bradyrhizobiaceae and Hyphomicrobiaceae being the most dominant within the Rhizobiales order. A decrease in Sphingomonadales was observed in all DWDSs from the inlet to outlet (Conventional = 6%, MIEX/Conventional = 6%, MIEX/Conventional/GAC = 28%) with the exception of the MF DWDS, where Sphingomonadales were detected in low abundances at the outlet (2%) but were not detected in the inlet. The MF DWDS inlet was instead dominated by Actinomycetales (72%) and Lactobacillales (8%). These OTUs were further identified as predominantly Propionibacterium and Streptococcus, which includes some pathogenic and human microbiome-associated species. Small percentages (<5%) of other important and potentially pathogenic genera such as Escherichia coli, Legionella, Clostridia and Bacilli were also identified in the biofilms (Table SM-5). The nitrifying bacteria Nitrospira was identified in small proportions in the biofilms (<2%), particularly in the MIEX/conventional inlet and the MF outlet biofilms. A general decrease in Actinobacteria, predominantly Actinomycetales and an unknown Actinobacteria
Fig. 4. Cluster dendogram of biofilm sample similarity based on the bacterial community identified (1940 sequences per jack-knifed subset) and corresponding pie charts showing percentage contribution of dominant taxa (i.e. taxa > 1% contribution to overall diversity). Taxa < 1% contribution and ‘unknown’ taxa were pooled as ‘other’ bacteria (yellow). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
190
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
Fig. 5. Percentage contribution of top 7 dominant taxa to the bacterial community of each sample at family phylogenetic level.
OTU, was also observed from the inlet to the outlet for all DWDS, with the largest decrease found in the MF DWDS (73%), which implies an association between Actinobacteria and highly treated drinking waters. 4. Discussion In this study, the MF DWDS treatment was the most effective at inhibiting biofilm formation, as the inlet of this DWDS had the lowest diversity, bacterial cell count and unique OTUs. However, sharp increases in biodiversity, OTU count and cell counts were observed within the MF system at a 1 km distance from the treatment application, resulting in communities more similar to that of the other DWDS treatment regimes. Additionally, the DWDS outlet biofilms were found to be more similar to one another, rather than their respective inlets, suggesting that a common biofilm community will eventually occur in DWDSs regardless of the initial treatment method applied. In summary, the biofilm communities located at the inlet of the distribution systems appear to be highly impacted by the type of treatment being applied but then shift towards a common, optimal community at a further distance downstream of the treatment application. Bacterial repopulation of the MF DWDS was most likely due to a small number of cells surviving initial decontamination efforts or contamination from holding tanks prior to distribution, as opposed to membrane failure. Whilst every care was taken to minimise contamination risk, it is possible that a small number of cells may have survived the initial pipe disinfection process prior to the beginning of the experiment or the disinfection stage prior to entering the DWDS. However, it must be noted that full-scale DWDSs are not sterile systems and have similar entry points for potential contamination, such as storage tanks. Therefore, the contamination risk of the experimental DWDSs in this study is no greater than that of a full-scale DWDS. It should also be stressed that although some taxa detected in the MF outlet were not detected in the MF inlet, this may be due to the abundance of those taxa being below the detectability threshold at the inlet, rather than contamination further down the pipeline, as rare sequences can often be obscured by more dominant taxa during PCR amplification. Variations in bacterial community structure were observed across the different treatments, with the most pronounced differ-
ences observed between the inlets of the DWDSs. The dominant taxa were identified as Sphingomonadales and Rhizobiales in all biofilms except the MF inlet, where the dominant taxa were Actinomycetales and Lactobacillales. Closer inspection of the Lactobacillales family revealed the majority of these OTUs were Streptococcus, a genus known to contain both pathogenic species and non-pathogenic species that occur naturally within the human gut, and are consequently used as an indicator of human faecal contamination (Pinto et al., 1999). The presence of Streptococcus in the MF DWDS is interesting, as this treatment appeared to be the most effective at removing the major bacterial groups observed in the other treatment streams, yet it harbored larger percentages of genera that include pathogenic species. However, as the number of cells present in the MF DWDS was much lower than that of the other three DWDS, the actual number of potentially pathogenic cells in the MF system were likely below infectivity threshold. Smaller percentages of other potentially pathogenic genera were also identified in the biofilms such as Enterobacteria, Legionella, Clostridia and Bacillus (Ashbolt, 2004). Of interest was the detection of Burkholderiales in both the inlet and outlet biofilms of all four systems, with an apparent increase of this group in the outlet biofilm communities. Burkholderiales includes the genus Burkholderia, which contains pathogenic species of particular significance in warmer climates (Inglis et al., 2000; Pumpuang et al., 2011; Mayo et al., 2011). Bacteria that are able to cause disinfectant decay in DWDSs through nitrification, such as Nitrospira, were also identified (Blackburne et al., 2007). The process of nitrification in chloraminated DWDSs leads to a loss of disinfectant residual and bacteriological contamination of the system, which causes deterioration in water quality and may pose a health hazard to consumers, therefore the presence of nitrifying bacteria in system biofilms must be considered when managing disinfectant dosing levels. It has been previously suggested that biofilms can take several years to become fully mature (Martiny et al., 2005; Liu et al., 2012). Consequently, the species observed in this 28 month study may be more representative of an immature biofilm in the earlier stages of establishment, and the dominant bacterial taxa observed may play an important role in DWDS biofilm colonisation. In three of the DWDSs (Conventional, MIEX/Conventional and MIEX/Conventional/GAC), the abundance of Sphingomonadales was higher in the inlets than the outlets. This suggests that Sphingomonadales
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
thrive in an environment where other bacteria are in low numbers, implicating them as opportunistic, colonising biofilm species. The MF treatment appears to prevent Sphinomonadales cells from penetrating the DWDS as they were not detected in the inlet of the MF DWDS; however, the number of Sphingomonadales cells increased at the MF outlet to levels similar to the inlets of the other DWDSs, indicating that although MF treatment hinders regrowth to a greater extent than organics removal alone, it is not capable of preventing a few surviving cells from colonising a DWDS. Sphingomonadales are a Gram-negative bacterial group defined by the presence of sphingolipids in the outer membrane of the cell wall. It has been proposed that the membrane of Gram-negative bacteria, such as Sphingomonadales, acts a protective barrier against antibiotics, pollutants and organic solvents (Ramos et al., 2002). The presence of a Gram-negative cell wall and a lipid bilayer could mean that Sphingomonadales are well protected from disinfection agents such as chloramine (Sun et al., 2013), making them an important biofilm species. In the outlets of the pipes, the Gramnegative group Burkholderiales were a dominant biofilm taxa. Some species of Burkholderiales are motile due to the presence of flagella (Garrity et al., 2005), potentially aiding attachment through migration towards nutrient rich biofilms. Therefore, Burkholderiales may be a species more likely to attach after the initial colonisation phase has taken place. Several studies have suggested that the reduction of organic carbon is essential for the inhibition of bacterial regrowth, by reducing the level of available nutrients (LeChevallier et al., 1990; Volk and LeChevallier, 1999). In this study, the reduction of organic carbon alone appears to stabilise bacterial biofilm regrowth, as the conventional, MIEX/conventional and MIEX/GAC DWDSs did not significantly change from the inlet to the outlet. This suggests that reducing the organic components of drinking water is an important step in inhibiting biofilm growth, especially in longer systems that may be prone to premature disinfectant decay. Other studies have also inferred the importance of reducing organic carbon via indirect measurements such as ATP, assimilable organic carbon and biodegradable organic carbon (van der Kooij, 1977; Magic-Knezev and van der Kooij, 2004), this study supports those findings using novel and high resolution techniques. In summary, this study illustrates the importance of effective disinfection post-treatment to prevent bacterial regrowth in DWDS, as even highly efficient treatments such as MF do not prevent colonisation of bacterial biofilms. Our results suggest that once the residual disinfectant has decayed, biofilms reach a common established community equilibrium regardless of the effectiveness of the treatment method at the beginning of the system, even in distances as short as 1 km. Moreover, this research demonstrates the power of performing metagenomic bacterial surveys on DWDSs and delivers a comprehensive overview of important biofilm taxa.
Acknowledgements This study was financially supported by the Australian Research Council (ARC) linkage grant LP0991985, Co-operative Research Centre for Water Quality and Treatment (CRC-WQT) project 2417, Water Research Australia (Water RA) project 1008, South Australian Water Corporation, United Water International, Grampians Wimmera Mallee Water, Water Corporation, Delft University of Technology, DCM Process Control and Orica Watercare. Laboratory work was carried out at the South Australian Water Corporation, Victoria Square, Adelaide and the Australian Centre for Ancient DNA at the University of Adelaide. We would like to thank Dr Julien Soubrier and Dr Laura Weyrich for assistance with the bioinformatic analysis and manuscript editing.
191
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2014.06.077.
References
Ashbolt, N.J., 2004. Microbial contamination of drinking water and disease outcomes in developing regions. Toxicology 198, 229–238. Beech, I.B., Sunner, J., 2004. Biocorrosion: towards understanding interactions between biofilms and metals. Curr. Opin. Biotechnol. 15, 181–186. Berry, D., Xi, C., Raskin, L., 2006. Microbial ecology of drinking water distribution systems. Curr. Opin. Biotechnol. 17, 297–302. Berry, D., Ben Mahfoudh, K., Wagner, M., Loy, A., 2011. Barcoded primers used in multiplex amplicon pyrosequencing bias amplification. Appl. Environ. Microbiol. 77, 7846–7849. Beumer, A., King, D., Donohue, M., Mistry, J., Covert, T., Pfaller, S., 2010. Detection of Mycobacterium avium subsp. paratuberculosis in drinking water and biofilms by quantitative PCR. Appl. Environ. Microbiol. 76, 7367–7370. Blackburne, R., Vadivelu, V.M., Yuan, Z., Keller, J., 2007. Kinetic characterisation of an enriched Nitrospira culture with comparison to Nitrobacter. Water Res. 41, 3033–3042. 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 high-throughput community sequencing data. Nat. Methods 7, 335–336. Characklis, W.G., Marshall, K.C., 1990. Biofilms. John Wiley & Sons, New York. Donlan, R.M., 2002. Biofilms: microbial life on surfaces. Emerg. Infect. Dis. 8, 881– 890. Douterelo, I., Sharpe, R.L., Boxall, J.B., 2013. Influence of hydraulic regimes on bacterial community structure and composition in an experimental drinking water distribution system. Water Res. 47, 503–516. Drury, B., Rosi-Marshall, E., Kelly, J.J., 2013. Wastewater treatment effluent reduces the abundance and diversity of benthic bacterial communities in urban and suburban rivers. Appl. Environ. Microbiol. 79, 1897–1905. Emtiazi, F., Schwartz, T., Marten, S.M., Krolla-Sidenstein, P., Obst, U., 2004. Investigation of natural biofilms formed during the production of drinking water from surface water embankment filtration. Water Res. 38, 1197–1206. Garrity, G., Bell, J., Lilburn, T., 2005. Family Burkholderiaceae. In: Garrity, G., Brenner, D.J., Staley, J.T., Krieg, N.R., Boone, D.R., De Vos, P., Goodfellow, M., Rainey, F.A., Schleifer, K. (Eds.), Bergey’s Manual of Systematic Biology, The Proteobacteria, vol. 2. Springer Science, New York, pp. 575–591. Haas, B.J., Gevers, D., Earl, A.M., Feldgarden, M., Ward, D.V., Giannoukos, G., 2011. Chimeric 16S rRNA sequence formation and detection in Sanger and 454pyrosequenced PCR amplicons. Genome Res. 21, 494–504. Hammes, F., Berney, M., Wang, Y., Vital, M., Köster, O., Egli, T., 2008. Flowcytometric total bacterial cell counts as a descriptive microbiological parameter for drinking water treatment processes. Water Res. 42, 269–277. Ho, L., Braun, K., Fabris, R., Hoefel, D., Morran, J., Monis, P., Drikas, M., 2012. Comparison of drinking water treatment process streams for optimal bacteriological water quality. Water Res. 46, 3934–3942. Hoefel, D., 2003. Enumeration of water-borne bacteria using viability assays and flow cytometry: a comparison to culture-based techniques. J. Microbiol. Methods 55, 585–597. Hoefel, D., Monis, P.T., Grooby, W.L., Andrews, S., Saint, C.P., 2005. Profiling bacterial survival through a water treatment process and subsequent distribution system. J. Appl. Microbiol. 99, 175–186. Holinger, E.P., Ross, K.A., Robertson, C.E., Stevens, M.J., Harris, J.K., Pace, N.R., 2014. Molecular analysis of point-of-use municipal drinking water microbiology. Water Res. 49, 225–235. Inglis, T.J., Garrow, S.C., Henderson, M., Clair, A., Sampson, J., O’Reilly, L., Cameron, B., 2000. Burkholderia pseudomallei traced to water treatment plant in Australia. Emerg. Infect. Dis. 6, 56–59. LeChevallier, M.W., Schulz, W., Lee, R.G., 1990. Bacterial nutrients in drinking water. AWWARF, Denver, Colorado. Lewis, K., 2001. Riddle of biofilm resistance. Antimicrob. Agents Chemother. 45, 999–1007. Liu, R., Yu, Z., Guo, H., Liu, M., Zhang, H., Yang, M., 2012. Pyrosequencing analysis of eukaryotic and bacterial communities in faucet biofilms. Sci. Total Environ. 435–436, 124–131. Lozupone, C., Knight, R., 2005. UniFrac: a new phylogenetic method for comparing microbial communities. Appl. Environ. Microbiol. 71, 8228–8235. Magic-Knezev, A., van der Kooij, D., 2004. Optimisation and significance of ATP analysis for measuring active biomass in granular activated carbon filters used in water treatment. Water Res. 38, 3971–3979. Martin, M., 2011. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet. J. 17, 10–12. Martiny, A.C., Albrechtsen, H.J., Arvin, E., Molin, S., 2005. Identification of bacteria in biofilm and bulk water samples from a nonchlorinated model drinking water distribution system: detection of a large nitrite-oxidizing population associated with Nitrospira spp. Appl. Environ. Microbiol. 71, 8611–8617.
192
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
Mayo, M., Kaestli, M., Harrington, G., Cheng, A.C., Ward, L., Karp, D., Jolly, P., Godoy, D., Spratt, B.G., Currie, B.J., 2011. Burkholderia pseudomallei in unchlorinated domestic bore water, tropical northern Australia. Emerg. Infect. Dis. 17, 1283– 1285. 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, 695–700. Pinto, B., Pierotti, G., Reali, D., 1999. Characterisation of ‘faecal streptococci’ as indicators of feacal pollution and distribution in the environment. Lett. Appl. Microbiol. 29, 258–263. Polz, M.F., Cavanaugh, C.M., 1998. Bias in template-to-product ratios in multitemplate PCR. Appl. Environ. Microbiol. 64, 3724–3730. Pumpuang, A., Chantratita, N., Wikraiphat, C., Saiprom, N., Day, N.P.J., Peacock, S.J., Wuthiekanun, V., 2011. Survival of Burkholderia pseudomallei in distilled water for 16 years. T. Roy. Soc. Trop. Med. H. 105, 598–600. Ramos, J.L., Duque, E., Gallegos, M.-T., Godoy, P., Ramos-Gonzalez, M.I., Rojas, A., Teran, W., Segura, A., 2002. Mechanisms of solvent tolerance in Gram-negative bacteria. Ann. Rev. Microbiol. 56, 743–768.
Schwartz, T., Hoffmann, S., Obst, U., 2003. Formation of natural biofilms during chlorine dioxide and u.v. disinfection in a public drinking water distribution system. J. Appl. Microbiol. 95, 591–601. Silvestry-Rodriguez, N., Bright, K.R., Slack, D.C., Uhlmann, D.R., Gerba, C.P., 2008. Silver as a residual disinfectant to prevent biofilm formation in water distribution systems. Appl. Environ. Microbiol. 74, 1639–1641. Storey, M., Kaucner, C.E., 2009. Understanding the growth of opportunistic pathogens within distribution systems. Co-operative research centre for water quality and treatment. Research report 79. Water Research Australia Ltd. ISBN: 1876616296. Sun, W., Liu, W., Cui, L., Zhang, M., Wang, B., 2013. Characterization and identification of a chlorine-resistant bacterium, Sphingomonas TS001, from a model drinking water distribution system. Sci. Total Environ. 458–460, 169– 175. van der Kooij, D., 1977. The occurence of Pseudomonas spp. in surface water and in tap water as determined on citrate media. J. Microbiol. 43, 187–197. Volk, C.J., LeChevallier, M.W., 1999. Impacts of the reduction of nutrient levels on bacterial water quality in distribution systems. Appl. Environ. Microbiol. 65, 4957–4966.
Contents lists available at ScienceDirect
Chemosphere journal homepage: www.elsevier.com/locate/chemosphere
Assessing the impact of water treatment on bacterial biofilms in drinking water distribution systems using high-throughput DNA sequencing Jennifer L.A. Shaw a,⇑, Paul Monis b, Rolando Fabris b, Lionel Ho b, Kalan Braun b, Mary Drikas b, Alan Cooper a a b
Australian Centre for Ancient DNA (ACAD), University of Adelaide, Darling Building, North Terrace, Adelaide, SA 5000, Australia Australian Water Quality Centre, SA Water Corporation, 250 Victoria Square, Adelaide, SA 5000, Australia
h i g h l i g h t s Four drinking water distribution systems were subjected to unique water treatments. Bacterial biofilm communities were characterised by sequencing the 16S rRNA gene. Initially, biofilm communities varied across different treatments. Biofilm biodiversity increased with distance from treatment application. A common community structure is eventually attained regardless of initial treatment.
a r t i c l e
i n f o
Article history: Received 11 February 2014 Received in revised form 20 June 2014 Accepted 20 June 2014
Handling Editor: O. Hao Keywords: Biofilm Metagenomics DNA sequencing Drinking water treatment Bacteria
a b s t r a c t Biofilm control in drinking water distribution systems (DWDSs) is crucial, as biofilms are known to reduce flow efficiency, impair taste and quality of drinking water and have been implicated in the transmission of harmful pathogens. Microorganisms within biofilm communities are more resistant to disinfection compared to planktonic microorganisms, making them difficult to manage in DWDSs. This study evaluates the impact of four unique drinking water treatments on biofilm community structure using metagenomic DNA sequencing. Four experimental DWDSs were subjected to the following treatments: (1) conventional coagulation, (2) magnetic ion exchange contact (MIEX) plus conventional coagulation, (3) MIEX plus conventional coagulation plus granular activated carbon, and (4) membrane filtration (MF). Bacterial biofilms located inside the pipes of each system were sampled under sterile conditions both (a) immediately after treatment application (‘inlet’) and (b) at a 1 km distance from the treatment application (‘outlet’). Bacterial 16S rRNA gene sequencing revealed that the outlet biofilms were more diverse than those sampled at the inlet for all treatments. The lowest number of unique operational taxonomic units (OTUs) and lowest diversity was observed in the MF inlet. However, the MF system revealed the greatest increase in diversity and OTU count from inlet to outlet. Further, the biofilm communities at the outlet of each system were more similar to one another than to their respective inlet, suggesting that biofilm communities converge towards a common established equilibrium as distance from treatment application increases. Based on the results, MF treatment is most effective at inhibiting biofilm growth, but a highly efficient post-treatment disinfection regime is also critical in order to prevent the high rates of post-treatment regrowth. Ó 2014 Elsevier Ltd. All rights reserved.
1. Introduction Biofilms are diverse microbial communities that grow on moisture-rich surfaces (Donlan, 2002). In oligotrophic environments, a biofilm habitat provides a selective advantage for ⇑ Corresponding author. Tel.: +61 8 8313 8245. E-mail address: [email protected] (J.L.A. Shaw). http://dx.doi.org/10.1016/j.chemosphere.2014.06.077 0045-6535/Ó 2014 Elsevier Ltd. All rights reserved.
microbes compared to a planktonic lifestyle. For example, established biofilms secrete mucus that free-floating, organic particles can adhere to, creating a stationary food source for biofilm bacteria (Characklis and Marshall, 1990). Multiple trophic levels exist within the biofilm community; the waste of some members is utilised as food by others, creating symbiotic relationships between inhabitants (Characklis and Marshall, 1990). Further, flowing channels are created within the biofilm structure to allow
186
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
increased circulation and access to nutrients, promoting cell survival and growth (Donlan, 2002). Biofilms within drinking water distribution systems (DWDSs) pose financial and functional problems, as they can impair taste, odour and colour of drinking water, and increase the risk of pathogen transmission to consumers (Berry et al., 2006). Biofilms also reduce effective pipe diameters and increase frictional drag, which results in the need for increased pumping power and costs. Some species can also corrode pipes, creating rough surfaces upon which other biofilm species can more easily attach (Beech and Sunner, 2004). Furthermore, controlling biofilms in DWDSs is problematic as microbes within a biofilm matrix are more resistant to disinfectants than planktonic microbes (Lewis, 2001). Disinfectants, such as silver, UV radiation, chlorine and chloramine have all proven ineffective against DWDS biofilm formation (Lewis, 2001; Schwartz et al., 2003; Silvestry-Rodriguez et al., 2008). Given that biofilms cause significant disruptions to DWDSs and provide a survival advantage to harmful microorganisms, understanding how different treatment strategies impact biofilm development is financially and medically relevant. However, most previous DWDS biofilm studies have only focused on a small range of bacterial taxa, roughly estimating biodiversity using DGGE techniques (Emtiazi et al., 2004), or detecting specific species of interest by qPCR (Beumer et al., 2010) and laboratory culturing. More recently, high-throughput metagenomic sequencing has enabled rapid and inexpensive examination of entire bacterial communities from multiple samples in a single DNA sequencing reaction (Drury et al., 2013; Holinger et al., 2014; Douterelo et al., 2013). This new, high-resolution approach allows for the taxonomic identification of manifold species from the DNA present and allows a more detailed analysis as a result. Sample specific, unique ‘barcodes’ are incorporated into each sequence, enabling multiple samples to be processed in one sequencing reaction, greatly reducing the time and costs associated with processing large numbers of samples. This study used high-throughput metagenomic DNA sequencing to investigate the impact of four different water treatments on bacterial biofilms within DWDSs. Four experimental DWDSs received water subjected to one of the following treatments: (1) conventional coagulation, (2) magnetic ion exchange resin (MIEX) plus conventional coagulation, (3) MIEX plus conventional coagulation and granular activated carbon (GAC), and (4) membrane filtration (MF). Conventional coagulation treatment was used as a baseline control for the study, as it is currently the most widely employed water treatment in Australia. Coagulation, MIEX and GAC treatments all reduce organic matter within DWDSs, limiting the amount of nutrients available to bacteria, and therefore inhibiting bacterial growth. The MF treatment functions as a physical barrier, preventing bacterial cells and other organic materials from entering distribution pipelines, therefore averting biofilm formation. Biofilm communities downstream of these four water treatments were compared at two locations (immediately after treatment, and 1 km downstream). The metagenomic sequencing approach enabled taxonomic identification of specific bacterial biofilm taxa and provided information about the effect of different water treatments on biofilm community structure.
2. Materials and methods 2.1. Experimental facility and treatment processes Four pilot DWDSs were installed on site at the Mount Pleasant water treatment plant (WTP), near Adelaide, South Australia. Full details of these experimental water treatment systems have been previously described (Ho et al., 2012). The feed water was supplied
Fig. 1. Four treatment methods for the experimental distribution systems in the order of hypothetical increasing water quality: Stream 1 – Conventional treatment; Stream 2 – MIEX plus conventional treatment; Stream 3 – MIEX plus conventional treatment plus GAC; Stream 4 – Direct Microfiltration and Nanofiltration. As previously described by Ho et al. (2012).
from the Mt. Pleasant WTP inlet, which is sourced from the River Murray. Each DWDS pipe was 1 km in length and generated a constant feed flow rate of approximately 250 L h1. To decontaminate the pipes prior to the experiment, each DWDS pipe was flushed with product water from the main WTP with a high level of chlorine (sodium hypochlorite) dosing, until a consistent chlorine residual of 20–25 mg L1 was exiting the systems and turbidity was similar to the inlet. Following this, each system was sealed under pressure, and the chlorinated water held in contact with the pipes overnight, during which pressure was monitored for the integrity testing. After 24 h, WTP product water was flushed through the pipes until chlorine residual at the outlets again matched the inlets (1.5 mg L1). Once all four systems demonstrated that water quality was unaffected by passage through the pipes, the project began, and water from each of the treatments began flowing through the pipes (Fig. 1). During the experiment, the product water from each treatment was dosed with chlorine (sodium hypochlorite) prior to distribution, to achieve a C.t factor of 30 mg min L1. No chlorine residual was present on entry to the DWDS, with the aim of replicating the last 1 km of a DWDS where residual disinfectant has decayed. 2.1.1. Treatment 1: Conventional treatment (coagulation and filtration) The conventional treatment comprised of coagulation, flocculation and dual media (sand/anthracite) filtration, utilising an upflow clarifier and a gravity-fed Perspex filter column. The coagulant employed was aluminium sulphate as Al2(SO4)318H2O, with the dose ranging from 20 to 160 mg L1. A coagulation pH of between 6.0 and 6.5 was maintained through addition of sodium hydroxide or sodium bicarbonate buffering, depending on source water alkalinity. In addition, a coagulant aid, either anionic polyacrylamide (LT20, BASF Chemicals, Australia) or high molecular weight polyDADMAC (LT425, BASF Chemicals, Australia), was dosed downstream of the coagulant. The conventional treatment system was selected as a baseline control, as it represents the most widely applied drinking water treatment process currently employed in Australia.
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
2.1.2. Treatment 2: MIEX and conventional treatment The second treatment consisted of high rate magnetic ion exchange contact (MIEX DOC process, Orica, Australia) coupled with conventional coagulation, flocculation and dual media (sand/anthracite) filtration (Section 2.1.1). The average MIEX resin dose applied during the study was 15 L kL1. A continuous stirred tank reactor with a cone settler operating at 10% resin regeneration was employed for the MIEX DOC process. Due to the ability of the MIEX DOC process to efficiently remove absorbable organic materials, the subsequent coagulation treatment is primarily a clarification step, following the main organic carbon removal by the MIEX resin. As such, the coagulant demand is reduced, leading to a lower and less variable aluminium dose range (10–80 mg L1) compared with the conventional coagulation treatment system alone. 2.1.3. Treatment 3: MIEX, conventional treatment & GAC The third treatment system utilised product water from treatment two (Section 2.1.2) with the addition of two parallel pilotscale F400 GAC filters (Calgon, USA). F400 is a bituminous coalbased GAC with effective granule size 0.55–0.75 mm, which is commonly applied in water and wastewater applications to remove organic contaminants. Filtration was achieved using packed bed columns with gravity-fed empty bed contact times of approximately 14 min at 125 L h1 for each column. 2.1.4. Treatment 4: MF MF consisted of microfiltration using a single submerged hollow fibre membrane module (Memcor CMF-S system, USA) for particulate removal, followed by a single FILMTEC NF 270-4040 spiral wound nanofiltration membrane (DOW, USA). The microfiltration process was operated at 1000 L h1 with 75% permeate recovery. The nanofiltration process was operated in dead-end configuration with 43% permeate recovery, producing 325 L h1. Nominal pore size for the microfiltration was 0.2 m, and the molecular weight cut-off for the nanofiltration was 270 Da. 2.2. Bacterial cell enumeration by flow cytometry Bacterial cells present in the water streams were enumerated using flow cytometry (FCM) (Hammes et al., 2008). Duplicate 500 L aliquots were collected on 29 occasions, on a weekly to fortnightly basis, for twelve months leading up to biofilm sample collection. Bacterial cells were stained with SYTO-9 from the BacLight bacterial viability kit (Molecular Probes, USA), and FCM analyses were conducted using a FACS Calibur flow cytometer (Becton Dickinson, USA), as previously described (Hoefel, 2003). Data were analysed using CellQuest software (Becton Dickinson, USA) and a Mann–Whitney U statistical test was performed using SPSS statistical software (IBM v.21) to determine if there were differences between samples. 2.3. Sample collection and sequencing The systems ran for 28 months prior to biofilm sample collection, from March 2010 to July 2012. Bacterial biofilms located inside the pipes of each DWDS were sampled post-treatment, under sterile conditions, both (a) immediately before distribution (inlet) and (b) at a 1-km distance from treatment application (outlet). The outlet samples represent biofilm communities that would occur closer to consumer taps, whereas inlet samples represent biofilms occurring where the water has recently undergone treatment. Biofilm sampling was carried out using a method adapted from Storey and Kaucner (2009). Flow to the system was stopped and soil excavated from around the buried pipes, enabling easy access to the pipes and preventing cross-contamination of soil into the system. The surfaces of the pipes were cleaned and physically
187
decontaminated with 100-ppm sodium hypochlorite solution and then rinsed with ultrapure water to prevent chlorinated water from contacting the contents of the pipe. Pipe sections were then cut and removed with a sterilised pipe cutter and immediately transferred to a sterile container, sealed in a zip-lock bag and placed into an ice-chest for transport to the laboratory. Within 4 h, the pipe sections were delivered to the laboratory and biofilm samples were recovered from pipe surfaces using a sterile cell scraper before being transferred into 1/4 strength Ringers solution. DNA was then extracted from 250 mg of each biofilm sample using the Ultraclean soil DNA extraction kit (MoBio Laboratories, Solana Beach, CA) following manufacturer’s recommendations. Bacterial 16S ribosomal RNA (rRNA) amplicon libraries were prepared from genomic DNA using a two-step nested real-time PCR (RT-PCR). A nested approach was adopted to reduce barcoded-primer bias (Berry et al., 2011) and minimise contributions of degraded contaminant DNA. RT-PCR was used to analyse the DNA melt curves prior to NGS sequencing. Melt curve analyses enable the level of laboratory contamination present in the samples to be assessed by comparing the melt profiles of the samples to that of the non-template control. All PCR reactions were prepared in a dedicated sterile hood. Prior to adding GoTaq, the PCR mastermix was subjected to ultraviolet radiation for fifteen min to reduce laboratory contamination. DNA was added in a second sterile PCR hood in a separate room. In the first step of the nested RT-PCR, universal primers 27F (50 -AGAGTTTGATCCTGGCTCAG-30 ) and 1492R (50 -TACCTTGTTACGACTT-30 ) were used to amplify a 1500 basepair (bp) segment of the hypervariable V3 region within the 16S rRNA gene (Hoefel et al., 2005). Four replicate 25 lL PCR reactions per sample were prepared and pooled post-PCR to reduce PCR bias (Polz and Cavanaugh, 1998). Each reaction contained 5 lL of DNA extraction template, 5% dimethylsulfoxide (DMSO), 3.3 M SYTO-9 dye, 0.4 U lL1 Go Taq (Promega), 0.2 mM dNTPs, 0.5 lM of each primer, 1x PCR buffer, 2.5 mM of MgCl2 and 4.4 L of nuclease free water. The 1500 bp fragment was amplified using the following parameters: 95 °C for 3 min, 35 cycles at 94 °C for 30 s, 50 °C for 60 s and 72 °C for 120 s. Next, a smaller 170 bp region from within the 1500 bp PCR product was amplified using barcoded Ion torrent fusion primers 341F (50 -CCATCTCATCCCTGCGTGTCTCCGACTCAGnnnnnnnCCTACGGGAGGCAGCAG-30 ) and 518R (50 -CCTCTCTATGGGCAGTCGGTGATATTACCGCGGCTGCTGG-30 ). The primer sequence from Muyzer et al. (1993) is underlined, and ‘n’ represents a sample specific barcode sequence. Replicate reactions were again prepared in quadruplicate, and each 25 L reaction contained: 5 lL of DNA extraction template, 5% DMSO, 3.3 M SYTO-9 dye, 0.4 U lL1 Go Taq, 0.2 mM dNTPs, 0.5 lM of each primer, 1x PCR buffer, 2.5 mM of MgCl2 and 4.4 L of nuclease free water. The second round of RT-PCR amplification was performed with the following parameters: 95 °C for 3 min, 35 cycles at 95 °C for 20 s, 65 °C for 20 s and 72 °C for 30 s. Replicate PCR reactions were again pooled together prior to purification (Ampure, Agencourt Bioscience). The eight purified samples were then quantified and blended together to equimolar concentrations to make the DNA library. The amplicon library was sequenced using the Ion Torrent Personal Genome Machine (Life technologies, Carlsbad, CA, USA). A combined total of 279,671 raw sequence reads were obtained (Table SM-1 in supplementary material). Of these reads, 236,192 had zero mismatches within the barcode and proximal primer sequence and were therefore demultiplexed into their respective samples. Raw sample sequences are available from the NCBI Sequence Read Archive (SRA accession number: SRR1425418; BioProject ID: PRJNA253081). Reads with mismatches present were discarded from further analysis. Quality (Phred threshold of 20 for 90% of sequence bp) and length filtering (90 bp) were applied using FastX tookit (FASTX-toolkit v0.0.13; http://hannonlab.cshl.edu/
188
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
fastx_toolkit), and barcodes and primer sequences were trimmed off using CutAdapt v1.1 (Martin, 2011). Post quality filtering, the mean number of reads per sample was 5756 ± 2572 (Table SM-1). The subsequent fastq file containing the quality-filtered sequences was then converted to a QIIME software compatible .fna file (Available from: http://genomics.azcc.arizona.edu/help. php3). Within QIIME (Caporaso et al., 2010), sequences that were 97% similar were clustered to form operational taxonomic units (OTUs) using Uclust, and aligned using Pynast. Chimeric sequences were detected and removed using Chimera slayer (Haas et al., 2011). A standard (default) threshold (0.8) was applied to assign taxonomy using Greengenes database (version gg_12_8; http://greengenes.lbl.gov). All samples were rarefied at 2,590 sequences (i.e. the lowest number of reads present within a single sample) to ensure an even read coverage for each sample and minimise diversity estimates becoming skewed by read abundance. Further exploratory and statistical analysis of the data was carried out using QIIME software. To assess the diversity of a sample, Chao species richness and Shannon-Weiner diversity estimates were calculated. To test for significant differences across samples, a two-sample t-test within the QIIME software was conducted, whereby 2590 sequences per sample were repeatedly sub-sampled 999 times by Monte Carlo permutations and compared. Sample similarity was assessed within UPGMA trees with 1,940 jackknifed sequences per sample (e = 10 permutations) and viewed using FigTree V1.4 software (tree.bio.ed.ac.uk/software). Phylogenetic distance between taxa was calculated using weighted UniFrac analyses and visualised by hierarchal clustering in UPGMA trees (Lozupone and Knight, 2005). 3. Results
inlet to the outlet (782%), followed by the MIEX/conventional DWDS (98%). The cell count in the MIEX/GAC DWDS stayed relatively stable throughout the system, showing only a 4% decrease in mean bacterial cells from the inlet to the outlet. Oddly, the conventional DWDS cell count decreased by 45%. These results suggest that regardless of the initial treatment applied, the bacterial cell count within each system shifts towards a common, optimal number further downstream of the treatment application. 3.2. Biofilm species diversity and richness To assess the complexity of the microbial biofilm communities, the number of OTUs (Fig. 3a), Shannon–Weaver diversity (Fig. 3b) and Chao species richness (Table SM-3) were calculated for each sample. Post-rarefaction, the number of OTUs per sample ranged from 198 to 487. An increase in the number of OTUs and species diversity was observed from inlet to outlet for all DWDSs. The MF inlet sample had the lowest OTU count and diversity, yet the largest increase from inlet to outlet was observed in the MF system. In contrast, the conventional system inlet had the largest diversity and OTU count, and was most stable from inlet to outlet. Each inlet biofilm was significantly different to the other inlets (P = <0.05; two sample, non-parametric t-test; 999 Monte Carlo permutations; Table SM-4), demonstrating that each treatment initially causes significant variations in biodiversity and species richness. However, with the exception of the MF DWDS, the biofilms at the outlets were not significantly different to one another (Table SM-4), indicating that biofilms begin to shift towards a common, community equilibrium further downstream from treatment application. Interestingly, the diversity and species richness of the MF outlet biofilm were not significantly different from the inlet of the MIEX/GAC DWDS (P = 0.961; two sample, non-parametric
3.1. FCM FCM-based enumeration of the bacterial cells revealed large differences between the four DWDS inlets, with most system inlets displaying significantly different cell counts (Mann–Whitney U test; Table SM-2). The MF DWDS inlet had considerably lower bacterial cell counts compared to the other inlet DWDSs, whereas the conventional DWDS inlet had the highest cell count (Fig. 2). However, cell counts at the four DWDS outlets appear to be more uniform. The MIEX/Conventional and MIEX/GAC DWDS outlet samples were not significantly different from the MF DWDS outlet (Mann– Whitney U test; P-value = 0.380 and 0.176 respectively). The MF DWDS displayed the sharpest increase in bacterial cells from the
(a)
(b)
Fig. 2. Mean active bacterial cell count (cells ml1) at the inlet and outlet of each experimental DWDS measured by flow cytometry (FCM) on a fortnightly basis over a 12 month period prior to biofilm samples being collected (error bars are shown as one standard deviation about the mean).
Fig. 3. (a) Number of unique operational taxonomic units identified for each sample (OTUs; 3% nucleotide cut off); (b) Shannon-Wiener diversity index for each sample. All samples were rarefied to 2590 sequencing depth previously. Line depicts change from inlet to outlet.
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
t-test; 999 Monte Carlo permutations), suggesting that although the MF outlet biofilm is significantly different to the other outlet biofilms at a 1 km distance, it may still be evolving towards the same common established equilibrium but at a slower rate. In summary, the MF treatment appears to hinder biofilm formation to a greater extent than the other treatments, yet without a disinfectant residual, the diversity and species richness increases sharply further down the pipe, eventually producing a biofilm community more similar to that of other DWDS treatments.
3.3. Biofilm community structure Biofilm community structure was compared by calculating phylogenetic distances in a weighted UniFrac UPGMA tree. Once again, the inlet of the MF DWDS contained the most unique biofilm community (Fig. 4). The inlets of three of the DWDS (conventional treatment, MIEX/conventional and MIEX/conventional/GAC) clustered together, illustrating that these treatment types have similar effects on the biofilm species. Interestingly, the outlet biofilms for all of the treatments clustered together in the UPGMA analysis, rather than with their respective inlet, demonstrating that the biofilms shift to a common community structure as distance from treatment application increases, regardless of the initial treatment applied. Further, the inlet and outlet of the MF DWDS were the only samples that were significantly different to one another (UniFrac Monte Carlo significance; P = 0.03), indicating the large extent to which the community changed downstream of the MF system in the absence of a chlorine residual.
189
3.4. Taxa identified Specific bacterial OTUs within each sample were identified (Fig. 5; Table SM-5), and in all biofilms, except the inlet of the MF DWDS, Alphaproteobacteria were the dominant taxa. From the inlets to the outlets, Alphaproteobacteria significantly decreased (Conventional = 4%, MIEX/Conventional = 24%, MIEX/Conventional/GAC = 35%), yet conversely, a 47% increase was observed in the MF DWDS. Upon closer inspection, Rhizobiales and Sphingomonadales were the dominant taxa within the Alphaproteobacteria class. Bradyrhizobiaceae and Hyphomicrobiaceae being the most dominant within the Rhizobiales order. A decrease in Sphingomonadales was observed in all DWDSs from the inlet to outlet (Conventional = 6%, MIEX/Conventional = 6%, MIEX/Conventional/GAC = 28%) with the exception of the MF DWDS, where Sphingomonadales were detected in low abundances at the outlet (2%) but were not detected in the inlet. The MF DWDS inlet was instead dominated by Actinomycetales (72%) and Lactobacillales (8%). These OTUs were further identified as predominantly Propionibacterium and Streptococcus, which includes some pathogenic and human microbiome-associated species. Small percentages (<5%) of other important and potentially pathogenic genera such as Escherichia coli, Legionella, Clostridia and Bacilli were also identified in the biofilms (Table SM-5). The nitrifying bacteria Nitrospira was identified in small proportions in the biofilms (<2%), particularly in the MIEX/conventional inlet and the MF outlet biofilms. A general decrease in Actinobacteria, predominantly Actinomycetales and an unknown Actinobacteria
Fig. 4. Cluster dendogram of biofilm sample similarity based on the bacterial community identified (1940 sequences per jack-knifed subset) and corresponding pie charts showing percentage contribution of dominant taxa (i.e. taxa > 1% contribution to overall diversity). Taxa < 1% contribution and ‘unknown’ taxa were pooled as ‘other’ bacteria (yellow). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
190
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
Fig. 5. Percentage contribution of top 7 dominant taxa to the bacterial community of each sample at family phylogenetic level.
OTU, was also observed from the inlet to the outlet for all DWDS, with the largest decrease found in the MF DWDS (73%), which implies an association between Actinobacteria and highly treated drinking waters. 4. Discussion In this study, the MF DWDS treatment was the most effective at inhibiting biofilm formation, as the inlet of this DWDS had the lowest diversity, bacterial cell count and unique OTUs. However, sharp increases in biodiversity, OTU count and cell counts were observed within the MF system at a 1 km distance from the treatment application, resulting in communities more similar to that of the other DWDS treatment regimes. Additionally, the DWDS outlet biofilms were found to be more similar to one another, rather than their respective inlets, suggesting that a common biofilm community will eventually occur in DWDSs regardless of the initial treatment method applied. In summary, the biofilm communities located at the inlet of the distribution systems appear to be highly impacted by the type of treatment being applied but then shift towards a common, optimal community at a further distance downstream of the treatment application. Bacterial repopulation of the MF DWDS was most likely due to a small number of cells surviving initial decontamination efforts or contamination from holding tanks prior to distribution, as opposed to membrane failure. Whilst every care was taken to minimise contamination risk, it is possible that a small number of cells may have survived the initial pipe disinfection process prior to the beginning of the experiment or the disinfection stage prior to entering the DWDS. However, it must be noted that full-scale DWDSs are not sterile systems and have similar entry points for potential contamination, such as storage tanks. Therefore, the contamination risk of the experimental DWDSs in this study is no greater than that of a full-scale DWDS. It should also be stressed that although some taxa detected in the MF outlet were not detected in the MF inlet, this may be due to the abundance of those taxa being below the detectability threshold at the inlet, rather than contamination further down the pipeline, as rare sequences can often be obscured by more dominant taxa during PCR amplification. Variations in bacterial community structure were observed across the different treatments, with the most pronounced differ-
ences observed between the inlets of the DWDSs. The dominant taxa were identified as Sphingomonadales and Rhizobiales in all biofilms except the MF inlet, where the dominant taxa were Actinomycetales and Lactobacillales. Closer inspection of the Lactobacillales family revealed the majority of these OTUs were Streptococcus, a genus known to contain both pathogenic species and non-pathogenic species that occur naturally within the human gut, and are consequently used as an indicator of human faecal contamination (Pinto et al., 1999). The presence of Streptococcus in the MF DWDS is interesting, as this treatment appeared to be the most effective at removing the major bacterial groups observed in the other treatment streams, yet it harbored larger percentages of genera that include pathogenic species. However, as the number of cells present in the MF DWDS was much lower than that of the other three DWDS, the actual number of potentially pathogenic cells in the MF system were likely below infectivity threshold. Smaller percentages of other potentially pathogenic genera were also identified in the biofilms such as Enterobacteria, Legionella, Clostridia and Bacillus (Ashbolt, 2004). Of interest was the detection of Burkholderiales in both the inlet and outlet biofilms of all four systems, with an apparent increase of this group in the outlet biofilm communities. Burkholderiales includes the genus Burkholderia, which contains pathogenic species of particular significance in warmer climates (Inglis et al., 2000; Pumpuang et al., 2011; Mayo et al., 2011). Bacteria that are able to cause disinfectant decay in DWDSs through nitrification, such as Nitrospira, were also identified (Blackburne et al., 2007). The process of nitrification in chloraminated DWDSs leads to a loss of disinfectant residual and bacteriological contamination of the system, which causes deterioration in water quality and may pose a health hazard to consumers, therefore the presence of nitrifying bacteria in system biofilms must be considered when managing disinfectant dosing levels. It has been previously suggested that biofilms can take several years to become fully mature (Martiny et al., 2005; Liu et al., 2012). Consequently, the species observed in this 28 month study may be more representative of an immature biofilm in the earlier stages of establishment, and the dominant bacterial taxa observed may play an important role in DWDS biofilm colonisation. In three of the DWDSs (Conventional, MIEX/Conventional and MIEX/Conventional/GAC), the abundance of Sphingomonadales was higher in the inlets than the outlets. This suggests that Sphingomonadales
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
thrive in an environment where other bacteria are in low numbers, implicating them as opportunistic, colonising biofilm species. The MF treatment appears to prevent Sphinomonadales cells from penetrating the DWDS as they were not detected in the inlet of the MF DWDS; however, the number of Sphingomonadales cells increased at the MF outlet to levels similar to the inlets of the other DWDSs, indicating that although MF treatment hinders regrowth to a greater extent than organics removal alone, it is not capable of preventing a few surviving cells from colonising a DWDS. Sphingomonadales are a Gram-negative bacterial group defined by the presence of sphingolipids in the outer membrane of the cell wall. It has been proposed that the membrane of Gram-negative bacteria, such as Sphingomonadales, acts a protective barrier against antibiotics, pollutants and organic solvents (Ramos et al., 2002). The presence of a Gram-negative cell wall and a lipid bilayer could mean that Sphingomonadales are well protected from disinfection agents such as chloramine (Sun et al., 2013), making them an important biofilm species. In the outlets of the pipes, the Gramnegative group Burkholderiales were a dominant biofilm taxa. Some species of Burkholderiales are motile due to the presence of flagella (Garrity et al., 2005), potentially aiding attachment through migration towards nutrient rich biofilms. Therefore, Burkholderiales may be a species more likely to attach after the initial colonisation phase has taken place. Several studies have suggested that the reduction of organic carbon is essential for the inhibition of bacterial regrowth, by reducing the level of available nutrients (LeChevallier et al., 1990; Volk and LeChevallier, 1999). In this study, the reduction of organic carbon alone appears to stabilise bacterial biofilm regrowth, as the conventional, MIEX/conventional and MIEX/GAC DWDSs did not significantly change from the inlet to the outlet. This suggests that reducing the organic components of drinking water is an important step in inhibiting biofilm growth, especially in longer systems that may be prone to premature disinfectant decay. Other studies have also inferred the importance of reducing organic carbon via indirect measurements such as ATP, assimilable organic carbon and biodegradable organic carbon (van der Kooij, 1977; Magic-Knezev and van der Kooij, 2004), this study supports those findings using novel and high resolution techniques. In summary, this study illustrates the importance of effective disinfection post-treatment to prevent bacterial regrowth in DWDS, as even highly efficient treatments such as MF do not prevent colonisation of bacterial biofilms. Our results suggest that once the residual disinfectant has decayed, biofilms reach a common established community equilibrium regardless of the effectiveness of the treatment method at the beginning of the system, even in distances as short as 1 km. Moreover, this research demonstrates the power of performing metagenomic bacterial surveys on DWDSs and delivers a comprehensive overview of important biofilm taxa.
Acknowledgements This study was financially supported by the Australian Research Council (ARC) linkage grant LP0991985, Co-operative Research Centre for Water Quality and Treatment (CRC-WQT) project 2417, Water Research Australia (Water RA) project 1008, South Australian Water Corporation, United Water International, Grampians Wimmera Mallee Water, Water Corporation, Delft University of Technology, DCM Process Control and Orica Watercare. Laboratory work was carried out at the South Australian Water Corporation, Victoria Square, Adelaide and the Australian Centre for Ancient DNA at the University of Adelaide. We would like to thank Dr Julien Soubrier and Dr Laura Weyrich for assistance with the bioinformatic analysis and manuscript editing.
191
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2014.06.077.
References
Ashbolt, N.J., 2004. Microbial contamination of drinking water and disease outcomes in developing regions. Toxicology 198, 229–238. Beech, I.B., Sunner, J., 2004. Biocorrosion: towards understanding interactions between biofilms and metals. Curr. Opin. Biotechnol. 15, 181–186. Berry, D., Xi, C., Raskin, L., 2006. Microbial ecology of drinking water distribution systems. Curr. Opin. Biotechnol. 17, 297–302. Berry, D., Ben Mahfoudh, K., Wagner, M., Loy, A., 2011. Barcoded primers used in multiplex amplicon pyrosequencing bias amplification. Appl. Environ. Microbiol. 77, 7846–7849. Beumer, A., King, D., Donohue, M., Mistry, J., Covert, T., Pfaller, S., 2010. Detection of Mycobacterium avium subsp. paratuberculosis in drinking water and biofilms by quantitative PCR. Appl. Environ. Microbiol. 76, 7367–7370. Blackburne, R., Vadivelu, V.M., Yuan, Z., Keller, J., 2007. Kinetic characterisation of an enriched Nitrospira culture with comparison to Nitrobacter. Water Res. 41, 3033–3042. 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 high-throughput community sequencing data. Nat. Methods 7, 335–336. Characklis, W.G., Marshall, K.C., 1990. Biofilms. John Wiley & Sons, New York. Donlan, R.M., 2002. Biofilms: microbial life on surfaces. Emerg. Infect. Dis. 8, 881– 890. Douterelo, I., Sharpe, R.L., Boxall, J.B., 2013. Influence of hydraulic regimes on bacterial community structure and composition in an experimental drinking water distribution system. Water Res. 47, 503–516. Drury, B., Rosi-Marshall, E., Kelly, J.J., 2013. Wastewater treatment effluent reduces the abundance and diversity of benthic bacterial communities in urban and suburban rivers. Appl. Environ. Microbiol. 79, 1897–1905. Emtiazi, F., Schwartz, T., Marten, S.M., Krolla-Sidenstein, P., Obst, U., 2004. Investigation of natural biofilms formed during the production of drinking water from surface water embankment filtration. Water Res. 38, 1197–1206. Garrity, G., Bell, J., Lilburn, T., 2005. Family Burkholderiaceae. In: Garrity, G., Brenner, D.J., Staley, J.T., Krieg, N.R., Boone, D.R., De Vos, P., Goodfellow, M., Rainey, F.A., Schleifer, K. (Eds.), Bergey’s Manual of Systematic Biology, The Proteobacteria, vol. 2. Springer Science, New York, pp. 575–591. Haas, B.J., Gevers, D., Earl, A.M., Feldgarden, M., Ward, D.V., Giannoukos, G., 2011. Chimeric 16S rRNA sequence formation and detection in Sanger and 454pyrosequenced PCR amplicons. Genome Res. 21, 494–504. Hammes, F., Berney, M., Wang, Y., Vital, M., Köster, O., Egli, T., 2008. Flowcytometric total bacterial cell counts as a descriptive microbiological parameter for drinking water treatment processes. Water Res. 42, 269–277. Ho, L., Braun, K., Fabris, R., Hoefel, D., Morran, J., Monis, P., Drikas, M., 2012. Comparison of drinking water treatment process streams for optimal bacteriological water quality. Water Res. 46, 3934–3942. Hoefel, D., 2003. Enumeration of water-borne bacteria using viability assays and flow cytometry: a comparison to culture-based techniques. J. Microbiol. Methods 55, 585–597. Hoefel, D., Monis, P.T., Grooby, W.L., Andrews, S., Saint, C.P., 2005. Profiling bacterial survival through a water treatment process and subsequent distribution system. J. Appl. Microbiol. 99, 175–186. Holinger, E.P., Ross, K.A., Robertson, C.E., Stevens, M.J., Harris, J.K., Pace, N.R., 2014. Molecular analysis of point-of-use municipal drinking water microbiology. Water Res. 49, 225–235. Inglis, T.J., Garrow, S.C., Henderson, M., Clair, A., Sampson, J., O’Reilly, L., Cameron, B., 2000. Burkholderia pseudomallei traced to water treatment plant in Australia. Emerg. Infect. Dis. 6, 56–59. LeChevallier, M.W., Schulz, W., Lee, R.G., 1990. Bacterial nutrients in drinking water. AWWARF, Denver, Colorado. Lewis, K., 2001. Riddle of biofilm resistance. Antimicrob. Agents Chemother. 45, 999–1007. Liu, R., Yu, Z., Guo, H., Liu, M., Zhang, H., Yang, M., 2012. Pyrosequencing analysis of eukaryotic and bacterial communities in faucet biofilms. Sci. Total Environ. 435–436, 124–131. Lozupone, C., Knight, R., 2005. UniFrac: a new phylogenetic method for comparing microbial communities. Appl. Environ. Microbiol. 71, 8228–8235. Magic-Knezev, A., van der Kooij, D., 2004. Optimisation and significance of ATP analysis for measuring active biomass in granular activated carbon filters used in water treatment. Water Res. 38, 3971–3979. Martin, M., 2011. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet. J. 17, 10–12. Martiny, A.C., Albrechtsen, H.J., Arvin, E., Molin, S., 2005. Identification of bacteria in biofilm and bulk water samples from a nonchlorinated model drinking water distribution system: detection of a large nitrite-oxidizing population associated with Nitrospira spp. Appl. Environ. Microbiol. 71, 8611–8617.
192
J.L.A. Shaw et al. / Chemosphere 117 (2014) 185–192
Mayo, M., Kaestli, M., Harrington, G., Cheng, A.C., Ward, L., Karp, D., Jolly, P., Godoy, D., Spratt, B.G., Currie, B.J., 2011. Burkholderia pseudomallei in unchlorinated domestic bore water, tropical northern Australia. Emerg. Infect. Dis. 17, 1283– 1285. 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, 695–700. Pinto, B., Pierotti, G., Reali, D., 1999. Characterisation of ‘faecal streptococci’ as indicators of feacal pollution and distribution in the environment. Lett. Appl. Microbiol. 29, 258–263. Polz, M.F., Cavanaugh, C.M., 1998. Bias in template-to-product ratios in multitemplate PCR. Appl. Environ. Microbiol. 64, 3724–3730. Pumpuang, A., Chantratita, N., Wikraiphat, C., Saiprom, N., Day, N.P.J., Peacock, S.J., Wuthiekanun, V., 2011. Survival of Burkholderia pseudomallei in distilled water for 16 years. T. Roy. Soc. Trop. Med. H. 105, 598–600. Ramos, J.L., Duque, E., Gallegos, M.-T., Godoy, P., Ramos-Gonzalez, M.I., Rojas, A., Teran, W., Segura, A., 2002. Mechanisms of solvent tolerance in Gram-negative bacteria. Ann. Rev. Microbiol. 56, 743–768.
Schwartz, T., Hoffmann, S., Obst, U., 2003. Formation of natural biofilms during chlorine dioxide and u.v. disinfection in a public drinking water distribution system. J. Appl. Microbiol. 95, 591–601. Silvestry-Rodriguez, N., Bright, K.R., Slack, D.C., Uhlmann, D.R., Gerba, C.P., 2008. Silver as a residual disinfectant to prevent biofilm formation in water distribution systems. Appl. Environ. Microbiol. 74, 1639–1641. Storey, M., Kaucner, C.E., 2009. Understanding the growth of opportunistic pathogens within distribution systems. Co-operative research centre for water quality and treatment. Research report 79. Water Research Australia Ltd. ISBN: 1876616296. Sun, W., Liu, W., Cui, L., Zhang, M., Wang, B., 2013. Characterization and identification of a chlorine-resistant bacterium, Sphingomonas TS001, from a model drinking water distribution system. Sci. Total Environ. 458–460, 169– 175. van der Kooij, D., 1977. The occurence of Pseudomonas spp. in surface water and in tap water as determined on citrate media. J. Microbiol. 43, 187–197. Volk, C.J., LeChevallier, M.W., 1999. Impacts of the reduction of nutrient levels on bacterial water quality in distribution systems. Appl. Environ. Microbiol. 65, 4957–4966.