To assess human adenoviruses (HAdVs) removal in an advanced wastewater treatment facility and compare two parallel tertiary treatment methods for the removal of HAdVs.
To assess human adenoviruses (HAdVs) removal in an advanced wastewater treatment facility and compare two parallel tertiary treatment methods for the removal of HAdVs.
Tangential flow ultrafiltration was used to concentrate the water samples, and HAdVs were precipitated by polyethylene glycol. HAdVs were detected only by TaqMan real-time PCR, and HAdV genotype was determined by DNA sequence. HAdVs were detected in 100% of primary clarification influent, secondary clarification effluent and granular media (GM) filtration effluent samples but only in 31·2% of membrane filtration (MF) effluent and 41·7% of final effluent (FE) samples, respectively. The average HAdVs loads were significantly reduced along the treatments but HAdVs were still present in FE. Comparison of two parallel treatments (GM vs MF) showed that MF was technically superior to GM for the removal of HAdVs.
These findings indicate that adenoviruses are not completely removed by treatment processes. MF is a better treatment for removal of adenoviruses than GM filtration. Because only qPCR was used, the results only indicate the removal of adenovirus DNA and not the infectivity of viruses.
Presence of HAdVs in FE by qPCR suggests a potential public health risk from exposure to the treated wastewater and using the FE for recreational or water reuse purposes should be cautious.
Adenoviruses belong to the family Adenoviridae and are a group of double-stranded DNA, nonenveloped viruses, with 52 immunologically distinct types grouped into six species (A–F) (Lu and Erdman 2006). While infections with these viruses commonly result in respiratory illness, serotypes 40 and 41 in the subgenus F are recognized as important causative viral agents of acute gastroenteritis in children (Nguyen et al. 2007; Zhang et al. 2011) and are also associated with outbreaks of acute gastroenteritis (Chiba et al. 1983; Shimizu et al. 2007; Mattner et al. 2008). These two serotypes are transmitted through the faecal-oral route. People infected with enteric adenoviruses can shed 1010 copies of adenoviruses per gram of stool (Lion et al. 2010), leading to high viral loads in sewage (Kuo et al. 2010). High concentrations of human adenoviruses (HAdVs) in wastewater have the potential to impact drinking and recreational water sources. This is particularly problematic because of variable viral removal efficiencies of water treatment facilities and water reclamation centres (Zhang and Farahbakhsh 2007; Carducci et al. 2008). Additionally, HAdVs have been proposed as a suitable indicator of faecal contamination to study viral prevalence in wastewater and to assess the efficacy of viral removal in wastewater treatment processes (Myrmel et al. 2006; Carducci et al. 2008). Therefore, monitoring adenoviruses in wastewater treatment plants would demonstrate the infectious potential of adenoviruses to humans and appears to provide insightful information of faecal contamination status in treated wastewater.
Water shortage is an increasing problem in the United States as well as internationally. In some areas, treated wastewater is routinely added to surface water or groundwater that serve as sources for drinking water treatment plants. However, the safety of reclaimed water and the efficiencies of different wastewater treatment facilities are unknown. The wastewater treatment facility in this study is an advanced water reclamation plant and the largest facility in Gwinnett County that serves 800 000 inhabitants in Georgia (USA) with a treatment capacity of 24 million gallons (90·8 million litres) of wastewater per day at the time of this study. This facility uses conventional primary [mechanical screening, grit removal and primary clarification (PC)] and secondary treatment [activated sludge and secondary clarification (SC)], as well as nutrient removal (biological phosphorus removal with chemical polishing). Following these treatments, the effluent wastewater was split, with a portion of the flow (15%) treated by granular media (GM) filtration and the rest of the flow (85%) treated by membrane filtration (MF) at the time of this study. The two parallel streams were merged following these processes and then treated by pre-ozonation, carbon adsorption and ozone disinfection. The final effluent (FE) from the water reclamation centre was sent through a pipeline to a man-made lake that serves as a drinking water source and recreational purpose for a large metropolitan area (Fig. 1). Currently, this facility has no system in place for routine monitoring of waterborne viruses.
The purpose of this study was to examine the human adenovirus removal efficiencies from wastewater by the advanced treatment processes. Wastewater samples were collected from different treatment points in the plant and were evaluated for HAdVs. HAdV genotypes were determined, and HAdV log removals were estimated for the different treatment processes. This study also compared the efficacy of two parallel treatment methods, MF and GM filtration, for the removal of HAdVs.
Water samples were collected biweekly for an entire year from the water reclamation centre between December 2007 and November 2008. Water samples were collected at five locations throughout the treatment process: (i) PC influent, (ii) SC effluent, (iii) GM filtration effluent [sand (effective size: 0·4–0·5 mm; uniformity coefficient: <2); anthracite (effective size: 1–1·2 mm; uniformity coefficient: <1·3); and carbon (effective size: 0·2–0·3 mm; uniformity coefficient: <1·5)], (iv) MF effluent [Zenon ZeeWeed® 500c membrane, 0·04 micron size, (GE Water, Feasterville-Trevose, PA, USA)], and (v) FE. Figure 1 outlines the treatment processes at the water reclamation facility and shows the sampling locations. Sampling collection from the GM and MF locations alternated, with each sample location tested once every four weeks during opposite sampling periods. Sample volumes of 100-l were collected in five 20-l polyethylene cubitainers (Cole-Parmer Instrument Co, Vernon Hills, IL, USA) for GM, MF and FE samples. Water samples were characterized (data not shown) using the following water quality parameters: pH, temperature and turbidity. Sample pH was measured with a Fisher Scientific Accumet® Research AR25 pH/mV/°C/ISE Meter (Fisher Scientific, Pittsburgh, PA, USA). Turbidity was measured using a Hach Model 2100N Laboratory Turbid Meter (Hach, Loveland, CO, USA).
Tangential flow ultrafiltration (UF) was used to concentrate the 20-l (SC) and 100-l (GM, MF, FE) water samples within 4–8 h of sampling. Prior to the UF, 10 mg l−1 of sodium polyphosphate (NaPP) (Sigma-Aldrich, St. Louis, MO, USA) was added to the samples to act as a dispersant. FE samples were dechlorinated by adding sodium thiosulfate (Fisher Scientific, Fair Lawn, NJ, USA) to achieve a final concentration of 32 mg l−1 of water. The UF apparatus was set up as described by Liu et al. (2012). Briefly, a hollow-fiber Fresenius Optiflux F200NR polysulfone ultrafilter (Fresenius Medical Care North America, Waltham, MA, USA) was connected to a peristaltic pump (Cole Parmer Instrument Co, Vernon Hills, IL, USA), pressure gauge and flow meter using silicon tubing (Masterflex; Cole-Parmer Instrument Co). The 100-l and 20-l water samples were concentrated to a retentate volume of 50–100 ml. Immediately following concentration, an elution step was performed to remove HAdVs that may have adsorbed onto the inner surfaces of the tubing and ultrafilter. A 500 ml elution solution containing 0·01% Tween 80, 0·01% NaPP and 0·001% antifoam A Y-30 emulsion (Sigma-Aldrich) in 1× PBS was circulated throughout the system. An additional backflushing procedure was used to desorb adenoviruses from the inner filter surfaces. The backflush solution consisted of 0·05% Tween 80, 0·01% NaPP and 0·001% antifoam Y-30 emulsion. This was pumped at 650 ml min−1 through the permeate port of the ultrafilter with the input port closed and output port open. Approximately 200 ml of the backflush solution was collected through the output port. The backflush samples were merged with the previous retentate/elution samples (Fig. 2).
HAdVs were precipitated from the merged retentate/elution and backflush samples by adding 12% polyethylene glycol 8000 (PEG 8000) (Sigma, St. Louse, MO, USA), 0·9 mol l−1 sodium chloride, 1% bovine serum albumin (Sigma) and incubated for 2 h at 4°C, pH 7–7·4. After centrifugation at 10 000 g for 30 min, the pelleted material was resuspended in 3 ml with 0·01 mol l−1 phosphate-buffered saline (PBS, Dulbecco's modification) containing 0·01% (v/v) Tween 80 and 0·001% antifoam Y-30 emulsion (v/v) (Fig. 2).
A 500 μl aliquot of the PEG concentrated sample was mixed with an equal volume of Vertrel (Miller-Stephenson, Danbury, CT, USA) and vortexed well. Samples were incubated at 4°C for 2 h, followed by centrifugation at 12 000 g for 15 min at 4°C. The supernatant was removed and used for adenovirus DNA extraction per QIAamp DNA Mini Kit (Qiagen, Valencia, CA, USA) in accordance with the manufacturer's instructions. To increase adenovirus DNA yield, each supernatant sample was divided into three parts. Each part was individually lysed, and the three lysates were later combined in a single QIAamp Mini column prior to the wash procedure. DNA was eluted from the column with 50 μl of the supplied elution buffer. The extracted DNA samples were aliquoted and stored at −80°C until analysed by adenovirus PCR assays.
A known concentration (0·34 μg ul−1) of type-2 (species C) adenovirus DNA (Invitrogen, Carlsbad, CA, USA) was obtained, and the concentration of the adenovirus DNA was converted to genome equivalents (GEQGEQ) via multiplication by the Avogadro's constant. The molecular concentration of the adenovirus DNA was serially diluted (20–20 000 GEQ) and incorporated into the adenovirus real-time PCR assay to produce a standard curve of GEQ vs CT value. The standard curve was used to quantify adenoviruses in the concentrated wastewater samples.
Adenoviruses were detected by a TaqMan real-time PCR assay using the OneStep PCR Kit (Qiagen) and adenovirus broadly reactive primers and probe (Jothikumar et al. 2005). A 25 μl PCR reaction mixture was prepared with 2 μl of adenovirus DNA template, 14·15 μl of water, 1·0 μl of dNTP mixture (10 mmol l−1), 0·6 μl of each primer (l0 μmol l−1), 0·4 μl of the TaqMan probe (l0 μmol l−1), 0·25 μl (40 U μl−1) of DNase inhibitor (Promega, Madison, WI, USA), 5·0 μl of 5× Qiagen OneStep buffer, 1 μl of Qiagen PCR enzyme. The amplification reactions were performed on a Stratagene MX 3000P system (Agilent Technologies, Inc., Santa Clara, CA, USA). PCR was performed at 95°C for 15 min to activate the HotStar Taq DNA polymerase. Subsequently, a total of 45 amplification cycles were carried out, each consisting of 95°C for 10 s, 55°C for 30 s and 72°C for 15 s. Positive and negative controls were run alongside all samples.
Conventional PCR was performed using HotStar Taq Plus Master Mix Kit (Qiagen) and primers that corresponded to the conserved region within the hexon gene of human adenovirus (Allard et al. 2001). For this reaction, 3 μl of DNA was added to 47 μl of PCR mixture containing 10 μl of 5× PCR buffer, 2 μl of dNTP mixture (10 mmol l−1), 2·0 μl of HotStar Taq Plus DNA polymerase, 2·5 μl of each primer (10 μmol l−1), 0·5 μl of DNase inhibitor and 27·5 μl of water. An initial denaturation and activation step at 95°C for 5 min was followed by 35 cycles at 94°C for 30 s, annealing at 55°C for 30 s and extension at 72°C for 1 min. The final cycle had an extension time of 5 min at 72°C. After amplification, 10 μl of the PCR products (301 bp) were visualized on 1·5% agarose gels.
PCR amplicons from the conventional PCR assay were sent to the GeneWiz Inc (South Plainfield, NJ, USA) to determine adenovirus sequences of amplicons using the conventional PCR primer set. All adenovirus nucleotide sequences were cleaned with the EditSeq program in the DNASTAR software package (Madison, WI, USA). The cleaned adenovirus sequences were aligned against reference adenovirus sequences representing species A–F from the GenBank database using the Clustal W program implemented in the MEGA 4 program (http://www.megasoftware.net). The phylogenetic tree was constructed using Neighbour-Joining method in the MEGA 4 package with 1,000 pseudoreplicate data sets. The GenBank accession numbers for previously represent adenovirus species include: species A (AB330112, type 31; AB330093, type 12; DQ149610, type 18); species B & E (AB330116, type 35; X76551, type 7; EF494650 type 3; AB330065, type 4); species D (AB361058, type 8; AB330132, type 51; AB330133, type 19); species C (AB330082, type 1; EU867481, type 2); species F (X51782, type 40; AB330121, type 40; AB330122, type 41).
The adenovirus limit of detection was determined by spiking a total of 106 plaque-forming units (pfu) that is equivalent to 7·3 × 108 GEQ of adenovirus DL307 (species C, serotype 5) (courtesy of Dr. Gooding, Emory University) into a 100-l dechlorinated tap water sample. This was then concentrated using the same UF method and PEG concentration technique as the wastewater samples. Virus DNA was extracted and detected by the previously described real-time PCR assay. If a positive result was obtained, the quantity of viruses seeded was reduced 10-fold, and the experiment was repeated. Processing of the seeded water sample was repeated in this manner until the real-time PCR assay was not able to detect the presence of adenovirus DNA. The lowest concentration of adenovirus seeded into the 100-l sample that produced a positive result was deemed to be the limit of detection for this procedure. The seeding experiment with the lowest detectable adenovirus concentration was performed twice using the same experimental conditions to confirm the end point of the detection limit.
Four pfu log of adenovirus (DL307) was spiked into a 100-l tap water sample to test the recovery of the entire procedure including UF, virus precipitation with PEG and real-time PCR. The same experiment and spiking level were performed at least three times.
The adenovirus concentration for each collected wastewater sample was estimated using the following equation described by Kuo et al. (2010). HAdV GEQ/per litre of wastewater = copies/PCR reaction × (1/2 μl) × 50 μl ×(1/500 μl) × 3000 μl × (1/100 l), where 2 represents the adenovirus DNA volume (μl) used for real-time PCR; 50 represents the total volume (μl) of adenovirus DNA extracted from 500 μl of the PEG concentrated sample; and 3000 μl represents the total concentrated sample volume from a 100-l sample, or a 20-l sample (for SC location), or a 1 l sample for PC. The mean HAdV concentration of the individual sampling site was calculated from all samples over a one-year period. The HAdV log reduction for each sampling site was calculated by subtracting the mean log-transformed HAdV concentration at each site from the mean log-transformed HAdV concentration in samples of primary clarifier influent.
The detection limit of the entire testing procedure including UF, PEG concentration and real-time PCR was determined by spiking a known concentration of the adenovirus DL307 (Species C, serotype 5) into a 100-l of dechlorinated tap water. Adenovirus spiking load started from 106 pfu (7·3 × 108 GEQ) and was gradually reduced to 103 pfu (7·3 × 105 GEQ). The spiking titre of 104 (7·3 × 106 GEQ) consistently gave positive PCR signal, and no real-time PCR signal was detected at the level of 103 pfu (7·3 × 105 GEQ). Therefore, 104 pfu (7·3 × 106 GEQ) of adenovirus is the limit of the detection in 100-l of water in this study. Recovery was determined at least in triple experiments and was calculated using the following equation: (total recovered adenoviruses in pfu/total spiked adenoviruses in pfu) × 100% and the average recovery was 2·6%.
A total of 105 wastewater samples (23 PC, 24 SC, 18 GM, 16 MF, and 24 FE) were collected from the water reclamation centre between December 2007 and November 2008. HAdVs were detected in 100% of PC (23/23), SC (24/24) and GM (18/18) samples but only in 31·2% (5/16) of MF and 41·7% (10/24) of FE samples (Table 1). As expected, the highest load of HAdVs was detected in PC samples, with an average of 6·1-log GEQ per litre of influent after the mechanical screens and grit removal. HAdV loads were significantly reduced along the treatment processes, with mean log loads ranging from 3·1 GEQ in SC to 2·9 in GM, 0·9 in MF and 1·0 GEQ in FE, respectively (Table 1). Figure 3 shows the ranges of adenovirus loads and the median log loads of positive adenovirus samples at each sampling site.
|Location||Volume (l)||No||No positive (%)||Mean log||Log removal|
The mean log removals of HAdV from primary clarifier influent to SC, GM, MF and FE were 3·0 GEQ, 3·2, 5·2 and 5·1, respectively (Table 1). Comparison of the two parallel treatment methods (GM filtration and MF) indicated that the MF performed better (P < 0·001) than the GM filtration process in terms of HAdV detection rate and adenovirus load (Table 1), suggesting that MF is a more effective method for HAdV removal in wastewater.
No obvious seasonal trend in adenovirus occurrence was observed at the five sampling sites (Table 2). Adenoviruses were consistently detected in every sample of PC, SC and GM each month, and the mean log loads were relatively stable from month-to-month at each sampling site. Although the adenovirus detection rates and mean log loads were higher in MF and FE samples in the last 5 months of our study (July to November 2008) compared with the first 7 months (December 2007–June 2008), the stable viral concentrations observed in the PC samples suggest that there was no seasonal adenovirus trend in this water reclamation plant over the course of the study year.
A total of 30 samples (7 PC, 16 SC, 6 GM and 1 FE) positive for HAdVs by real-time PCR were amplified using a conventional PCR, and the PCR amplicons were sequenced using a primer set spanned the hexon gene of the adenovirus genome. Sequences obtained from this study were aligned with HAdV reference sequences downloaded from the GenBank database, and the phylogenetic analysis indicates that all the HAdV sequences in this study belonged to HAdV type 41 (data not shown).
Water reuse is a sustainable strategy to manage finite water resources and provides for the water needs of growing populations. However, a major concern associated with water reuse is the potential for incomplete removal or inactivation of pathogens and the potential public health risks from exposure to these waters. The water reclamation centre in this study has no system for monitoring pathogenic viruses in the FE as this is not required by state guidelines.
In this study, human adenovirus was used as a representative enteric virus to evaluate the efficiency of viral removal in this plant and to compare the efficacy of two parallel treatment methods. Adenoviruses were selected for evaluation because they are commonly detected in a variety of water sources (Choi and Jiang 2005; van Heerden et al. 2005; Sedmak et al. 2005; Xagoraraki et al. 2007) and are currently included in the contaminant candidate list of the U.S. Environmental Protection Agency (Mena and Gerba 2009). We used a broadly reactive TaqMan PCR assay previously shown to be suitable for quantitative detection of adenovirus DNA from all six species (A–F) (Jothikumar et al. 2005). The presence of HAdVs varied between sampling locations, with 100% detection in PC, SC, GM, and 31·2% and 41·7% detection in MF and FE, respectively. Moreover, the mean loads of HAdVs ranged from 6·1 log of GEQ l−1 in PC to 3·1 log GEQ l−1 in SC, 2·9 log GEQ l−1 in GM, 0·9 log GEQ l−1 in MF and 1·0 log GEQ l−1 in FE. The HAdV loads in this study for PC and FE samples are consistent with those reported by a recent study in New Zealand (Hewitt et al. 2011) that measured 3·25–8·62 log GEQ l−1 in influent and 2·97–6·96 GEQ l−1 in effluent samples from multiple wastewater treatment plants. In comparison with a recent study (Fong et al. 2010) in Michigan, all the corresponding sampling points in the present study showed slightly higher HAdV loads. Consistent with previous reports (Katayama et al. 2008; Fong et al. 2010; Kuo et al. 2010), we also found that HAdVs detection rates and mean loads showed no clear seasonal trends.
Ten of 24 FE samples were positive for HAdVs, and the mean load was 1·0 log GEQ l−1. Because molecular assays were used to test viral nucleic acid, the infectivity and potential health risk of the HAdVs we detected is not known. As the FE is discharged into a lake that serves as the source for the drinking water plant in the County, our results should be interpreted with caution and emphasize the need for appropriate water treatment to protect human health.
Comparison of the mean HAdVs log removal levels indicated that the MF treatment process produced the greatest reduction (5·2 log) in adenovirus load, and this was significantly higher than the mean HAdVs removal (3·2 log) by the GM filtration method. Additionally, 100% of the GM samples were positive for adenovirus compared with only 31·2% of the MF samples. These results suggest that MF is more effective than GM filtration for adenovirus removal. During the study period, 85% of the plant's influent flow was treated by MF, while 15% of the influent was treated by GM filtration. If the entire flow was directed to MF, the viral removal from the combined FE would be improved, and the adenovirus prevalence would be further reduced.
Previous laboratory studies indicated that adenoviruses were the most UV-resistant waterborne pathogens (Gerba et al. 2002; Hijnen et al. 2006). Because qPCR was used for testing HAdV inactivation by UV in this study, the results provide little information about HAdV inactivation. Furthermore, adenovirus 41 was the only serotype detected in this study, while several adenovirus serotypes were reported in studies of wastewater in other locations (Kuo et al. 2010; Kokkinos et al. 2011). For example, in Patras, Greece, adenovirus types 8, 41 and 40 were detected in treated municipal wastewater (Kokkinos et al. 2011). In Michigan, United States, Kuo et al. (2010) reported that adenovirus species A (types 12 and 31), C (types 2 and 1) and F (serotype 41 only) were found in untreated and treated wastewater. Because details about the treatment processes were not described in these other studies, we are not able to determine whether advanced treatment methods at the water reclamation centre in this study contribute to the inactivation of some susceptible adenovirus serotypes but is not sufficient to destroy the DNA of the more resistant serotype 41 as the previous laboratory studies demonstrated (Ko et al. 2005; Nwachuku et al. 2005). Although adenovirus serotyping patterns varied between studies, species F adenovirus was consistently dominant in wastewater, and serotype 41 was more frequently detected than type 40 within species F. This is consistent with reports that serotype 41 was the most prevalent adenovirus serotype detected in faecal specimens from children with acute gastroenteritis in the world (Sdiri-Loulizi et al. 2009; Dey et al. 2011).
In this study, real-time PCR was used to quantify adenoviruses in treated wastewater samples, and this molecular-based assay is not able to distinguish between infectious and noninfectious virus particles. Despite this disadvantage, the molecular assay is still superior to the problematic and time-consuming culturing method for adenovirus type 40 and 41. Additionally, the UF technique we used to concentrate adenoviruses from large volumes of wastewater likely also concentrated PCR inhibitors present in the samples. These may result in reduced or inhibited PCR amplification and, as a consequence, may lead to an underestimation of the viral load in the samples. While monitoring environmental inhibitors during virus concentration procedure from water has been documented, the removal of these inhibitors remains a challenge (Xagoraraki et al. 2007). Given the possible presence of PCR inhibitors in these samples, the adenovirus occurrence and concentration may be underestimated.
In conclusion, the water reclamation centre in this study provides a practical opportunity to study the feasibility of water reuse. The results from this study demonstrate that HAdVs are consistently present at every stage of wastewater treatment. Although each treatment process effectively reduced the adenovirus load, adenoviruses were still detected in the FE. Comparison of two parallel tertiary treatment processes showed that MF is more effective in removing adenoviruses than GM filtration. This research also demonstrated that the UF can be effectively used in conjunction with secondary concentration techniques and TaqMan real-time PCR to enable sensitive detection of human adenovirus in large-volume reclaimed water samples.
Funding for this project was provided by the Gwinnett County Department of Water Resources. The authors are grateful for the assistance provided by the staff of the F. Wayne Hill Water Resources Center in Gwinnett County, Georgia, USA. The authors have no conflict of interest to declare.