Distribution of human fecal marker GB-124 bacteriophages in urban sewage and reclaimed water of São Paulo city, Brazil

The water demand associated with new patterns of economic development, consumption and increasing pollution levels are compromising the availability of this resource for various purposes, including drinking water supply. To prevent future environmental impacts and health public risks, new approaches to evaluate effluent sanitary quality are emerging (Diston et al. 2012; Rusiñol et al. 2014).

Traditional microbiological indicators (fecal indicators bacteria (FIB), e.g. thermotolerant and fecal coliforms, Escherichia coli and fecal enterococci) are not appropriate to evaluate the water contamination caused by other pathogens, such as protozoan and enteric viruses (Harwood et al. 2014). Moreover, they are not suitable to distinguish between human and animal sources of contamination (Harwood et al. 2014; Rusiñol et al. 2014). Determining sources of contamination is an important strategy in water management and this approach is often called microbial source tracking (MST) (Ahmed et al. 2010; Nnane et al. 2011; Araújo et al. 2014; Rusiñol et al. 2014).

Viruses are one of the major concerns in wastewater reclamation due to their high resistance to conventional wastewater treatment processes; their high rate of survival under adverse environmental conditions; and their serious implication in major outbreaks of waterborne diseases, especially those caused by enteric viruses, such as the norovirus, rotavirus, hepatitis A and E viruses, human enteroviruses, and adenoviruses (Bofill-Mas et al. 2013).

However, monitoring of multiple pathogens is not feasible and, therefore, researchers have focused on the study of viral indicators that could be used together with bacterial indicators for assessing human fecal contamination in the environment (Ahmed et al. 2010; Bofill-Mas et al. 2013; Rusiñol et al. 2014). Human adenovirus, polyomavirus and bacteriophages (F-specific, somatic coliphages and Bacteroides spp. phages) are considered potential candidates to assess human viral contamination in the environment (Payan et al. 2005; Ebdon et al. 2012; Bofill-Mas et al. 2013; Rusiñol et al. 2014; Purnell et al. 2015), and molecular methods (notably polymerase chain reaction and variants) are commonly used to track these microorganisms (Ahmed et al. 2010; Bofill-Mas et al. 2013; Rusiñol et al. 2014).

Bacteriophages infecting a specific host strain (Bacteroides fragilis GB-124) have been described as potential viral markers to evaluate human fecal pollution in different environmental matrices worldwide (Payan et al. 2005; Nnane et al. 2011; Diston et al. 2012; Ebdon et al. 2012; Jofre et al. 2014; McMinn et al. 2014). Most bacteriophages infecting different Bacteroides species have been reported to belong to the Siphoviridae family and are recognised to contain a dsDNA genome, a capsid composed of an icosahedral head, and a flexible tail (Jofre et al. 2014).

Advantages to the use of Bacteroides GB-124 phages to tracking human faecal pollution include their morphology, environmental survival, and persistence to treatment processes, characteristics similar to viruses of concern, such as enteric viruses (Diston et al. 2012; Ebdon et al. 2012; Jofre et al. 2014). Relatively high concentrations of GB-124 phages have been reported in municipal wastewaters worldwide and temporal and geographic stability were already observed (Ebdon et al. 2012; Jofre et al. 2014; McMinn et al. 2014). Additionally, assays using specific host strains have been developed, allowing the identification of human or animal sources of contamination (Payan et al. 2005; Ebdon et al. 2012).

Methods based on bacteriophage detection using culture assays provide a more reliable analysis of the infectivity of the virus strains detected. A simple and low-cost method to detect these viruses can be useful for laboratories, environmental sanitation companies or environmental regulatory agencies that need routine monitoring.

However, the occurrence of Bacteroides fragilis GB-124 seems to vary among distinct geographic regions because it depends on epidemiological and environmental factors (Payan et al. 2005). Therefore, it is necessary to identify distinct occurrence patterns of host strains under the local conditions (Payan et al. 2005; Ebdon et al. 2012; McMinn et al. 2014).

Knowledge about the occurrence of bacteriophages that infect Bacteroides host strain GB-124 in wastewaters, treated effluents and reclaimed water is limited in Brazil. Therefore, the aim of this study was to perform an environmental surveillance of these viruses in raw sewage and treated effluents (including reclaimed water) from four wastewater treatment plants (WWTPs) of São Paulo city and to evaluate bacteriophage removal at the different treatment processes. Physicochemical variables were also measured, and statistical analyses were performed to verify the influence of these parameters on virus detection.

The four WWTPs selected in this study are located in different regions of São Paulo metropolitan area, Brazil, and receive the sewage of approximately eight million people. São Paulo city is the capital of São Paulo state and is located in the southeast region of the country, being one of the most densely populated regions with approximately 21 million inhabitants.

These WWTPs receive mainly domestic sewage and comprise primary, secondary (activated sludge) and tertiary (filtration or ultrafiltration/reverse osmosis and chlorination) processes (Table 1). The WWTP-3 receives contributions from three collection networks, including two domestic sewage systems and one of a large chemical industry of nitrocellulose. The sewage collection network is separate from the urban drainage network.

Raw sewage, secondary effluent and reclaimed water were collected into sterile polypropylene bottles monthly during one year (April 2015–March 2016), totalling 144 samples (36 samples from each WWTP). Reclaimed water samples were dechlorinated with sodium thiosulphate (10%) after collection. The samples were maintained under refrigeration (4 °C) and transported to the laboratory for processing. The sample processing occurred up to 15 h after each collection.

For phage analysis, a raw sewage sample of WWTP-3 was composed in the laboratory considering the flow rate of each of the three individual influents discharging into this plant, because there is not a collect point resulting from the mixture of these influents before primary treatment in the respective WWTP.

Bacteroides fragilis GB-124 phages were enumerated using a double-layer agar technique, according to ISO 10705-4 (ISO 2001) and as described elsewhere (Ebdon et al. 2012). In brief, 12 mL screw-topped tubes containing Bacteroides phage recovery medium (BPRM) broth were inoculated with 1 mL of host bacterium (GB-124) and incubated at 37 ± 0.2 °C for 18 h. One millilitre of bacteria was inoculated into BPRM broth and incubated at 37 ± 0.2 °C for approximately 3 h to achieve an optimal optical density for phage detection (approximately 0.33 at 620 nm).

The bacterial suspension was then used immediately. Samples were passed through a 0.22 μm syringe driven polyvinylidene difluoride filter (Millipore, USA) to remove background bacterial contamination. Filtered samples of raw sewage were also diluted (10-fold dilution) to avoid excessive lysis plaques on the agar. One millilitre of the samples was then combined with 1 mL of host (GB-124) in test tubes containing 2.5 mL of semisolid BPRM agar, poured into 90 mm Petri dishes with a monolayer of BPRM agar and allowed to solidify. The Petri dishes were inverted and incubated at 37 ± 0.2 °C for 18–24 h under anaerobic conditions using anaerobic jars and commercial anaerobic gas packs (BBL Gas Pack Plus, Maryland, USA).

The presence of phages resulted in the production of visible plaques (zones of lysis) in a confluent lawn of the host growth. A suspension of reference phage (B-14) known to infect the host strain was also included in all analyses as a positive control. Negative controls (B. fragilis GB-124 culture) were also included in all experiments. All samples were analysed in duplicate, and the results were expressed as the mean number of plaque-forming units (PFU) per litre.

Water temperature, pH, conductivity and residual chlorine were measured in the field according to Standard Methods (methods 2550 B, 4500 H+ and 2510B, respectively) employing a pH metre (Mettler Toledo), conductivity metre (WTW) and colorimeter (Hach). The levels of total organic carbon, and turbidity were analysed at CETESB Chemistry Laboratory using methods 5310C, and 2130, respectively, recommended by Standard Methods (APHA/AWWA/WEF 2012).

The datasets of secondary and tertiary effluents resulted in the occurrence of several left-censored observation samples with concentrations below the theoretical detection limit (TDL ≤1 PFU/mL). In this case, for statistical analysis, the value considered was the TDL divided by the square root of 2 (Croghan & Egeghy 2003). The high concentration of points below the detection limit creates a non-standard distribution; however, it can be adjusted to a known distribution. The log (GB-124) was used as the response variable. It was noted that this variable follows the Zeros Adjusted Inverse Gaussian (ZAIG) distribution (Heller et al. 2006).

For achievement of the proposed objectives and considering the unique features of the data involved, the statistical modelling methodology Generalized Additive Models for Location, Scale and Shape (GAMLSS) was used (Stasinopoulos & Rigby 2007). A saturated model was started (with all the variables and factors present) and progressed by removing, one by one, those variables whose contribution was not significant in explaining variations in densities of GB-124 phages (Figure 1).

The final model was validated according to the hypotheses assumed: normal distribution and independence among residuals. The R packages used in this methodology were the GAMLSS family (Rigby & Stasinopoulos 2005). Statistical significance was obtained when the adjusted model produced p ≤ 0.001 and all model coefficients showed p < 0.10.

The geometric means of the physicochemical parameters obtained during the sampling period are shown in Table 2. Water temperature did not vary greatly throughout the year in the WWTPs analysed, with values ranging from 24.1 to 25.5 °C for raw sewage samples, except for WWTP-3, which had a mean temperature of 35.1 °C for the chemical industry wastewater (Table 2). Higher values of water temperature were also observed in treated effluents and reclaimed water from WWTP-3 compared to the water temperature from the other WWTPs (Table 2).

Mean values of pH in all stages of the WWTPs were near neutrality (pH = 7.0), with low standard deviation (Table 2). The water temperature and pH results in the treated effluents of all WWTPs were in compliance with standards established by the national legislation for effluent discharges in receiving waters (water temperature below 40 °C and pH of 6.0–9.0) (CONAMA, Resolution n°. 430 2011).

The conductivity results varied widely, but were in accordance with values commonly found in wastewaters and effluents from WWTPs that receive domestic and industrial sewage.

Higher levels of total organic carbon (TOC) were observed in raw sewage from the industrial wastewaters from WWTP-3, while in the treated effluents, higher mean concentrations were found in WWTP-2 (Table 2).

In general, lower levels of turbidity and TOC were observed in reclaimed water samples from WWTP-4, which operates with membrane bioreactor (MBR) and reverse osmosis (Table 2), indicating a better performance of this system for pollutant removal.

Residual mean chlorine values were also within the range of standards commonly found for the disinfection of effluents in tertiary treatment, and lower concentrations were found in reclaimed water samples from WWTP-4 which uses chlorine dioxide as a disinfecting agent (Table 2).

GB-124 phages were recovered in 100% of the raw sewage samples from all the WWTPs (Table 3). Viral loads varied from 3 × 10⁴ (min) to 2.7 × 10⁶ (max) PFU/L in WWTP-1, from 2.3 × 10⁴ (min) to 1.32 × 10⁶ (max) PFU/L in WWTP-2, from 7.5 × 10³ (min) to 1.01 × 10⁶ (max) PFU/L in WWTP-3 and from 4.7 × 10⁴ (min) to 1.3 × 10⁶ (max) PFU/L in WWTP-4, with geometric mean viral loads (PFU/L) from 2.09 × 10⁵ (±SD 3.42) (WWTP-1), 2.91 × 10⁵ (±SD 3.45) (WWTP-2), 8.1 × 10⁴ (±SD 3.49) (WWTP-3) and 2.18 × 10⁵ (±SD 3.0) (WWTP-4) (Table 3).

Lower frequencies of detection were observed for secondary effluent and reclaimed water samples (Table 3). In secondary effluents, GB-124 phages were detected in 50% of WWTP-1, in 41.6% of WWTPs 2 and 4 samples and only in 16.6% of WWTP-3 samples (Table 3). Discharge of chemical sewage entering directly in the system could contribute to viral inactivation. Geometric mean viral loads obtained in treated effluents were as follows: 2.53 × 10³ (±SD 4.68), 1.43 × 10³ (±SD 2.92), 1.03 × 10³ (±SD 3.34) and 1.16 × 10³ (±SD 2.65) (PFU/L) for WWTPs 1, 2, 3 and 4, respectively (Table 3).

The highest percentage of phage detection in reclaimed water was observed in the system composed of sand-anthracite and chlorination treatment from WWTP-1 (Table 3). Only two samples were positive (16.6%) for WWTP-2 and 3 (Table 3). No GB-124 phages were recovered from reclaimed water samples from WWTP-4, which operates with MBR, reverse osmosis and disinfection with chlorine dioxide (Table 3), suggesting a better virus-removal efficiency by this system. Geometric mean viral loads obtained after adjustment of left-censored data set in reclaimed water samples were 9.81 × 10² (±SD 2.36), 8.45 × 10² (±SD 2.08), 8.91 × 10² (±SD 1.75) and 7.07 × 10² (±SD 1.0) for WWTPs 1, 2, 3 and 4, respectively (Table 3).

Figure 2 shows the median and interquartile range of log (GB-124) phages values achieved in each stage of WWTP. Statistical analysis showed that the stage of treatment has a significant impact on the concentrations of GB-124 (p < 0.001), while there was no significant difference in concentrations of phages among the WWTPs.

Mean virus removal efficiency achieved by activated sludge processes varied from 1.89 log₁₀ (min) (WWTP-3) to 2.31 log₁₀ (max) (WWTP-2) (Figure 2). Tertiary treatments were able to increase the virus removal efficiency up to 0.41 log₁₀ (WWTP-1) (Figure 2).

Water temperature, TOC and conductivity were statistically significant for GB-124 phage occurrence in the different treatment stages of WWTPs (Figure 3), while turbidity and pH did not present significant contributions. GB-124 phage occurrence decreased with increased water temperature in the raw sewage and treated effluents of the WWTPs (p ≤ 0.0021), while at higher values of conductivity, the densities of GB-124 phages increased in raw sewage and in the secondary treatment stage in all WWTPs (p ≤ 0.091) (Figure 3). In raw sewage samples, the detection rates of phages decreased with higher levels of TOC (p ≤ 0.0046), while the opposite occurred in treated effluent samples from WWTPs with higher levels of TOC (p ≤ 0.100) (Figure 3), although TOC concentrations below 100 mg/L were observed in these samples.

Analysis of GB-124 phages in environmental samples has been considered a suitable approach in MST studies since they present a narrow host range and are highly specific to identify human fecal contamination (Nnane et al. 2011; Ebdon et al. 2012; McMinn et al. 2014).

Similar to data from other studies, environmental surveillance of GB-124 phages demonstrated that these viruses are widely disseminated in municipal wastewaters from São Paulo city, being detected throughout the year. Moreover, no significant difference was obtained among viral concentrations present in raw sewage of different WWTPs located in different regions and receiving different inflows of sewage.

The viral loads obtained are within the range of those of other studies where the presence of GB-124 phages was demonstrated in municipal wastewaters in Europe (Ebdon et al. 2012; Jofre et al. 2014; Purnell et al. 2015) and the USA (McMinn et al. 2014), suggesting that these microorganisms could also be potential indicators of human fecal pollution in water impacted with urban sewage in São Paulo city.

Although correlations among physicochemical parameters and concentrations of phages observed in different stages of the WWTPs should be interpreted with caution, results have suggested that the density of phages can be dependent on wastewater temperature.

The stability of GB-124 phages in sewage at different temperatures has been evaluated (McMinn et al. 2014). In general, phages persist longer at lower temperature (5 °C) than at higher temperature (i.e. 20 and 35 °C). After 21 days at 20 °C, no GB-124 phages could be detected by McMinn et al. (2014) in sewage samples, and inactivation was achieved after 7 days at 35 °C. Therefore, although GB-124 phages may be a potential viral marker to assessment of human fecal pollution of receiving waters, their applicability to evaluate the contamination of water downstream from a wastewater treatment plant in hot climate regions should be confirmed.

GB-124 phages were also observed after secondary treatment by activated sludge processes. Virus removal efficiency in activated sludge processes is recognized to achieve 0.7–2.9 log (Zhang et al. 2016). In this study, virus removal efficiencies ranged from 1.89 to 2.31 log₁₀ and were similar to other studies that reported virus removal efficiencies achieved by activated sludge processes (Zhang & Farahbakhsh 2007; Carducci et al. 2008). Carducci et al. (2008) found an average reduction of 2.2 log of somatic coliphages in activated sludge processes, and Zhang & Farahbakhsh (2007) observed a range of 1.5–2.3 log removal for the same types of viruses.

Although phage densities still remain in effluents from activated sludge processes, the importance of this treatment in virus removal cannot be disregarded. However, the presence of cultivable viruses in these effluents suggests a potential risk of dissemination of other enteric viruses into the receiving waters.

A fraction of these effluents is further treated by tertiary treatments in WWTPs to generate reclaimed water that is used by City Hall for urban and industrial purposes, such as washing streets and monuments, cooling boilers, and other non-potable reuses.

Mixed-media (sand, gravel, anthracite, coal) filtration is commonly used as a tertiary treatment to enhance the reduction of pathogens. In this study, sand-anthracite filters followed by chlorination are used as tertiary treatment in three WWTPs. As few reclaimed water samples were positive for GB-124 phages, results were adjusted to a known distribution using a statistical model, providing a more realistic set of data. GB-124 phage removal achieved by tertiary treatments varied from 0.06 log₁₀ in WWTP-3 up to 0.41 log₁₀ for WWTP-1. These results are in accordance with other studies that reported typical removals from sand filtration to be 0- to 1-log₁₀ for enteric viruses and ∼0.14- to 2-log₁₀ for coliphages (USEPA 2015).

Interestingly, in some samples from WWTP-2, reductions in GB-124 phages were not observed: the number present in effluent from tertiary treatment were greater than the number found in secondary treatment effluent. Ebdon et al. (2012) also observed that phage concentrations in treated effluent samples from WWTPs were not significantly different from concentrations detected in raw sewage samples, suggesting that the behaviour of these phages under different environmental conditions needs to be more understood.

In Brazil, chlorination is the most common disinfection process used and, in general, the cheaper technology. Higher resistance of viruses to disinfection by chlorination is recognised (Durán et al. 2003; Jofre et al. 2014), and the presence of interfering substances, such as organic matter and suspended solids in treated effluents can reduce the disinfection efficacy of chlorine (Carducci et al. 2008). Durán et al. (2003) demonstrated that bacteriophages infecting Bacteroides fragilis and selected somatic coliphages are more resistant to chlorination than bacterial indicators and other enteric viruses in sewage effluents.

The Environmental Company of the São Paulo State (CETESB), which is a regulatory agency, perform periodic monitoring (twice a year) in WWTPs using FIB (thermotolerant coliforms (TC) and E. coli) to evaluate reuse water quality produced by sanitation companies. Data on the presence of fecal indicators (i.e. TC and E. coli) performed in 2015 and 2016 demonstrated that these microorganisms were not detected in effluents from tertiary treatments of the respective WWTPs analysed in this study.

MBRs and reverse osmosis have been described as more effective technologies in virus removal compared to other tertiary treatments (Chaudhry et al. 2015; Pype et al. 2016). In MBRs, several factors could contribute in virus removal, including differences in morphology, attachments to biomass in the mixed liquor, inactivation by enzymes or predation, retention by the fouling cake layer and backwashing of membranes (Chaudhry et al. 2015; Purnell et al. 2015).

Zhang & Farahbakhsh (2007) found that MBR achieved 5.8-log₁₀ removal of coliphages, and more than 4.0 log₁₀ was achieved for adenoviruses and noroviruses removal in a study conducted by Chaudhry et al. (2015). In a full-scale MBR study, Purnell et al. (2015) reported a 5.3 log₁₀ reduction in indigenous somatic coliphages and a 3.8 log removal for GB-124 phages.

Although there are few studies about virus removal by reverse osmosis treatment, this system can present high virus-removal efficiency, achieving more than 5.0 log₁₀ for enteric viruses (Pype et al. 2016). However, operational problems or damage to the membranes could allow the passage of viruses, and therefore, continuous monitoring is recommended. In this study, no GB-124 phages were detected in reclaimed water treated by MBR coupled with reverse osmosis and disinfection by chlorine dioxide. Studies have demonstrated that chlorine dioxide concentrations of 0.1–1.0 mg/L are more effective in inactivating bacteria and viruses in water than chlorine (USEPA 2015).

Although a low frequency of GB-124 phage detection was observed in reclaimed water samples, the data obtained in this study indicate that these viruses could survive even after tertiary treatment by filtration and disinfection processes. In this way, although non-potable reuse is practiced, some applications such as street washing and irrigation of public gardens could pose some risk either by direct or indirect contact or by inhalation of aerosols containing considerable numbers of viable viral particles because a positive correlation between the detection rates of the studied phages and other pathogenic viruses in effluents from WWTPs have been observed (Ebdon et al. 2012).

It is also important to emphasise that the method used here to recover GB-124 phages uses 1 mL of samples to detect these viruses, which could be unsuitable to detect bacteriophages from reclaimed water samples or cleaner water (Harwood et al. 2013).

The absence of positivity of GB-124 phages in MBR effluents was already reported by other authors, although other enteric viruses and coliphages had been detected in the same samples (Purnell et al. 2016). The same authors also discuss that negative results for GB-124 phages in effluents from MBR systems could be associated with the detection limit of the used method, but different behaviour characteristics, such as attachment properties onto biomass in the mixed liquor, could also explain the absence of GB-124 phages in treated effluents (Purnell et al. 2015, 2016). Nevertheless, in the present study, the MBR system is coupled with reverse osmosis and final disinfection by chlorine dioxide, which could provide better virus removal efficiency. In this way, negative results obtained in effluents from the MBR/reverse osmosis system can be more associated with the performance of this system in virus removal than the limitations of the method used to detect phages. The same hypothesis would not be suitable for effluents from sand-anthracite filters, as the performance for virus removal is lower than MBR/reverse osmosis systems.

While the behaviour of GB-124 phages is not completely understood during the treatment stages in WWTPs (Ebdon et al. 2012; Purnell et al. 2016), it is more reasonable to consider that these bacteriophages could not be a good viral marker to assess human fecal pollution in these environmental matrices using the present methodology. Larger sample volumes or the use of methods to concentrate viruses could be applied to increase the probability of bacteriophage detection in these environmental matrices (Harwood et al. 2013). Further studies should be carried out using different methods to detect GB-124 phages and comparisons with other viral markers to evaluate their potential to tracking human fecal pollution in reclaimed or reuse water.

Although GB-124 phages were less abundant in reclaimed water samples, the test is a simple performance analysis, and then these bacteriophages could be considered complementary low-cost viral markers for tracking sources of human fecal pollution in municipal wastewaters and impacted receiving waters, as described previously (Nnane et al. 2011; Ebdon et al. 2012; McMinn et al. 2014). Therefore, the use of this marker should be encouraged, and other studies involving GB-124 phages detection or bacteriophages that infect other host strains of Bacteroides spp. in several environmental matrices, under different local conditions, could provide important information to policy makers for orientation or implementation of future water reuse and wastewater regulation programmes.

This study was supported by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) (grant number N^o 2013/26586-1) and Environmental Company of the São Paulo State (CETESB). The authors wish to acknowledge the staff of Environmental Company of the São Paulo State – CETESB, especially Carlos Jesus Brandão, Vivian Baltazar Rocha, Marcelo Oliveira, Fernando de Caires and Waldemar Rodrigues for conducting sampling and field analysis.

The authors declare that they have no conflicts of interest.