Background
Currently, enteric viruses are considered to be the main etiological agents of waterborne diseases, accounting for 30-90% of gastroenteritis worldwide [
1]. Enteric viruses are frequently aggregated in the environment [
2], and due to the small size of the particles (0.5 – 1.0 μm), they are not efficiently retained in the filtration stage at Water Treatment Plants (WTPs) [
3]. Disinfection is therefore critical for reducing the infectious virus concentrations in source water.
According to the guidance manual published in 1991 by the Environmental Protection Agency of the United States (US EPA), a 4log
10 (99.99%) removal or inactivation of enteric viruses by filtration and/or disinfection is recommended. The EPA also recommends values for the contact time - Ct (disinfectant concentration (mg/L) x time (min)) of 4, 6 and 8 to achieve inactivation of 2log
10, 3log
10 and 4log
10, respectively, using free chlorine [
4]. However, the values established in this manual were based on studies with hepatitis A in buffered demand free water at 5°C. As the water quality can significantly affect the effectiveness of the disinfection by free chlorine [
5], it is unclear whether these recommended Ct values are sufficient to inactivate other viral pathogens in different water matrices.
Enteric viruses are generally more resistant to environmental conditions and conventional water treatment using chlorination and filtration than enteropathogenic bacteria, and there is no potential for replication in the environment because the viruses are obligatory intracellular parasites. Although virus degradation is expected to occur, the amount of virus that remains is more meaningful than the amount of remaining bacteria that can re-grow after being excreted. There have been virus-related outbreaks with the consumption of water in compliance with bacterial standards [
6].
The human adenovirus (HAdV) belongs to the Adenoviridae family, genus
Mastadenovirus, comprising 57 serotypes [
7]. HAdV has been indicated as a potential marker of human fecal contamination in water [
6]. The current contaminant candidate list of the aquatic environment (CCL3) considers the adenovirus as a high priority emerging contaminant present in drinking water and a candidate contamination marker of the aquatic environment [
8].
HAdV has been extensively detected in environmental matrices. In 2005, Choi and Jiang [
9] found that 16% of the river samples in California, USA were positive for HAdV (10
2 - 10
4 gc/L). Albinana-Gimenez et al. in 2009 [
10] described that 90% of the river water samples in Barcelona, Spain were HAdV positive (10
1 - 10
4 gc/L). Dong et al. in 2010 [
11] detected HAdV in 100% of the sewage samples (1.87 × 10
3 - 4.6 × 10
6 gc/L) and in 83.33% of the recreational water samples (1.7 × 10
1 – 1.19 × 10
3 gc/L) in New Zealand. Win-Jones et al., in 2011 [
12] found that 60.6% of the European recreational and fresh water samples were positive for HAdV, with a mean value of 3.260 × 10
3 gc/L. In 2012, Fongaro et al. [
13] described an HAdV presence in 96% of the samples collected in the Peri Lagoon, Brazil (1.73 × 10
6 - 2.41 × 10
8 gc/L) and Garcia et al. (2012) [
14] described a presence of HAdV in 100% of the river water samples in Brazil, with an average of 10
7 gc/L. In the same year, Ye et al. [
15] described 100% HAdV positive for river and drinking water samples in Wuhan, China (10
2 - 10
4 gc/L).
Several studies have evaluated the inactivation efficiency of HAdV by free chlorine in buffer [
3,
16,
17] in waters from rivers and lakes [
5], groundwater [
3], seawater [
18] and sewage [
19]. However, the methods chosen to evaluate the HAdV infectivity are often time-consuming. The plaque assay has long been considered a standard method, although it can require 5 to 12 days to achieve results [
5,
17-
19]. Other methods are based on genome detection, such as PCR or the observation of a cytopathic effect.
As an alternative, recombinant adenoviruses (rAdV) can be used as a viral model to study the water disinfection procedures. rAdV are defective in their replication, as they lack the early gene, E1, which is involved in viral gene transcription, DNA replication, and the inhibition of host cell apoptosis [
20]. Thus, rAdV replication is weakened in this condition, unless the replication occurs in permissive cell lines that express the E1 gene products, such as the Human Embryonic Kidney (HEK) 293A cells [
21]. rAdV replication can, therefore, be directly monitored by fluorescence methods, based on the expression of the green fluorescent protein (GFP) that is encoded by a gene incorporated into the viral DNA. HEK 293A cells infected with rAdV provide a novel reporter for viral infectivity assays, enabling the use of rapid (24 h) and quantitative methods of monitoring GFP expression in individual cells, such as fluorescence microscopy.
In this context, the goal of the present study was to evaluate the viral inactivation in water collected from two Water Treatment Plants after the filtration (non-disinfected) by subsequent free chlorine addition, using the recombinant adenovirus as a model. Buffered demand free (BDF) water was used as the control. This study also evaluates the treated water quality throughout the water distribution network in relation to the concentration of human adenovirus and total coliforms.
Discussion
The application of recombinant adenovirus provides a versatile system for therapeutic applications and gene expression studies, including gene transfer
in vitro, gene therapy and vaccine therapy [
20]. Despite this well-established use, we describe herein a novel application of rAdV in the environmental virology field, especially in disinfection assessment studies. The current study shows, for the first time, the efficiency of free chlorine disinfection of rAdV in water samples undergoing treatment for human consumption. The temperatures of 15°C and 20°C mimic the temperature range of the natural waters in south Brazil during the winter and summer seasons [
23], respectively, and the pH conditions, which were not modified. We believe that it is very important to select temperatures that assess the real conditions that occur in the real environment. Comparison techniques based on genome detection (qPCR) and infectivity (cell culture) are also important because risk assessment studies based on genomic copy detection are encouraged [
24-
28], and other studies have reported that free chlorine can damage genetic material [
29,
30].
The use of rAdV proved to be economical, convenient and fast for several reasons: it does not require the use of primary and secondary antibodies; it decreases the possibility of overestimating the viral titers due to non-specific binding; it avoids cell loss during the washing stages commonly performed in immunodetection techniques; and it is faster (24 h) than the conventional plaque assay method (7 to 10 days), described by Cromeans et al. (2008) [
31].
Regarding the GFP fluorescence stability in chlorine solutions, according to Mazzola et al. (2006) [
32], who evaluated the GFP stability in chlorinated water and buffered solutions, the main conclusion was that GFP is a suitable fluorescent marker for monitoring disinfection effectiveness. They observed that, with constantly stirred solutions, the GFP fluorescence decreased abruptly after contact with chlorine in concentrations greater than 150 ppm, and the GFP fluorescence intensity was reduced by 42% in the initial 30 s of contact with a 70 ppm phosphate buffered chlorinated solution. Webb et al. (2001) [
33] exposed
Aureobasidium pullulans cells expressing GFP to chlorinated solutions (25–150 ppm) and observed that the loss of GFP fluorescence was highly correlated with a decrease of the number of viable cells. Casey and Nguyen (1995) [
34] exposed
Escherichia coli cells also expressing GFP and observed the same result as Webb et al. (2001). In the present study, the chlorine concentrations employed were 0.2 ppm and 0.5 ppm, much lower than the values described above. Therefore, we can conclude that the GFP fluorescence itself was not affected by this low concentration of applied chlorine, and the lack of fluorescence is certainly due to a lack of rAdV replication. This phenomenon was also proven by the same effect of the chlorine on viral disinfection using non-recombinant human adenovirus, which was previously described in the literature [
3,
5,
17].
Viral purification is essential for the experiments of disinfection by free chlorine because viral suspensions contain considerable amounts of organic matter that consumes free chlorine, preventing its virucidal and bactericidal action [
18]. This work was the first to use chromatography as a method of purification and proved to be comparable to studies using other forms of purification, with comparable and adequate Ct values [
5,
17] because the concentrations of disinfectant did not vary significantly in the presence of the purified virus stock (P > 0.05).
No significant difference in the disinfection efficiency was observed (P > 0.05) between the tested temperatures (15°C and 20°C). However, the pH variation exerted a great influence on the disinfection efficiency: the Ct for the 4log
10 disinfection at BDF pH 8.0 (1.87) was 10 times greater than the Ct at BDF pH 6.9 (0.187). This result is due to residual free chlorine in both pHs; at pH 8.0 there is approximately 25% HOCl and 75% of the hypochlorite ion (HCl
+), and at pH 6.9, approximately 80% is HOCl, and 20% is HCl
+. According to AWWA (2006) [
35], the germicidal efficiency of HOCl is approximately 100 times greater than HCl
+, which explains the observed results; therefore, the pH of the water can cause a variability in the disinfection efficiency.
Nevertheless, fresh water submitted to water treatment is constantly influenced by geological features. It is well known that the levels of chemicals in soils reflect the levels at the source rock, except in cases with anthropogenic influence [
36,
37], and the geology has a great influence on the chemical characteristics of the soil and surface water [
38]. Moreover, it has been demonstrated that the pH values can vary in different water bodies, such as rivers, water reservoirs or estuaries, depending on the season [
39-
41] and the daily basis [
40,
42] and spatially (variation throughout the water layer or sampling sites) [
42,
43]. Furthermore, air pollutants, such as carbon dioxide (CO
2), have a great influence on the pH of water because air pollutants can enter the water through biological metabolism involving organic carbon and through equilibrium with the atmosphere. Once in the water, CO
2 reacts and forms bicarbonate (HCO
3
−) and carbonate (CO
3
2−), decreasing the pH. Therefore, more air pollution results in more CO
2 in the water and higher water acidity [
44]. Some other factors can affect the disinfection efficiency, such as antioxidants from commercial hygiene products, which reduce hypochlorite to chloride ions and decrease the free chlorine available for disinfection [
45,
46]. In addition, it is well known that the temperature range can influence the pH, as well as the CO
2 solubility and may vary on a daily basis [
47]. Altogether, these factors can affect the pH and change the disinfection dynamics. Thus, it is essential to carefully control the pH in water treatment plants throughout the process, especially before the addition of chlorine, due to its great influence on the disinfection performance.
It is possible to observe that the inactivation curves for all of the experimental conditions, except for the BDF pH 8.0, were characterized by two phases: an initial phase in which the inactivation occurred rapidly (approximately 2log
10 in 2 seconds), followed by a phase with a lower rate of inactivation, which may be designated the “tailing phase.” Page et al. (2009) [
16] described that the loss of disinfection efficiency observed during the tailing phase is most likely due to the rapid change of specific chemical moieties on the viral structure that preferentially react with HOCl. Thus, some authors propose that the HOCl-mediated transformation of proteins, which is due to the high reactivity with proteins and their abundance in biological systems, plays a key role in the loss of the biological function of this form of the free residual chlorine, leading to the formation of the tailing phase [
48]. As the main feature, adenovirus capsids are composed of proteins (fibers, pentons and hexons) that are physically exposed to the disinfectant. These proteins contain functional groups, such as amines and thiols, that react with free chlorine, leading to a loss of the biological function of the disinfectant [
49].
The inactivation curve in the BDF pH 8.0 experiments may be associated with damage involving secondary oxidizing agents [
16]. In addition, at this pH, the HOCl concentration is approximately 25% [
35], making the disinfection slower with no biphasic behavior observed.
The qPCR assay has already proven to be fast and specific for the detection and quantification of rAdV genomes. However, this technique does not provide sufficient information about inactivated viruses compared with the fluorescence microscopy technique after cell culture. The time necessary for the assays was determined by the disinfection achievement. Therefore, once the 4log
10 of disinfection was achieved, the experiment was considered concluded, although the viral genomic copies did not show a significant log reduction. In fact, some studies had performed viral disinfection studies employing PCR, and some studies indeed showed a reduction of the viral copies. Nevertheless, they observed the same profile: the genome integrity decreased more slowly than the viral viability [
18,
19,
30,
50-
54]. Although some studies have reported that free chlorine can damage the viral genetic material [
30] by interacting with the amine group of nucleotides [
29], it is suggested that the extent of DNA damage caused by free chlorine is not sufficient to detect viral inactivation by the qPCR technique, which often results in very small amplicons [
55]. Even with qualitative PCR using primer sets that generate greater amplicons (400 bp to 1,215 bp), the genome integrity is not correlated with the viability of HAdV because the PCR products are generated even when higher concentrations of chlorine are used [
30]. This result suggests that the ability of free chlorine to cause damage in the viral genome is limited [
30], and viruses with lesions in the capsid proteins caused by chlorine may still contain their genomes that are protected from the inactivation procedures. Thus, the viral nucleic acids detected by PCR or qPCR can be derived from infective and non-infective damaged viruses and from free nucleic acids from lysed viruses, and the results obtained by cell culture, when possible, are more representative of the actual health risk. Therefore, risk assessment studies based on genomic copy detection are inadequate, overestimating the actual risk of consuming drinking water treated with chlorine.
The absence of viable HAdV in the output of both WTPs and the presence of infectious virus in the distribution network suggest that the water treatment is efficient for the inactivation of HAdV; however, the water is re-contaminated during its distribution and/or storage. Thus, low concentrations of residual free chlorine throughout the water distribution system are not sufficient to inactivate high viral loads that are accidentally re-introduced after the water treatment process [
3].
It is postulated that biofilms in the drinking water distribution networks may play a role in the accumulation, protection and dissemination of pathogens [
56]. The formation of biofilms has been described to provide bacteria much greater resistance to free chlorine, and it has also been shown that viruses can adsorb into biofilms [
57]. The detection of viable HAdV in the distribution network may be due to viral aggregation and adsorption by particles, which have previously been reported as increasing the resistance to chlorine and the environment [
2,
3] or adsorption into biofilms, also protecting them from the action of the free chlorine disinfectant.
As described in the literature, the human adenovirus is rapidly inactivated by free chlorine. However, it is difficult to make a direct comparison of the Ct values due to variations in the experimental conditions, especially the technique used for the viral purification [
5]. Thurston-Enriquez et al. (2003) [
3], Kahler et al. (2010) [
5], and Cromeans et al. (2010) [
17] reported Ct values similar to those observed in the present study, except for BDF at pH 8.0, which was reported previously as 0.24 to inactivate 4log
10 [
3], in contrast to the 1.87 value observed in this study. However, the Ct value of 0.24 that was described by Thurston-Enriquez et al. (2003) [
3] was calculated by the model, whereas the Ct observed in the experiment (not modeled) was 36.09 for the 4log
10 reduction. Taken together, these data indicate the high susceptibility of the human adenovirus to free chlorine. Because the present work employed rAdV, and the Ct values reported in the literature for HAdV are comparable, this result confirms the applicability of rAdV as a model for HAdV for studies of free chlorine disinfection and discards the need to perform all experiments with HAdV in parallel in this study. By comparing the Ct values recommended by the EPA [
4] with those predicted by modeling, the values are lower than those recommended in all of the experimental conditions, providing a margin of safety for free chlorine water treatment in terms of the human adenovirus.
The Chick-Watson model was chosen as it best fits the reactors in the batch mode or ideal piston, in which the longitudinal dispersion is equal to zero [
58]. As the experiments were performed in 10 or 40 mL, the longitudinal dispersion is disregarded. The values of
k (inactivation constant rate) calculated by the Chick-Watson model are in agreement with what was observed: the inactivation was faster in MQ, followed by LP/BDF pH 6.9 and finally BDF pH 8.0 (Table
2 and Figures
2,
3,
4 and
5) because the
k value is directly proportional to the inactivation rate.
The constants of the Chick-Watson model described herein (k, k’) were determined by conducting bench experiments. These constants are considered characteristic for the rAdV inactivation kinetics, the specific conditions of the pH and temperature, and the composition of the environmental matrices. Therefore, these constants can be used to calculate the Ct at other chlorine concentrations, without the need to perform additional bench experiments.
Authors’ contributions
MAN and CRMB designed the research. MAN carried out the collection of water samples and concentration of treated water samples, physicochemical parameters and fecal contamination analysis, reagents and glassware treatments, cytotoxicity tests, disinfection assays, microscopy fluorescence and qPCR assays. MEM and MAN carried out the kinetic modeling and statistical analysis.CDS performed the plaque assay. CRMB conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.