Inactivation of particle-associated viral surrogatesby ultraviolet light
Michael R. Templeton*, Robert C. Andrews, and Ron Hofmann
Department of Civil Engineering, University of Toronto
35 St. George Street, Toronto, Ontario, CanadaM5S 1A4
*Corresponding Author: Tel. +1-416-978-3220, Fax. +1-416-978-3674, Email address:.
Submitted for publication in Water Research
Submitted 10 February 2005
Re-submitted with revisions 11 June 2005
Abstract
Thisstudyinvestigatedwhether colloid-sized particles can enmesh and protect viruses from 254-nm ultraviolet (UV) light and sought to determine the particle characteristics (e.g. size, chemical composition) that are most relevant in causing a protective effect. Two viral surrogates (MS2 coliphage and bacteriophage T4), three types of particles (kaolin clay, humic acid powder, and activated sludge), two coagulants (alum and ferric chloride), two filtration conditions (none and 0.45 µm), and two UV doses (40 and 80 mJ/cm2 for MS2 coliphage; 2 and 7 mJ/cm2 for bacteriophage T4) were considered in a series of bench-scale UV collimated beam experiments.Transmission electron microscopy was used to qualitatively confirm the phage particle-association after coagulation. Humic acid and activated sludge flocparticles shielded both viral surrogatesto a statistically significant degree(with > 99% confidence) relative to particle-free control conditions, while the kaolin clay particles provided no significant protection. The results of the study suggest that particles < 2 µm in diameter are large enough to protect viruses from UV light and that particulate chemical composition (e.g.UV-absorbing organic content) may be a critical factor in the survival ofparticle-associated viruses during UV disinfection.
Keywords: Ultraviolet (UV) disinfection; waterborne viruses; particle-association.
1. Introduction
Most of the existing ultraviolet (UV)disinfection dose-response data reported in the literature wasobtained from bench-scale UV exposures of pure-culture microorganisms in particle-free water. However this may not be representative of the inactivation of particle-associated microorganisms that may be encountered in practice. Coliform bacteria have been shown to be frequently particle-associated in wastewater (Parker and Darby, 1995; Loge et al., 1999; Ormeci and Linden, 2002) and viruses are predominantly particle-associated in natural waters and wastewater (Bitton, 1975; Moore et al., 1975; Wellings et al., 1976; Gerba et al., 1978; Sakoda et al., 1997; Meschke and Sobsey, 1998).This microbial particle-association has been attributed to electrostatic attraction (Gerba, 1984) and hydrophobic interactions (Wait and Sobsey, 1983; Mamane-Gravetz and Linden, 2005). In addition to becoming associated with particles via these attachment mechanisms, microorganisms can also be released into the environment already occluded within fecal particulates (Hejkal et al., 1979). A concern is that particle-associated pathogens may be protected from UV disinfection, resulting in lower microbial inactivation than would be expected from currently reporteddisinfection dose-response relationships.
Coliform bacteria, which are typicallybetween 1 and 10µm in size, have been shown to be shielded during the UV disinfection of wastewater by being enmeshed within particles greater than 10 µm in diameter (Qualls et al., 1985; Emerick et al., 2000). In most well-operated conventional drinking water treatment facilities, particles of this size are likely to be removed by filtration upstream of UV disinfection or may notbe present in the source water at all. Viruses, however, are one to two orders of magnitude smaller than bacteria and may therefore be protected from UV disinfection by much smaller particles (e.g. particles < 10 µm in diameter) that may routinely pass through even well-operated filters in water treatment facilities. Further, it has beenreported thatparticles can protect coliform bacteria and viruses from chemical disinfectants (Stagg et al., 1977; Hejkal et al., 1979; Babich and Stotzky, 1980; Ormeci and Linden, 2002). It is unknown whether particle-enmeshmentsimilarly impacts the UV disinfection of viruses.
Recent studies have considered the effect of low levels of turbidity and particle counts on UV disinfection and have reported no negative impacts (Batch et al., 2004; Passantino et al., 2004). However,the challenge organisms in these studies were spiked into test waters withoutsteps taken to encourage particle-association (e.g. via coagulation). As such, these studies may only suggest the impact of turbidity on dose-response as it relates to the impact of UV light scattering by particles, rather than the impact of particle-association of microorganisms (Swaim et al., 2005). Further, while it has been shown that some particles have a greater effect on UV dose-response kinetics than others (Passantino et al., 2004), it is unclear which specific particle characteristics (e.g. particle size, composition, structure) enhance the survival of particle-associated microorganisms during UV disinfection.
Previous disinfection studies have emphasized the importance of accounting for particle-associated organismsby using elution methods prior to microbial enumeration (Parker and Darby, 1995; Ormeci and Linden, 2002; Borst and Selvakumar, 2003).These methods typically involve addition of a chemical eluent (e.g. protein-rich beef extract solution for virus elution) to break electrostatic and hydrophobic bonds, combined with application of physical shearing forces (e.g. blending, shaking, or sonication) to break apart particles and release enmeshed organisms (Gerba, 1984; Chauret et al., 1999; Parker and Darby, 1995; Ormeci and Linden, 2002).Microbial analysis techniques that do not include an elution step do not account forparticle-associated organisms and may therefore underestimate the organism concentrations (Borst and Selvakumar, 2003).
Since most human viruses are typically present in very diffuse concentrations in natural waters, viral surrogates such as bacteriophages are often used in environmental virology and disinfection studies (Havelaar et al., 1991; Payment and Franco, 1993; Hot et al., 2003). However,since the particle adsorption behavior of viruses depends on virus size, surface charge, and morphology (Gerba, 1984)a single viral surrogate may not adequately represent the adsorptive behavior of all human viruses. Gerba (1984) divided human viruses into categories based on their empirically observed soil adsorption behavior. MS2 coliphage, a 25 nm diameter icosahedral phage with an isoelectric point (pI) of 3.9 (Gerba, 1984; Sakoda et al., 1997), and bacteriophage T4, a 200 nm-long tailed phage with a pI of 4.2 (Sakoda et al., 1997), were categorized by Gerba (1984) in the twolargest separate groups of viruses. For the research described herein it was therefore decided to use both MS2 coliphage and bacteriophage T4, in anticipation that the adsorption behavior of a broad range of human viruses could be represented.
The goals of this study were:(i) to determine whether it is possible for particles < 10 µm in diameter to enmesh and protect viral surrogates from UV disinfection and (ii) to determine what specific particles characteristics (e.g. size, composition) influence the survival of particle-associated viruses during UV disinfection.
2. Materials and methods
2.1 Mixing of particles, phages, and coagulants
A standard jar test apparatus (Phipps and Bird, Richmond, VA, USA) was used to sequentially mix the particle, phage, and coagulant into two liters of buffered water (pH 5.5) in square jars. Each liter of buffer water was prepared by mixing 630 mL of a solution consisting of 0.20 M boric acid (H3BO3) and 0.05 M citric acid (H3C6H5O7∙H2O),with 370 mL of 0.10 M tertiary sodium phosphate (Na3PO4∙12H2O). The solutions were prepared in distilled water. Synthetic water matrices were considered in this study so that the particle characteristics (type, concentration) and pH conditions could be controlled in order to answer fundamental questions concerning the interactions of particles and viruses and the effect of various particle characteristics on UV disinfection, as stated in the research objectives.
The pH of the buffered water (5.5) was selected on the basis of jar tests conducted beforehand as the optimum for alum floc formation with kaolin clay particles, as quantified by turbidity reduction. The samebuffered water (pH 5.5) was used in all trials.While the optimum pH would differ for ferric chloride coagulation or for particles other than kaolin, the water matrix pH was maintained constant among all trials in orderto limit the number of experimental variables. Further, a confounding factor that had to be considered was that the electrostatic attraction of viruses to particle surfaces is pH-dependent, and so this variable was controlled by maintaining constant pH. Measurement of the pH before and after addition of the particle, phages, and coagulant to the buffered water confirmed that the pH did not change over the course of the trials.
The pH of 5.5 used in these experiments is on the low end of the pH range typically encountered in drinking water treatment, however this pH was chosen in order to optimize alum floc formation, as described above. Also, as the pH of the water decreases and approaches the isoelectric points of the phages (pI = 3.9 for MS2 and pI = 4.2 for T4), the electrostatic repulsive forces between the phages and the particles are reduced and particle-association is therefore encouraged. As such, the pH was selected so as to be as close as possible to the isoelectric points of the phages without being beyond the realm of realistic pH values encountered in treatment practice.
Three particleadditives were used: kaolin clay particles (Mallinckrodt Baker, Phillipsburg, NJ, USA), humic acidpowder (Aldrich Chemical Company, Milwaukee, WI, USA), and return activated sludge particles obtained from the North Toronto Sewage Treatment Plant (Toronto, ON, Canada). Kaolin clay and Aldrich humic acid have been used in numerous previous studies to represent inorganic colloids and organic matter present in source waters, respectively (Bitton et al., 1972; Nasser et al., 1995; Barbeau et al., 2004). Also, humic substances have been reported to be a major portion of the dissolved organic carbon in secondary wastewater effluent and to be one of the most UV-absorbing organic substances (Bitton et al., 1972; Qualls et al., 1983). The activated sludge particles were selected to representslightly larger particles for comparison purposes, as well as torepresent thosethat may typically be present in wastewater.
Two viral surrogates were used: MS2 coliphage and bacteriophage T4. The initial spiked phage concentrations in the jars were between 105-106 plaque-forming units (PFU) per milliliter in all trials. Both the phage and particleconcentrations used in this study were much higher than typicalvalues in source waters encountered in drinking water treatment. Elevated phage concentrations were necessary to have countable numbers of surviving phages post-UV exposure, such that log inactivation values could be calculated. At these phage concentrations, elevated particle concentrations were required to achieve a consistently measurable fraction of the total phage population as particle-associated, such that the effect of this particle-association on UV disinfection (if any) could be quantitatively defined.In other words, the fraction of the phage population that was particle-associated was maximized so that the relative number of free-floating phages would be reduced, which allowed the impact of the particle-associated fraction on subsequent UV disinfection to be more easily identified.
As a numerical example, considera particle-free sample containing 10,000 free-floating phages, and a particle-laden sample containing 10,000 free-floating phages and only 100 particle-associated phages which are completely protected from UV light and able to survive (i.e. 100/10,100 x 100% = ~1% of the total population is particle-associated). Assume that a UV dose is applied that is high enough to cause 4-log reduction of free-floating phage, but that all 100 particle-associated phages are completely protected and survive the UV dose. Therefore, the inactivation in the particle-free sample will be 4-log (i.e. since essentially all 10,000 free-floating phages will be inactivated), whereas the inactivation in the particle-laden sample will be log (10,100/100) = 2-log. Therefore, the more phages that are particle-associated and that are completely protected from the UV light, the more evident will be the particle-shielding effect. Since not all phages that are determined to beparticle-associated prior to UV exposure are actually completely protected from UV light (i.e. some phages may only be surface-attached, or may be associated with non-UV-absorbing particles), it is desirable to have as high a percentage of the initial phage population being particle-associated as possible for the purposes of this experiment.
The kaolin clay particles and Aldrich humic acid were added to the buffered water to achieve concentrations of 200 and 150 mg/L, respectively. For the trials with return activated sludge, which hada mean suspended solids concentration of 4700 ± 800 mg/L (n = 16), 100 mL of the sludgesuspension was added to 1.9 L of buffered water. The return activated sludge was mostly organic, having a mean volatile suspended solid concentration of 3400 ± 700 mg/L (n = 16). Suspended solids and volatile suspended solids were measured using Standard Methods 2540-D and 2540-E, respectively (APHA, 1998). As explained above, these particle concentrations were selected such that either a consistently measurable percentage of the phage population was particle-associated after coagulation, or such that the percentage of the phage population was high enough that a statistically significant reduction in log inactivation (statistical significance is defined below) was being observed. As such, these particle concentrations resulted in initial turbidities (i.e. prior to coagulation and flocculation) in the range of 70-100 NTU, as measured using a Hach 2100 N turbidimeter. However, it is important to note that despite theelevated particle concentrations used in this study, the results can still be extrapolated to lower particle concentration conditions that are representative of those encountered in water treatment practice, due to the factors that are incorporated in the UV dose calculation method that was used. This is explained furtherin the Results and Discussion section below.
Two coagulants common to the drinking water industry, alum and ferric chloride, were considered. It was anticipated that differences in the UV absorbance of these coagulants – i.e. iron is more UV-absorbing(Cairns et al., 1993) – as well as differences in the structure of the floc that they form, would result in differences in the degree to which enmeshed phages would be protected from UV light. Alum and ferric chloride were applied at 30 mg/L and 15 mg/L, respectively. These coagulant doses were determined in jar tests beforehand as those that resulted in optimized floc formation with kaolin clay particles, as quantified by turbidity reduction. The same coagulant doses were used for the trials with the other particles (humic acid, activated sludge) in order to minimize the number of experimental variables under consideration.
Each jar test combined one particle additivetype with one viral surrogatetype and one coagulant. First, the particle additive and two liters of buffered water weremixed at 100 rpm for five minutes. The phage was then added and mixed at 100 rpm for three minutes. Lastly, the coagulant was added and mixed at 100 rpm for one minute, followed by slow mixing at 30 rpm for 20 minutes to promote floc formation. These flocculationsteps were similar to those used in previous coagulation studies (Krasner and Amy, 1995; Nasser et al., 1995), but with the usual final settling period omitted such that suspended particles could be drawn from the supernatantimmediately following flocculation. Supernatant samples were withdrawn slowly from the jars using sterile wide-bore pipettes, in an attempt to minimize floc break-up. The suspensions were mixed gently at 5 rpm while the supernatant samples were withdrawn. The jar tests for each possible combination of particle, phage, and coagulant were conducted in a randomized order. Each jar test particle-phage-coagulant condition was conducted twice, with three 20 mL supernatant samples per trial exposed to thecollimated UV beam (i.e. in total there were six inactivation data points generated per phage-particle-coagulant condition). Control trials were also conducted in which only the phage was added to the buffered water (i.e. no particle or coagulant added). In addition, trials were conducted in which only the phage and particle were added (i.e. no coagulant). All experiments were conducted at 20 ± 1 ºC.
2.2 UV collimated beam exposures
Twenty milliliter supernatant samples were exposed to UV light using a low pressure UV collimated beam. The standard UV dose measurement and calculation methods for collimated beam experiments as described by Bolton and Linden (2003) were followed. The samples were contained in 8.5 centimeter diameter Petri dishes and gently stirred by a one centimeter long magnetic stir bar during the exposures. Samples were transferred from the jars to the Petri dishes using sterile wide-bore pipettes. The distribution of UV intensity across the exposure surface (i.e. along a 0.5 cm by 0.5 cm grid) was measured using an IL1700 radiometer equipped with an SUD240 sensor (International Light, Newburyport, MA, USA) and was incorporated into the average UV dose calculation, as described in detail in Bolton and Linden (2003). The UV intensity was 0.24 mW/cm2 at the center of the exposure surface.A CE3055 spectrophotometer mounted with an integrating sphere sensor (Cecil Instruments Ltd., Cambridge, UK) was used to measure the UV absorbance of the water samples. The integrating sphere sensor captures light that is reflected off particles (which is still able to disinfect) and accounts for it in the UV absorbance measurement. Without this sensor, only the light passing directly through the sample at 180º to the direction of the incident light beam would be measured. Therefore the UV absorbance for samples containing particles is overestimated when using a spectrophotometer without an integrating sphere sensor (Christensen and Linden, 2003).
Two UV doses were considered for each phage, to determine whether the protective effect caused by particle-association was UV dose-dependent. It was hypothesized that the protection of phages might decrease as the UV dose was increased and more energy was applied to the floc particles. Doses were selected to result in approximately 2- and 4-log inactivation of the phages under the control (particle-free and coagulant-free) condition. MS2 coliphage samples were exposed to either 40 or 80 mJ/cm2. Bacteriophage T4 was found to be a far more sensitive phage than MS2 coliphage, and therefore samples containing T4 phage were exposed to only 2 or 7 mJ/cm2. In each case, the protective effect of particle-association was assessed by comparing the log inactivation that was achieved in the samples containing particles with the log inactivation of free-floating phages in the control conditions, via a t-test at the 99% confidence level.
2.3 Phage elution method
A phage elution method was used to release phages from particles for enumeration. The method was adapted from those previously described by Parker and Darby (1995)and Chauret et al.(1999) and involved addition of a 2% (w/v) beef extract solution (adjusted to pH 9.0 using NaOH) and blending at 20,000 rpm for three minutes. A comparison of parallel eluted and non-eluted samples was used to quantify the percentage of total phage population that was particle-associated. The non-eluted sample was used to enumerate the free-floating phages, while the eluted sample enumerated the sum of the free-floating and particle-associated phages. Subtracting the non-eluted count from the eluted count gave the number of particle-associated phages. This calculation was performed on samples both before and after UV exposure. The total phage populations (i.e. the sum of free-floating and particle-associated phages) were used in the log inactivation calculations. Initial spike and recovery experiments showed that this method yielded high phage recoveries (> 95%) and did not affect the viability of the phages. The blending did not alter the temperature of the samples by more than 2 ºC in any trial. The same elution method was used for both MS2 coliphage and bacteriophage T4.