crossmark

Elution Is a Critical Step for Recovering Human Adenovirus 40 from Tap Water and Surface Water by Cross-Flow Ultrafiltration Hang Shi,a Irene Xagoraraki,a

Kristin N. Parent,b Merlin L. Bruening,c

Volodymyr V. Tarabaraa

Department of Civil and Environmental Engineering, Michigan State University, East Lansing, Michigan, USAa; Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, Michigan, USAb; Department of Chemistry, Michigan State University, East Lansing, Michigan, USAc

ABSTRACT

This paper examines the recovery of the enteric adenovirus human adenovirus 40 (HAdV 40) by cross-flow ultrafiltration and interprets recovery values in terms of physicochemical interactions of virions during sample concentration. Prior to ultrafiltration, membranes were either blocked by exposure to calf serum (CS) or coated with a polyelectrolyte multilayer (PEM). HAdV 40 is a hydrophobic virus with a point of zero charge between pH 4.0 and pH 4.3. In accordance with predictions from the extended Derjaguin-Landau-Verwey-Overbeek theory, the preelution recovery of HAdV (rpre) from deionized water was higher with PEMPEM CS coated membranes (rpre ⴝ 74.8% ⴞ 9.7%) than with CS-blocked membranes (rpre ⴝ 54.1% ⴞ 6.2%). With either membrane PEM CS type, the total virion recovery after elution (rpost) was high for both deionized water (rpost ⴝ 99.5% ⴞ 6.6% and rpost ⴝ 98.8% ⴞ PEM CS 7.7%) and tap water (rpost ⴝ 89% ⴞ 15% and rpost ⴝ 93.7% ⴞ 6.9%). The nearly 100% recoveries suggest that the polyanion (sodium polyphosphate) and surfactant (Tween 80) in the eluent disrupt electrostatic and hydrophobic interactions between the CS virion and the membrane. Addition of EDTA to the eluent greatly improved the elution efficacy (rpost ⴝ 88.6% ⴞ 4.3% and PEM rpost ⴝ 87.0% ⴞ 6.9%) with surface water, even when the organic carbon concentration in the water was high (9.4 ⴞ 0.1 mg/liter). EDTA likely disrupts cation bridging between virions and particles in the feed water matrix or the fouling layer on the membrane surface. For complex water matrices, the eluent composition is the most important factor for achieving high virion recovery. IMPORTANCE

Herein we present the results of a comprehensive physicochemical characterization of HAdV 40, an important human pathogen. The data on HAdV 40 surface properties enabled rigorous modeling to gain an understanding of the energetics of virion-virion and virion-filter interactions. Cross-flow filtration for concentration and recovery of HAdV 40 was evaluated, with postelution recoveries from ultrapure water (99%), tap water (⬃91%), and high-carbon-content surface water (⬃84%) being demonstrated. These results are significant because of the very low adenovirus recoveries that have been reported, to date, for other methods. The recovery data were interpreted in terms of specific interactions, and the eluent composition was designed accordingly to maximize HAdV 40 recovery.

P

athogenic microorganisms cause many waterborne diseases, and the World Health Organization reports that water supply contamination leads to 842,000 deaths annually (1). Along with caliciviruses (e.g., norovirus), enteroviruses (e.g., poliovirus, coxsackievirus, and echovirus), and hepatitis A viruses, human adenovirus (HAdV) is one of the viral pathogens in EPA contaminant candidate lists 3 and 4 and is the second-leading cause of childhood gastroenteritis worldwide (2–4). Most cases of adenovirusassociated gastroenteritis stem from HAdV serotypes 40 and 41 (5), which are relatively resistant to environmental stressors (e.g., they survive longer in the environment than poliovirus, hepatitis A virus, and fecal indicator bacteria [6]) and treatment with UV irradiation (7). Monitoring HAdV 40 and 41 levels is essential to better understand their fate in the environment and in treatment systems, and eventually to prevent HAdV outbreaks. However, the low concentrations of HAdV in various natural waters, i.e., 101 to 104 genome copies (GC)/liter in river and lake water (8–10) and 103 to 105 GC/liter in wastewater treatment effluent (11, 12), make detection challenging and highlight the importance of sample concentration for rapid and reliable virus detection. Virus adsorption-elution (VIRADEL), the standard method for virus concentration, includes adsorption of virions on either electropositive or electronegative filters and subsequent elution of

4982

aem.asm.org

the adsorbed virions. Although recoveries by VIRADEL depend on the specific filter and eluent and vary greatly as a function of water composition, recoveries of adenoviruses are low relative to those of other viruses. For example, Gibbons et al. reported high recoveries of coliphages (96%) and noroviruses (100%) but very low recoveries of HAdV 41 (⬍3%) when they concentrated viruses from seawater using a NanoCeram filter and 3% (wt/vol) beef extract as the eluent (13). In another study, the coxsackievirus B5 recovery from tap water was 72%, whereas the recovery of HAdV 2 from the same water was only 39% (14). Low recoveries

Received 21 March 2016 Accepted 3 June 2016 Accepted manuscript posted online 10 June 2016 Citation Shi H, Xagoraraki I, Parent KN, Bruening ML, Tarabara VV. 2016. Elution is a critical step for recovering human adenovirus 40 from tap water and surface water by cross-flow ultrafiltration. Appl Environ Microbiol 82:4982–4993. doi:10.1128/AEM.00870-16. Editor: E. G. Dudley, Pennsylvania State University Address correspondence to Volodymyr V. Tarabara, [email protected]. Supplemental material for this article may be found at http://dx.doi.org/10.1128 /AEM.00870-16. Copyright © 2016, American Society for Microbiology. All Rights Reserved.

Applied and Environmental Microbiology

August 2016 Volume 82 Number 16

Elution Is Critical for Human Adenovirus Recovery

of adenovirus may occur because fibers associated with each penton base of the capsid facilitate the physical entrapment of virus particles in the filter matrix (13, 14). McMinn achieved a relatively high recovery of HAdV 40 with a 1MDS filter (using the VIRADEL method), but the standard deviation was also high (49% ⫾ 32%) (15). Another virus concentration method, cross-flow ultrafiltration (UF) (also called tangential-flow filtration), relies on size exclusion rather than adsorption to concentrate virions during filtration and might give higher recoveries than those obtained with the VIRADEL method. A UF membrane with pores smaller than the virion allows passage of water and low-molecular-weight compounds while maintaining virions in the recirculating retentate stream. The cross-flow should minimize membrane fouling and virion deposition onto the membrane. Nevertheless, significant virion deposition on membranes in cross-flow ultrafiltration systems is still observed (16–19). Similar to VIRADEL, elution of virions that adsorb to the UF membrane may be accomplished by using reagents, such as sodium polyphosphate (NaPP) and Tween 80, to disrupt electrostatic and hydrophobic interactions between virions and the membrane (18, 20, 21). UF can concentrate a variety of viruses, including bacteriophages (e.g., MS2 [20, 22], ␾X174 [20, 23], and P22 [19]) and human-pathogenic viruses (e.g., poliovirus [16, 24], echovirus [18], and norovirus [25]). Cross-flow filtration has the following three major advantages over VIRADEL: (i) the lower cost of UF membranes than VIRADEL filters, (ii) simultaneous concentration of multiple types of pathogens, and (iii) removal of small compounds that may inhibit quantitative PCR (qPCR) analysis. However, a fraction of the virions may still adsorb to the membrane, decreasing recoveries (26). To enhance recovery, UF membranes are typically coated with proteinaceous solutions (e.g., calf serum [CS]) to block potential virion adsorption sites and reduce virion adhesion (16, 17, 20, 21). However, long coating times and possible contamination during storage and transportation limit the application of CS-blocked membranes for rapid or field sampling (27). In a previous study, to improve virion recovery during cross-flow UF, we coated membranes with a polyelectrolyte multilayer (PEM) film prepared via rapid (⬍1 h) layer-by-layer adsorption of polycations (chitosan [CHI]) and polyanions (heparin [HE]). Compared with traditional CS-blocked membranes, the PEMcoated membranes showed higher recoveries of bacteriophage P22 from both deionized (DI) water and a membrane bioreactor effluent (19). Despite the need for improved HAdV concentration and recovery, cross-flow UF has not been evaluated comprehensively for this application. Pei et al. employed cross-flow UF to recover viruses, including HAdV 2, from tap water, but they coupled this process with VIRADEL. They reported a 42.4% recovery of HAdV 2 by VIRADEL only, whereas they did not determine the recovery of HAdV 2 by cross-flow UF (22). Skraber et al. determined the recovery of MS2, ␾X174, and poliovirus by cross-flow UF. They also processed environmental samples containing HAdV, but they did not know the initial concentration of HAdV and thus could not determine HAdV recovery (23). Nestola et al. reported HAdV 5 recovery from a prefiltered virus stock after virus harvest (28). Rhodes et al. examined recoveries of multiple viruses, including enteroviruses, ␾X174, and HAdV 41, from tap and river water by cross-flow ultrafiltration; postelution recoveries of HAdV 41 were 69.4% ⫾ 12.4% from tap water and 55.8% ⫾ 7.9% from river

August 2016 Volume 82 Number 16

water, while preelution recoveries were not reported (29). They concluded that in addition to size exclusion, other factors have an impact on the virus recovery process. Characterization of adenovirus physicochemical properties is important for designing efficient methods for their concentration. Point-of-zero-charge (PZC) values for adenoviral serotypes are available for HAdV 2 (PZC of 3.5 to 4.0 [30]) and HAdV 5 (PZC of ⬃4.5 [31]) but not for the enteric viruses HAdV 40 and HAdV 41. Moreover, no studies have quantified the surface energies of HAdV 40 and HAdV 41 virions. Recently, Wong et al. computed the extended Derjaguin-Landau-Verwey-Overbeek (XDLVO) interaction energies of HAdV 2 (30), but the calculations used surface energy parameters of surrogates. The goal of the present work was to improve the recovery of HAdV from water. We built on our earlier experimental and modeling study (19) to pursue the following two specific objectives: (i) to elucidate specific physicochemical mechanisms that control HAdV 40 adhesion to a membrane surface and (ii) to evaluate cross-flow UF for concentration and recovery of HAdV 40 from several water matrices. We employed two types of membranes to examine the effects of membrane properties on virion recovery. The approach included comprehensive physicochemical characterization of HAdV 40, evaluations of virion-virion and virionmembrane interactions using the XDLVO theory, and crossflow filtration tests with feed waters spiked with HAdV 40. The premise of the study was that understanding virion interactions can guide process design for improving preelution and postelution recoveries. MATERIALS AND METHODS Reagents and water samples. All reagents were of American Chemical Society (ACS) analytical-grade or higher purity. Heparin (sodium salt), chitosan (medium molecular weight), calf serum, EDTA (0.5 M in H2O), diiodomethane, sodium chloride, and sodium polyphosphate (NaPP) were purchased from Sigma-Aldrich (St. Louis, MO). Tween 80 was obtained from Fisher Scientific (Pittsburgh, PA), and glycerol was purchased from J. T. Baker (Center Valley, PA). All solutions were prepared using DI water with a resistivity of 18.2 M⍀ · cm (Barnstead NANOpure system). Tap water was supplied by the East Lansing-Meridian Water and Sewer Authority (East Lansing, MI) and used immediately after collection. Samples of surface water were collected from Lake Lansing at the boat ramp in Lake Lansing Park South (Haslett, MI) in October 2013 and April 2014. Surface water samples were filtered through 0.45-␮m membranes (mixed cellulose esters; Millipore, Billerica, MA) immediately after collection, stored at 4°C, and used within 1 week. All samples were characterized in terms of pH (Orion 3-Star Plus meter; Thermo Electron Corp., Waltham, MA), conductivity (Orion 550 conductivity meter; Thermo Electron Corp., Waltham, MA), total organic carbon (TOC) content, and concentrations of key cations. The TOC in each water sample was measured at least in triplicate (model 1010 analyzer; OI Analytical, College Station, TX). Concentrations of cations (Na⫹, K⫹, Mg2⫹, and Ca2⫹) were determined by inductively coupled plasma mass spectrometry (iCAP Q quadrupole ICP-MS; Thermo Scientific, Waltham, MA). Propagation, purification, and characterization of HAdV 40. HAdV 40 (ATCC VR-931) was purchased from ATCC (Manassas, VA) and propagated in A549 cells. The concentration of propagated virus, as measured by quantitative real-time PCR (qPCR), was 109 to 1010 GC/ml for all propagated batches. Virions were further purified using CsCl density gradient centrifugation following the protocol described by Mautner (32). See Section S1 in the supplemental material for a detailed description of virus growth and purification procedures. HAdV 40 is a human pathogen and should be manipulated following biosafety level 2 (BSL-2) criteria. HAdV 40 from the purified stock was used in all virion characteriza-

Applied and Environmental Microbiology

aem.asm.org

4983

Shi et al.

tion tests. The hydrodynamic diameter of the virion was determined by dynamic light scattering (DLS) (90 Plus; Brookhaven Instruments Corp., Holtsville, NY), and the ␨-potential of the virion was measured using phase analysis light scattering (ZetaPALS) (Brookhaven Instruments Corp., Holtsville, NY). To study the effects of pH on the aggregation state and ␨-potential of HAdV 40 virions, the solution pH was adjusted by the addition of NaOH and HCl, and a 1 mM NaCl solution served as the background electrolyte at all pH values except 7.6. The pH of an unbuffered NaCl solution could not be stabilized at pH 7.6, so a solution of 10 mM Tris-HCl and 1 mM EDTA (the recommended buffer for the storage of purified HAdV 40 [32]) was used as the background electrolyte. Transmission electron microscopy (TEM) was used to image HAdV 40. A few drops of purified virus stock were applied to a carbon-coated Formvar grid (Ted Pella, Redding, CA). The sample was rested on the grid for 1 min and then washed away by drops of water. Uranyl formate (1%) was then used to stain the grid, and excess stain was removed with filter paper. The grid was air-dried prior to TEM imaging. Images were recorded using a JEM-2200FS (JEOL, Tokyo, Japan) microscope with an in-column energy filter operated at 200 kV and equipped with a Direct Electron DE-20 camera. The surface energy of HAdV 40 was determined using a previously described protocol (19). Briefly, a virion lawn with ⬃7 layers of virions was prepared on a 30-kDa polyethersulfone membrane (Omega, Pall Corp., East Hills, NY) by filtering purified virion stock through the membrane. The virion-coated membrane was dried at room temperature until the water contact angle on the virion lawn plateaued, indicating formation of a monolayer of strongly bound water that resists evaporation. The contact angles of three probe liquids (DI water, glycerol, and diiodomethane) on the virion lawn were then measured with a goniometer (model 250; Ramé-Hart, Succasunna, NJ) by the sessile drop method to determine surface energy components of HAdV 40. XDLVO energies of virion-virion and virion-membrane interactions. XDLVO energy profiles were calculated (see Section S6 in the supplemental material for details) using the surface properties (charge and energy) of HAdV 40 and membranes. Membranes were characterized in our previous study (19). Membrane preparation. Prior to modification, polyethersulfone membranes (Omega; Pall Corp., East Hills, NY) with a molecular weight cutoff of 30,000 were soaked in 0.1 M NaOH for 3 h and then in DI water for 24 h at 4°C, with water exchange after 12 h, as recommended by the manufacturer. To coat a membrane with a PEM film, 1-mg/ml HE and 1-mg/ml CHI solutions were prepared in 0.15 M NaCl and adjusted to pH 5 using 1 M HCl and 1 M NaOH. Both polyelectrolyte solutions were stored at 4°C until use. The surface of the polyethersulfone membrane was alternately exposed to each polyelectrolyte solution for 5 min, with a 1-min DI water rinse between exposures. The deposition was initiated with the negatively charged HE solution. Alternating deposition of HE and CHI continued until 4.5 bilayers were obtained, with HE as the outermost (terminal) layer. To block a membrane with calf serum, the membrane was placed in a cross-flow filtration cell, and 500 ml of 5% (vol/vol) calf serum was circulated over the membrane overnight at 1.0 liter/min at room temperature, with no pressure applied across the membrane. After blocking, the membrane was rinsed with DI water twice at the same rate for 10 min. The PEM-coated and CS-blocked membranes were characterized in terms of charge and surface energy as described earlier (19). Virus concentration and recovery tests. Virions were recovered from DI water, tap water, and surface water. Prior to recovery experiments, 1-liter water samples spiked with 1 ml of HAdV 40 stock (virion concentration of ⬃106 to 107 GC/ml in the diluted solution) were recirculated in the filtration system without a membrane and with the permeate port blocked. The virus concentration in the feed tank was quantified after 1 h of recirculation (approximate time needed for filtration) and compared with that initially in the feed. No difference in virus concentration was observed, indicating no loss of virions to the filtration apparatus. Figure

4984

aem.asm.org

S3 in the supplemental material shows a schematic of the cross-flow filtration apparatus. The membrane was placed in the cross-flow cell (CF042P; Sterlitech, Kent, WA) and compacted for 90 min at 40 lb/in2 (⬃276 kPa). A 1-liter water sample was spiked with 1 ml of HAdV 40 stock (resulting in a virion concentration of ⬃106 to 107 GC/ml in the diluted solution), and 5 ml was collected to quantify virions in the feed sample. The remaining sample (996 ml) was pressurized using compressed N2 gas and delivered to the membrane cell by use of a peristaltic pump (model 621 CC; Watson Marlow, Wilmington, MA) at 2.0 liters/min. A digital flowmeter (model 101-7; McMillan, Georgetown, TX) monitored the cross-flow rate. The transmembrane pressure was maintained at ⬃20 lb/ in2 (⬃138 kPa) and continually monitored by a pressure transducer (0.5to 5.5-V output; Cole Parmer, Vernon Hills, IL). The ratios of cross-flow flux to initial permeate flux (Jcf/Jp) ranged from 5 ⫻ 103 to 6 ⫻ 103. Permeate was collected into a flask placed on a digital balance (Adventurer Pro AV8101C; Ohaus, Parsippany, NJ), and the permeate mass was recorded in real time to calculate the flux. The data from the pressure transducer, digital flowmeter, and digital balance were transmitted to a computer via a data acquisition module (PCI-6023E/SC-2345; National Instruments, Austin, TX). Filtration was stopped when 900 ml of sample was filtered. The remaining 96 ml of retentate was analyzed for virus content so that preelution recovery could be calculated. After filtration, virions were eluted from the membrane by using an eluent containing 0.01% NaPP and 0.01% Tween 80 (pH 6.5). For virion concentration and recovery from lake water, another eluent, containing 0.01% NaPP, 0.01% Tween 80, and 0.01% EDTA (pH 7), was also employed. Elution used 100 ml of eluent and was performed at a cross-flow rate of 2.0 liters/min for 20 min, with no transmembrane pressure applied. The feed, permeate, retentate, and eluate were all analyzed for virion concentration. Samples were also taken during membrane compaction to ensure that virions did not contaminate the feed water prior to the experiment. Virus quantification and recovery. HAdV 40 DNA in water samples was extracted by use of a MagNA Pure compact system (Roche Diagnostics USA, Indianapolis, IN) immediately after sample collection. qPCR was used to determine the HAdV 40 concentration (see Section S2 in the supplemental material for a description of the qPCR procedures). Possible effects of organic compounds in the feed water on virion quantification by qPCR were evaluated by performing inhibition tests. Virion preelution recovery (rpre), postelution recovery (rpost), and log removal (LRV) were calculated using the following equations: CrVr rpre ⫽ (1) CfVf rpost ⫽

CrVr ⫹ CeVe CfVf

LRV ⫽ log

10

冉

Cf ⫹ Cr 2Cp

(2)

冊

(3)

where Cf, Cr, Ce, and Cp represent virion concentrations in feed, retentate, eluate, and permeate samples, respectively, and Vf, Vr, Ve, and Vp are the volumes of these samples. In calculating the LRV (equation 3), the virus concentration in feed was calculated by averaging the initial concentration of HAdV 40 in feed prior to filtration and the concentration of the virus in retentate after filtration, i.e., ¯ Cf ⫽ 1/2(Cf ⫹ Cr). Thus, the reported virus removal (LRV) value is the average over the entire filtration process.

RESULTS AND DISCUSSION

Characteristics of source waters. Table 1 provides water quality parameters for DI water, tap water, and Lake Lansing water (sampled in spring and fall). The biggest differences among the waters were (i) much lower values for TOC, conductivity, and cation concentrations in DI water and (ii) high TOC and Ca2⫹ concentrations in the Lake Lansing water samples. Size, ␨-potential, and surface charge density of HAdV 40. Co-

Applied and Environmental Microbiology

August 2016 Volume 82 Number 16

Elution Is Critical for Human Adenovirus Recovery

TABLE 1 Water quality characteristics

Water source

pH range

Conductivity (␮S/cm)

Deionized water Tap water Lake water (fall) Lake water (spring)

5.7–6.0 7.5–8.0 7.0–7.5 7.5–7.8

NAa 319.5 ⫾ 27.6 336.3 ⫾ 12.7 369.3 ⫾ 32.0

a

TOC concn (mg/liter) NAa 1.1 ⫾ 0.1 7.5 ⫾ 0.2 9.4 ⫾ 0.1

Cation concn (mg/liter) Na⫹

K⫹

Mg2⫹

Ca2⫹

0.18 ⫾ 0.00 22.18 ⫾ 2.33 16.42 ⫾ 2.53 16.46 ⫾ 2.60

0.02 ⫾ 0.00 9.32 ⫾ 0.32 10.91 ⫾ 0.11 11.54 ⫾ 0.20

0.04 ⫾ 0.00 1.86 ⫾ 0.09 1.47 ⫾ 0.03 1.50 ⫾ 0.03

0.06 ⫾ 0.00 24.37 ⫾ 1.83 30.33 ⫾ 0.54 41.09 ⫾ 0.72

NA, not available (below detection limit).

liphages commonly serve as surrogates for human viruses (33). However, studies on virus removal and adsorption show that coliphages (e.g., MS2 [34, 35], ␾X174 [34, 35], and PRD1 [34]) are not suitable models for HAdV. Moreover, unique properties of serotypes 40 and 41 of enteric HAdV (e.g., the presence of short and long fibers on the capsid [36, 37], fastidiousness in cell culture [38–40] [though cultivation improvement was reported with a modified host cell line {41, 42}], and difficulty of isolation [40, 43]) make them different from other HAdV serotypes. This uniqueness makes detailed characterization of HAdV 40 and 41 necessary. Aggregation state and ␨-potential as functions of pH were reported for HAdV 2 (30) but not for HAdV 40. In this study, we measured the size of HAdV 40 by dynamic light scattering and compared the data with the results of TEM imaging. We also determined the ␨-potential and surface charge density of the virion as functions of pH. Based on TEM images (Fig. 1), the diameter of HAdV 40 is ⬃80 nm. Figure 2 shows the number-based hydrodynamic diameter distribution of HAdV 40 as a function of pH, and Table 2 summarizes the hydrodynamic diameters over a wide range of pH values. At pH 5.8 (unadjusted), pH 6.7, and pH 7.6, the number-based h average hydrodynamic diameters of HAdV 40 (d¯HAdV ) ranged from 94 ⫾ 3 nm to 102 ⫾ 9 nm, agreeing with typical size values reported earlier for HAdVs of different serotypes (44–46). The small increase in the hydrodynamic diameter relative to the diameter determined by TEM may have stemmed from fibers on the HAdV 40 surface (30, 47) that we did not take into account when estimating HAdV 40 diameters from TEM images. Such fibers

FIG 1 Transmission electron microscopy image of HAdV 40.

August 2016 Volume 82 Number 16

should, however, decrease particle diffusivity in light-scattering experiments. Additionally, shrinkage of HAdV 40 during negative staining may have decreased the diameters observed in TEM images (48). The polydispersity indexes of the particle diameters in the HAdV 40 suspensions at pH 5.8, 6.7, and 7.6 were 0.06 to 0.07, indicating minimal aggregation. However, between pH 4.0 and 4.7, both average size and polydispersity increased, consistent with aggregation. Aggregates of up to 1 ␮m appeared at pH 4.0 and 4.3 in several measurements. In contrast to the aggregation between pH 4.0 and 4.7, at pH 2.8, the virion hydrodynamic diameter was only 112 ⫾ 20 nm, suggesting minimal aggregation. This is consistent with studies showing that the enteric adenoviruses remain infective under the acidic conditions in the gut (49, 50). A high positive ␨-potential at low pH (Fig. 3) probably minimizes aggregation. At pH 9.7, the average hydrodynamic diameter of HAdV 40 (80 ⫾ 5 nm) was significantly lower than the values measured at pH 5.8 to 7.6 and close to the size determined by TEM imaging. A previous study observed inactivation of adenovirus after liming (51, 52), and the reduced size at pH 9.7 may have stemmed from virus degradation. Size reduction at pH 10 also occurred with HAdV 2 (30). Figure 3 presents the virion ␨-potentials determined from electrophoretic mobilities (see Fig. S9 in the supplemental material) by use of an approximate expression derived by Ohshima (53); the expression provides an accurate (⬍1% error) estimate for ␨-potentials for arbitrary values of ␬a, where a is the radius of the virion and ␬ is the Debye-Hückel parameter (see Section S9). Due to the small virion diameters, Smoluchowski’s expression for electrophoretic mobility, applicable for ␬a values of ⬎⬎1, is not accurate in our system. Thus, we used Ohshima’s expression. The ␨-potential decreased from ⫺22.0 mV to ⫺30.5 mV when the pH increased from 5.8 to 7.6. The negative surface charge on the virion likely minimizes aggregation, but for several measurements in this pH range, some aggregates (300 to 350 nm; 1.2% ⫾ 0.5% of the total population) appeared, possibly as an artifact of the dialysis process (32). Relatively small absolute values of the ␨-potential between pH 4.0 and 4.7 probably allow for the extensive aggregation described above. As Fig. 3 shows, the PZC for HAdV 40 is between pH 4.0 and 4.3. Favier et al. predicted the theoretical PZC for each major capsid protein of HAdV 40 by using the primary sequences, and they showed that the PZC values for the hexon, penton base, long fiber protein, and short fiber protein were ⬃6.7, 5.8, 7.5, and 7.8, respectively (49). All of these values are higher than the HAdV 40 PZC we determined, which is consistent with prior observations that the PZC for an intact virion is lower than the PZC values for its structural protein components (54). PZC values were reported to be in the range of 3.5 to 4.0 for HAdV 2 (30) and 4.5 for HAdV 5 (31), consistent with the low PZC of HAdV 40.

Applied and Environmental Microbiology

aem.asm.org

4985

Shi et al.

FIG 2 Normalized number-based size distribution for HAdV 40 suspension as a function of pH. Vertical red dashed lines indicate the average modal diameter (99 nm) (see Table S2 in the supplemental material) in the buffer recommended for the storage of purified HAdV 40 (10 mM Tris-HCl and 1 mM EDTA, pH 7.6). Vertical blue dashed-dotted lines denote the average diameter of HAdV virions (⬃80 nm) as determined by TEM. See Fig. S6 in the supplemental material for the size distribution for an HAdV 40 suspension in tap water.

At pH values at which virions aggregate, the measured ␨-potential is that for an aggregate of virions, not a single virion. In this case, we calculated virion ␨-potentials using an expression developed by Makino and Ohshima (55); the expression connects the surface charge density of a particle with its ␨-potential, diameter, and Debye length (see Section S10 in the supplemental material). The calculation assumes that surface charge density is an intensive property and does not depend on the aggregation state of the virion in the same solution. The average ␨-potential of HAdV 40 in tap water (pH 7.5 to 8.0) was ⫺17.7 mV, which is much higher than the ␨-potential (⫺30.5 mV) of HAdV 40 in a buffered solution with a similar pH (pH 7.6). The difference may stem from specific adsorption of divalent or multivalent ions onto virions in tap water.

Surface energy components of HAdV 40. To determine the surface energy components of HAdV 40, we applied the procedure we developed previously for examination of P22 bacteriophage (19). The technique includes measuring the contact angles of several liquids on a lawn of virions. As Table 3 shows, the apolar surface energy component (␥LW) of HAdV 40 was 41.6 mJ/m2, a value typical for biological materials (56). The electron donor (␥⫺) and electron acceptor (␥⫹) components of surface energy were 14.7 mJ/m2 and 0.01 mJ/m2, respectively, making the polar component of HAdV 40 surface energy (␥AB ⫽ 2兹␥⫹␥⫺) equal to 0.84 mJ/m2. Very small values of ␥⫹ were also reported for tobacco mosaic virus (56), bacteriophage P22 (19), and multiple proteins (56), suggesting that this is a common characteristic of

TABLE 2 Size distribution parameters for HAdV 40 suspension at different pH valuesa Value for water matrixb Parameter

1 mM NaCl

pH h d¯HAdV (nm) Polydispersity

2.8 112 ⫾ 20 0.22 ⫾ 0.06

a b c

4.0 284 ⫾ 99 0.31 ⫾ 0.04

4.3 238 ⫾ 69 0.31 ⫾ 0.01

4.7 149 ⫾ 3 0.10 ⫾ 0.01

5.8 94 ⫾ 3 0.06 ⫾ 0.01

6.7 95 ⫾ 3 0.07 ⫾ 0.02

Bufferc

Tap water

1 mM NaCl

7.6 103 ⫾ 9 0.07 ⫾ 0.01

7.5–8.0 109 ⫾ 14 0.11 ⫾ 0.01

9.7 80 ⫾ 5 0.12 ⫾ 0.03

Additional parameters (modal diameter and half-width at half-maximum) are given in Table S3 in the supplemental material. Shaded areas denote pH values where significant aggregation occurs. Buffer recommended for the storage of purified HAdV 40 (10 mM Tris-HCl and 1 mM EDTA).

4986

aem.asm.org

Applied and Environmental Microbiology

August 2016 Volume 82 Number 16

Elution Is Critical for Human Adenovirus Recovery

FIG 3 ␨-Potential of HAdV 40 as a function of pH.

proteins and nonenveloped viruses. Based on values of ␥LW, ␥⫹, and ␥⫺ for water and HAdV 40, the calculated polar adhesion energy between HAdV 40 and water is 79.7 mJ/m2, which is insufficient to overcome the energy of cohesion of water (⫺102.0 mJ/ m2). As a result, virion-virion interactions for HAdV 40 in water have a negative interfacial free energy (⌬Gvwv ⫽ ⫺30.4 mJ/m2), so HAdV 40 is hydrophobic. XDLVO energy profiles of virion-virion interfacial interactions in aqueous media. Several studies suggest that the classic Derjaguin-Landau-Verwey-Overbeek theory does not accurately predict the interactions between colloids and various surfaces in aqueous media because the calculations do not account for acidbase interactions that dominate at short separation distances (57, 58). In the present study, we employ the extended DLVO (XDLVO) model to evaluate virion-virion and virion-membrane interfacial interactions. Inputs to the model include the surface characteristics of HAdV 40 (see “Size, ␨-potential, and surface

TABLE 3 Contact angles, calculated surface energy parameters, and free energy of interfacial virion-virion interactions in water for HAdV 40a Parameter

Value

Contact angle (°) with indicated probe liquid H2O Glycerol Diiodomethane

68 ⫾ 2 64 ⫾ 1 36 ⫾ 2

Surface energy parameter (mJ/m2) ␥LW ␥⫹ ␥⫺ ␥AB ␥TOT

41.6 0.01 14.7 0.8 42.4

Free energy of interfacial virion-virion interactions in water (⌬Gvwv [mJ/m2])

⫺30.4

a Note that corresponding surface energies and free energies of interaction for CSblocked and PEM-coated membranes were reported previously (19).

August 2016 Volume 82 Number 16

FIG 4 XDLVO energy profiles of virion-virion interfacial interactions for HAdV 40 at different pHs.

charge density of HAdV 40” and “Surface energy components of HAdV 40”) and membranes (19). Figure 4 shows XDLVO energy profiles of virion-virion interfacial interactions at different pH values. For all pH values except 7.6, the calculations were performed for HAdV in 1 mM NaCl. For pH 7.6, size and ␨-potential values used in XDLVO calculations were for virions in the recommended storage buffer (1 mM EDTA plus 10 mM Tris). For pH values of ⬎5.8, the height of the energy barrier increases with pH, with the sole exception of pH 7.6. The deviation from the trend is due to the different background solution (10 mM Tris plus 1 mM EDTA instead of 1 mM NaCl) (see “Size, ␨-potential, and surface charge density of HAdV 40”). The negligible aggregation of virions for pH values of ⱖ5.8 is due to the high energy barrier (24.5 kT to 52.6 kT). At pH 4.7, where the energy barrier is 8.0 kT, significant aggregation occurs. At pH 4.0 and pH 4.3, the energy barrier decreases to below 1.2 kT, leading to further aggregation. Due to the high positive ␨-potential (29.4 mV) at pH 2.8, the energy barrier is 20.7 kT and the virion suspension is stable. XDLVO energies of virion-membrane interfacial interactions in membrane feed solutions made with DI water and tap water. XDLVO calculations for virion-membrane interactions in DI water were performed for an ionic strength of 0.2 mM; this is the approximate ionic strength determined from conductivity measurements of DI water spiked with HAdV 40 stock. However, the streaming potential of the membrane could be measured only with 1 mM NaCl electrolytes, so this value was used to approximate the membrane charge in the HAdV-spiked DI water. The approximation is reasonable because the electrophoretic mobilities of virions in DI water (⫺1.65 ⫾ 0.19 ␮m · S⫺1 · V⫺1 · cm at pH 5.8) and in 1 mM NaCl solution (⫺1.72 ⫾ 0.48 ␮m · S⫺1 · V⫺1 · cm at pH 5.8) were not statistically different.

Applied and Environmental Microbiology

aem.asm.org

4987

Shi et al.

FIG 5 XDLVO energies of interfacial interaction for HAdV 40 with PEM-coated (A) and CS-blocked (B) membranes in aqueous medium.

At separation distances of ⬎5 nm, van der Waals or electrostatic interactions dominate the XDLVO energy of interfacial interaction between HAdV 40 and membranes in both DI and tap water (see Fig. S7 and S8 in the supplemental material). However, as the separation distance increases, the magnitude of each interaction decreases. The magnitude of van der Waals interaction energy, for example, decreases to ⬍0.5 kT as the separation distance increases to approximately 15 nm. Over a shorter range (from the minimum equilibrium cutoff distance of ⬃0.16 nm to 0.7 nm), however, the acid-base interaction energy is significantly greater than both van der Waals and electrostatic interaction energies. In 1 mM NaCl at pH 5.8 (unadjusted), a PEM-coated membrane is negatively charged (␨ ⫽ ⫺7.0 ⫾ 3.0 mV) (19). Due to the hydrophilicity of the PEM as well as a repulsive electrostatic interaction between the virion and the membrane surface, the secondary minimum in the membrane-virion interaction energy profile is shallow (⫺4.2 kT) (Fig. 5A, solid line). The same membrane carries a positive charge (␨ ⫽ 5.6 ⫾ 0.4 mV) in tap water. The charge reversal may stem from specific adsorption of various cations from tap water onto the membrane surface, and a similar charge reversal occurred with latex particles as the solution ionic strength increased (59, 60). For the membrane-virion interaction energy profile for tap water, electrostatic attraction coupled with van der Waals attraction at large separation distances and a predominantly repulsive acid-base interaction at small separation distances yields a secondary minimum of ⫺10.3 kT at a separation distance of 3.2 nm (Fig. 5A, dashed line). This secondary minimum may lead to reversible adsorption of HAdV 40 onto the membrane surface during tap water filtration. Even though the PZC of bovine serum albumin, a major component of calf serum, is in the range of 4.7 to 5.6 (61–63), CSblocked membranes in a 1 mM NaCl solution at pH 5.8 (unadjusted) carry a weak positive charge (␨ ⫽ 3.0 ⫾ 2.0 mV) (19). The small positive charge on these membranes may stem from other components in calf serum, such as bovine IgG (PZC range of 7.5 to 8.3 [64]). The XDLVO energy profile for the interaction of HAdV 40 virions and a CS-blocked membrane shows a secondary mini-

4988

aem.asm.org

mum of ⫺48.2 kT at a separation distance of 2.4 nm (Fig. 5B, solid line), which suggests deposition of a significant amount of virions. In tap water, the ␨-potential of CS-blocked membranes is 2.8 ⫾ 1.0 mV, similar to that in 1 mM NaCl. Compared to experiments performed with DI water, the reduced negative charge of HAdV 40 in tap water (Fig. 3) leads to less virion-membrane electrostatic attraction, and the depth of the secondary minimum in tap water is only ⫺7.9 kT for interactions of the virion and the CS-blocked membrane (Fig. 5B, dashed line). Virus recovery from DI water. Preelution virion recovery (see equation 1) after cross-flow UF of DI water spiked with virions was significantly greater with a PEM-coated membrane (74.8% ⫾ 9.7%) than with a CS-blocked membrane (54.1% ⫾ 6.2%). The higher recovery likely stems from the negative charge and hydrophilicity of the PEM-coated membrane (19). However, the ⬍100% recovery indicates that some virion adsorption takes place on the PEM-modified membrane, even though the secondary minimum in the XDLVO energy profile is shallow (Fig. 5A, solid line). This adsorption may stem from microscale attraction (56, 65), whereas XDLVO theory describes only macroscopic interactions. For example, in experiments on Burkholderia cepacia adhesion, Hwang et al. reported bacterial adhesion in the absence of a secondary minimum in the XDLVO energy profile and suggested that the adhesion stems from cell appendages, such as pili and flagella (66). HAdV 40 fibers, which consist of long, thin shafts terminated with globular knobs, may extend beyond the characteristic length of macroscopic repulsion (approximately the same as the Debye length) to achieve microscopic attraction between a virion and a membrane. Interactions such as those between electron donor sites on the fiber knob and electron acceptor sites on the membrane may overcome the macroscopic repulsion and lead to local attraction. Debye lengths for HAdV 40 at different pH values in 1 mM NaCl (see Table S2 in the supplemental material) are shorter than the lengths of both longer and shorter fibers (⬃30 nm and ⬃18 nm, respectively [36]) on HAdV 40. In HAdV-spiked DI water, the Debye length is ⬃23 nm (see Table S2), which is still shorter than the 30-nm HAdV fibers. Other possible explanations

Applied and Environmental Microbiology

August 2016 Volume 82 Number 16

Elution Is Critical for Human Adenovirus Recovery

FIG 6 Recoveries of HAdV 40 from deionized water (A) and tap water (B) by use of CS-blocked and PEM-coated membranes.

for the incomplete preelution recovery with PEM-coated membranes include local electrostatic attraction due to charge heterogeneity and the presence of non-XDLVO interactions (e.g., bridging by multivalent cations). The relatively low virion recovery from DI water after crossflow UF with CS-blocked membranes likely stems from the secondary minimum in the XDLVO energy profile (Fig. 5B, solid line). The secondary minimum is sufficiently deep (⫺48.2 kT) to capture virions with an average thermal energy of ⬃0.5 kT (67). The fact that preelution recovery after cross-flow UF is higher with PEM-coated membranes than with CS-blocked membranes is consistent with our previous findings for cross-flow filtration of bacteriophage P22 in DI water, where PEM-coated membranes gave significantly higher preelution recovery (80%) than that with CS-blocked membranes (30%) (19). After filtration, we eluted virions to increase recovery. For both PEM-coated and CS-blocked membranes, eluents containing 0.01% NaPP and 0.01% Tween 80 give high postelution recoveries (see equation 2) of 99.5% ⫾ 6.6% for PEM-coated membranes and 98.8% ⫾ 7.7% for CS-blocked membranes. During elution, adsorption of NaPP on the membrane likely increases the negative charge on the surface to enhance electrostatic repulsion between the negatively charged virion and the membrane. Additionally, Tween 80, a nonionic surfactant, should minimize hydrophobic interactions between the membrane surface and adsorbed virions. Thus, both NaPP and Tween 80 may help to release virions from the membrane. The relatively high feed concentration of adenovirus (⬃106 to 7 10 GC/ml) in our study may have led to a higher recovery and a smaller variance than those for filtration of lower virion concentrations, because at high concentrations, the virion may saturate sorption sites. However, we had to use a relatively high virion concentration to accurately quantify the low concentration of virions in the permeate (and to quantify both virion removal and the fraction of virions adsorbed on the membrane). The HAdV 40 removal rate was high with both CS-blocked membranes (LRV values of up to 2.17) and PEM-coated membranes (LRV values of up to 4.64), so the permeate concentration was much lower than the feed concentration. Virus recovery from tap and surface water. We also evaluated

August 2016 Volume 82 Number 16

the recovery of HAdV 40 from tap water and from lake water collected in two different seasons. Because organic molecules in surface waters may inhibit qPCR detection of virions (68, 69), we initially performed inhibition tests (see Section S5 in the supplemental material for details). Briefly, DI, tap, and surface water samples were spiked with HAdV 40 from the stock suspension, viral DNA was extracted, and the virion concentration in each water matrix was determined by qPCR. Consistent with minimal qPCR inhibition, we observed no differences in virion concentrations in the three water matrices. With tap water, we saw no statistically significant difference in preelution virion recovery after cross-flow UF with PEM-coated membranes (40.5% ⫾ 9.9%) and CS-blocked membranes (38.3% ⫾ 9.3%). Based on XDLVO predictions for CS-blocked membranes (Fig. 5B), the preelution recovery from tap water should have been higher than that from DI water. However, the opposite occurred—for CS-blocked membranes, the average preelution recovery from tap water was ⬃16% lower than that from DI water (Fig. 6). The discrepancy suggests that nonXDLVO effects (e.g., steric interactions [70] or Ca2⫹ bridging [71]) should be considered. Another possible cause for the discrepancy is virion adsorption to dissolved and suspended species present in the tap water (TOC concentration ⫽ 1.1 ⫾ 0.1 mg/liter) and on the membrane surface. Permeate flux declined ⬃15% to 20% during cross-flow UF of tap water (see Fig. S4 in the supplemental material), suggesting the formation of a fouling layer on the membrane surface. The preelution recovery from tap water after cross-flow UF with the PEM-coated membrane was ⬃34% lower than that from DI water (Fig. 6). In addition to the possible effects described above for the CS-blocked membranes, charge reversal of the PEM membrane in tap water may have decreased the preelution recovery. Preelution recovery from tap water had a larger variance, with values ranging from 30% to 50% for both types of membranes. Such variances likely stem from variations in tap water quality over several months of sample collection and tests. Surface water is a complex matrix containing naturally occurring organic and inorganic species, both dissolved and colloidal. Preelution recoveries from surface water collected in the fall were ⬃40% with both PEM- and CS-coated membranes, similar to

Applied and Environmental Microbiology

aem.asm.org

4989

Shi et al.

FIG 7 Recoveries of HAdV 40 from surface water by use of CS-blocked (A) and PEM-coated (B) membranes.

recoveries from tap water but significantly lower than recoveries from DI water. The TOC concentration in surface water collected in the fall was 7.5 ⫾ 0.2 mg/liter. For surface water, permeate flux declined more than 50% over 90 min of cross-flow UF (see Fig. S4 in the supplemental material) with membranes of each type. Foulants formed a cake layer and masked the antiadhesive properties of the membrane surface, leading to the deposition of virions on membranes. Virions could also be adsorbed to various components of the feed water (e.g., humic acid, clay, or silica particles) and codeposited on the membrane. Postelution recovery was higher for tap water (Fig. 6B) than for surface water, suggesting that fouling also decreased the effectiveness of the elution process. Many studies have shown that divalent cations, such as Ca2⫹ (30.3 ⫾ 0.5 mg/liter in the surface water collected in the fall), enhance deposition of virions on natural organic matter via innersphere complexation with carboxyl groups on both surfaces (72, 73) and on clay and silica by screening the charge (74). Calcium bridging may also occur between carboxyl groups on natural organic matter and on the HAdV 40 capsid (e.g., carboxylic groups on fiber knobs), leading to virion loss to the membrane surface. Bridging by other cations may also occur (e.g., by iron [75], copper [76], and aluminum [77]). In general, virion-foulant interactions rather than virion-membrane interactions likely govern recovery from complex waters by UF. This is consistent with the result that preelution recoveries from surface water were not statistically different for the two membrane types (Fig. 7). We also analyzed virion recovery from the surface water collected from the same lake in early spring, after the ice melted. Flux decline also occurred for both membranes during filtration of spring surface water (see Fig. S4 in the supplemental material), and preelution recoveries with both membranes were ⬃20%. The lower recovery from spring surface water than from fall surface water may have stemmed from higher concentrations of Ca2⫹ (41.1 ⫾ 0.7 mg/liter) and TOC (9.4 ⫾ 0.1 mg/liter) (Table 1). Postelution recoveries with CS-blocked membranes were 61.0% ⫾ 2.8% and 34.9% ⫾ 10.1% for water samples collected in the fall and spring, respectively. For PEM-coated membranes, postelution recoveries were 62.4% ⫾ 2.2% and 41.6% ⫾ 2.0% for fall and spring samples. Thus, virions adsorbed from surface water

4990

aem.asm.org

on both membranes were not eluted as effectively (Fig. 7) as those adsorbed from DI or tap water (Fig. 6) with an eluent containing NaPP and Tween 80 only. To increase the elution efficiency, we added 0.01% (wt/wt) EDTA to the eluent to complex Ca2⫹ (78). EDTA (1 mM) is a component of the storage buffer for HAdV 40, so it should not inactivate the virus (32). With EDTA in the eluent, the postelution recovery from spring and fall surface water, averaged over all PEM-coated and CS-blocked membranes, increased to 84.3% ⫾ 4.5%. The increased elution efficiency is consistent with the hypothesis that calcium binding decreases virion recovery from surface water. In a previous study, we showed that a PEM consisting of a weak polycation and a strong polyanion disassembles at low pH (79). Because HAdV 40 tolerates acidic conditions, such a PEM may enable a simple additional processing step to improve the overall recovery: disassembling the PEM after elution may release trapped virions to potentially achieve 100% recovery. Virus removal. As Fig. 8 shows, in tests with HAdV 40-spiked waters of all types, virion removal (see equation 3) by PEM-coated membranes was typically an order of magnitude higher than re-

FIG 8 Removal of HAdV 40 by cross-flow ultrafiltration with CS-blocked membranes and PEM-coated membranes.

Applied and Environmental Microbiology

August 2016 Volume 82 Number 16

Elution Is Critical for Human Adenovirus Recovery

moval by CS-blocked membranes. However, the loss of 1% of virions to the permeate by CS-blocked membranes should not significantly decrease recoveries. The low LRV by CS-blocked membranes might result from an uneven calf serum distribution on the membrane surface and resultant incomplete blockage. The blocking procedure involves recirculating a calf serum solution over the membrane surface and may be viewed as poorly controlled membrane fouling. In contrast, coating a membrane with multiple layers of polyelectrolytes is inherently more reproducible. There was no consistent correlation between LRV values and the extent of membrane fouling. We showed recently that HAdV 40 removal does not correlate simply with the extent of fouling alone. Pore blockage by humic acid could enhance virion removal, whereas cake formation by silica colloids could decrease virion removal, and the opposing effects may be compensatory when both foulants are present (80). Surface water contains colloids and organic macromolecules of a range of sizes, which may explain the similar LRV values for surface water and DI water. Conclusions. Physicochemical characterization of HAdV 40 showed that this relatively large virus (hydrodynamic diameter of 102.5 ⫾ 8.6 nm at pH 7.6 in 10 mM Tris plus 1 mM EDTA) is hydrophobic (⌬Gvwv ⫽ ⫺30.4 mJ/m2) and has a point of zero charge between pH 4.0 and pH 4.3. The extended Derjaguin-Landau-Verwey-Overbeek theory predicted that hydrophilic and negatively charged (␨ ⫽ ⫺7.0 ⫾ 3.0 mV at pH 5.8 in 1 mM NaCl) PEM-coated membranes should be advantageous for recovering HAdV from DI water in comparison with positively charged CScoated membranes (␨ ⫽ 3.0 ⫾ 2.0 mV at pH 5.8 in 1 mM NaCl). We tested the validity of the XDLVO prediction in filtration experiments. The preelution recovery of HAdV from DI water (ionic strength of ⬃0.2 mM when spiked with HAdV stock) after PEM ⫽ cross-flow UF was higher with PEM-coated membranes (rpre CS 74.8% ⫾ 9.7%) than with CS-blocked membranes (rpre ⫽ 54.1% ⫾ 6.2%). Although preelution recovery from tap water was lower, virion elution from both PEM- and CS-coated membranes with an aqueous solution of sodium polyphosphate and Tween 80 PEM CS ⫽ 99.5% ⫾ 6.6% and rpost ⫽ was effective for both DI water (rpost PEM CS 98.8% ⫾ 7.7%) and tap water (rpost ⫽ 88.8% ⫾ 15.3% and rpost ⫽ 93.7% ⫾ 6.9%). The nearly 100% efficacy of elution indicates that polyanions and surfactants (e.g., sodium polyphosphate and Tween 80) in the eluent can disrupt electrostatic and hydrophobic interactions between the virion and the membrane. Pre- and postelution recoveries from surface waters were significantly CS PEM CS PEM and rpre values as low as ⬃21% and rpost and rpost lower (rpre values of ⬍65%) and showed no statistically significant difference between the two membrane types. However, addition of EDTA to CS the eluent greatly increased the elution efficacy (rpost ⫽ 88.6% ⫾ PEM 4.3% and rpost ⫽ 87.0% ⫾ 6.9%), possibly by eliminating cation bridging between virions and other components of the feed water matrix in suspension or in the fouling layer on the membrane surface. Interestingly, the membrane choice is not very important for achieving high virion recoveries. For tap water, the postelution virion recovery was nearly 100% for both PEM- and CS-coated membranes, despite the higher removal of virions with PEMcoated membranes (LRVPEM ⫽ 2.82 ⫾ 0.32 and LRVCS ⫽ 1.96 ⫾ 0.53 for tap water). For more complex water matrices, such as surface water, the composition of the eluent is the most important

August 2016 Volume 82 Number 16

factor for achieving high virion recovery. Evidently, the recovery of HAdV depends on its interactions with other components of the feed water (in suspension or deposited on the membrane surface as a fouling layer), not on virion-membrane interactions. An eluent that includes a polyanion (sodium polyphosphate), a nonionic surfactant (Tween 80), and a chelating agent (EDTA) recovers HAdV effectively (⬃88%), even from high-TOC (9.4 ⫾ 0.1 mg/liter) surface water. ACKNOWLEDGMENTS We thank Jason Schrad for his assistance with TEM imaging and two anonymous reviewers for their comments and suggestions. This material is based upon work supported by the National Science Foundation Partnerships for International Education and Research program under grant IIA-1243433.

FUNDING INFORMATION This work, including the efforts of Hang Shi, Irene Xagoraraki, Kristin N. Parent, Merlin L. Bruening, and Volodymyr V. Tarabara, was funded by National Science Foundation (NSF) (IIA-1243433).

REFERENCES 1. WHO. 2016. Water sanitation health. Burden of disease and costeffectiveness estimates. WHO, Geneva, Switzerland. http://www.who.int /water_sanitation_health/diseases/burden/en/. Accessed 12 February 2016. 2. Crabtree KD, Gerba CP, Rose JB, Haas CN. 1997. Waterborne adenovirus: a risk assessment. Water Sci Technol 35:1– 6. 3. Mena K, Gerba C. 2009. Waterborne adenovirus. Rev Environ Contam Toxicol 198:133–167. http://dx.doi.org/10.1007/978-0-387-09647-6_4. 4. Fongaro G, do Nascimento MA, Rigotto C, Ritterbusch G, da Silva AD, Esteves PA, Barardi CRM. 2013. Evaluation and molecular characterization of human adenovirus in drinking water supplies: viral integrity and viability assays. Virol J 10:166. http://dx.doi.org/10.1186/1743-422X-10 -166. 5. Jiang SC. 2006. Human adenoviruses in water: occurrence and health implications: a critical review. Environ Sci Technol 40:7132–7140. http: //dx.doi.org/10.1021/es060892o. 6. Enriquez CE, Hurst CJ, Gerba CP. 1995. Survival of the enteric adenovirus-40 and adenovirus-41 in tap, sea, and waste-water. Water Res 29: 2548 –2553. http://dx.doi.org/10.1016/0043-1354(95)00070-2. 7. Thurston-Enriquez JA, Haas CN, Jacangelo J, Riley K, Gerba CP. 2003. Inactivation of feline calicivirus and adenovirus type 40 by UV radiation. Appl Environ Microbiol 69:577–582. http://dx.doi.org/10.1128/AEM.69 .1.577-582.2003. 8. Choi S, Jiang SC. 2005. Real-time PCR quantification of human adenoviruses in urban rivers indicates genome prevalence but low infectivity. Appl Environ Microbiol 71:7426 –7433. http://dx.doi.org/10.1128/AEM .71.11.7426-7433.2005. 9. Xagoraraki I, Kuo DHW, Wong K, Wong M, Rose JB. 2007. Occurrence of human adenoviruses at two recreational beaches of the Great Lakes. Appl Environ Microbiol 73:7874 –7881. http://dx.doi.org/10.1128/AEM .01239-07. 10. Albinana-Gimenez N, Miagostovich MP, Calqua B, Huguet JM, Matia L, Girones R. 2009. Analysis of adenoviruses and polyomaviruses quantified by qPCR as indicators of water quality in source and drinking-water treatment plants. Water Res 43:2011–2019. http://dx.doi.org/10.1016/j .watres.2009.01.025. 11. Fong TT, Phanikumar MS, Xagoraraki I, Rose JB. 2010. Quantitative detection of human adenoviruses in wastewater and combined sewer overflows influencing a Michigan river. Appl Environ Microbiol 76:715– 723. http://dx.doi.org/10.1128/AEM.01316-09. 12. Osuolale O, Okoh A. 2015. Incidence of human adenoviruses and hepatitis A virus in the final effluent of selected wastewater treatment plants in Eastern Cape Province, South Africa. Virol J 12:98. http://dx.doi.org/10 .1186/s12985-015-0327-z. 13. Gibbons CD, Rodriguez RA, Tallon L, Sobsey MD. 2010. Evaluation of positively charged alumina nanofibre cartridge filters for the primary concentration of noroviruses, adenoviruses and male-specific coliphages

Applied and Environmental Microbiology

aem.asm.org

4991

Shi et al.

14.

15. 16. 17.

18.

19.

20.

21.

22.

23.

24. 25.

26. 27.

28.

29.

30.

from seawater. J Appl Microbiol 109:635– 641. http://dx.doi.org/10.1111 /j.1365-2672.2010.04691.x. Ikner LA, Soto-Beltran M, Bright KR. 2011. New method using a positively charged microporous filter and ultrafiltration for concentration of viruses from tap water. Appl Environ Microbiol 77:3500 –3506. http://dx .doi.org/10.1128/AEM.02705-10. McMinn BR. 2013. Optimization of adenovirus 40 and 41 recovery from tap water using small disk filters. J Virol Methods 193:284 –290. http://dx .doi.org/10.1016/j.jviromet.2013.06.021. Berman D, Rohr ME, Safferman RS. 1980. Concentration of poliovirus in water by molecular filtration. Appl Environ Microbiol 40:426 – 428. Morales-Morales HA, Vidal G, Olszewski J, Rock CM, Dasgupta D, Oshima KH, Smith GB. 2003. Optimization of a reusable hollow-fiber ultrafilter for simultaneous concentration of enteric bacteria, protozoa, and viruses from water. Appl Environ Microbiol 69:4098 – 4102. http://dx .doi.org/10.1128/AEM.69.7.4098-4102.2003. Polaczyk AL, Narayanan J, Cromeans TL, Hahn D, Roberts JM, Amburgey JE, Hill VR. 2008. Ultrafiltration-based techniques for rapid and simultaneous concentration of multiple microbe classes from 100-L tap water samples. J Microbiol Methods 73:92–99. http://dx.doi.org/10.1016 /j.mimet.2008.02.014. Pasco EV, Shi H, Xagoraraki I, Hashsham SA, Parent KN, Bruening ML, Tarabara VV. 2014. Polyelectrolyte multilayers as anti-adhesive membrane coatings for virus concentration and recovery. J Membr Sci 469:140 –150. http://dx.doi.org/10.1016/j.memsci.2014.06.032. Hill VR, Kahler AM, Jothikumar N, Johnson TB, Hahn D, Cromeans TL. 2007. Multistate evaluation of an ultrafiltration-based procedure for simultaneous recovery of enteric microbes in 100-liter tap water samples. Appl Environ Microbiol 73:4218 – 4225. http://dx.doi.org/10.1128/AEM .02713-06. Liu PB, Hill VR, Hahn D, Johnson TB, Pan Y, Jothikumar N, Moe CL. 2012. Hollow-fiber ultrafiltration for simultaneous recovery of viruses, bacteria and parasites from reclaimed water. J Microbiol Methods 88:155– 161. http://dx.doi.org/10.1016/j.mimet.2011.11.007. Pei L, Rieger M, Lengger S, Ott S, Zawadsky C, Hartmann NM, Selinka HC, Tiehm A, Niessner R, Seidel M. 2012. Combination of crossflow ultrafiltration, monolithic affinity filtration, and quantitative reverse transcriptase PCR for rapid concentration and quantification of model viruses in water. Environ Sci Technol 46:10073–10080. http://dx.doi.org /10.1021/es302304t. Skraber S, Gantzer C, Helmi K, Hoffmann L, Cauchie HM. 2009. Simultaneous concentration of enteric viruses and protozoan parasites: a protocol based on tangential flow filtration and adapted to large volumes of surface and drinking waters. Food Environ Virol 1:66 –76. http://dx.doi .org/10.1007/s12560-009-9011-z. Belfort G, Rotem Y, Katzenelson E. 1975. Virus concentration using hollow fiber membranes. Water Res 9:79 – 85. http://dx.doi.org/10.1016 /0043-1354(75)90155-4. Francy DS, Stelzer EA, Brady AM, Huitger C, Bushon RN, Ip HS, Ware MW, Villegas EN, Gallardo V, Lindquist HD. 2013. Comparison of filters for concentrating microbial indicators and pathogens in lake water samples. Appl Environ Microbiol 79:1342–1352. http://dx.doi.org/10 .1128/AEM.03117-12. Ikner LA, Gerba CP, Bright KR. 2012. Concentration and recovery of viruses from water: a comprehensive review. Food Environ Virol 4:41– 67. http://dx.doi.org/10.1007/s12560-012-9080-2. Hill VR, Polaczyk AL, Hahn D, Narayanan J, Cromeans TL, Roberts JM, Amburgey JE. 2005. Development of a rapid method for simultaneous recovery of diverse microbes in drinking water by ultrafiltration with sodium polyphosphate and surfactants. Appl Environ Microbiol 71: 6878 – 6884. http://dx.doi.org/10.1128/AEM.71.11.6878-6884.2005. Nestola P, Martins DL, Peixoto C, Roederstein S, Schleuss T, Alves PM, Mota JPB, Carrondo MJT. 2014. Evaluation of novel large cut-off ultrafiltration membranes for adenovirus serotype 5 (Ad5) concentration. PLoS One 9:e115802. http://dx.doi.org/10.1371/journal.pone.0115802. Rhodes ER, Huff EM, Hamilton DW, Jones JL. 2016. The evaluation of hollow-fiber ultrafiltration and celite concentration of enteroviruses, adenoviruses and bacteriophage from different water matrices. J Virol Methods 228:31–38. http://dx.doi.org/10.1016/j.jviromet.2015.11.003. Wong K, Mukherjee B, Kahler AM, Zepp R, Molina M. 2012. Influence of inorganic ions on aggregation and adsorption behaviors of human adenovirus. Environ Sci Technol 46:11145–11153. http://dx.doi.org/10.1021 /es3028764.

4992

aem.asm.org

31. Trilisky EI, Lenhoff AM. 2007. Sorption processes in ion-exchange chromatography of viruses. J Chromatogr A 1142:2–12. http://dx.doi.org/10 .1016/j.chroma.2006.12.094. 32. Mautner V. 2007. Growth and purification of enteric adenovirus type 40, p 145–156. In Wold WSM, Tollefson AE (ed), Adenovirus methods and protocols. Humana Press, Totowa, NJ. 33. Mesquita MMF, Emelko MB. 2014. Bacteriophages as surrogates for the fate and transport of pathogens in source water and in drinking water treatment processes, p 57– 80. In Kurtboke I (ed), Bacteriophages. Intech, Rijeka, Croatia. 34. Abbaszadegan M, Mayer BK, Ryu H, Nwachuku N. 2007. Efficacy of removal of CCL viruses under enhanced coagulation conditions. Environ Sci Technol 41:971–977. http://dx.doi.org/10.1021/es061517z. 35. Bellou MI, Syngouna VI, Tselepi MA, Kokldnos PA, Paparrodopoulos SC, Vantarakis A, Chrysikopoulos CV. 2015. Interaction of human adenoviruses and coliphages with kaolinite and bentonite. Sci Total Environ 517:86 –95. http://dx.doi.org/10.1016/j.scitotenv.2015.02.036. 36. Kidd AH, Chroboczek J, Cusack S, Ruigrok RWH. 1993. Adenovirus type-40 virions contain 2 distinct fibers. Virology 192:73– 84. http://dx.doi .org/10.1006/viro.1993.1009. 37. Song JD, Liu XL, Chen DL, Zou XH, Wang M, Qu JG, Lu ZZ, Hung T. 2012. Human adenovirus type 41 possesses different amount of short and long fibers in the virion. Virology 432:336 –342. http://dx.doi.org/10.1016 /j.virol.2012.05.020. 38. Gary GW, Jr, Hierholzer JC, Black RE. 1979. Characteristics of noncultivable adenoviruses associated with diarrhea in infants: a new subgroup of human adenoviruses. J Clin Microbiol 10:96 –103. 39. De Jong JC, Kapsenberg JG, Muzerie CJ, Wermenbol AG, Kidd AH, Wadell G, Firtzlaff RG, Wigand R. 1983. Candidate adenoviruses 40 and 41: fastidious adenoviruses from human infant stool. J Med Virol 11:215– 231. http://dx.doi.org/10.1002/jmv.1890110305. 40. De Jong JC, Bijlsma K, Wermenbol AG, Verweij-Uijterwaal MW, Van der Avoort H, Wood DJ, Bailey AS, Osterhaus A. 1993. Detection, typing, and subtyping of enteric adenoviruses 40 and 41 from fecal samples and observation of changing incidences of infections with these types and subtypes. J Clin Microbiol 31:1562–1569. 41. Sherwood V, Burgert H-G, Chen Y-H, Sanghera S, Katafigiotis S, Randall RE, Connerton I, Mellits KH. 2007. Improved growth of enteric adenovirus type 40 in a modified cell line that can no longer respond to interferon stimulation. J Gen Virol 88:71–76. http://dx.doi.org/10.1099 /vir.0.82445-0. 42. Kim M, Lim MY, Ko G. 2010. Enhancement of enteric adenovirus cultivation by viral transactivator proteins. Appl Environ Microbiol 76: 2509 –2516. http://dx.doi.org/10.1128/AEM.02224-09. 43. Tiemessen CT, Kidd AH. 1995. The subgroup F adenoviruses. J Gen Virol 76:481– 497. http://dx.doi.org/10.1099/0022-1317-76-3-481. 44. Wilhelmi I, Roman E, Sanchez-Fauquier A. 2003. Viruses causing gastroenteritis. Clin Microbiol Infect 9:247–262. http://dx.doi.org/10.1046/j .1469-0691.2003.00560.x. 45. Kennedy MA, Parks RJ. 2009. Adenovirus virion stability and the viral genome: size matters. Mol Ther 17:1664 –1666. http://dx.doi.org/10.1038 /mt.2009.202. 46. Nemerow GR, Stewart PL, Reddy VS. 2012. Structure of human adenovirus. Curr Opin Virol 2:115–121. http://dx.doi.org/10.1016/j.coviro .2011.12.008. 47. Gutierrez L, Mylon SE, Nash B, Nguyen TH. 2010. Deposition and aggregation kinetics of rotavirus in divalent cation solutions. Environ Sci Technol 44:4552– 4557. http://dx.doi.org/10.1021/es100120k. 48. Putaux JL, Buleon A, Borsali R, Chanzy H. 1999. Ultrastructural aspects of phytoglycogen from cryo-transmission electron microscopy and quasielastic light scattering data. Int J Biol Macromol 26:145–150. http://dx.doi .org/10.1016/S0141-8130(99)00076-8. 49. Favier AL, Burmeister WP, Chroboczek J. 2004. Unique physicochemical properties of human enteric Ad41 responsible for its survival and replication in the gastrointestinal tract. Virology 322:93–104. http://dx .doi.org/10.1016/j.virol.2004.01.020. 50. Lu Z-Z, Zou X-H, Dongi L-X, Qu J-G, Song J-D, Wang M, Guo L, Hung T. 2009. Novel recombinant adenovirus type 41 vector and its biological properties. J Gene Med 11:128 –138. http://dx.doi.org/10.1002/jgm.1284. 51. Hansen JJ, Warden PS, Margolin AB. 2007. Inactivation of adenovirus type 5, rotavirus Wa and male specific coliphage (MS2) in biosolids by lime stabilization. Int J Environ Res Public Health 4:61– 67. http://dx.doi .org/10.3390/ijerph2007010010.

Applied and Environmental Microbiology

August 2016 Volume 82 Number 16

Elution Is Critical for Human Adenovirus Recovery

52. Bean CL, Hansen JJ, Margolin AB, Balkin H, Batzer G, Widmer G. 2007. Class B alkaline stabilization to achieve pathogen inactivation. Int J Environ Res Public Health 4:53– 60. http://dx.doi.org/10.3390 /ijerph2007010009. 53. Ohshima H. 1994. A simple expression for Henry’s function for the retardation effect in electrophoresis of spherical colloidal particles. J Colloid Interface Sci 168:269 –271. http://dx.doi.org/10.1006/jcis.1994.1419. 54. Norrby E. 1968. Biological significance of structural adenovirus components. Curr Top Microbiol Immunol 43:1– 43. 55. Makino K, Ohshima H. 2010. Electrophoretic mobility of a colloidal particle with constant surface charge density. Langmuir 26:18016 –18019. http://dx.doi.org/10.1021/la1035745. 56. van Oss CJ. 2006. Interfacial forces in aqueous media. CRC Press, Boca Raton, FL. 57. van Oss CJ. 2003. Long-range and short-range mechanisms of hydrophobic attraction and hydrophilic repulsion in specific and aspecific interactions. J Mol Recognit 16:177–190. http://dx.doi.org/10.1002/jmr.618. 58. Brant JA, Childress AE. 2002. Assessing short-range membrane-colloid interactions using surface energetics. J Membr Sci 203:257–273. http://dx .doi.org/10.1016/S0376-7388(02)00014-5. 59. Martin-Molina A, Maroto-Centeno JA, Hidalgo-Alvarez R, QuesadaPerez M. 2008. Charge reversal in real colloids: experiments, theory and simulations. Colloids Surf A 319:103–108. http://dx.doi.org/10.1016/j .colsurfa.2007.09.041. 60. Schneider C, Hanisch M, Wedel B, Jusufi A, Ballauff M. 2011. Experimental study of electrostatically stabilized colloidal particles: colloidal stability and charge reversal. J Colloid Interface Sci 358:62– 67. http://dx.doi .org/10.1016/j.jcis.2011.02.039. 61. Kanal KM, Fullerton GD, Cameron IL. 1994. A study of the molecular sources of nonideal osmotic-pressure of bovine serum-albumin solutions as a function of pH. Biophys J 66:153–160. http://dx.doi.org/10.1016 /S0006-3495(94)80773-8. 62. Ge SR, Kojio K, Takahara A, Kajiyama T. 1998. Bovine serum albumin adsorption onto immobilized organotrichlorosilane surface: influence of the phase separation on protein adsorption patterns. J Biomater Sci Polym Ed 9:131–150. http://dx.doi.org/10.1163/156856298X00479. 63. Barbosa LRS, Ortore MG, Spinozzi F, Mariani P, Bernstorff S, Itri R. 2010. The importance of protein-protein interactions on the pH-induced conformational changes of bovine serum albumin: a small-angle X-ray scattering study. Biophys J 98:147–157. http://dx.doi.org/10.1016/j.bpj .2009.09.056. 64. Josephson RV, Mikolajick EM, Sinha DP. 1972. Gel isoelectric focusing of selected bovine immunoglobulins. J Dairy Sci 55:1508 –1510. http://dx .doi.org/10.3168/jds.S0022-0302(72)85705-9. 65. Hori K, Matsumoto S. 2010. Bacterial adhesion: from mechanism to control. Biochem Eng J 48:424 – 434. http://dx.doi.org/10.1016/j.bej.2009 .11.014. 66. Hwang G, Kang S, Gamal El-Din M, Liu Y. 2012. Impact of conditioning films on the initial adhesion of Burkholderia cepacia. Colloids Surf B 91: 181–188. http://dx.doi.org/10.1016/j.colsurfb.2011.10.059. 67. Hahn MW, O’Melia CR. 2004. Deposition and reentrainment of Brown-

August 2016 Volume 82 Number 16

68. 69.

70.

71.

72. 73.

74. 75. 76.

77.

78. 79.

80.

ian particles in porous media under unfavorable chemical conditions: some concepts and applications. Environ Sci Technol 38:210 –220. http: //dx.doi.org/10.1021/es030416n. Sutlovic D, Gojanovic MD, Andelinovic S, Gugic D, Primorac D. 2005. Taq polymerase reverses inhibition of quantitative real time polymerase chain reaction by humic acid. Croat Med J 46:556 –562. Albers CN, Jensen A, Baelum J, Jacobsen CS. 2013. Inhibition of DNA polymerases used in Q-PCR by structurally different soil-derived humic substances. Geomicrobiol J 30:675– 681. http://dx.doi.org/10.1080 /01490451.2012.758193. Yuan BL, Pham M, Nguyen TH. 2008. Deposition kinetics of bacteriophage MS2 on a silica surface coated with natural organic matter in a radial stagnation point flow cell. Environ Sci Technol 42:7628 –7633. http://dx .doi.org/10.1021/es801003s. Kim J, Shan WQ, Davies SHR, Baumann MJ, Masten SJ, Tarabara VV. 2009. Interactions of aqueous NOM with nanoscale TiO2: implications for ceramic membrane filtration-ozonation hybrid process. Environ Sci Technol 43:5488 –5494. http://dx.doi.org/10.1021/es900342q. Pham M, Mintz EA, Nguyen TH. 2009. Deposition kinetics of bacteriophage MS2 to natural organic matter: role of divalent cations. J Colloid Interface Sci 338:1–9. http://dx.doi.org/10.1016/j.jcis.2009.06.025. Gutierrez L, Nguyen TH. 2012. Interactions between rotavirus and Suwannee River organic matter: aggregation, deposition, and adhesion force measurement. Environ Sci Technol 46:8705– 8713. http://dx.doi.org/10 .1021/es301336u. Tong MP, Shen Y, Yang HY, Kim H. 2012. Deposition kinetics of MS2 bacteriophages on clay mineral surfaces. Colloids Surf B 92:340 –347. http: //dx.doi.org/10.1016/j.colsurfb.2011.12.017. Rose AL, Waite TD. 2003. Kinetics of iron complexation by dissolved natural organic matter in coastal waters. Marine Chem 84:85–103. http: //dx.doi.org/10.1016/S0304-4203(03)00113-0. Karlsson T, Persson P, Skyllberg U. 2006. Complexation of copper(II) in organic soils and in dissolved organic matter—EXAFS evidence for chelate ring structures. Environ Sci Technol 40:2623–2628. http://dx.doi.org /10.1021/es052211f. Guan X-H, Shang C, Chen G-H. 2006. ATR-FTIR investigation of the role of phenolic groups in the interaction of some NOM model compounds with aluminum hydroxide. Chemosphere 65:2074 –2081. http: //dx.doi.org/10.1016/j.chemosphere.2006.06.048. Li QL, Elimelech M. 2004. Organic fouling and chemical cleaning of nanofiltration membranes: measurements and mechanisms. Environ Sci Technol 38:4683– 4693. http://dx.doi.org/10.1021/es0354162. Ahmadiannamini P, Bruening ML, Tarabara VV. 2015. Sacrificial polyelectrolyte multilayer coatings as an approach to membrane fouling control: disassembly and regeneration mechanisms. J Membr Sci 491:149 – 158. http://dx.doi.org/10.1016/j.memsci.2015.04.041. Yin Z, Tarabara VV, Xagoraraki I. 2015. Human adenovirus removal by hollow fiber membranes: effect of membrane fouling by suspended and dissolved matter. J Membr Sci 482:120 –127. http://dx.doi.org/10.1016/j .memsci.2015.02.028.

Applied and Environmental Microbiology

aem.asm.org

4993