Elution Is a Critical Step for Recovering Human Adenovirus 40 fromTap 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 andinterprets recovery values in terms of physicochemical interactions of virions during sample concentration. Prior to ultrafiltra-tion, membranes were either blocked by exposure to calf serum (CS) or coated with a polyelectrolyte multilayer (PEM). HAdV 40is a hydrophobic virus with a point of zero charge between pH 4.0 and pH 4.3. In accordance with predictions from the extendedDerjaguin-Landau-Verwey-Overbeek theory, the preelution recovery of HAdV (rpre) from deionized water was higher with PEM-coated membranes (rpre

PEM � 74.8% � 9.7%) than with CS-blocked membranes (rpreCS � 54.1% � 6.2%). With either membrane

type, the total virion recovery after elution (rpost) was high for both deionized water (rpostPEM � 99.5% � 6.6% and rpost

CS � 98.8% �7.7%) and tap water (rpost

PEM � 89% � 15% and rpostCS � 93.7% � 6.9%). The nearly 100% recoveries suggest that the polyanion (so-

dium polyphosphate) and surfactant (Tween 80) in the eluent disrupt electrostatic and hydrophobic interactions between thevirion and the membrane. Addition of EDTA to the eluent greatly improved the elution efficacy (rpost

CS � 88.6% � 4.3% andrpost

PEM � 87.0% � 6.9%) with surface water, even when the organic carbon concentration in the water was high (9.4 � 0.1 mg/li-ter). EDTA likely disrupts cation bridging between virions and particles in the feed water matrix or the fouling layer on themembrane surface. For complex water matrices, the eluent composition is the most important factor for achieving high virionrecovery.

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-virionand virion-filter interactions. Cross-flow filtration for concentration and recovery of HAdV 40 was evaluated, with postelutionrecoveries 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 tomaximize HAdV 40 recovery.

Pathogenic microorganisms cause many waterborne diseases,and the World Health Organization reports that water supply

contamination leads to 842,000 deaths annually (1). Along withcaliciviruses (e.g., norovirus), enteroviruses (e.g., poliovirus, cox-sackievirus, and echovirus), and hepatitis A viruses, human ade-novirus (HAdV) is one of the viral pathogens in EPA contaminantcandidate lists 3 and 4 and is the second-leading cause of child-hood gastroenteritis worldwide (2–4). Most cases of adenovirus-associated 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, hepatitisA virus, and fecal indicator bacteria [6]) and treatment with UVirradiation (7). Monitoring HAdV 40 and 41 levels is essential tobetter understand their fate in the environment and in treatmentsystems, 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) and103 to 105 GC/liter in wastewater treatment effluent (11, 12), makedetection challenging and highlight the importance of sampleconcentration for rapid and reliable virus detection.

Virus adsorption-elution (VIRADEL), the standard methodfor virus concentration, includes adsorption of virions on eitherelectropositive or electronegative filters and subsequent elution of

the adsorbed virions. Although recoveries by VIRADEL dependon the specific filter and eluent and vary greatly as a function ofwater composition, recoveries of adenoviruses are low relative tothose of other viruses. For example, Gibbons et al. reported highrecoveries of coliphages (96%) and noroviruses (100%) but verylow recoveries of HAdV 41 (�3%) when they concentrated vi-ruses from seawater using a NanoCeram filter and 3% (wt/vol)beef extract as the eluent (13). In another study, the coxsackievirusB5 recovery from tap water was 72%, whereas the recovery ofHAdV 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 acritical step for recovering human adenovirus 40 from tap water and surfacewater 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, tarabara@msu.edu.

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.

crossmark

4982 aem.asm.org August 2016 Volume 82 Number 16Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

of adenovirus may occur because fibers associated with each pen-ton base of the capsid facilitate the physical entrapment of virusparticles in the filter matrix (13, 14). McMinn achieved a relativelyhigh recovery of HAdV 40 with a 1MDS filter (using the VIRADELmethod), 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 exclu-sion rather than adsorption to concentrate virions during filtra-tion and might give higher recoveries than those obtained with theVIRADEL method. A UF membrane with pores smaller than thevirion allows passage of water and low-molecular-weight com-pounds while maintaining virions in the recirculating retentatestream. The cross-flow should minimize membrane fouling andvirion deposition onto the membrane. Nevertheless, significantvirion deposition on membranes in cross-flow ultrafiltration sys-tems is still observed (16–19). Similar to VIRADEL, elution ofvirions that adsorb to the UF membrane may be accomplished byusing reagents, such as sodium polyphosphate (NaPP) and Tween80, to disrupt electrostatic and hydrophobic interactions betweenvirions and the membrane (18, 20, 21). UF can concentrate avariety 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 advan-tages over VIRADEL: (i) the lower cost of UF membranes thanVIRADEL filters, (ii) simultaneous concentration of multipletypes of pathogens, and (iii) removal of small compounds thatmay inhibit quantitative PCR (qPCR) analysis. However, a frac-tion of the virions may still adsorb to the membrane, decreasingrecoveries (26). To enhance recovery, UF membranes are typicallycoated with proteinaceous solutions (e.g., calf serum [CS]) toblock potential virion adsorption sites and reduce virion adhesion(16, 17, 20, 21). However, long coating times and possible con-tamination during storage and transportation limit the applica-tion of CS-blocked membranes for rapid or field sampling (27). Ina previous study, to improve virion recovery during cross-flowUF, we coated membranes with a polyelectrolyte multilayer(PEM) film prepared via rapid (�1 h) layer-by-layer adsorptionof polycations (chitosan [CHI]) and polyanions (heparin [HE]).Compared with traditional CS-blocked membranes, the PEM-coated membranes showed higher recoveries of bacteriophageP22 from both deionized (DI) water and a membrane bioreactoreffluent (19).

Despite the need for improved HAdV concentration and re-covery, cross-flow UF has not been evaluated comprehensively forthis application. Pei et al. employed cross-flow UF to recover vi-ruses, including HAdV 2, from tap water, but they coupled thisprocess with VIRADEL. They reported a 42.4% recovery of HAdV2 by VIRADEL only, whereas they did not determine the recoveryof HAdV 2 by cross-flow UF (22). Skraber et al. determined therecovery of MS2, �X174, and poliovirus by cross-flow UF. Theyalso processed environmental samples containing HAdV, but theydid not know the initial concentration of HAdV and thus couldnot determine HAdV recovery (23). Nestola et al. reported HAdV5 recovery from a prefiltered virus stock after virus harvest (28).Rhodes et al. examined recoveries of multiple viruses, includingenteroviruses, �X174, and HAdV 41, from tap and river water bycross-flow ultrafiltration; postelution recoveries of HAdV 41 were69.4% � 12.4% from tap water and 55.8% � 7.9% from river

water, while preelution recoveries were not reported (29). Theyconcluded that in addition to size exclusion, other factors have animpact on the virus recovery process.

Characterization of adenovirus physicochemical properties isimportant for designing efficient methods for their concentration.Point-of-zero-charge (PZC) values for adenoviral serotypes areavailable 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 ofHAdV 40 and HAdV 41 virions. Recently, Wong et al. computedthe extended Derjaguin-Landau-Verwey-Overbeek (XDLVO) in-teraction energies of HAdV 2 (30), but the calculations used sur-face energy parameters of surrogates.

The goal of the present work was to improve the recovery ofHAdV from water. We built on our earlier experimental and mod-eling study (19) to pursue the following two specific objectives: (i)to elucidate specific physicochemical mechanisms that controlHAdV 40 adhesion to a membrane surface and (ii) to evaluatecross-flow UF for concentration and recovery of HAdV 40 fromseveral water matrices. We employed two types of membranes toexamine the effects of membrane properties on virion recovery.The approach included comprehensive physicochemical charac-terization of HAdV 40, evaluations of virion-virion and virion-membrane interactions using the XDLVO theory, and cross-flow filtration tests with feed waters spiked with HAdV 40. Thepremise of the study was that understanding virion interac-tions can guide process design for improving preelution andpostelution recoveries.

MATERIALS AND METHODSReagents and water samples. All reagents were of American ChemicalSociety (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 ob-tained from Fisher Scientific (Pittsburgh, PA), and glycerol was purchasedfrom J. T. Baker (Center Valley, PA). All solutions were prepared using DIwater with a resistivity of 18.2 M� · cm (Barnstead NANOpure system).

Tap water was supplied by the East Lansing-Meridian Water andSewer Authority (East Lansing, MI) and used immediately after collec-tion. Samples of surface water were collected from Lake Lansing at theboat ramp in Lake Lansing Park South (Haslett, MI) in October 2013 andApril 2014. Surface water samples were filtered through 0.45-�m mem-branes (mixed cellulose esters; Millipore, Billerica, MA) immediately aftercollection, stored at 4°C, and used within 1 week. All samples were char-acterized in terms of pH (Orion 3-Star Plus meter; Thermo ElectronCorp., Waltham, MA), conductivity (Orion 550 conductivity meter;Thermo Electron Corp., Waltham, MA), total organic carbon (TOC) con-tent, and concentrations of key cations. The TOC in each water samplewas measured at least in triplicate (model 1010 analyzer; OI Analytical,College Station, TX). Concentrations of cations (Na�, K�, Mg2�, andCa2�) were determined by inductively coupled plasma mass spectrometry(iCAP Q quadrupole ICP-MS; Thermo Scientific, Waltham, MA).

Propagation, purification, and characterization of HAdV 40. HAdV40 (ATCC VR-931) was purchased from ATCC (Manassas, VA) andpropagated in A549 cells. The concentration of propagated virus, as mea-sured by quantitative real-time PCR (qPCR), was 109 to 1010 GC/ml for allpropagated batches. Virions were further purified using CsCl density gra-dient centrifugation following the protocol described by Mautner (32).See Section S1 in the supplemental material for a detailed description ofvirus growth and purification procedures. HAdV 40 is a human pathogenand should be manipulated following biosafety level 2 (BSL-2) criteria.

HAdV 40 from the purified stock was used in all virion characteriza-

Elution Is Critical for Human Adenovirus Recovery

August 2016 Volume 82 Number 16 aem.asm.org 4983Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

tion tests. The hydrodynamic diameter of the virion was determined bydynamic light scattering (DLS) (90 Plus; Brookhaven Instruments Corp.,Holtsville, NY), and the -potential of the virion was measured usingphase analysis light scattering (ZetaPALS) (Brookhaven InstrumentsCorp., Holtsville, NY). To study the effects of pH on the aggregation stateand -potential of HAdV 40 virions, the solution pH was adjusted by theaddition of NaOH and HCl, and a 1 mM NaCl solution served as thebackground electrolyte at all pH values except 7.6. The pH of an unbuf-fered NaCl solution could not be stabilized at pH 7.6, so a solution of 10mM Tris-HCl and 1 mM EDTA (the recommended buffer for the storageof purified HAdV 40 [32]) was used as the background electrolyte.

Transmission electron microscopy (TEM) was used to image HAdV40. A few drops of purified virus stock were applied to a carbon-coatedFormvar grid (Ted Pella, Redding, CA). The sample was rested on the gridfor 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 filterpaper. The grid was air-dried prior to TEM imaging. Images were re-corded using a JEM-2200FS (JEOL, Tokyo, Japan) microscope with anin-column energy filter operated at 200 kV and equipped with a DirectElectron DE-20 camera.

The surface energy of HAdV 40 was determined using a previouslydescribed protocol (19). Briefly, a virion lawn with �7 layers of virionswas prepared on a 30-kDa polyethersulfone membrane (Omega, PallCorp., East Hills, NY) by filtering purified virion stock through the mem-brane. The virion-coated membrane was dried at room temperature untilthe water contact angle on the virion lawn plateaued, indicating formationof a monolayer of strongly bound water that resists evaporation. Thecontact angles of three probe liquids (DI water, glycerol, and diiodometh-ane) on the virion lawn were then measured with a goniometer (model250; Ramé-Hart, Succasunna, NJ) by the sessile drop method to deter-mine surface energy components of HAdV 40.

XDLVO energies of virion-virion and virion-membrane interac-tions. XDLVO energy profiles were calculated (see Section S6 in the sup-plemental material for details) using the surface properties (charge andenergy) of HAdV 40 and membranes. Membranes were characterized inour previous study (19).

Membrane preparation. Prior to modification, polyethersulfonemembranes (Omega; Pall Corp., East Hills, NY) with a molecularweight cutoff of 30,000 were soaked in 0.1 M NaOH for 3 h and then inDI water for 24 h at 4°C, with water exchange after 12 h, as recom-mended 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 MNaCl and adjusted to pH 5 using 1 M HCl and 1 M NaOH. Bothpolyelectrolyte solutions were stored at 4°C until use. The surface ofthe polyethersulfone membrane was alternately exposed to each poly-electrolyte solution for 5 min, with a 1-min DI water rinse betweenexposures. The deposition was initiated with the negatively chargedHE solution. Alternating deposition of HE and CHI continued until4.5 bilayers were obtained, with HE as the outermost (terminal) layer.To block a membrane with calf serum, the membrane was placed in across-flow filtration cell, and 500 ml of 5% (vol/vol) calf serum wascirculated over the membrane overnight at 1.0 liter/min at room tem-perature, with no pressure applied across the membrane. After block-ing, the membrane was rinsed with DI water twice at the same rate for10 min. The PEM-coated and CS-blocked membranes were character-ized in terms of charge and surface energy as described earlier (19).

Virus concentration and recovery tests. Virions were recovered fromDI water, tap water, and surface water. Prior to recovery experiments,1-liter water samples spiked with 1 ml of HAdV 40 stock (virion concen-tration of �106 to 107 GC/ml in the diluted solution) were recirculated inthe filtration system without a membrane and with the permeate portblocked. The virus concentration in the feed tank was quantified after 1 hof recirculation (approximate time needed for filtration) and comparedwith that initially in the feed. No difference in virus concentration wasobserved, indicating no loss of virions to the filtration apparatus. Figure

S3 in the supplemental material shows a schematic of the cross-flow fil-tration 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 dilutedsolution), and 5 ml was collected to quantify virions in the feed sample.The remaining sample (996 ml) was pressurized using compressed N2 gasand delivered to the membrane cell by use of a peristaltic pump (model621 CC; Watson Marlow, Wilmington, MA) at 2.0 liters/min. A digitalflowmeter (model 101-7; McMillan, Georgetown, TX) monitored thecross-flow rate. The transmembrane pressure was maintained at �20 lb/in2 (�138 kPa) and continually monitored by a pressure transducer (0.5-to 5.5-V output; Cole Parmer, Vernon Hills, IL). The ratios of cross-flowflux to initial permeate flux (Jcf/Jp) ranged from 5 103 to 6 103.Permeate was collected into a flask placed on a digital balance (AdventurerPro AV8101C; Ohaus, Parsippany, NJ), and the permeate mass was re-corded in real time to calculate the flux. The data from the pressure trans-ducer, digital flowmeter, and digital balance were transmitted to a com-puter via a data acquisition module (PCI-6023E/SC-2345; NationalInstruments, Austin, TX). Filtration was stopped when 900 ml of samplewas filtered. The remaining 96 ml of retentate was analyzed for viruscontent so that preelution recovery could be calculated.

After filtration, virions were eluted from the membrane by using aneluent containing 0.01% NaPP and 0.01% Tween 80 (pH 6.5). For virionconcentration and recovery from lake water, another eluent, containing0.01% NaPP, 0.01% Tween 80, and 0.01% EDTA (pH 7), was also em-ployed. Elution used 100 ml of eluent and was performed at a cross-flowrate of 2.0 liters/min for 20 min, with no transmembrane pressure applied.The feed, permeate, retentate, and eluate were all analyzed for virion con-centration. Samples were also taken during membrane compaction toensure that virions did not contaminate the feed water prior to the exper-iment.

Virus quantification and recovery. HAdV 40 DNA in water sampleswas extracted by use of a MagNA Pure compact system (Roche Diagnos-tics USA, Indianapolis, IN) immediately after sample collection. qPCRwas used to determine the HAdV 40 concentration (see Section S2 in thesupplemental material for a description of the qPCR procedures). Possi-ble effects of organic compounds in the feed water on virion quantifica-tion by qPCR were evaluated by performing inhibition tests. Virion pree-lution recovery (rpre), postelution recovery (rpost), and log removal (LRV)were calculated using the following equations:

rpre �CrVr

CfVf(1)

rpost �CrVr � CeVe

CfVf(2)

LRV � log 10�Cf � Cr

2Cp� (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 thevolumes of these samples. In calculating the LRV (equation 3), the virusconcentration in feed was calculated by averaging the initial concentrationof HAdV 40 in feed prior to filtration and the concentration of the virus inretentate after filtration, i.e., C̄f � 1/2(Cf � Cr). Thus, the reported virusremoval (LRV) value is the average over the entire filtration process.

RESULTS AND DISCUSSIONCharacteristics of source waters. Table 1 provides water qualityparameters for DI water, tap water, and Lake Lansing water (sam-pled in spring and fall). The biggest differences among the waterswere (i) much lower values for TOC, conductivity, and cationconcentrations in DI water and (ii) high TOC and Ca2� concen-trations in the Lake Lansing water samples.

Size, �-potential, and surface charge density of HAdV 40. Co-

Shi et al.

4984 aem.asm.org August 2016 Volume 82 Number 16Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

liphages commonly serve as surrogates for human viruses (33).However, studies on virus removal and adsorption show that co-liphages (e.g., MS2 [34, 35], �X174 [34, 35], and PRD1 [34]) arenot suitable models for HAdV. Moreover, unique properties ofserotypes 40 and 41 of enteric HAdV (e.g., the presence of shortand long fibers on the capsid [36, 37], fastidiousness in cell culture[38–40] [though cultivation improvement was reported with amodified host cell line {41, 42}], and difficulty of isolation [40,43]) make them different from other HAdV serotypes. Thisuniqueness makes detailed characterization of HAdV 40 and 41necessary. Aggregation state and -potential as functions of pHwere reported for HAdV 2 (30) but not for HAdV 40. In this study,we measured the size of HAdV 40 by dynamic light scattering andcompared the data with the results of TEM imaging. We also de-termined the -potential and surface charge density of the virionas functions of pH.

Based on TEM images (Fig. 1), the diameter of HAdV 40 is �80nm. Figure 2 shows the number-based hydrodynamic diameterdistribution of HAdV 40 as a function of pH, and Table 2 summa-rizes 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

average hydrodynamic diameters of HAdV 40 (d̄HAdVh ) ranged

from 94 � 3 nm to 102 � 9 nm, agreeing with typical size valuesreported earlier for HAdVs of different serotypes (44–46). Thesmall increase in the hydrodynamic diameter relative to the diam-eter determined by TEM may have stemmed from fibers on theHAdV 40 surface (30, 47) that we did not take into account whenestimating HAdV 40 diameters from TEM images. Such fibers

should, however, decrease particle diffusivity in light-scatteringexperiments. Additionally, shrinkage of HAdV 40 during negativestaining may have decreased the diameters observed in TEM im-ages (48). The polydispersity indexes of the particle diameters inthe 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 and4.7, both average size and polydispersity increased, consistent withaggregation. Aggregates of up to 1 �m appeared at pH 4.0 and 4.3in several measurements. In contrast to the aggregation betweenpH 4.0 and 4.7, at pH 2.8, the virion hydrodynamic diameter wasonly 112 � 20 nm, suggesting minimal aggregation. This is con-sistent with studies showing that the enteric adenoviruses remaininfective under the acidic conditions in the gut (49, 50). A highpositive -potential at low pH (Fig. 3) probably minimizes aggre-gation. At pH 9.7, the average hydrodynamic diameter of HAdV40 (80 � 5 nm) was significantly lower than the values measured atpH 5.8 to 7.6 and close to the size determined by TEM imaging. Aprevious study observed inactivation of adenovirus after liming(51, 52), and the reduced size at pH 9.7 may have stemmed fromvirus degradation. Size reduction at pH 10 also occurred withHAdV 2 (30).

Figure 3 presents the virion -potentials determined from elec-trophoretic mobilities (see Fig. S9 in the supplemental material)by use of an approximate expression derived by Ohshima (53); theexpression provides an accurate (�1% error) estimate for -po-tentials for arbitrary values of �a, where a is the radius of the virionand � is the Debye-Hückel parameter (see Section S9). Due to thesmall virion diameters, Smoluchowski’s expression for electro-phoretic mobility, applicable for �a values of 1, is not accuratein our system. Thus, we used Ohshima’s expression.

The -potential decreased from �22.0 mV to �30.5 mV whenthe pH increased from 5.8 to 7.6. The negative surface charge onthe virion likely minimizes aggregation, but for several measure-ments in this pH range, some aggregates (300 to 350 nm; 1.2% �0.5% of the total population) appeared, possibly as an artifact ofthe dialysis process (32). Relatively small absolute values of the-potential between pH 4.0 and 4.7 probably allow for the exten-sive aggregation described above.

As Fig. 3 shows, the PZC for HAdV 40 is between pH 4.0 and4.3. Favier et al. predicted the theoretical PZC for each majorcapsid protein of HAdV 40 by using the primary sequences, andthey showed that the PZC values for the hexon, penton base, longfiber 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 40PZC we determined, which is consistent with prior observationsthat the PZC for an intact virion is lower than the PZC values forits structural protein components (54). PZC values were reportedto be in the range of 3.5 to 4.0 for HAdV 2 (30) and 4.5 for HAdV5 (31), consistent with the low PZC of HAdV 40.

TABLE 1 Water quality characteristics

Water source pH rangeConductivity(�S/cm)

TOC concn(mg/liter)

Cation concn (mg/liter)

Na� K� Mg2� Ca2�

Deionized water 5.7–6.0 NAa NAa 0.18 � 0.00 0.02 � 0.00 0.04 � 0.00 0.06 � 0.00Tap water 7.5–8.0 319.5 � 27.6 1.1 � 0.1 22.18 � 2.33 9.32 � 0.32 1.86 � 0.09 24.37 � 1.83Lake water (fall) 7.0–7.5 336.3 � 12.7 7.5 � 0.2 16.42 � 2.53 10.91 � 0.11 1.47 � 0.03 30.33 � 0.54Lake water (spring) 7.5–7.8 369.3 � 32.0 9.4 � 0.1 16.46 � 2.60 11.54 � 0.20 1.50 � 0.03 41.09 � 0.72a NA, not available (below detection limit).

FIG 1 Transmission electron microscopy image of HAdV 40.

Elution Is Critical for Human Adenovirus Recovery

August 2016 Volume 82 Number 16 aem.asm.org 4985Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

At pH values at which virions aggregate, the measured -po-tential is that for an aggregate of virions, not a single virion. In thiscase, we calculated virion -potentials using an expression devel-oped by Makino and Ohshima (55); the expression connects thesurface 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 intensiveproperty and does not depend on the aggregation state of thevirion in the same solution.

The average -potential of HAdV 40 in tap water (pH 7.5 to8.0) 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 ofdivalent or multivalent ions onto virions in tap water.

Surface energy components of HAdV 40. To determine thesurface energy components of HAdV 40, we applied the procedurewe developed previously for examination of P22 bacteriophage(19). The technique includes measuring the contact angles of sev-eral liquids on a lawn of virions. As Table 3 shows, the apolarsurface energy component (�LW) of HAdV 40 was 41.6 mJ/m2, avalue typical for biological materials (56). The electron donor(��) and electron acceptor (��) components of surface energywere 14.7 mJ/m2 and 0.01 mJ/m2, respectively, making the polar

component of HAdV 40 surface energy (�AB � 2�����) equalto 0.84 mJ/m2. Very small values of �� were also reported fortobacco mosaic virus (56), bacteriophage P22 (19), and multipleproteins (56), suggesting that this is a common characteristic of

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 forthe size distribution for an HAdV 40 suspension in tap water.

TABLE 2 Size distribution parameters for HAdV 40 suspension at different pH valuesa

Parameter

Value for water matrixb

1 mM NaCl Bufferc Tap water 1 mM NaCl

pH 2.8 4.0 4.3 4.7 5.8 6.7 7.6 7.5–8.0 9.7d̄HAdV

h (nm) 112 � 20 284 � 99 238 � 69 149 � 3 94 � 3 95 � 3 103 � 9 109 � 14 80 � 5Polydispersity 0.22 � 0.06 0.31 � 0.04 0.31 � 0.01 0.10 � 0.01 0.06 � 0.01 0.07 � 0.02 0.07 � 0.01 0.11 � 0.01 0.12 � 0.03a Additional parameters (modal diameter and half-width at half-maximum) are given in Table S3 in the supplemental material.b Shaded areas denote pH values where significant aggregation occurs.c Buffer recommended for the storage of purified HAdV 40 (10 mM Tris-HCl and 1 mM EDTA).

Shi et al.

4986 aem.asm.org August 2016 Volume 82 Number 16Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

proteins and nonenveloped viruses. Based on values of �LW, ��,and �� for water and HAdV 40, the calculated polar adhesionenergy between HAdV 40 and water is 79.7 mJ/m2, which is insuf-ficient to overcome the energy of cohesion of water (�102.0 mJ/m2). As a result, virion-virion interactions for HAdV 40 in waterhave a negative interfacial free energy (�Gvwv � �30.4 mJ/m2), soHAdV 40 is hydrophobic.

XDLVO energy profiles of virion-virion interfacial interac-tions in aqueous media. Several studies suggest that the classicDerjaguin-Landau-Verwey-Overbeek theory does not accuratelypredict the interactions between colloids and various surfaces inaqueous media because the calculations do not account for acid-base 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-membraneinterfacial interactions. Inputs to the model include the surfacecharacteristics of HAdV 40 (see “Size, -potential, and surface

charge density of HAdV 40” and “Surface energy components ofHAdV 40”) and membranes (19).

Figure 4 shows XDLVO energy profiles of virion-virion inter-facial interactions at different pH values. For all pH values except7.6, the calculations were performed for HAdV in 1 mM NaCl. ForpH 7.6, size and -potential values used in XDLVO calculationswere for virions in the recommended storage buffer (1 mM EDTAplus 10 mM Tris). For pH values of 5.8, the height of the energybarrier increases with pH, with the sole exception of pH 7.6. Thedeviation from the trend is due to the different background solu-tion (10 mM Tris plus 1 mM EDTA instead of 1 mM NaCl) (see“Size, -potential, and surface charge density of HAdV 40”). Thenegligible aggregation of virions for pH values of �5.8 is due tothe high energy barrier (24.5 kT to 52.6 kT). At pH 4.7, where theenergy barrier is 8.0 kT, significant aggregation occurs. At pH 4.0and pH 4.3, the energy barrier decreases to below 1.2 kT, leadingto further aggregation. Due to the high positive -potential (29.4mV) at pH 2.8, the energy barrier is 20.7 kT and the virion sus-pension is stable.

XDLVO energies of virion-membrane interfacial interac-tions in membrane feed solutions made with DI water and tapwater. XDLVO calculations for virion-membrane interactions inDI water were performed for an ionic strength of 0.2 mM; this isthe approximate ionic strength determined from conductivitymeasurements of DI water spiked with HAdV 40 stock. However,the streaming potential of the membrane could be measured onlywith 1 mM NaCl electrolytes, so this value was used to approxi-mate the membrane charge in the HAdV-spiked DI water. Theapproximation is reasonable because the electrophoretic mobili-ties of virions in DI water (�1.65 � 0.19 �m · S�1 · V�1 · cm at pH5.8) and in 1 mM NaCl solution (�1.72 � 0.48 �m · S�1 · V�1 · cmat pH 5.8) were not statistically different.

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

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

Parameter Value

Contact angle (°) with indicated probe liquidH2O 68 � 2Glycerol 64 � 1Diiodomethane 36 � 2

Surface energy parameter (mJ/m2)�LW 41.6�� 0.01�� 14.7�AB 0.8�TOT 42.4

Free energy of interfacial virion-virion interactions inwater (�Gvwv [mJ/m2])

�30.4

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

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

Elution Is Critical for Human Adenovirus Recovery

August 2016 Volume 82 Number 16 aem.asm.org 4987Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

At separation distances of 5 nm, van der Waals or electro-static interactions dominate the XDLVO energy of interfacial in-teraction between HAdV 40 and membranes in both DI and tapwater (see Fig. S7 and S8 in the supplemental material). However,as the separation distance increases, the magnitude of each inter-action decreases. The magnitude of van der Waals interaction en-ergy, for example, decreases to �0.5 kT as the separation distanceincreases to approximately 15 nm. Over a shorter range (from theminimum equilibrium cutoff distance of �0.16 nm to 0.7 nm),however, the acid-base interaction energy is significantly greaterthan both van der Waals and electrostatic interaction energies. In1 mM NaCl at pH 5.8 (unadjusted), a PEM-coated membrane isnegatively charged ( � �7.0 � 3.0 mV) (19). Due to the hydro-philicity of the PEM as well as a repulsive electrostatic interactionbetween the virion and the membrane surface, the secondary min-imum in the membrane-virion interaction energy profile is shal-low (�4.2 kT) (Fig. 5A, solid line). The same membrane carries apositive charge ( � 5.6 � 0.4 mV) in tap water. The chargereversal may stem from specific adsorption of various cationsfrom tap water onto the membrane surface, and a similar chargereversal occurred with latex particles as the solution ionic strengthincreased (59, 60). For the membrane-virion interaction energyprofile for tap water, electrostatic attraction coupled with van derWaals attraction at large separation distances and a predomi-nantly repulsive acid-base interaction at small separation dis-tances yields a secondary minimum of �10.3 kT at a separationdistance of 3.2 nm (Fig. 5A, dashed line). This secondary mini-mum may lead to reversible adsorption of HAdV 40 onto themembrane surface during tap water filtration.

Even though the PZC of bovine serum albumin, a major com-ponent of calf serum, is in the range of 4.7 to 5.6 (61–63), CS-blocked membranes in a 1 mM NaCl solution at pH 5.8 (unad-justed) carry a weak positive charge ( � 3.0 � 2.0 mV) (19). Thesmall positive charge on these membranes may stem from othercomponents in calf serum, such as bovine IgG (PZC range of 7.5 to8.3 [64]). The XDLVO energy profile for the interaction of HAdV40 virions and a CS-blocked membrane shows a secondary mini-

mum of �48.2 kT at a separation distance of 2.4 nm (Fig. 5B, solidline), 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 experimentsperformed with DI water, the reduced negative charge of HAdV 40in tap water (Fig. 3) leads to less virion-membrane electrostaticattraction, and the depth of the secondary minimum in tap wateris only �7.9 kT for interactions of the virion and the CS-blockedmembrane (Fig. 5B, dashed line).

Virus recovery from DI water. Preelution virion recovery (seeequation 1) after cross-flow UF of DI water spiked with virionswas significantly greater with a PEM-coated membrane (74.8% �9.7%) than with a CS-blocked membrane (54.1% � 6.2%). Thehigher recovery likely stems from the negative charge and hydro-philicity of the PEM-coated membrane (19). However, the�100% recovery indicates that some virion adsorption takes placeon the PEM-modified membrane, even though the secondaryminimum in the XDLVO energy profile is shallow (Fig. 5A, solidline). This adsorption may stem from microscale attraction (56,65), whereas XDLVO theory describes only macroscopic interac-tions. For example, in experiments on Burkholderia cepacia adhe-sion, Hwang et al. reported bacterial adhesion in the absence of asecondary minimum in the XDLVO energy profile and suggestedthat the adhesion stems from cell appendages, such as pili andflagella (66). HAdV 40 fibers, which consist of long, thin shaftsterminated with globular knobs, may extend beyond the charac-teristic length of macroscopic repulsion (approximately the sameas the Debye length) to achieve microscopic attraction between avirion and a membrane. Interactions such as those between elec-tron donor sites on the fiber knob and electron acceptor sites onthe membrane may overcome the macroscopic repulsion and leadto local attraction. Debye lengths for HAdV 40 at different pHvalues in 1 mM NaCl (see Table S2 in the supplemental material)are shorter than the lengths of both longer and shorter fibers (�30nm and �18 nm, respectively [36]) on HAdV 40. In HAdV-spikedDI water, the Debye length is �23 nm (see Table S2), which is stillshorter than the 30-nm HAdV fibers. Other possible explanations

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

Shi et al.

4988 aem.asm.org August 2016 Volume 82 Number 16Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

for the incomplete preelution recovery with PEM-coated mem-branes include local electrostatic attraction due to charge hetero-geneity and the presence of non-XDLVO interactions (e.g., bridg-ing by multivalent cations).

The relatively low virion recovery from DI water after cross-flow UF with CS-blocked membranes likely stems from the sec-ondary minimum in the XDLVO energy profile (Fig. 5B, solidline). The secondary minimum is sufficiently deep (�48.2 kT) tocapture virions with an average thermal energy of �0.5 kT (67).The fact that preelution recovery after cross-flow UF is higher withPEM-coated membranes than with CS-blocked membranes isconsistent with our previous findings for cross-flow filtration ofbacteriophage P22 in DI water, where PEM-coated membranesgave significantly higher preelution recovery (80%) than that withCS-blocked membranes (30%) (19).

After filtration, we eluted virions to increase recovery. For bothPEM-coated and CS-blocked membranes, eluents containing0.01% NaPP and 0.01% Tween 80 give high postelution recoveries(see equation 2) of 99.5% � 6.6% for PEM-coated membranesand 98.8% � 7.7% for CS-blocked membranes. During elution,adsorption of NaPP on the membrane likely increases the negativecharge on the surface to enhance electrostatic repulsion betweenthe negatively charged virion and the membrane. Additionally,Tween 80, a nonionic surfactant, should minimize hydrophobicinteractions between the membrane surface and adsorbed virions.Thus, both NaPP and Tween 80 may help to release virions fromthe membrane.

The relatively high feed concentration of adenovirus (�106 to107 GC/ml) in our study may have led to a higher recovery and asmaller variance than those for filtration of lower virion concen-trations, because at high concentrations, the virion may saturatesorption sites. However, we had to use a relatively high virionconcentration to accurately quantify the low concentration of vi-rions in the permeate (and to quantify both virion removal andthe fraction of virions adsorbed on the membrane). The HAdV 40removal rate was high with both CS-blocked membranes (LRVvalues of up to 2.17) and PEM-coated membranes (LRV values ofup to 4.64), so the permeate concentration was much lower thanthe feed concentration.

Virus recovery from tap and surface water. We also evaluated

the recovery of HAdV 40 from tap water and from lake watercollected in two different seasons. Because organic molecules insurface waters may inhibit qPCR detection of virions (68, 69), weinitially performed inhibition tests (see Section S5 in the supple-mental material for details). Briefly, DI, tap, and surface watersamples were spiked with HAdV 40 from the stock suspension,viral DNA was extracted, and the virion concentration in eachwater matrix was determined by qPCR. Consistent with minimalqPCR inhibition, we observed no differences in virion concentra-tions in the three water matrices.

With tap water, we saw no statistically significant difference inpreelution virion recovery after cross-flow UF with PEM-coatedmembranes (40.5% � 9.9%) and CS-blocked membranes(38.3% � 9.3%). Based on XDLVO predictions for CS-blockedmembranes (Fig. 5B), the preelution recovery from tap watershould have been higher than that from DI water. However, theopposite occurred—for CS-blocked membranes, the averagepreelution recovery from tap water was �16% lower than thatfrom DI water (Fig. 6). The discrepancy suggests that non-XDLVO effects (e.g., steric interactions [70] or Ca2� bridging[71]) should be considered. Another possible cause for the dis-crepancy is virion adsorption to dissolved and suspended speciespresent in the tap water (TOC concentration � 1.1 � 0.1 mg/liter)and on the membrane surface. Permeate flux declined �15% to20% during cross-flow UF of tap water (see Fig. S4 in the supple-mental material), suggesting the formation of a fouling layer onthe membrane surface.

The preelution recovery from tap water after cross-flow UFwith the PEM-coated membrane was �34% lower than that fromDI water (Fig. 6). In addition to the possible effects describedabove for the CS-blocked membranes, charge reversal of the PEMmembrane in tap water may have decreased the preelution recov-ery. Preelution recovery from tap water had a larger variance, withvalues ranging from 30% to 50% for both types of membranes.Such variances likely stem from variations in tap water qualityover several months of sample collection and tests.

Surface water is a complex matrix containing naturally occur-ring 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

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

Elution Is Critical for Human Adenovirus Recovery

August 2016 Volume 82 Number 16 aem.asm.org 4989Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

recoveries from tap water but significantly lower than recoveriesfrom DI water. The TOC concentration in surface water collectedin the fall was 7.5 � 0.2 mg/liter. For surface water, permeate fluxdeclined more than 50% over 90 min of cross-flow UF (see Fig. S4in the supplemental material) with membranes of each type. Fou-lants formed a cake layer and masked the antiadhesive propertiesof the membrane surface, leading to the deposition of virions onmembranes. Virions could also be adsorbed to various compo-nents of the feed water (e.g., humic acid, clay, or silica particles)and codeposited on the membrane. Postelution recovery washigher for tap water (Fig. 6B) than for surface water, suggestingthat 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 inner-sphere complexation with carboxyl groups on both surfaces (72,73) and on clay and silica by screening the charge (74). Calciumbridging may also occur between carboxyl groups on natural or-ganic matter and on the HAdV 40 capsid (e.g., carboxylic groupson fiber knobs), leading to virion loss to the membrane surface.Bridging by other cations may also occur (e.g., by iron [75], cop-per [76], and aluminum [77]). In general, virion-foulant interac-tions rather than virion-membrane interactions likely govern re-covery from complex waters by UF. This is consistent with theresult that preelution recoveries from surface water were not sta-tistically different for the two membrane types (Fig. 7).

We also analyzed virion recovery from the surface water col-lected from the same lake in early spring, after the ice melted. Fluxdecline also occurred for both membranes during filtration ofspring surface water (see Fig. S4 in the supplemental material),and preelution recoveries with both membranes were �20%. Thelower recovery from spring surface water than from fall surfacewater 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 inthe fall and spring, respectively. For PEM-coated membranes,postelution recoveries were 62.4% � 2.2% and 41.6% � 2.0% forfall and spring samples. Thus, virions adsorbed from surface water

on both membranes were not eluted as effectively (Fig. 7) as thoseadsorbed from DI or tap water (Fig. 6) with an eluent containingNaPP 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 acomponent of the storage buffer for HAdV 40, so it should notinactivate the virus (32). With EDTA in the eluent, the postelutionrecovery from spring and fall surface water, averaged over allPEM-coated and CS-blocked membranes, increased to 84.3% �4.5%. The increased elution efficiency is consistent with the hy-pothesis that calcium binding decreases virion recovery from sur-face water.

In a previous study, we showed that a PEM consisting of a weakpolycation and a strong polyanion disassembles at low pH (79).Because HAdV 40 tolerates acidic conditions, such a PEM mayenable a simple additional processing step to improve the overallrecovery: disassembling the PEM after elution may release trappedvirions to potentially achieve 100% recovery.

Virus removal. As Fig. 8 shows, in tests with HAdV 40-spikedwaters of all types, virion removal (see equation 3) by PEM-coatedmembranes was typically an order of magnitude higher than re-

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

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

Shi et al.

4990 aem.asm.org August 2016 Volume 82 Number 16Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

moval by CS-blocked membranes. However, the loss of 1% ofvirions to the permeate by CS-blocked membranes should notsignificantly decrease recoveries.

The low LRV by CS-blocked membranes might result from anuneven calf serum distribution on the membrane surface and re-sultant incomplete blockage. The blocking procedure involves re-circulating a calf serum solution over the membrane surface andmay be viewed as poorly controlled membrane fouling. In con-trast, coating a membrane with multiple layers of polyelectrolytesis inherently more reproducible.

There was no consistent correlation between LRV values andthe extent of membrane fouling. We showed recently that HAdV40 removal does not correlate simply with the extent of foulingalone. Pore blockage by humic acid could enhance virion removal,whereas cake formation by silica colloids could decrease virionremoval, and the opposing effects may be compensatory whenboth foulants are present (80). Surface water contains colloids andorganic macromolecules of a range of sizes, which may explain thesimilar LRV values for surface water and DI water.

Conclusions. Physicochemical characterization of HAdV 40showed that this relatively large virus (hydrodynamic diameter of102.5 � 8.6 nm at pH 7.6 in 10 mM Tris plus 1 mM EDTA) ishydrophobic (�Gvwv � �30.4 mJ/m2) and has a point of zerocharge between pH 4.0 and pH 4.3. The extended Derjaguin-Lan-dau-Verwey-Overbeek theory predicted that hydrophilic and neg-atively charged ( � �7.0 � 3.0 mV at pH 5.8 in 1 mM NaCl)PEM-coated membranes should be advantageous for recoveringHAdV from DI water in comparison with positively charged CS-coated membranes ( � 3.0 � 2.0 mV at pH 5.8 in 1 mM NaCl).

We tested the validity of the XDLVO prediction in filtrationexperiments. The preelution recovery of HAdV from DI water(ionic strength of �0.2 mM when spiked with HAdV stock) aftercross-flow UF was higher with PEM-coated membranes (rpre

PEM �74.8% � 9.7%) than with CS-blocked membranes (rpre

CS �54.1% � 6.2%). Although preelution recovery from tap water waslower, virion elution from both PEM- and CS-coated membraneswith an aqueous solution of sodium polyphosphate and Tween 80was effective for both DI water (rpost

PEM � 99.5% � 6.6% and rpostCS �

98.8% � 7.7%) and tap water (rpostPEM � 88.8% � 15.3% and rpost

CS �93.7% � 6.9%). The nearly 100% efficacy of elution indicates thatpolyanions and surfactants (e.g., sodium polyphosphate andTween 80) in the eluent can disrupt electrostatic and hydrophobicinteractions between the virion and the membrane. Pre- andpostelution recoveries from surface waters were significantlylower (rpre

CS and rprePEM values as low as �21% and rpost

CS and rpostPEM

values of �65%) and showed no statistically significant differencebetween the two membrane types. However, addition of EDTA tothe eluent greatly increased the elution efficacy (rpost

CS � 88.6% �4.3% and rpost

PEM � 87.0% � 6.9%), possibly by eliminating cationbridging between virions and other components of the feed watermatrix in suspension or in the fouling layer on the membranesurface.

Interestingly, the membrane choice is not very important forachieving high virion recoveries. For tap water, the postelutionvirion recovery was nearly 100% for both PEM- and CS-coatedmembranes, despite the higher removal of virions with PEM-coated membranes (LRVPEM � 2.82 � 0.32 and LRVCS � 1.96 �0.53 for tap water). For more complex water matrices, such assurface water, the composition of the eluent is the most important

factor for achieving high virion recovery. Evidently, the recoveryof HAdV depends on its interactions with other components ofthe feed water (in suspension or deposited on the membrane sur-face as a fouling layer), not on virion-membrane interactions. Aneluent that includes a polyanion (sodium polyphosphate), a non-ionic surfactant (Tween 80), and a chelating agent (EDTA) recov-ers HAdV effectively (�88%), even from high-TOC (9.4 � 0.1mg/liter) surface water.

ACKNOWLEDGMENTS

We thank Jason Schrad for his assistance with TEM imaging and twoanonymous reviewers for their comments and suggestions.

This material is based upon work supported by the National ScienceFoundation Partnerships for International Education and Research pro-gram under grant IIA-1243433.

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

REFERENCES1. WHO. 2016. Water sanitation health. Burden of disease and cost-

effectiveness estimates. WHO, Geneva, Switzerland. http://www.who.int/water_sanitation_health/diseases/burden/en/. Accessed 12 February2016.

2. Crabtree KD, Gerba CP, Rose JB, Haas CN. 1997. Waterborne adeno-virus: a risk assessment. Water Sci Technol 35:1– 6.

3. Mena K, Gerba C. 2009. Waterborne adenovirus. Rev Environ ContamToxicol 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 characteriza-tion of human adenovirus in drinking water supplies: viral integrity andviability 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 healthimplications: 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 adeno-virus-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 adeno-viruses 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. Occurrenceof 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, MatiaL, Girones R. 2009. Analysis of adenoviruses and polyomaviruses quan-tified by qPCR as indicators of water quality in source and drinking-watertreatment 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. Quantitativedetection of human adenoviruses in wastewater and combined seweroverflows 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 hepa-titis A virus in the final effluent of selected wastewater treatment plants inEastern 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 ofpositively charged alumina nanofibre cartridge filters for the primary con-centration of noroviruses, adenoviruses and male-specific coliphages

Elution Is Critical for Human Adenovirus Recovery

August 2016 Volume 82 Number 16 aem.asm.org 4991Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

from seawater. J Appl Microbiol 109:635– 641. http://dx.doi.org/10.1111/j.1365-2672.2010.04691.x.

14. Ikner LA, Soto-Beltran M, Bright KR. 2011. New method using a posi-tively charged microporous filter and ultrafiltration for concentration ofviruses from tap water. Appl Environ Microbiol 77:3500 –3506. http://dx.doi.org/10.1128/AEM.02705-10.

15. McMinn BR. 2013. Optimization of adenovirus 40 and 41 recovery fromtap water using small disk filters. J Virol Methods 193:284 –290. http://dx.doi.org/10.1016/j.jviromet.2013.06.021.

16. Berman D, Rohr ME, Safferman RS. 1980. Concentration of poliovirusin water by molecular filtration. Appl Environ Microbiol 40:426 – 428.

17. Morales-Morales HA, Vidal G, Olszewski J, Rock CM, Dasgupta D,Oshima KH, Smith GB. 2003. Optimization of a reusable hollow-fiberultrafilter 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.

18. Polaczyk AL, Narayanan J, Cromeans TL, Hahn D, Roberts JM, Am-burgey JE, Hill VR. 2008. Ultrafiltration-based techniques for rapid andsimultaneous concentration of multiple microbe classes from 100-L tapwater samples. J Microbiol Methods 73:92–99. http://dx.doi.org/10.1016/j.mimet.2008.02.014.

19. Pasco EV, Shi H, Xagoraraki I, Hashsham SA, Parent KN, BrueningML, Tarabara VV. 2014. Polyelectrolyte multilayers as anti-adhesivemembrane coatings for virus concentration and recovery. J Membr Sci469:140 –150. http://dx.doi.org/10.1016/j.memsci.2014.06.032.

20. Hill VR, Kahler AM, Jothikumar N, Johnson TB, Hahn D, CromeansTL. 2007. Multistate evaluation of an ultrafiltration-based procedure forsimultaneous 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.

21. 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.

22. Pei L, Rieger M, Lengger S, Ott S, Zawadsky C, Hartmann NM, SelinkaHC, Tiehm A, Niessner R, Seidel M. 2012. Combination of crossflowultrafiltration, monolithic affinity filtration, and quantitative reversetranscriptase PCR for rapid concentration and quantification of modelviruses in water. Environ Sci Technol 46:10073–10080. http://dx.doi.org/10.1021/es302304t.

23. Skraber S, Gantzer C, Helmi K, Hoffmann L, Cauchie HM. 2009.Simultaneous concentration of enteric viruses and protozoan parasites: aprotocol based on tangential flow filtration and adapted to large volumesof surface and drinking waters. Food Environ Virol 1:66 –76. http://dx.doi.org/10.1007/s12560-009-9011-z.

24. Belfort G, Rotem Y, Katzenelson E. 1975. Virus concentration usinghollow fiber membranes. Water Res 9:79 – 85. http://dx.doi.org/10.1016/0043-1354(75)90155-4.

25. Francy DS, Stelzer EA, Brady AM, Huitger C, Bushon RN, Ip HS, WareMW, Villegas EN, Gallardo V, Lindquist HD. 2013. Comparison offilters for concentrating microbial indicators and pathogens in lake watersamples. Appl Environ Microbiol 79:1342–1352. http://dx.doi.org/10.1128/AEM.03117-12.

26. Ikner LA, Gerba CP, Bright KR. 2012. Concentration and recovery ofviruses from water: a comprehensive review. Food Environ Virol 4:41– 67.http://dx.doi.org/10.1007/s12560-012-9080-2.

27. Hill VR, Polaczyk AL, Hahn D, Narayanan J, Cromeans TL, RobertsJM, Amburgey JE. 2005. Development of a rapid method for simultane-ous recovery of diverse microbes in drinking water by ultrafiltration withsodium polyphosphate and surfactants. Appl Environ Microbiol 71:6878 – 6884. http://dx.doi.org/10.1128/AEM.71.11.6878-6884.2005.

28. Nestola P, Martins DL, Peixoto C, Roederstein S, Schleuss T, Alves PM,Mota JPB, Carrondo MJT. 2014. Evaluation of novel large cut-off ultra-filtration membranes for adenovirus serotype 5 (Ad5) concentration.PLoS One 9:e115802. http://dx.doi.org/10.1371/journal.pone.0115802.

29. Rhodes ER, Huff EM, Hamilton DW, Jones JL. 2016. The evaluation ofhollow-fiber ultrafiltration and celite concentration of enteroviruses, ad-enoviruses and bacteriophage from different water matrices. J Virol Meth-ods 228:31–38. http://dx.doi.org/10.1016/j.jviromet.2015.11.003.

30. Wong K, Mukherjee B, Kahler AM, Zepp R, Molina M. 2012. Influenceof inorganic ions on aggregation and adsorption behaviors of human ad-enovirus. Environ Sci Technol 46:11145–11153. http://dx.doi.org/10.1021/es3028764.

31. Trilisky EI, Lenhoff AM. 2007. Sorption processes in ion-exchange chro-matography 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 andprotocols. Humana Press, Totowa, NJ.

33. Mesquita MMF, Emelko MB. 2014. Bacteriophages as surrogates for thefate and transport of pathogens in source water and in drinking watertreatment processes, p 57– 80. In Kurtboke I (ed), Bacteriophages. Intech,Rijeka, Croatia.

34. Abbaszadegan M, Mayer BK, Ryu H, Nwachuku N. 2007. Efficacy ofremoval of CCL viruses under enhanced coagulation conditions. EnvironSci Technol 41:971–977. http://dx.doi.org/10.1021/es061517z.

35. Bellou MI, Syngouna VI, Tselepi MA, Kokldnos PA, PaparrodopoulosSC, Vantarakis A, Chrysikopoulos CV. 2015. Interaction of human ad-enoviruses and coliphages with kaolinite and bentonite. Sci Total Environ517:86 –95. http://dx.doi.org/10.1016/j.scitotenv.2015.02.036.

36. Kidd AH, Chroboczek J, Cusack S, Ruigrok RWH. 1993. Adenovirustype-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 andlong 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 noncul-tivable adenoviruses associated with diarrhea in infants: a new subgroupof 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 and41: 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, Vander Avoort H, Wood DJ, Bailey AS, Osterhaus A. 1993. Detection,typing, and subtyping of enteric adenoviruses 40 and 41 from fecal sam-ples and observation of changing incidences of infections with these typesand 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 entericadenovirus type 40 in a modified cell line that can no longer respond tointerferon 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 cul-tivation 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 Virol76:481– 497. http://dx.doi.org/10.1099/0022-1317-76-3-481.

44. Wilhelmi I, Roman E, Sanchez-Fauquier A. 2003. Viruses causing gas-troenteritis. 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 viralgenome: 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 adeno-virus. 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 andaggregation kinetics of rotavirus in divalent cation solutions. Environ SciTechnol 44:4552– 4557. http://dx.doi.org/10.1021/es100120k.

48. Putaux JL, Buleon A, Borsali R, Chanzy H. 1999. Ultrastructural aspectsof phytoglycogen from cryo-transmission electron microscopy and quasi-elastic 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 physicochem-ical properties of human enteric Ad41 responsible for its survival andreplication 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, HungT. 2009. Novel recombinant adenovirus type 41 vector and its biologicalproperties. J Gene Med 11:128 –138. http://dx.doi.org/10.1002/jgm.1284.

51. Hansen JJ, Warden PS, Margolin AB. 2007. Inactivation of adenovirustype 5, rotavirus Wa and male specific coliphage (MS2) in biosolids bylime stabilization. Int J Environ Res Public Health 4:61– 67. http://dx.doi.org/10.3390/ijerph2007010010.

Shi et al.

4992 aem.asm.org August 2016 Volume 82 Number 16Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from

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 re-tardation effect in electrophoresis of spherical colloidal particles. J ColloidInterface Sci 168:269 –271. http://dx.doi.org/10.1006/jcis.1994.1419.

54. Norrby E. 1968. Biological significance of structural adenovirus compo-nents. Curr Top Microbiol Immunol 43:1– 43.

55. Makino K, Ohshima H. 2010. Electrophoretic mobility of a colloidalparticle 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, BocaRaton, FL.

57. van Oss CJ. 2003. Long-range and short-range mechanisms of hydropho-bic attraction and hydrophilic repulsion in specific and aspecific interac-tions. J Mol Recognit 16:177–190. http://dx.doi.org/10.1002/jmr.618.

58. Brant JA, Childress AE. 2002. Assessing short-range membrane-colloidinteractions 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, Quesada-Perez M. 2008. Charge reversal in real colloids: experiments, theory andsimulations. 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. Experi-mental study of electrostatically stabilized colloidal particles: colloidal sta-bility 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 molecularsources of nonideal osmotic-pressure of bovine serum-albumin solutionsas 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 albuminadsorption onto immobilized organotrichlorosilane surface: influence ofthe phase separation on protein adsorption patterns. J Biomater Sci PolymEd 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-inducedconformational changes of bovine serum albumin: a small-angle X-rayscattering 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 focusingof 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 tocontrol. 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 conditioningfilms 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-

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.

68. Sutlovic D, Gojanovic MD, Andelinovic S, Gugic D, Primorac D. 2005.Taq polymerase reverses inhibition of quantitative real time polymerasechain reaction by humic acid. Croat Med J 46:556 –562.

69. Albers CN, Jensen A, Baelum J, Jacobsen CS. 2013. Inhibition of DNApolymerases used in Q-PCR by structurally different soil-derived humicsubstances. Geomicrobiol J 30:675– 681. http://dx.doi.org/10.1080/01490451.2012.758193.

70. Yuan BL, Pham M, Nguyen TH. 2008. Deposition kinetics of bacterio-phage MS2 on a silica surface coated with natural organic matter in a radialstagnation point flow cell. Environ Sci Technol 42:7628 –7633. http://dx.doi.org/10.1021/es801003s.

71. Kim J, Shan WQ, Davies SHR, Baumann MJ, Masten SJ, Tarabara VV.2009. Interactions of aqueous NOM with nanoscale TiO2: implications forceramic membrane filtration-ozonation hybrid process. Environ SciTechnol 43:5488 –5494. http://dx.doi.org/10.1021/es900342q.

72. Pham M, Mintz EA, Nguyen TH. 2009. Deposition kinetics of bacterio-phage MS2 to natural organic matter: role of divalent cations. J ColloidInterface Sci 338:1–9. http://dx.doi.org/10.1016/j.jcis.2009.06.025.

73. Gutierrez L, Nguyen TH. 2012. Interactions between rotavirus and Su-wannee River organic matter: aggregation, deposition, and adhesion forcemeasurement. Environ Sci Technol 46:8705– 8713. http://dx.doi.org/10.1021/es301336u.

74. Tong MP, Shen Y, Yang HY, Kim H. 2012. Deposition kinetics of MS2bacteriophages on clay mineral surfaces. Colloids Surf B 92:340 –347. http://dx.doi.org/10.1016/j.colsurfb.2011.12.017.

75. Rose AL, Waite TD. 2003. Kinetics of iron complexation by dissolvednatural organic matter in coastal waters. Marine Chem 84:85–103. http://dx.doi.org/10.1016/S0304-4203(03)00113-0.

76. Karlsson T, Persson P, Skyllberg U. 2006. Complexation of copper(II) inorganic soils and in dissolved organic matter—EXAFS evidence for che-late ring structures. Environ Sci Technol 40:2623–2628. http://dx.doi.org/10.1021/es052211f.

77. Guan X-H, Shang C, Chen G-H. 2006. ATR-FTIR investigation of therole of phenolic groups in the interaction of some NOM model com-pounds with aluminum hydroxide. Chemosphere 65:2074 –2081. http://dx.doi.org/10.1016/j.chemosphere.2006.06.048.

78. Li QL, Elimelech M. 2004. Organic fouling and chemical cleaning ofnanofiltration membranes: measurements and mechanisms. Environ SciTechnol 38:4683– 4693. http://dx.doi.org/10.1021/es0354162.

79. Ahmadiannamini P, Bruening ML, Tarabara VV. 2015. Sacrificial poly-electrolyte multilayer coatings as an approach to membrane fouling con-trol: disassembly and regeneration mechanisms. J Membr Sci 491:149 –158. http://dx.doi.org/10.1016/j.memsci.2015.04.041.

80. Yin Z, Tarabara VV, Xagoraraki I. 2015. Human adenovirus removal byhollow fiber membranes: effect of membrane fouling by suspended anddissolved matter. J Membr Sci 482:120 –127. http://dx.doi.org/10.1016/j.memsci.2015.02.028.

Elution Is Critical for Human Adenovirus Recovery

August 2016 Volume 82 Number 16 aem.asm.org 4993Applied and Environmental Microbiology

on March 27, 2020 by guest

http://aem.asm

.org/D

ownloaded from