MEMBRANE-BASED CONCENTRATION AND RECOVERY OF
VIRUSES FROM COMPLEX WATER MATRICES
By
Hang Shi

A DISSERTATION
Submitted to
Michigan State University
In partial fulfillment of the requirements
for the degree of
Environmental Engineering - Doctor of Philosophy
2017

ABSTRACT
MEMBRANE-BASED CONCENTRATION AND RECOVERY OF VIRUSES
FROM COMPLEX WATER MATRICES
By
Hang Shi
Prevention of waterborne disease outbreaks relies on the efficient detection of
pathogens in drinking and recreational water. Development of sample concentration
technologies that ensure fast and high recovery of pathogens from aquatic samples is
crucial for timely detection. The most effective approaches to sample concentration and
virus recovery employ membrane filtration and rely on controlling physicochemical
interactions between the virus and the filter. The two main goals of the present work
were to understand the reasons for the poor efficiency (typically below 30%) of the
current methods in recovering human adenovirus and to propose alternative strategies
that facilitate concentration and recovery of this important human virus. The first part of
the dissertation is devoted to the study on how common methods of virus propagation
(broth-based and agar-based) and purification (polyethylene glycol precipitation,
centrifugal diafiltration and CsCl density gradient centrifugation) affect physicochemical
properties of virions. Experimental data for bacteriophage MS2 showed that results of
virus size, charge, and hydrophobicity measurements depend strongly on the methods
and protocols used to grow and purify the virus. The optimal sample preparation
protocol was determined to consist of broth-based growth followed by purification via
CsCl density gradient centrifugation. This method was then used to measure
physicochemical properties of human adenovirus 40 (HAdV40) and employ these
values to calculate the energy of virion-virion and virion-membrane interactions. The

second part of the dissertation describes an experimental study on the recovery of
HAdV40 from three water matrices (spiked deionized water, tap water, and high organic
content surface water) by crossflow ultrafiltration. Prior to ultrafiltration, membranes
were either blocked by calf serum or coated with a polyelectrolyte multilayer to minimize
virus adsorption on the membrane surface. The multilayer was designed to be
antiadhesive with respect to HAdV 40 using the virus-membrane interaction energy
calculations performed earlier. In the sample concentration tests, HAdV 40 was
recovered from ultrapure water, tap water, and surface water with very high post-elution
recoveries of ~99%, ~91%, and ~84%, respectively. The obtained recovery data were
interpreted in terms of physicochemical interactions of HAdV40 virions with the
membrane and how components of the eluent disrupt specific interactions between
HAdV40 and the membrane to maximize HAdV40 recovery. Results on HAdV40
concentration indicate that the composition of the eluent is the most important factor for
achieving high virus recovery and can be designed to efficiently recover viruses even
from highly complex water matrices.

Copyright by
HANG SHI
2017

ACKNOWLEDGEMENTS

Pursuing a PhD has not all been smooth sailing. It is full of ups and downs. There are
many people I would like to acknowledge for helping me complete this milestone.

To my advisor, Dr. Volodymyr Tarabara, thank you for your continued support, constant
guidance, your effort to translate my fragmented thoughts when I had some new
exciting data. You are always willing to help me improve not only my technical skills, but
also my presentation skills and allow me the opportunity to expand my horizons from
which I will benefit throughout my life. Your encouragement and advice will continue to
be invaluable to me for my future career.

I would like to thank my committee members, Dr. Kristin Parent, Dr. Merlin Bruening, Dr.
Wei Zhang and Dr. Irene Xagoraraki for their suggestions and critical comments.
Especially for Dr. Parent, thank you for always being available for offering help which
really expedites my research.

To the former and current lab members in our research group, working with all of you is
a great experience in my life, which not only gives me a chance to be exposed to other
topics in this interdisciplinary filed, but also helps myself improve important skills
required to become a solid team player.

v

I would like to thank Lori Larner, research administrator in our department and Joseph
Nguyen, lab technician in our department. Thank you for your administrative and
technical support.

To my parents, without your unconditional love and support, I would never have had a
chance to come to this country for graduate study. I always feel loved and I love you
both very much.

I would also like to thank Yu and my naughty hyper dog Mei-Chou-Chou for their
company and bring laughter into my life in these tough years.

Last, I would like to thank the source of funding, The National Science Foundation
(NSF) Partnerships for International Research and Education (PIRE) program under
Grant IIA-1243433 that makes my research financially possible.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................ x
LIST OF FIGURES ......................................................................................................... xii
CHAPTER ONE .............................................................................................................. 1
Dissertation Overview ..................................................................................................... 1
CHAPTER TWO .............................................................................................................. 3
Membrane-based methods of virus concentration from water: A review of process
parameters and their effects on virus recovery ............................................................... 3
Abstract ....................................................................................................................... 3
2.1 Introduction ........................................................................................................... 4
2.1.1 The need for timely detection of viral pathogens in water .............................. 4
2.1.2 The challenge of virus detection in complex water matrices .......................... 6
2.1.3 Sample concentration as a key step in the detection process ....................... 7
2.1.4 The scope and structure of this review .......................................................... 8
2.2 Currently practiced methods of primary concentration: Principles, advantages,
limitations............................................................................................................ 10
2.2.1 VIRus ADsorption and ELution .................................................................... 10
2.2.2 Crossflow ultrafiltration ................................................................................ 13
2.3 Parameters influencing the performance of membrane-based methods of virus
concentration ...................................................................................................... 15
2.3.1 Virus properties............................................................................................ 16
2.3.2 Membrane properties ................................................................................... 18
2.3.3 Design of the filtration process: Hydrodynamic conditions ........................... 21
2.3.4 Composition of the water sample................................................................. 24
2.3.4.1 Effect of water composition on VIRADEL ............................................. 24
2.3.4.2 Effect of water composition on CFUF ................................................... 26
2.3.5 Eluent composition and elution protocol ...................................................... 29
2.3.5.1 Elution in VIRADEL .............................................................................. 29
2.3.5.2 Elution in CFUF .................................................................................... 32
2.3.6 Initial virus content in the water sample ....................................................... 34
2.4 Research needs and recommendations for future work ...................................... 35
2.4.1 Method development: Standard procedures ................................................ 36
2.4.2 Data reporting and analysis ......................................................................... 37
2.4.3 Multi-method testing and controls ................................................................ 38
2.4.4 Virus interactions with components of a concentrated water sample ........... 40
2.4.5 Elution and sacrificial coatings as strategies for highly fouling water matrices
..................................................................................................................... 41
APPENDIX .................................................................................................................... 42
REFERENCES .............................................................................................................. 76
CHAPTER THREE ........................................................................................................ 88
vii

The choice of virus propagation and purification methods affects results of virus size
and surface charge measurements ............................................................................... 88
Abstract ..................................................................................................................... 88
3.1 Introduction ......................................................................................................... 89
3.2 Material and Methods ......................................................................................... 94
3.2.1 Virus propagation ......................................................................................... 95
3.2.1.1 Virus propagation in double agar overlay ............................................. 95
3.2.1.2 Virus propagation in broth .................................................................... 96
3.2.2 Virus purification .......................................................................................... 96
3.2.2.1 Virus purification by CsCl density gradient centrifugation ..................... 97
3.2.2.2 Virus purification by PEG precipitation ................................................. 98
3.2.2.3 Virus purification by centrifugal diafitration ........................................... 98
3.2.3 Hydrodynamic diameter and electrophoretic mobility measurements ........ 101
3.2.4 Surface tension and hydrophobicity determination .................................... 101
3.3 Results and Discussion ..................................................................................... 103
3.3.1 Effects of the virus propagation method on the hydrodynamic size and
surface charge of MS2 purified by three different procedures .................... 103
3.3.1.1 Effect on MS2 hydrodynamic size ...................................................... 103
3.3.1.2 Effect on MS2 charge ......................................................................... 107
3.3.2 Effects of the virus purification method on virus size and charge
measurements ........................................................................................... 110
3.3.3 Effects of virus purification methods on surface tension parameters and
hydrophobicity of bacteriophages............................................................... 116
3.4 Summary........................................................................................................... 119
3.5 Conclusions ...................................................................................................... 120
APPENDIX .................................................................................................................. 122
REFERENCES ............................................................................................................ 125
CHAPTER FOUR ........................................................................................................ 130
Elution is a critical step for human adenovirus 40 recovery from tap and surface water
by crossflow ultrafiltration ............................................................................................ 130
Abstract ................................................................................................................... 130
Importance .............................................................................................................. 131
4.1 Introduction ....................................................................................................... 131
4.2 Materials and Methods ...................................................................................... 136
4.2.1 Reagents and water samples .................................................................... 136
4.2.2 Human adenovirus 40: Propagation, purification and characterization ...... 137
4.2.3 XDLVO energy of virion-virion and virion-membrane interactions ............. 139
4.2.4. Membrane preparation ............................................................................. 139
4.2.5. Virus concentration and recovery tests ..................................................... 140
4.2.6 Virus quantification and recovery ............................................................... 141
4.3 Results and Discussion ..................................................................................... 142
4.3.1 Characteristics of source waters ................................................................ 142
4.3.2 Physicochemical properties of HAdV 40: Size, 𝜻-potential and surface charge
density........................................................................................................ 144
4.3.3 Surface energy components of HAdV 40 ................................................... 151
viii

4.3.4. XDLVO energy profile of virion-virion interfacial interaction in aqueous
media ......................................................................................................... 152
4.3.5 XDLVO energy of virion-membrane interfacial interaction in membrane feed
solutions: DI water and tap water. .............................................................. 154
4.3.6 Virus recovery from DI water ..................................................................... 157
4.3.7 Virus recovery from tap and surface water ................................................ 161
4.3.8 Virus removal ............................................................................................. 166
4.4 Conclusions ...................................................................................................... 167
APPENDIX .................................................................................................................. 170
REFERENCES ............................................................................................................ 190
CHAPTER FIVE .......................................................................................................... 196
Conclusions................................................................................................................. 196

ix

LIST OF TABLES

Table 1. Norovirus recovery by VIRADEL as a function of water matrix and VIRADEL
process parameters........................................................................................ 43
Table 2. Norovirus recovery by crossflow ultrafiltration as a function of water matrix and
ultrafiltration process parameters. .................................................................. 52
Table 3. Adenovirus recovery by VIRADEL as a function of water matrix and VIRADEL
process parameters........................................................................................ 54
Table 4. Adenovirus recovery by crossflow ultrafiltration as a function of water matrix
and ultrafiltration process parameters. ........................................................... 60
Table 5. Primary recovery of bacteriophage MS2 by VIRADEL as a function of water
matrix and VIRADEL process parameters...................................................... 63
Table 6. Primary recovery of bacteriophage MS2 by crossflow ultrafiltration as a function
of water matrix and ultrafiltration process parameters. ................................... 69
Table 7. The propagation and purification methods evaluated in this study. ................. 94
Table 8. Summary of average hydrodynamic diameter and polydispersity index for MS2
in storage solution recorded for the six different preparations evaluated in this
study............................................................................................................. 105
Table 9. Measured contact angles, calculated surface energy parameters and free
energy of interfacial interaction when immersed in water for broth-propagated
P22 bacteriophage as a function of the purification method. ........................ 117
Table 10. Water quality characteristics........................................................................ 143
Table 11. Parameters of HAdV 40 size distributions at different pH values a. Shaded
(gray) areas denote pH values where significant aggregation occurs. ......... 148
Table 12. Contact angles, calculated surface energy parameters and the enerinterfacial
interaction when immersed in water for HAdV 40. ....................................... 152
Table 13. Primer and probe sequences used in qPCR for HAdV 40 detection ........... 174
Table 14. Charge characteristics of HAdV 40 at different pH values. Shaded (gray)
areas denote pH values where significant aggregation occurs and
𝜁𝐻𝐴𝑑𝑉𝑎𝑔𝑔 ≠ 𝜁𝐻𝐴𝑑𝑉. ................................................................................... 188

x

Table 15. Additional parameters of HAdV 40 size distribution at different pH values.
Shaded (gray) areas denote pH values where significant aggregation occurs.
..................................................................................................................... 189

xi

LIST OF FIGURES

Figure 1. Schematic graph of purification with CsCl density gradient, polyethylene glycol
precipitation and centrifugal diafiltration. ...................................................... 100
Figure 2. Size distribution of MS2 propagated by two different methods (in broth and on
double layer agar overlay) and purified by CsCl density gradient centrifugation
(a), PEG precipitation (b) and centrifugal ultrafiltration (c). Solution: 1mM
NaCl, pH 5.7. ................................................................................................ 104
Figure 3. Electrophoretic mobility and zeta potential as function of pH for MS2
propagated by two different methods (in broth and on double layer agar
overlay) and purified by CsCl density gradient centrifugation (a), PEG
precipitation (b) and centrifugal ultrafiltration (c). ......................................... 109
Figure 4. Size distribution (a, c) and electrophoretic mobility (b, d) of broth-propagated
MS2 (a, b) and P22 (c, d) phages purified by CsCl density gradient
centrifugation, PEG precipitation and centrifugal ultrafiltration. Solution used
in size measurements: 1mM NaCl, pH 5.7. Data for MS2 overlaps with parts of
Figures 2 and 3 and are shown here for easier comparisons. ...................... 111
Figure 5. Average hydrodynamic diameter as function of pH for MS2 purified by CsCl
density gradient (a), PEG precipitation (b) and centrifugal ultrafiltration (c). 123
Figure 6. (a) Average hydrodynamic diameter as function of pH for MS2 purified by
CsCl density gradient, PEG precipitation and centrifugal ultrafiltration......... 124
Figure 7. Transmission electron microscopy image of HAdV 40. ................................ 144
Figure 8. Normalized number-based size distribution of HAdV 40 as a function of pH.
Vertical red dashed lines indicate the average modal diameter (99 nm, see
APPENDIX, Table 15) in the buffer recommended for the storage of purified
HAdV 40 (10 mM Tris-HCl and 1 mM EDTA, pH 7.6). Vertical blue dash-dot
lines denote the average HAdV diameter (~ 80 nm) as determined by TEM.
The APPENDIX shows the HAdV 40 size distribution in tap water (Figure 20).
..................................................................................................................... 147
Figure 9. 𝜁-potential of HAdV 40 as a function of pH. ................................................. 149
Figure 10. XDLVO energy profile of virion-virion interfacial interactions in 1 mM NaCl a
for HAdV 40 at different pH values. (For pH 7.6, size and zeta potential values
used in XDLVO calculations were for virions in the recommended buffer (1 mM
EDTA + 10mM Tris), and not in 1 mM NaCl.) ............................................... 153

xii

Figure 11. XDLVO energy of interfacial interaction of HAdV 40 in aqueous media with a
PEM-coated membrane (A) and a CS-blocked membrane (B). The energies
were calculated from surface parameters determined at pH 5.8. ................. 156
Figure 12. Recovery of HAdV 40 with CS-blocked and PEM-coated membranes from
HAdV-spiked deionized water (A) and tap water (B). ................................... 159
Figure 13. Recovery of HAdV 40 from surface water after crossflow UF with CS-blocked
membranes (A) and PEM-coated membranes (B). ...................................... 163
Figure 14. Removal of HAdV 40 by crossflow ultrafiltration with calf-serum blocked
membranes and membranes coated by a polyelectrolyte multilayer. ........... 166
Figure 15. HAdV 40 band obtained by CsCl density gradient centrifugation. .............. 172
Figure 16. Transmission electron microscopy image of HAdV 40 viruses. .................. 173
Figure 17. Crossflow filtration apparatus. .................................................................... 174
Figure 18. Normalized permeate flux as function of filtration time for CS-blocked
membranes (A) and PEM-coated membranes (B). ...................................... 175
Figure 19. Log concentration of DNA detected by qPCR in feed and retentate water
matrices spiked with the same concentrations of virions. ............................. 177
Figure 20. Normalized number-based size distribution of HAdV 40 in tap water (pH = 7.5
to 8.0). The vertical dashed red line indicates the average modal diameter
(98.5 nm) in the buffer recommended for the storage of purified HAdV 40 (10
mM Tris-HCl and 1 mM EDTA, pH 7.6). The vertical dash-dot blue line
denotes the average HAdV diameter (~ 80 nm) determined from TEM. ...... 182
Figure 21. XDLVO energies of interfacial interaction of HAdV 40 with CS-blocked
membranes in (A) DI water (𝐼 ≅ 0.2 mM after spiking with HAdV 40) and (B)
tap water. ..................................................................................................... 183
Figure 22. XDLVO energies of interfacial interaction of HAdV 40 with PEM-coated
membranes in (A) DI water (𝐼 ≅ 0.2 mM after spiking with HAdV 40) and (B)
tap water. ..................................................................................................... 184
Figure 23. Electrophoretic mobility of HAdV 40 as a function of pH. ........................... 185

xiii

CHAPTER ONE
Dissertation Overview

Pathogenic viruses are a major cause of waterborne diseases. To prevent disease
outbreaks, rapid and reliable detection of viruses in water is essential. The detection
process typically requires concentration of the source water sample to increase the
virus titer. VIRADEL (VIRus ADsorption-ELution) and CFUF (crossflow ultrafiltration) are
two methods that are commonly used for concentrating large volume environmental
samples. Although fast concentration of all viruses in a sample is ideal, engineering
challenges such as membrane fouling and irreversible adsorption of viruses to
components of the filtration unit limit the concentration speed and percent recovery.
Better understanding the fate of viruses throughout the sample concentration process
and identifying reasons for virus loss can guide process design for improving virus
recoveries. The material presented in this dissertation divided into Chapter where each
Chapter is a manuscript either already published or in preparation for submission.

Chapter 2 is a comprehensive review of the literature on membrane-based methods of
virus concentration from aqueous media. Published data on noroviruses, adenoviruses
and bacteriophage MS2 are critically analyzed to establish how parameters of the
concentration process impact virus recovery. The review highlights the importance of
physicochemical interactions of virions during sample concentration and concludes by
identifying research needs.

1

The work presented in Chapter 3 is based on the premise that physicochemical
interactions between the virus and the filter govern virus recovery. Indeed, VIRADEL
and CFUF are designed to harness these interactions to achieve higher and more
reproducible recoveries. It is important therefore to quantify key physicochemical
properties of a virus accurately. The Chapter focuses on the impact of different sample
preparation methods on virus physicochemical characterization. Two virus propagation
methods (broth-based and double agar overlay growth) and three purification methods
(PEG precipitation, centrifugal diafiltration, CsCl density gradient centrifugation) were
systematically evaluated with bacteriophages MS2 and P22.

Chapter 4 builds on the findings presented in Chapter 3. Using the optimal sample
preparation method, the study offers the first comprehensive physicochemical
characterization of human adenovirus 40 (HAdV 40), an important human pathogen.
The data on HAdV 40 surface properties was used to calculate the energy of virionvirion and virion-membrane interactions. In a set of cross-flow filtration tests, HAdV 40
was recovered from ultrapure water (99%), tap water (~91%), and high-carbon-content
surface water with post-elution recoveries of 99%, ~91%, and ~84%, respectively. The
composition of the eluent was designed to disrupt specific interactions between HAdV40
and the membrane and maximize HAdV 40 recovery.

Finally, Chapter 5 presents overarching conclusions drawn based on the material
described in Chapters 2, 3, and 4.

2

CHAPTER TWO
Membrane-based methods of virus concentration from water: A
review of process parameters and their effects on virus recovery

Abstract

This review focuses on membrane-based methods of virus concentration from aqueous
media. The discussion is centered on two most commonly used sample concentration
techniques: virus adsorption and elution (VIRADEL) and crossflow ultrafiltration (CFUF).
We summarize all published VIRADEL and CFUF studies on the concentration of two
human viral pathogens that are difficult to recover and quantify: noroviruses (17 reports
total) and adenoviruses (15 reports total). We also include in the discussion all studies
published since 2001 on the concentration of MS2, a bacteriophage commonly used in
fate and transport studies as a surrogate of human viruses. The reported recoveries
are analyzed in the context of key concentration process parameters including virus and
membrane properties, initial virus content in the sample, details of the filtration protocol,
as well as composition of the water and the eluent used to collect viruses adsorbed on
the membrane surface. We conclude by identifying knowledge gaps and outlining a
possible research agenda for improving the science and practice of virus separation for
detection.

3

2.1 Introduction

2.1.1 The need for timely detection of viral pathogens in water

Waterborne diseases are infections transmitted through contact with contaminated
waters. The main etiological agents of waterborne disease outbreaks are viruses,
bacteria, protozoa, fungi and helminths 1. The reservoirs of concern are recreational
waters as well as surface and ground waters used as sources of drinking water.
Developments in environmental regulation and progress in drinking water treatment
technologies since the beginning of the 20th century led to a decrease in the occurrence
of waterborne disease outbreaks resulting from exposure to contaminated drinking
water 2. As a result, such outbreaks have become relatively uncommon in the U.S. and
other developed countries 3. An average of 30 disease outbreaks associated with
drinking water were reported every year to the Center for Disease Control during the
2001-2012 period in the U.S. 4-9. Actual outbreak occurrence might have been higher
as it has been suggested that only 10% to 50% of outbreaks are actually reported 2.
Within this time span, the number of affected persons per year varied from the minimum
of 431 (2011-2012) to the maximum of 4,128 (2007-2008) while the number of resulting
deaths per year varied from 3 (2011-2012) to 14 (2007-2008) 4-9. It is estimated that
since 2001 from 6% to 16% of the outbreaks, for which the etiological agent was
identified, have been caused by viruses 4-9. In addition to underreporting, viruses are
also typically more difficult to detect as an etiologic agent. The difficulty is caused by a
combination of factors including the typically very low concentrations of viruses in water

4

10;

an overlap in clinical symptoms leading to possible misinterpretation of viral etiology

for that of bacterial or protozoan origin 11; and complexity of working with noroviruses, a
leading documented cause of water-borne disease of viral origin 12, 13.

Rapid detection of waterborne pathogens with high sensitivity and reproducibility is
essential. Although modern water treatment technologies are efficient at removing
pathogens, complete removal cannot be guaranteed 14 and the consequences of even
one incident can be dramatic. Since 1940s and 1950s when first studies demonstrated
that waterborne diseases can be of viral etiology 15, multiple events have highlighted the
risks that such outbreaks entail. More recently, contaminated water supply system led
to the 1991 outbreak of hepatitis E virus in Kanpur, India which affected approximately
79,091 persons 16, The 1999 outbreak of aseptic meningitis associated with multiple
enterovirus subtypes in Romania affected approximately 6,000 persons

17.

During

2008–2014, a total of 133 cruise ship acute gastroenteritis outbreaks were reported in
US and among outbreaks in which specimens available for testing, 97% were caused
by norovirus 18. As of May 2017, the latest norovirus outbreak on a cruise ship occurred
in March 2017, in which 157 passengers (out of 2,016 total) and 25 crew members (out
of 881 total) reported being ill 19. The 2004 multiple pathogens outbreak associated with
groundwater along the shoreline of South Bass Island off the coast of Lake Erie, Ohio,
led to approximately 1,450 reported cases of gastroenteritis

20, 21.

A study published a

year before the South Bass Island outbreak showed that infective viruses were present
in 5% of groundwater samples taken in 35 different states 22. As the susceptibility to

5

infections among the general population increases 3, future outbreaks can be expected
to have only more devastating effects.

2.1.2 The challenge of virus detection in complex water matrices

Contrary to the trend observed for drinking water, occurrence of waterborne disease
outbreaks caused by the exposure to contaminated recreational water in the U.S. is on
the rise from an average of 68 outbreaks resulting in about 3,215 illnesses per year for
the 2001-2006 period to an average of 101 outbreaks resulting in about 5,693 illnesses
per year for the 2007-2012 period 23-28. Virus-associated outbreaks accounted for 1% to
10% of these incidents, with norovirus being the most prevalent viral etiological agent
identified 23-28. A recent review of published literature on waterborne disease outbreaks
associated with recreational water also identified norovirus to be the largest cause of
viral outbreaks followed by adenoviruses and echoviruses

29.

In swimming pools, where

almost half (48%) of outbreaks occurred, 68% of those outbreaks were associated with
inadequate disinfection. In lakes and ponds, the second largest affected recreational
type of water, water quality surveillance is the most effective means of preventing
outbreaks 29.

With direct and indirect water reuse gaining broader acceptance 30, 31, the potential for
drinking water contamination increases and so does the need for monitoring the
microbiological quality of water treated for reuse. The complex compositions of
recreational water and treated wastewater complicate the detection of pathogens in

6

such matrices. Various dissolved and suspended species in water get concentrated
along with target pathogens, foul the filters, slow down the concentration process, and
interfere with assays performed on the concentrates.

2.1.3 Sample concentration as a key step in the detection process

Reliable virus preconcentration and recovery from water remains one of the most
significant challenges in virus detection 32. Typical virus concentrations are 101 to 104
particles/L in surface water 33, 34 and 105 or more particles/L in primary and secondary
treated sewage effluent 35; in drinking water, virus concentrations <1 particle per 100 L
represent a significant risk of illness 36. Thus, to reliably detect viruses in treated
waters, large volumes (10 to 1,000 L) of source water must be concentrated to volumes
amenable to rapid assays 37, 38. Although fast concentration of all viruses in a sample is
ideal, engineering challenges such as membrane fouling and irreversible adsorption of
viruses to components of the filtration unit increase time required for concentration and
limit recovery. The variability and complexity of water may result in virus recoveries that
are unacceptably low, poorly reproducible or both. These problems often prohibit
definitive association of waterborne viruses with specific disease outbreaks.

Primary concentration techniques employed currently include VIRADEL (VIRus
ADsorption-ELution), which is the EPA-approved method of virus concentration (Tables
1, 3, and 5; see APPENDIX), crossflow ultrafiltration (CFUF; Tables 2, 4, and 6),
ultracentrifugation 39, 40, centrifugal ultrafiltration 41, membrane-based electro-separation

7

42,

microfluidic evaporation 43 and flocculation/re-dissolution/ultrafiltration 44. Only the

first two methods are practical for concentrating large volume environmental samples
and it is one of these two methods that is usually used for primary concentration. At the
secondary concentration step the sample volume is reduced further to enable analyses
such as quantitative polymerase chain reaction (qPCR). The primary concentration
step is the bottleneck because of both its longer duration and the typically larger loss
(i.e. lower recovery) of target pathogens.

Both VIRADEL and CFUF techniques perform well only with a limited range of viruses.
The most difficult challenge that VIRADEL and UF-based methods share is poor
reproducibility of elution 45, which is subject to multiple interferences 46-55. Changes in
pH 46, 47, concentration 54-56 and type 55, 56 of salts present in the sample, as well as the
composition and loading of dissolved organics 46-48, 51, 52 lead to significant variations in
virus recovery.

2.1.4 The scope and structure of this review

The scope of the paper includes VIRADEL and CFUF as the two main membranebased methods of sample concentration for virus recovery. The review builds on and
complements the recent reviews by Ikner et al. 57, Cashdollar and Wymer 15 and Gibson
and Borchardt 58 by focusing on how various parameters of the concentration process
affect virus recovery. The discussion distinguishes the primary and the secondary
concentration stages and centers on the former as the bottleneck in the overall sample

8

concentration sequence. Virus recoveries reported in the literature are analyzed in the
context of concentration process parameters with a special attention to the
reproducibility of the recovery data. We summarize, in a tabulated format, all published
VIRADEL and CFUF studies on the concentration of noroviruses (13 VIRADEL studies
with 77 different concentration conditions and 4 CFUF studies with 8 different
concentration conditions) and adenoviruses (10 VIRADEL studies with 42 different
concentration conditions and 5 CFUF studies with 15 different concentration conditions),
both important human pathogens with relatively low recoveries

15.

In support of the

discussion, we also summarize all recent (published since 2001) VIRADEL and CFUF
studies on the concentration of MS2 (6 VIRADEL studies with 35 different concentration
conditions and 11 CFUF studies with 47 different concentration conditions), perhaps the
most studied bacteriophage and a human virus surrogate commonly used in virus fate
and transport studies.

The structure of this paper is as follows. The following section (Section 2) describes the
VIRADEL and CFUF as applied to virus concentration from large water samples.
Principles underpinning the two processes as well as their advantages and limitations
are discussed in detail. Section 3 outlines the various parameters affecting the
performance of VIRADEL and CFUF. Here we describe main physicochemical
interactions that determine the efficiency of virus recovery and propose possible
approaches to improving the sample concentration process. Section 4 identifies
knowledge gaps and outlines a possible research agenda for improving the science and
practice of virus separation for detection.

9

2.2 Currently practiced methods of primary concentration: Principles,
advantages, limitations.

2.2.1 VIRus ADsorption and ELution

The VIRADEL method includes two steps: (1) adsorption of viruses by passing the
sample (up to 1,000 L for drinking water) through a microfilter, which can be either
electropositive 59-68 or electronegative 56, 60, 64, 67, 69, 70; and (2) elution of viruses from the
filter using a pH adjusted (usually basic) solution, the eluent. While the intended
mechanism for virus removal at the first stage of the VIRADEL process is adsorption,
straining by smaller pores of the microfilter can contribute to the overall removal. We
also note that although the terms “electropositive” and “electronegative” have been
broadly used to describe the charge on the filters used to concentrate viruses, the
usage is unfortunate for two reasons. First, the charge is generally pH dependent.
Second, the term electronegativity has an established and distinctly different meaning
(the ability of an atom to attract electrons) in the chemistry literature.

In 1960, Metcalf used a membrane filter to perform a VIRADEL-like separation to
concentrate influenza virus from bacterial mixtures. In these tests, instead of eluting
viruses from the membrane, the membrane was crushed to pulp and resuspended for
virus detection by hemagglutination tests and infectivity titrations

71;

the author noted

that “on several occasions, …, essentially comparable recoveries could be achieved
simply by washing the membrane surface.” The first application of the VIRADEL dates

10

back to 1967 when Wallis and Melnick concentrated poliovirus from saline-diluted crude
harvest using an electronegative filter and a variety of eluents 72.

Commonly used electronegative filters include MF-Millipore™ Membrane Filters made
of a mixture of cellulose nitrate and cellulose acetate
made of glass fiber bound by an epoxy resin

55, 61, 67, 73-76

54, 60, 77, 78.

and Filterite filters

Due to the negative charge of

most viruses at pH values typical for environmental water samples, the use of
electronegative filters requires sample preconditioning either by acidification of the
sample or by addition of multivalent cation salts to facilitate virus adsorption to the filter
56, 64

(see section 3.2). Sample preconditioning is often time-consuming and can limit

the sample volume, especially if field-collected samples need to be sent to a remote
laboratory. It may also be difficult to determine the dosage of acid and salts needed to
pre-condition a complex environmental sample. Furthermore, sample acidification
might affect virus integrity and infectivity 61, 79, 80 and confound accurate assessment of
microbiological quality of the water.

A significant advantage of electropositive filters is that sample preconditioning can be
avoided. Zeta Plus 1MDS microfilter (formerly marketed as Zeta Plus Virosorb 1MDS)
made of charge-modified glass and cellulose was the only electropositive filter that the
U.S. EPA originally recommended for the VIRADEL process

81.

Because of the filter’s

high cost ($200 to $300 15) and vulnerability with respect to fouling by colloids, several
research groups have evaluated other electropositive filters as alternatives 59-62, 68, 82.
NanoCeram (a thermally bonded blend of microglass fibers and cellulose infused with

11

nanoalumina fibers in a non-woven matrix) is another positively charged filter used in
virus concentration 68, 74, 83-85; relatively inexpensive ($40 15), NanoCeram was found to
be as efficient as Virosorb 68. In 2010, U.S. EPA added NanoCeram to the list of
recommended VIRADEL filters while pointing out that the NanoCeram is more
susceptible to clogging than the 1MDS 86. Other electropositive filters used in the
VIRADEL process include Zeta Plus S series depth filters (surface-modified cellulose
and filter aid mixtures) 60, 77, 87, lifeASSURE 045SP filter (Nylon 6,6 based; formerly
marketed as Zetapor 045SP) 88, MK filter (molded waste fibers coated with melamine
resin) 62, oiled sodocalcic glass wool filter 85, 89 and electronegative filters pretreated by
a solution of multivalent cations 67, 76, 90 or a cationic polymer 61. Zeta Plus 50S filter
appeared to be as efficient as Virosorb in recovering viruses from various types of water
60;

Rose et al. obtained higher recoveries from primary effluent with Zeta Plus 50 S than

with 1MDS and Zeta Plus 30 S, but saw no significant differences among these filters in
virus recovery from a secondary effluent. MK filter was found to give recoveries that
were both lower and more variable 62. Diatomaceous earth filters treated with a cationic
polymer (e.g. Nalco 7111) showed improved coliphage recovery from trickling filter
effluent samples 61 when compared with Virosorb filters; however, unless a drop-bydrop elution (see section 3.5.1) was performed the coliphage recovery from sewage
was still below 50% (data not available for tap water samples).

Tables 1 and 3 summarize recoveries reported for norovirus and adenovirus along with
the experimental conditions used in the concentration tests. VIRADEL recoveries of
MS2, the microorganism most commonly employed in virus concentration studies, are

12

provided in Table 5. The data clearly indicate that virus recovery by VIRADEL can be
low and poorly reproducible 91. Recoveries reported for adenovirus (e. g. 0.02% from a
mixture of DI and tap water 66) and certain other viruses such as astrovirus (0.5% from
tap water 55) are especially low. Borrego et al. observed that less than 35% of the
indigenous E. coli phages present in tap water spiked with sewage were adsorbed on
the Virosorb 1-MDS filter and less than 12% of those were recovered by elution 61
yielding the overall recovery of ≲ 4 %. Concentrating human adenovirus 41 by a
NanoCeram filter, Gibbons et al. reported high retention (> 99%) by the filter but very
low efficiency of elution (< 3%) with 3% beef extract solution used as the eluent

83.

Low

elution recoveries were also observed by Abdelzaher et al. who showed 35% recovery
of the coliphage spiked in seawater and adsorbed on a negatively charged cellulose
membrane 79. Although no study has definitively identified mechanisms behind the low
elution efficiency of certain viruses by VIRADEL, it was hypothesized that virion capsid
structure may play a role in virus entrapment by a filter matrix resulting in hindered
elution 83, 84, 92.

2.2.2 Crossflow ultrafiltration

CFUF (also known as tangential ultrafiltration) has emerged as a promising alternative
to VIRADEL 66, 93-99. During CFUF, viruses are maintained in the recirculated retentate
by an ultrafilter with a nominal pore size significantly smaller than the virus. In 1933,
Clark et al. used a cotton membrane to concentrate poliomyelitis virus suspension from
the growth medium by a factor of up to 165 100. It appears that the first tests on

13

concentrating water samples spiked with viruses were performed almost 40 years later.
In 1971, Sweet et al. used an osmotic ultrafiltration system to concentrate poliovirus
with an asymmetric cellulose acetate membrane and either MgSO4 or sucrose providing
osmotic driving force 101. Then Morrow employed pressure-driven CFUF to concentrate
the inactivated virus of foot-and-mouth disease; the method was proposed for possible
applications in “vaccine production as a preliminary step in purification” 102. In 1974,
Belfort et al. employed cellulose acetate hollow fiber as an ultrafilter to concentrate
poliovirus from water also using applied transmembrane pressure as the driving force

93.

CFUF has several advantages over VIRADEL. First, low molecular weight solutes,
which can potentially inhibit the downstream detection by molecular methods and cell
culture infectivity assays, pass through the membrane and are not concentrated
together with pathogens 45. Second, crossflow minimizes membrane fouling and virus
deposition onto the membrane, and thus could increase virus recovery. Third, CFUF
can simultaneously concentrate multiple pathogen types

103-109.

Fourth, the potential for

virus inactivation due to pH changes in VIRADEL (e.g. pre-acidification of the sample
and elution with high or low pH) is minimized. While some viruses are acid-tolerant and
some are base-tolerant, many can be inactivated when exposed to low pH 110-112 or high
pH 111, 113, 114. Even for the same virus, different serotypes could have different response
to low or high pH 115. The loss of infectivity is caused by an irreversible change in the
virus capsid; mechanisms hypothesized to be responsible for such changes include
proteolytic cleavage 115, dissociation of capsid 111 and a decrease in its stability 110.

14

Including secondary concentration in their analysis, Hill et al. showed that CFUF gave
significantly higher and more reproducible (58% Âą 8% vs 33 Âą 18%) total recoveries of
echovirus 1 than VIRADEL 99. Yet, virus deposition onto membranes was still observed
in CFUF 95, 103, 106, 116. To further recover the viruses adsorbed on the filters, elution may
be performed after filtration using the approach similar to that of VIRADEL
121.

66, 103-106, 116-

The molecular weight cutoff (MWCO) of membranes used to concentrate viruses by

CFUF is typically in the 10 kDa to 100 kDa range ensuring virus retention of 2 log or
higher. The membranes are made of polysulfone 104-107, polyethersulfone 116, 117, 121-123,
polyacrylonitrile 103, or cellulose triacetate 124, all of which are polymers commonly used
to make porous membranes. In practice, hollow fiber membranes assembled in large
modules have been the most commonly used configuration

93, 94, 99, 103, 106, 125, 126.

These

modules have the advantage of a high membrane packing density allowing for a fast
filtration of large volumes of water. Such filtration systems are suitable for field
applications and can be reused to make sample concentration more cost-effective.

2.3 Parameters influencing the performance of membrane-based methods of virus
concentration

As discussed in section 2, VIRADEL is a two-stage process that necessarily includes
filtration and elution steps. While essential in VIRADEL, elution is optional in CFUF and
can be employed if preelution recovery by CFUF is insufficiently high. In a number of
studies on CFUF-based sample concentration elution was not performed 95, 107, 123, 127.
During the elution step, both VIRADEL and CFUF aim at disrupting virus-membrane

15

interactions to recover attached viruses. During the filtration step, however, the two
methods rely on opposite strategies: VIRADEL is designed to maximize virus adsorption
onto the filter while CFUF employs anti-adhesive membranes to concentrate viruses in
the recirculating retentate stream. Thus, conditions that increase virus recovery by
VIRADEL may have an opposite effect during CFUF. In this section, the effects of
various parameters on the virus recovery are discussed by contrasting their impact on
VIRADEL and CFUF. We overview published data and, when possible, suggest reasons
for study-to-study discrepancies and possible hypotheses that might be helpful to test.

2.3.1 Virus properties

Virus-surface interactions depend on physicochemical properties of viruses, including
size, surface charge, isoelectric point (pI), surface energy and morphology. Because
VIRADEL and CFUF rely on different principles for virus recovery, the same virus
property may have an opposite effect on the recovery by these two methods.
Analogously, one filter may yield different recoveries for different viruses.

In VIRADEL, strongly charged viruses should adsorb to the oppositely charged filter
during the filtration step but may be difficult to “strip off” the filter during elution;
Polaczyk et al. illustrated this with MS2 (pI = 3.9) and ÎŚX174 (pI = 6.6) bacteriophages
in experiments conducted with 1MDS filter at pH ranging from 6.5 to 7 where MS2
carries a more negative surface charge 128. When using an electronegative filter,
adjusting pH to render viruses positively charged will also lower the negative charge on

16

the filter and possibly reverse the filter’s charge to positive. The window of pH values
where the virus and the filter have electrical charges of opposite signs depends on pI
values of the virus and the filter and can be quite narrow. The closer the pI values are,
the narrower this pH range is.

CFUF employs membranes that resist virus adhesion. In the case of non-enveloped
viruses, for instance, membranes that exhibit low protein binding should show lower
adhesion of viruses. Generally, membranes that present a highly hydrophilic surface are
more fouling resistant and, therefore, preferable for CFUF

129-131.

At pH values typical

for environmental samples most viruses are negatively charged and by choosing an
electronegative membrane one can rely on electrostatic repulsion to reduce virus loss to
the membrane. However, more apolar viruses could still be adsorbed onto membranes
due to hydrophobic attraction hence the need for a hydrophilic membrane surface.
Hydrophobicity can be operationally defined in terms of the water contact angle on a
virus lawn 116, 121, 132 or virus partitioning coefficient in hydrophobic interaction
chromatography 133; these relatively simple metrics give an aggregate assessment of
virus hydrophobicity, but the insight they provide is limited. Some of the contradictions
reported in the literature can be due to the simplistic interpretation of, and lack of
quantitative data on virus hydrophobicity. Determining individual surface energy
components 116, 121, 132 for both viruses and membranes may help understand why
surfaces of similar water contact angle and charge show different adsorption behavior.

17

Virus morphology is another factor that can affect recovery by both VIRADEL and
CFUF. For example, fibers associated with each penton base of the adenovirus capsid
could facilitate virion entrapment in the filter matrix making adenovirus difficult to elute
83, 84, 92.

Indeed, low elution efficiencies of adenoviruses were observed in several

VIRADEL studies though virus retention by filters was high

83, 84.

Fiber or spike

structures on the capsid surface can also sterically hinder virus adsorption to a
membrane surface 134, 135. However, in the presence of macroscopic repulsion between
the virus and the membrane, the protrusions on the capsid may extend beyond the
Debye layer leading to microscopic attraction 121. An important question that needs to
be answered is whether aggregate measures of the charge and surface energy of a
virus (i.e. its zeta potential and hydrophobicity) can adequately describe virus adhesion.
A general hypothesis that could be tested is that the local distribution of these properties
along the virus surface affects virus adhesion.

2.3.2 Membrane properties

Due to the negative net charge of most viruses at the pH typical for tap and natural
water samples (pH 6 to 8), electropositive filters are favored and have been extensively
used in VIRADEL-type concentrations. With electronegative filters, either in-situ charge
modification of filters or preconditioning of samples is possible with the former often
being a more practicable approach, especially for large volume samples. Modifying
negatively charged surfaces with positively charged amines

136

or multivalent cations (e.

g. Mg2+ 90, Fe3+ 90, Al3+ 67, 70, 90) enhances virus adsorption at pH above its pI. The

18

specific type of the VIRADEL filter may also affect recovery. In experiments on
norovirus concentration by electropositive NanoCeram and 1MDS filters (exp. 18-21,
Table 1) 68, higher recoveries were obtained with NanoCeram for both tap water (3.6% Âą
0.6% vs 1.2% Âą 1.4%) and river water (12.2% Âą 16.3% vs 0.4% Âą 1.8%).

In the case of CFUF, covering or “blocking” the membrane surface with a layer of
protein (e.g. using solutions of beef extract 95, 137, calf serum 105, 116 or bovine serum
albumin 137), glycine 95 or sodium polyphosphate 104, 106 minimizes virus adhesion. The
concept was first proposed in 1965 by Cliver who pretreated microfiltration membranes
with serum and gelatin solution to minimize adsorption of two enteroviruses (poliovirus
type 1 and Coxsackievirus B-2) to membrane surface 138. The method was later used
for poliovirus recovery by CFUF in a study by Berman et al. where the membrane was
pre-treated with either 1% glycine or 3% flocculated beef extract

95.

The term “blocking”

refers to the blockage of potential virus adsorption sites on the membrane surface and
not to the membrane pore blockage. Pore blockage is undesirable as it leads to a
decline in permeate flux and prolongs sample concentration. Ideally, the pore size of
the membrane should be small enough to ensure no virus loss to the membrane pore
space or the permeate and yet sufficiently large so that the blockage of adsorption sites
by the blocking agent does not lead to a significant loss in permeate flux. Molecular
blocking agents such as sodium polyphosphate (NaPP) (e. g. 600 Da

104, 106)

and

glycine (75 Da) with molecular weights smaller than the MWCO of the membranes
present an attractive choice: such molecules adsorb onto walls of ultrafiltration
membrane pores but do not plug them.

19

The most common choices of blocking agents have been solutions of calf serum, bovine
serum albumin and beef extract. Membrane pretreatment with 3% beef extract helped
increase murine norovirus recovery by a polysulfone ultrafilter more than ten-fold (exp.
2-5, Table 2) 107 and MS2 bacteriophage recovery almost two-fold (see APPENDIX;
exp. 23 and 24, Table 6) 107. Only two papers, to our knowledge, reported that blocking
a membrane with proteins had no effect on (see APPENDIX; exp. 31-36, Table 6) 118 or
decreased 126 virus recovery. A possible reason for the discrepancy is that blocking
agents are selected without regard to type of surface they modify. A useful hypothesis
to explore is that the blocking efficiency is governed by membrane-blocking agent
interaction; the more energetically preferable such interaction is in comparison to the
interaction between blocking molecules, the better “blockage” is likely to be.

Protein-blocked membrane may not be practical for field sampling because of a
possible contamination of the membrane by bacteria during storage and transport
106.

104,

Furthermore, because blocking is time-consuming (usually done overnight), the

process is not optimal for rapid detection. Hill et al. showed that 15-min blocking of an
UF membrane (MWCO 15 to 20 kDa) with NaPP solution resulted in the microbial
recovery similar to that obtained in the control experiments with calf serum as the
blocking agent (see APPENDIX; exp. 6-14, Table 6) 104. While membrane blocking has
been assumed to minimize virus adsorption to the membrane, all evidence for such an
effect appears to be indirect and based on the observed higher values of virus recovery
by blocked filters. For example, it was suggested that NaPP increases electrostatic

20

repulsion between the membrane and microorganisms 104 but surface charge data were
not provided to support the statement. We suggest testing the hypothesis that the
mechanism responsible for blocking is a combination of electrostatic and acid-base
repulsive interactions between the virus and the blocked membrane.

Layer-by-layer coating by polyelectrolytes has been proposed as an alternative to the
blocking process 116, 121. Coating a membrane by a polyelectrolyte multilayer (PEM) film
can produce surfaces with a broad range of properties, where variability is achieved by
choosing various polyelectrolyte combinations and deposition condition (pH and ionic
strength of deposition solutions, number of polyelectrolyte layers). This is in contrast to
blocking by proteins, which can be viewed as intentional, but poorly controlled, fouling of
the filter. Membranes coated with PEMs were shown to yield higher recoveries of
bacteriophage P22 116 and human adenovirus 40 121 from DI water compared with the
traditional blocking by beef extract (control). An analysis of virus-membrane interaction
energy pointed to the membrane’s higher negative charge and hydrophilicity as the
reasons for higher recovery by PEM-coated membranes.

2.3.3 Design of the filtration process: Hydrodynamic conditions

Sample concentration is one of the most time-consuming steps in the overall detection
process. Concentration times depend on many factors such as, for example, feed water
turbidity and NOM content, both of which contribute to membrane fouling and, therefore,
affect the rate of concentration. For example, concentrating 100 L of tap water to 400

21

mL by CFUF using a hollow fiber cartridge takes 2 h, and further volume reduction to a
few mL for qPCR analysis requires another 1 to 3 h 106. Because virus concentrations
of less than one genomic copy per 100 L of water represents a significant risk of illness
36,

virus monitoring in drinking water often requires testing up to 1,000 L

64.

One

approach to minimizing virus adsorption in CFUF is to increase the crossflow rate.
However, the shear stress on the virus can potentially disrupt its infectivity. Indeed, a
decrease in infective virus titer in the retentate of a crossflow filtration unit has been
reported 126, 139-141 although in at least one study such a trend was not observed

142.

In

two studies, virus aggregation was initially considered but eventually ruled out as the
reason for the observed loss of infectivity 140, 141. Belfort et al. noticed no significant
difference in poliovirus recovery by filtration through a non-blocked cellulose acetate
membrane when tripling the crossflow rate; this observation, however, was based on
only one filtration test performed at the low crossflow rate 93. Polaczyk et al. observed
no statistical difference in recoveries of MS2 and echovirus 1 when operating a
membrane pretreated with NaPP at two different permeate flow rates

106

although the

ratio of corresponding fluxes, 𝐽𝑥 ⁄𝐽𝑝 , remained within a narrow range from 5,800 to
6,500. Pasco et al. 116 evaluated a large range of operational conditions by varying
𝐽𝑥 ⁄𝐽𝑝 from 2,000 to 12,000 to concentrate P22 bacteriophage; pre-elution recovery with
PEM-coated membranes significantly improved from 32% to 85% with an increase in
𝐽𝑥 ⁄𝐽𝑝 while no statistically significant change in recovery over the same range of
hydrodynamic conditions was observed with membranes blocked by calf serum.
Rhodes et al. concentrated MS2 and poliovirus from tap water with hollow fiber
ultrafilters 127 and observed increased recoveries at lower permeate fluxes: an increase

22

from 64.7% Âą 4.9% to 98.7% Âą 5.3% for MS2 and an increase from 62.9% Âą 4.7% to
104.5% Âą 20.7% for poliovirus (see APPENDIX; exp. 29-30, Table 6).

The effect of permeate flux on the virus recovery by VIRADEL was also evaluated and,
as with CFUF, results varied. In VIRADEL, lower flow rate at the filtration stage
increases virus residence time within the filter and decreases permeate drag;
theoretically, this should translate into more effective virus capture. Sobsey and Glass
reported that poliovirus adsorption to Zeta Plus 50 filter from tap increased significantly
(p = 0.006) from 62% to 86% when the permeate flux decreased 10 times from 15 to 1.5
mL/(min¡cm2) 77. Jin et al. used Al(OH)3-based electropositive granule media packed
into a filter cartridge to concentrate poliovirus from tap water and obtained recoveries of
71% and <50% at flow rates of 300 and 500 mL/min, respectively (filtration fluxes of
~6.0 mL/(min¡cm2) and ~9.9 mL/(min¡cm2) based on the cartridge diameter reported by
the authors) 143. However, two other studies on poliovirus recovery from tap water
showed no significant correlation between virus adsorption and permeate flux

62, 68.

Operating the concentration unit at a higher permeate flux would hasten the process;
however, the larger permeate flux may decrease virus retention by the membrane (due
to increased drag) and elution efficiency (due to virus entrapment on or within the
membrane or the fouling layer). Moreover, a larger permeate flux typically translates
into more membrane fouling, which can eventually offset the gains in the concentration
rate. Adjusting hydrodynamics conditions inside the membrane channel can help
optimize the process to achieve faster filtration without affecting virus recovery. To

23

better understand how recovery depends on the crossflow and permeate fluxes and
understand the reasons behind the often contradictory findings cited above, one needs
to comprehensively assess adsorption kinetics in a filter as a function of hydrodynamic
drag and surface forces.

2.3.4 Composition of the water sample

2.3.4.1 Effect of water composition on VIRADEL
Spiking samples with humic or fulvic acids decreased virus recoveries by VIRADEL with
both electropositive and electronegative filters 144, 145. Complex water matrices were
also shown to give lower virus recoveries by VIRADEL than simpler water types did.
For example, Haramoto et al. achieved higher recoveries of human norovirus from
MilliQ water, tap water, and bottled water than from river and pond water (exp. 29-33,
Table 1) 67. Noroviruses were recovered with higher efficiency from bore hole water
than from other water types such as open well, river, food processing, and rain waters
(exp. 68-77; Table 1) 146. Karim et al. concentrated poliovirus with two electropositive
filters (1MDS and NanoCeram) and observed higher recovery from tap water in
comparison with river samples for both filters 68. In a study by Dan et al., infectious
recovery and genome recovery of adenovirus were higher from RO water samples
compared with those from sea water and treated sewage (exp. 17-25, Table 3) 75. AbdElmaksoud et al. seeded multiple pathogens into tap water with different added
amounts of soil and found that recoveries of different viruses were significantly different
only in the water matrix with low turbidity 147. However, opposite trends were also

24

reported in several studies. Victoris et al. concentrated norovirus and astrovirus from
multiple water matrices and obtained higher recoveries from mineral and river water
than from tap water (exp. 6-17, Table 1) 55. Hata et al. showed that virus recoveries
from tap and river were comparable to or even higher than those from Milli-Q water 76.
Similarly, the recovery of murine norovirus from ground water was higher than that from
surface water but the lowest recovery was obtained from ultrapure DI water (exp. 62-65,
Table 1) 148. Notably, in most studies the comparisons were drawn between tests with
different feed waters but not against a certain control such as DI water.

Not all water constituents decrease virus recovery. Amending water samples with
certain chemicals has been explored as a means to improve the process. For example,
norovirus recovery from tap and river water was highest from samples with 5 mM MgCl 2
added to the water. For mineral and sea water, the highest recovery of norovirus was
obtained from samples amended with 25 mM MgCl2 and 50 mM MgCl2, respectively
(exp. 6-17, Table 1) 55. The pH of the water sample was shown to have a significant
influence on adenovirus recovery, which increased from <5% at pH 4.5 to 55% Âą 19% at
pH 3.5 (exp. 17-25, Table 3) 149. Further, effects of the water matrix on virus recovery
may vary for different types of filters used. Increasing concentration of salt in feed water
increased virus adsorption on negatively charged nitrocellulose filter, but decreased
adsorption on positively charged 1MDS (control: salt-free sample) 54. However, virus
adsorption onto another commonly used positively charged filter, NanoCeram, was not
affected by high salt content in the samples

74.

25

2.3.4.2 Effect of water composition on CFUF

Many studies showed that increasing water complexity can also affect virus recovery by
CFUF. This is to be expected as surface properties, permeability, and selectivity of the
membrane affected by dissolved and suspended components and various salts in the
feed water. Belfort et al. showed that poliovirus recovery by a polysulfone hollow fiber
membrane decreased from 77% in DI water to 52% in tap water 94. Shi et al. reported
higher pre-elution recoveries of human adenovirus 40 from DI water (~75%) than from
tap (~39%) and lake water (~21%) 121. Kahler et al. showed that virus recovery
decreased as water turbidity increased up to 26 NTU

119.

In at least one study,

however, water complexity did not affect virus recovery as there was no significant
difference between drinking and surface waters in recovery values for infective MS2 and
PRD1 124; this result, however, may not be directly compared with those studies where
qPCR was used to quantify viruses as infective recovery should not be expected to
follow same trends as total recovery (see section 4.2 on recommendations for improving
study-to-study comparability). Addition of 0.1% NaPP as a sample amendment was
associated with significantly higher recovery of MS2 when filters were blocked with fetal
bovine serum (see APPENDIX; exp. 2-5, Table 6; control: no amendment added) 104.

Ionic strength was found to correlate with the loss of virus infectivity in tap water

140, 141,

although the correlation between sample conductivity and loss of infectivity was not
observed in more complex water matrices 126. This can be attributed to masking of the
ions’ effect by the complex chemistry of background solution. In addition to crossflow-

26

induced shear stress, virus morphology and presence of hydrophobic surfaces in the
concentration unit have also been suggested as factors contributing to the loss of viral
infectivity 139, 140. Another factor that might influence infectivity assays, and thus
recovery of infective viruses by crossflow filtration is the presence of natural organic
matter or cations which have been shown to affect virus aggregation

150.

Effects of the water matrix on virus recovery also depend on the virus type. This is not
unexpected as different viruses should interact with dissolved species in the water
differently. Rhodes et al. examined multiple virus recoveries with hollow fiber
ultrafiltration from river and tap water and found that recoveries for adenovirus (exp. 4
and 5, Table 4), poliovirus and bacteriophage ÎŚX174 were higher from tap water than
those from river water while opposite trends were observed for enterovirus 70,
echovirus 7 and coxsackievirus B4 120. Even for the same type of water, variations in
membrane operation conditions and in water composition can greatly affect virus
recovery 125, 126. T1 phage recovery from surface water from 6 different locations varied
from 34% to 75% 125, and PP7 recoveries from storm water from 21 different locations
varied from 10% to 98 % (when assessed by qPCR), and from 25% to 90% (when
assessed by cell culture infectivity assays) 126. Water samples from the same site, but
collected in different seasons, virus recoveries using the same ultrafiltration membrane
may vary due to differences in the ionic makeup and TOC content (exp. 6-11, exp. 1314, Table 4) 121. Again, unless specified otherwise, recoveries were compared between
tests with waters of different compositions but not against a common control. Since
standard deviations of recoveries were not provided in some of the cited studies it is

27

unclear if the variations were due to differences in the water composition or to an
inherent irreproducibility of the combination of the recovery and detection methods (see
section 4.2). When determining virus recovery from drinking water from 8 different
locations, Hill et al. 105 did perform replicate measurements and found no statistically
significant correlations between the 7 parameters tested (pH, turbidity, alkalinity,
conductivity, iron concentration, TOC, and DOC) and recoveries of MS2 or ÎŚX174
phages. A likely reason for the absence of correlations is that the parameters tested (𝑥𝑖 )
are either not independent variables or do not form the complete variable set or both so
that the recovery, 𝑟, cannot be described as 𝑟 = ∑7𝑖=1 𝑥𝑖 .

Common to all studies summarized in the section is the dependence of virus recovery
on the complexity of water matrix. Because of the lack of controls that can make interstudy comparisons possible, more specific generalizable conclusions cannot be drawn.
Yet the very variability of the data points to a general hypothesis that the recovery is
governed by interactions between water components and depend on the virus and
membrane properties only inasmuch as their interactions with the water constituencies
is concerned. Indeed, dissolved and suspended species in the feed water can alter the
surface properties of both the membrane and the virus

151.

In addition, these various

constituents in the sample can be concentrated simultaneously with viruses and can
affect the downstream detection method

152,

especially for samples with high total

organic carbon (TOC) content.

28

2.3.5 Eluent composition and elution protocol

2.3.5.1 Elution in VIRADEL

Elution is an essential step in virus concentration by VIRADEL. Low virus recovery
reported in multiple VIRADEL studies stemmed from low elution efficacy 79, 83, 84. For
electropositive filters, a commonly used eluent is beef extract solution (1.5%-3%) in the
pH range from 7.5 to 9.5 61, 62, 67, 68, 74, 85, 88, 90, 149, sometimes buffered with glycine 62, 68,
74, 85.

It has been suggested that beef extract improves virus elution by disrupting

hydrophobic interactions between the virus and the membrane and facilitates
bioflocculation during the following secondary concentration

70, 149;

this interpretation is

consistent with the use of beef extract as a blocking agent (see section 3.2). The
alkaline pH increases the net negative charge of both the filter surface and viruses
thereby enhancing the electrostatic repulsion and promoting virus desorption

77, 87, 153.

Addition of 0.1% NaPP to the beef extract-based eluent helped increase recoveries for
MS2 (see APPENDIX; exp. 5 and 7, Table 5) 128. Unfortunately, organic components of
the beef extract solution inhibit qPCR analysis 55, 70, 76, 154, 155; yet, beef extract is still
commonly used as an eluent. Farrah and Bitton determined that 4 M urea buffered with
0.05 M lysine at pH 9 was an efficient eluent for an electropositive filter 156 when
recovering viruses from sludge 157 and treated wastewater 87. Similar to beef extract,
urea likely interferes with hydrophobic interactions between the virus and the filter while
lysine is used to buffer the solution at a pH at which both the filter and virus have net
negative charges 87. Another eluent considered was 0.05 M glycine at pH 11.5, but the

29

corresponding elution efficacy was poor compared with that of 3% beef extract at pH 9.5
for both coliphages 60 and animal viruses 59, 60. The lower recovery with the 0.05 M
glycine-only eluent could be attributed to virus inactivation at the higher pH

158, 159.

Compared with beef extract-only eluent, solution of NaPP (1.0%) and glycine (0.05 M)
in phosphate buffer (3.8 mM Na2HPO4, 6.5 mM KH2PO4; pH 7.5) was very effective in
recovering MS2 (see APPENDIX; exp. 21-30, Table 5) 84. Although the eluents have
been mostly designed by a trial-end-error method, the common trend appears to be that
eluent components are selected to target specific types of virus-filter interactions.

For electronegative filters, samples are usually acidified to induce a positive charge on
the virus. Divalent cations are also often added to the sample, instead of or in addition
to lowering its pH. Alkaline beef extract or glycine solution have been applied to elute
virus adsorbed to electronegative filters as well 60, 79, 149. These filters remain negatively
charged in such eluent, but the charge sign of adsorbed virions is reversed to negative
allowing for an efficient elution. In the study by Katayama et al. as well as in several
more recent studies with electronegative filters, 1 mM NaOH (pH 10.5 to 10.8) was used
as the eluent

55, 67, 70, 74, 76, 160;

in these tests, pretreatment was first performed either by

soaking the filters in solution with multivalent cations (Mg2+, Al3+) or by adding those
cations to the water. Cations can enhance virus adsorption to membranes by acting as
bridges between the virus and the membrane 161. After filtration, membranes were
rinsed with 0.5 mM H2SO4 prior to elution with 1 mM NaOH. During the acid rinse,
cations were removed while viruses attained a positive charge and directly adsorbed to
electronegative filter. During the elution the virus charge reverses back to negative and

30

the viruses detach from the electronegative filter 70. However, including the acid rinse
step may inactivated viruses 88 (also see section 2.2).

The optimal choice of the eluent appears to be filter-specific even for membranes of like
charges (e. g. NanoCeram and 1MDS, which are both electropositive). Lee et al.
evaluated several elution buffers (1.5% beef extract, 0.05 M glycine, both with and
without 0.01% Tween 80) to determine the recoveries of human norovirus (exp. 45-48;
Table 1) 162. Without 0.01% Tween 80, the recoveries were higher with 1MDS than with
NanoCeram. At the same time, when 0.01% Tween 80 was added to the eluent, the
recoveries from NanoCeram were significantly higher than those from 1MDS

In addition to the eluent type, the elution protocol itself may influence VIRADEL
recovery. The standard elution procedure involves immersing the membrane in the
eluent and then filtering the eluent through the membrane. The membrane-eluent
contact time varies from study to study and ranges from 10 s to multiple hours (e.g.
overnight exposure) 55, 61, 62, 68, 85, 87, 90, 149. Working with bacteriophages in a trickling
filter effluent, Borrego et al. reported that the slow (0.5 mL/min) drop-by-drop elution
yielded slightly higher recovery in comparison to the elution performed at larger
flowrates (~ 1 mL/s)

61.

To our knowledge, the work by Borrego et al. is the only one

where the effect of elution flow rate was explored. Further studies are warranted to
better understand the nature of the likely kinetics or mass transfer limitations on
recovery (also see discussion in section 3.5.2).

31

Successive elutions, rather than a single elution, were performed in several studies to
maximize recoveries 68, 74, 85. Karim et al. tested poliovirus recoveries from tap water
with NanoCeram by varying eluent-filter contact time and obtained the highest recovery
(~77%) using a two-step elution with 1 min and 15 min immersion times during the first
and the second elution, respectively 68. Gibbons et al. achieved significantly higher
recovery of norovirus (~99%) by recirculating the eluent than by passing it only once
through the membrane (only 0.1% ) (exp. 38 and 39, Table 1) 83. Yet, in the same
study, eluent recirculation had no statistically significant effect on adenovirus recovery
(exp. 8-11, Table 3) 83. Sun et al. tested three elution methods when concentrating
adenovirus from seawater: (1) direct filtration of 1 mM NaOH; (2) shaking membrane in
1 mM NaOH on a vortex mixer and (3) magnetically stirring a 1 mM NaOH solution with
suspended membrane coupons; stirring yielded the highest recovery of ~80% (exp. 4042, Table 3) 160. We recommend a systematic evaluation of virus desorption behavior,
including both kinetics and equilibrium (isotherms) tests, to understand how the
desorption (elution) protocol can be optimized. We suggest testing common eluents first
to take advantage of the wealth of ad-hoc data already available in the literature.

2.3.5.2 Elution in CFUF

During CFUF, viruses are concentrated and recovered in the retentate stream. Size
exclusion and electrostatic repulsion are main mechanisms of virus rejection while
hydrophilicity of the membrane surface helps minimize virus adhesion to the membrane
surface. However, significant virus deposition on the membrane was observed in many

32

studies prompting the use of elution 95, 103, 106, 116 either as backflush 104, 106, 163 or as
crossflow rinse in the absence of a transmembrane pressure 103, 105, 106, 116-118, 120-122. Pei
et al. recovered 31 Âą 8 % of MS2 from 10 L tap water using CFUF followed by elution
with pure water 122. Other researchers eluted ultrafilters with eluents commonly used in
VIRADEL. Sylvain et al. eluted MS2 (73% to 84%) and ÎŚX174 (62% to 80%) using 1%
beef extract buffered with 0.4% glycine at pH 9.5

117.

Morales-Morales et al. eluted

bacteriophages T1 and PP7 using 0.05 M glycine in the retentate

103.

Albinana-

Gimenez et al. used 1 mM NaOH as the elutent and reported low recoveries for
adenovirus (3% to 6%) and norovirus (<1%) 66. A small amount (typically 0.01%) of
NaPP coupled with 0.01% Tween 80 has been used to elute bacteriophages (MS2
106,

105,

ÎŚX174 105, 120 and P22 116) and human viruses (poliovirus 120, coxsackievirus 120,

adenovirus 120, echovirus 106, 120 and enterovirus 120) from ultrafilters. It was suggested
that NaPP likely increases the electrostatic repulsion between viruses and the
membrane 104. In contrast, Tween 80 as a non-ionic surfactant can disrupt hydrophobic
attraction between virus and membrane 84, 104. In our previous study on human
adenovirus 40 recovery by CFUF, we showed that virus elution using aqueous solution
of 0.01% NaPP and 0.01% Tween 80 as the eluent produced nearly 100% recoveries
from both DI water and tap water 121. However, the same eluent only gave <65% postelution recovery of the virus from a surface water. Ethylenediaminetetraacetic acid
(EDTA) in the eluent gave a significantly higher post-elution recovery (84.3% Âą 4.5%) of
adenovirus from high TOC water (exp. 12 and 15, Table 4) 121. It was hypothesized that
EDTA improved elution by eliminating cation bridging 161 between viruses and natural
organic matter in water or deposited on the membrane. As with other parameters, the

33

mostly heuristic search for an optimal eluent and elution protocol yielded large amount
of data that is often contradictory and difficult to generalize. To design the elution
process deliberately one should consider specific interactions that resist virus
desorption and aim at disrupting them by chemical or hydrodynamic means. Including
appropriate controls (e.g. eluents without one of a constituents) will be needed.
Unfortunately, most referenced studies compared recoveries obtained with different
eluents in a pair-wise fashion without manipulating the composition of one eluent or
adjusting the elution procedure to obtain baselines.

2.3.6 Initial virus content in the water sample

When evaluating a concentration method with respect to a particular water type, viruses
are normally spiked into water to give an a priori known virus load in the sample. This is
done to factor out the initial concentration of the virus, known to affect removal values
164.

Indeed, in most VIRADEL studies virus recovery from samples with higher virus

content was found to be higher. Li et al. demonstrated this for both electropositive
(NanoCeram) and electronegative (Millipore) microfilters when recovering infectious
adenovirus 5 seeded at 102 PFU/L and 103 PFU/L levels in sea water, RO-treated sea
water, and treated sewage (exp. 17-23; Table 3) 75. Polaczyk et al. also found that
mean recovery of MS2 decreased from 92% to 44% when reducing the feed
concentration from 106 PFU/L to 103 PFU/L (see APPENDIX; exp. 6-8, Table 5); at the
same time, the decrease in ÎŚX174 recovery observed in this study was not statistically
significant 128. At least two studies, however, showed an opposite trend. A lower

34

influent MS2 seed titer (104 versus 108 PFU) resulted in a higher recovery when using
the optimal eluent (1.0% NaPP, phosphate buffer (3.8 mM Na2HPO4, 6.5 mM KH2PO4;
pH 7.5)), 0.05 M Gly, pH 9.3) (see APPENDIX; exp. 20 and 21, Table 5) 84. Working
with murine norovirus, Cashdollar et al. found the lower titer in the feed to give a higher
recovery (exp. 64 and 65; Table 1) 148. The effect of the feed concentration on virus
recovery was also evaluated in CFUF. Liu et al showed higher recoveries of MS2 and
ÎŚX174 for higher seeding levels in 100 L reclaimed water samples

118.

However, in

another study on the concentration of multiple virus types tap water (also 100 L
samples), Polaczyk et al. did not observe a statistically significant effect of the seeding
level on recovery 106.

2.4 Research needs and recommendations for future work

Two overarching conclusions can be drawn based on the already very large and
growing number of studies on concentrating virus-containing water samples. First,
physicochemical interactions between the virus and the filter govern virus recovery.
Indeed, the primary concentration methods are designed to harness these interactions
to achieve higher and more reproducible recoveries. Second, the effects of various
parameters of the concentration process are complex and, in many cases, appear to be
interrelated. Understanding virus-filter interactions and the effects - individual and
combined - of various parameters is important for the design of an effective membranebased concentration process. In this section, we identify knowledge gaps and offer
several recommendations for future research.

35

2.4.1 Method development: Standard procedures

Interlaboratory and interstudy comparisons of virus recovery data are challenging. The
difficulties in comparing results with different water matrices and target viruses are
further confounded by a large variability in the methods of concentration and in
operational parameters of each method. Filters, blocking agents and procedures, flow
rates, concentration ratios, eluents, and other process parameters vary broadly from
one laboratory to another and often between different studies within one lab. The
variability reflects the largely heuristic nature of the work. Consequently, despite the
large number of published studies it is difficult to generalize findings, identify trends and
establish guidelines. In their recent review, Cashdollar and Wymer chose to limit their
statistical analysis to mean recoveries because “the majority of the studies were
concerned only with performance of a single filter type and do not afford direct
comparison of tap water versus environmental samples, electropositive versus
electronegative filters, nor sample volume” 15.

Establishing a standard protocol for evaluating concentration methods would make
comparisons easier, save effort and resources. For example, a panel of virus
surrogates (e. g. bacteriophages) can be selected to comprehensively cover the range
of virus properties (size, charge, morphology) to generate better datasets for regulated
and emerging human viruses. Similarly, standard aqueous solutions can be designed
to represent important water types such as tap water, groundwater, surface water, sea
water and treated wastewater effluent. The variability in the water composition – both

36

temporal and between water types and sources – would remain an uncertainty but
experimenting with standards can provide guidelines and help with selecting optimal
concentration procedures.

2.4.2 Data reporting and analysis

A more unified approach to the data reporting is also recommended. Many publications
on virus concentration by CFUF only include total recovery and leave out the virus loss
to the filter at the preelution stage. Similarly, few publications on VIRADEL indicate the
amount of viruses not retained by the filter at the filtration stage. To better understand
the fate of viruses throughout the sample concentration process and identify reasons for
virus loss, it is recommended that both preelution recovery (or virus retention in
VIRADEL) and post-elution recovery are measured and reported. The same
recommendation was also made by Cashdollar and Wymer

15.

Given the large sample-

to-sample variability in recovery values measured for most environmental samples, the
data should be accompanied by a statistical analysis to give an estimate of the
reproducibility of the measurements and, preferably, provide sufficient basis for risk
assessment

165.

For human enteric viruses, we recommend that whenever possible

culture assays are performed to report infectious virus recovery in addition to the total
viral particle count. While molecular assays are fast and relatively accurate, their
results account for non-infectious microorganisms that do not reflect the pathogenicity of
microorganisms detected. Culture-based assays, although slow and prone to
experimental error, would help ascertain the extent to which a concentration method

37

inactivates target viruses. Culture-based assays are performed routinely in studies with
bacteriophages; when working with human viruses, however, many researchers have
only used qPCR for quantifying concentrations and recoveries.

2.4.3 Multi-method testing and controls

Virus concentration is a complicated process that could be affected by many
parameters. In some studies, rigorous controls were employed. For example, virus
concentration with unmodified membrane was used as a control when effect of
membrane coating was evaluated 95, 107, 118. To test the effects of humic acid and fulvic
acid on recovery, experiments performed without these macromolecules could be
considered as appropriate controls 144, 145. In most of the studies, however, the
comparison of recoveries was performed on a test-to-test basis with one or more
variable differing between the tests; the differences have often being in a complex
aggregate variable (e.g. composition of a multicomponent eluent solution, feed water
type 67, 68, 75, 120) and not between samples and a control. A number of studies with
multiple parameters were structured to optimize parameters sequentially. For example,
the elution protocol 68 and hydrodynamic conditions 116 were optimized in tests on
recovery from DI water and then were employed when concentrating pathogens from
more complex matrices such as tap water 68 and MBR effluent 116. While serving a
purpose, this approach is not designed to gain mechanistic, generalizable insights that
might be possible in the presence of rigorous controls.

38

From the point of view of establishing appropriate controls, the dependence of virus
recovery on the virus load in the source water is of special concern because the virus
load is the endpoint of the detection process and generally not known until after the
samples are concentrated and analyzed. The dependence complicates the
interpretation of results and calls for additional procedures that can help converge on an
accurate estimate of the virus content in the sample. Additional tests may involve single
or multiple dilutions of the source water sample, preparatory tests with a specific water
type seeded with a virus at different concentrations, or application of an alternative
concentration method to help verify and validate the initial estimate. Recovery control
tests should be performed to confirm the efficiency of the concentration method and to
verify if the sample inhibits downstream detection. A discussion of recovery control
options can be found in the review by Gibson and Borchardt

58.

When additional

experiments are not practical, an “educated guess” might be possible and warranted.
Once the analysis is performed and an estimate of the virus load in the initial sample is
available, one can consider how vulnerable the concentration method is with respect to
the virus loss and make a qualitative allowance for the lower recovery. Typically,
although not always 84, 106, 148, virus losses are higher for samples with a lower virus load
(see section 3.6).

39

2.4.4 Virus interactions with components of a concentrated water sample

Dissolved and suspended constituents of the sample water matrix have been reported
to affect virus recovery by both VIRADEL 67, 68, 75, 144, 145, 147 and CFUF 94, 116, 119-121.
During CFUF, the target virus is concentrated along with other components of the feed
magnifying their effect on virus recovery. To assess potential virus losses during
filtration, virus adsorption and aggregation behavior in such concentrated solutions
needs to be understood.

Virus loss to the fouling layer in CFUF should be mitigated at higher crossflow-topermeate flow ratios 116. Because lower permeate fluxes mean longer concentration
times, an increase in the crossflow rate is the preferred option. While it helps to
decrease membrane fouling and virus adhesion, the larger crossflow shear rate may
also affect virus infectivity. How susceptible various types of viruses are to the shear
stress (see section 3.3) is a question that should be explored.

The issue of membrane fouling is important for both CFUF and VIRADEL because of
their reliance on membrane-virus interactions (attractive in VIRADEL and repulsive in
CFUF). For complex water matrices and with the exception of the very early stages of
filtration, the filter surface is coated by foulants making the original surface properties of
the filter irrelevant in terms of further virus capture and permeate flux decline. It is the
properties of the components of water matrix and of the fouling layer they form on the
filter surface that largely determine virus recovery. To optimize virus concentration by

40

VIRADEL and CFUF, one needs to consider virus interaction not only with the
membrane surface, but also with the potential foulants. Because viruses adsorbed early
in the filtration process are the most difficult to elute, there is still a benefit in designing
the surface of the pristine filter to adsorb viruses reversibly.

2.4.5 Elution and sacrificial coatings as strategies for highly fouling water
matrices
Elution is an essential step in VIRADEL. It is also typically performed in CFUF as
preelution recoveries are often unacceptably low. Elution aims to disrupt favorable
virus-membrane and virus-foulant interactions to release the virus to the bulk solution
(permeate in VIRADEL and retentate in CFUF). Components of the eluent should not
inactivate virus or interfere with the further analysis by qPCR and cell culture infectivity
assay. For example, given that beef extract inhibits qPCR (see section 3.5), one should
explore other eluents as alternatives to the beef extract solution. The choice of the
eluent depends on the context. In the case of a specific virus in a high purity aqueous
solution, the eluent composition can be tailored to the virus-filter combination at hand.
Detecting a specific viral threat in drinking water (e. g. a virus surrogate used to test
integrity of a water treatment system) is an example of the corresponding application.
In contrast, screening for multiple unknown viruses in a complex water matrix calls for a
non-specific eluent capable of disrupting fouling layers and releasing viruses from
organics-rich aggregates. Filter backflush with an eluent or a sacrificial membrane
coating that can be disassembled to remove the fouling layer with adsorbed or
embedded viruses can possibly be used to maximize post-elution and total recoveries.

41

APPENDIX

42

APPENDIX

Table 1. Norovirus recovery by VIRADEL as a function of water matrix and VIRADEL process parameters.
Exp.
#
1

VirusA

HNoV
(GI)

Water
typeB
ground I

2

ground II

3

tap

4

HNoV
(GII)

ground II

6

tap

8

sea
HNoV
(GII)

river

9

bottled

10

tap

Filter (+/-)D
/TreatmentE

EluentF

Primary Secondary
Total
Ref.
recoveryG concentra- recoveryG,
,%
tionH
%
45

1.1×107
GC/L

33
none

ground I

5

7

Sample
Sample
Virus
amendme concentrationC volume,
nt
L

1.3×103
GC/L
(1.9 to 50)×106
GC/L

5 mM
MgCl2

25 mM
MgCl2

4×105
GC/L
1.8×105
GC/L
3.3×106
GC/L
2.2×106
GC/L

20

glass wool (+) /
none

3% BE
0.5 M Gly
pH 9.5∆

not
reported

PEG
precipitation

30

89

16
10 to 20

32
3Âą1
cellulose
esters(-) /
none

2

0.5 mM
H2SO4
 1 mM
NaOH╧

1Âą1
not
reported

centrifugal
UF

18 Âą 9
17 Âą 7

4×105
GC/L

2Âą3

43

55

Table 1 (cont’d).
Exp.
#

VirusA

11

Water
typeB
sea

12

river

13

bottled

14

Sample
Sample
Virus
C
amendme concentration volume,
nt
L

HNoV
(GII)

25 mM
MgCl2

15

sea

3.5×105
GC/L

16

bottled

17

river

50 mM
MgCl2

9Âą1

none

20

23

DI
tap

23 Âą 17
not
reported

centrifugal
UF

6

4.1×105
GC/L

6Âą5

3.5×105
GC/L

4Âą3

1.2×106
GC/L

1MDS (+)
/ none

10 or 100

NanoCeram (+)
/ none

(2.9 to 6.2)×105
GC/L

1MDS (+)
/ none

0.5

(2.2 to 5.8)×105
GC/L

1.5% BE
0.05 M Gly
pH 9.0∆

not
reported

Celite
adsorptionelution

1Âą1
68

12 Âą 16
0Âą2

3% BE
pH 9

77 to 107
14 to 46

44

55

4Âą1

1MDS (+)
/ none
none

2Âą2

1.6×106
GC/L

river

HNoV
(GII)

0.5 mM
H2SO4
 1 mM
NaOH╧

NanoCeram (+)
/ none

19

22

cellulose
esters(-) /
none

2

tap

21

Primary Secondary
Total
Ref.
G
recovery concentra- recoveryG,
,%
tionH
%
5Âą7

18

HNoV

EluentF

1.3×106
GC/L
1.4×106
GC/L
4.6×106
GC/L

tap

Filter (+/-)D
/TreatmentE

not
performed

same as
primary
recovery

67

Table 1 (cont’d).
Exp.
#

VirusA

Water
typeB

Sample
Sample
Virus
C
amendme concentration volume,
nt
L

24

river

(7.9 to 13)×105
GC/L

25

DI

(6.5 to 57)×105
GC/L

none
26

bottled

(3.8 to 6.7)×105
GC/L

27

tap

(5.8 to 6.2)×105
GC/L

river

(1.7 to 11)×105
GC/L

28

HNoV
(GII)

30

bottled

31

tap

32

river

(2.7 to 5.5)×105
GC/L

pond

6.9)×105

35

HNoV
(GII)

1MDS (+)
/ none

3% BE
pH 9

0.25

(4.3 to 4.5)×105
GC/L
25 mM
MgCl2

0.5

cellulose esters
(-)
/ AlCl3

pH 3.5

36 to 63
0.25

0.5 mM
H2SO4
 1 mM
NaOH

none

24 to 45
172 to 200

not
performed

same as
primary
recovery

67

138 to195

0.5
cellulose esters
(-)
/ none

7.9×105 to
6.0×107
GC/L

1.2×103
GC/L

13 to 24

25 to 95

55 to 104

11 to 18
0.25

(4.5 to
GC/L

DI + tap

Primary Secondary
Total
Ref.
G
recovery concentra- recoveryG,
,%
tionH
%

54 to 154

(5.3 to
GC/L

DI

34

EluentF

19)×106

29

33

Filter (+/-)D
/TreatmentE

38 to 39

10

glass wool (+)
/ none

45

3% BE
0.05 M Gly
pH 9.5

not
reported

3% BE
flocculation

0 to 19
66

<1

Table 1 (cont’d).
Exp.
#

VirusA

Water
typeB

Sample
Sample
Virus
C
amendme concentration volume,
nt
L

glass wool (+)
/ none

36
HNoV
(GII)

DI + tap

none

2.4×102
GC/L

Filter (+/-)D
/TreatmentE

50
MK (+)
/ none

37

EluentF

3% BE
0.05 M Gly
pH 9.5
1% BE
0.25 M Gly
pH 9.5

Primary Secondary
Total
Ref.
G
recovery concentra- recoveryG,
,%
tionH
%
<1
not
reported

38

<0.1

39

3% BE
0.1 M Gly
pH 9.5†

111 Âą 28

8.8×104
GC/L

NanoCeram (+)
/ none

not
performed

40

3% BE
0.1 M Gly
0.01% TW
80
pH 9.5†

88 Âą 24

41

3% BE
0.1 M Gly
0.1% TW 80
pH 9.5†

119 Âą 26

1.5% BE
pH 9.8

not
reported

42
43

HNoV
(GII)

sea

none

DI
none
tap

9.7×104
GC/L

40

10

NanoCeram (+)
/ none

46

66

<1

3% BE
0.1 M Gly
pH 9.5∆

HNoV
(GII)

3% BE
flocculation

1.5% BE,
0.1% FeCl3,
flocculation

same as
primary
recovery

83

42 Âą 8
166

29 Âą 15

Table 1 (cont’d).
Exp.
#

VirusA

Water
typeB

44

HNoV
(GII)

river

Sample
Sample
Virus
C
amendme concentration volume,
nt
L
none

9.7×104
GC/L

10

45

46
HNoV
(GII)

DI

none

1.0×105
GC/L

1

47

48

49

HNoV
(GII)

lake

pH 6.5 to
7.0

52

10

EluentF

NanoCeram (+)
/ none

1.5% BE
pH 9.8

1.5% BE,
0.05 M Gly,
1MDS (+)
pH 9.5∆
/ none
1.5% BE,
0.05 M Gly,
0.01% TW
80 pH 9.5∆
1.5% BE,
0.05 M Gly,
NanoCeram (+)
pH 9.5∆
/ none
1.5% BE,
0.05 M Gly,
0.01% TW
80 pH 9.5∆
3% BE
Glasswool (+)
0.5 M Gly
/ none
pH 9.5

Primary Secondary
Total
Ref.
G
recovery concentra- recoveryG,
,%
tionH
%
not
reported

1.5% BE,
0.1% FeCl3,
flocculation

18 Âą 3

166

not
performed

same as
primary
recovery

162

PEG
precipitation

2

167

61 Âą 11

68 Âą 40

27 Âą 13

86 Âą 26

not
reported

5×104
GC/L

50

51

(4.6±8.1)×103
GC/L

Filter (+/-)D
/TreatmentE

HNoV
(GI)

irrigation

5 mM AlCl3
pH 3.5

5×105
GC/L

42 Âą 7
10

Filterite (-)
/ none

5×106
GC/L

10% TPB,
0.05 M Gly,
pH 10

not
reported

PEG
precipitation

26 Âą 5
168

43 Âą 1

47

Table 1 (cont’d).
Exp.
#

VirusA

Water
typeB

Sample
Sample
Virus
C
amendme concentration volume,
nt
L

EluentF

Primary Secondary
Total
Ref.
G
recovery concentra- recoveryG,
,%
tionH
%

5×104
GC/L

53

54

Filter (+/-)D
/TreatmentE

14 Âą 1
organic
flocculation

5×105
GC/L

HNoV
(GI)

13 Âą 0

55

5×106
GC/L

16 Âą 1

56

5×104
GC/L

38 Âą 1

57

irrigation

5 mM AlCl3
pH 3.5

5×105
GC/L

10

Filterite (-)
/ none

10% TPB,
0.05 M Gly,
pH 10

not
reported

PEG
precipitation

22 Âą 7

5×106
GC/L

38 Âą 1

59

5×104
GC/L

15 Âą 1

60

5×105
GC/L

61

5×106
GC/L

58
HNoV
(GII)

organic
flocculation

15 Âą 0

16 Âą 1

48

168

Table 1 (cont’d).
Exp.
#

VirusA

62

Water
typeB

Sample
Sample
Virus
C
amendme concentration volume,
nt
L

Filter (+/-)D
/TreatmentE

EluentF

Primary Secondary
Total
Ref.
G
recovery concentra- recoveryG,
,%
tionH
%

ground

30
5×105
PFU/L

63

surface
none

10

MNoV

1.5% BE,
NanoCeram (+)
0.05 M Gly,
/ none
pH 9∆

not
reported

100
PFU/L

64

1.5% BE
flocculation

centrifugal
UF

6
148

<1

DI
30
PFU/L

65

66

none

67

tap

68

bore hole

69

2.6×107
GC/L

DI
MNoV
(S7)

HNoV
(GI)

rain

2 to 16

5

cellulose ester
(-)
/AlCl3

0.5 mM
H2SO4
 1 mM
NaOH

2.2×107
GC/L

5mM
MgCl2
pH 3.5

106
GC/L

45
not
performed

same as
primary
recovery

76

31

5

cellulose esters
(-)
/ none

5

49

0.5 mM
H2SO4

Tr alk
buffer∆

not
reported

PEG
precipitation

16 to 30

11 to 15

146

Table 1 (cont’d).
Exp.
#

VirusA

70

71

Water
typeB

open well
HNoV
(GI)

food
processing

73

bore hole

74

106
GC/L

river

72

75

Sample
Sample
Virus
C
amendme concentration volume,
nt
L

5mM
MgCl2
pH 3.5

HNoV
(GII)

open well

107
GC/L

EluentF

Primary Secondary
Total
Ref.
G
recovery concentra- recoveryG,
,%
tionH
%

2

4 to 5

5

6 to 10

1

4 to 12

5

rain

Filter (+/-)D
/TreatmentE

cellulose esters
(-)
/ none

0.5 mM
H2SO4

Tr alk
buffer∆

not
reported

PEG
precipitation

23 to 32
146

5

17 to 25

2

10 to 11

76

river

5

13 to 15

77

food
processing

1

14 to 24

50

Table 1 (cont’d).
Notes for Table 1:
A

HNoV (GI): human norovirus genogroup I; HNoV (GII): human norovirus genogroup II; MNoV: murine norovirus.

B

DI: deionized water; DI+tap: mixture of equal volumes of DI water and tap water.

C

GC/L: genome copies per liter; PFU/L: plaque forming units per liter.

D

(+): electropositive filter; (-): electronegative filter.

E

Filter pretreatment, if any.

F

BE: beef extract; Gly: glycine; TW 80: Tween 80; TPB: tryptose phosphate broth; Tr alk buffer: 0.05 M KH2PO4, 1 M NaCl, 0.1% (v/v) Triton
X-100, pH 9.2. The arrow () indicates sequential application of eluents. In most studies, the elution was performed by filtering the eluent.
Additional features of the elution protocol include:
∆

Filters were soaked in the eluent prior to elution. See each study for specific contact time.

†

Eluent is recirculated.

╧

Filter was placed feed side down in a Petri dish containing NaOH and shaken for 10 min.

G

Recoveries were rounded to the nearest integer.

H

The arrow () indicates the sequence of secondary concentration steps.

51

Table 2. Norovirus recovery by crossflow ultrafiltration as a function of water matrix and ultrafiltration process parameters.
Exp.
#

VirusA

1

HNoV
(GII)

Water
typeB

DI + tap

Sample
amendm
entC

none

Virus
concentrationD

1.2×103
GC/L

Sample
volume,
L

10

2

Filter
/ MWCOE
/TreatmentF
Infilco
Degremont
filter
/ (NA)
/ 0.1% BSA
PS
/ 15 - 20 kDa
/ none

EluentG

1 mM NaOH╧

Primary Secondary
Total
Ref.
recovery8, concentra- recoveryH
%
tion
,%
not
reported

centrifugal
UF

<1

66

not
performed

same as
primary
recovery

107

5Âą6

DI
PS
/ 15 - 20 kDa
/ 3% BE

3
MNoV1

none

2×108
GC/L

5

4
DI +
0.05 M Gly
0.14 M NaCl
5

6

tap
MNoV1

7

0.01%
NaPP
surface

“High seeding
level”
10 to 100
PFU/L

100

63 Âą 30
none

PS
/ 15 - 20 kDa
/ none

6Âą5

PS
/ 15 - 20 kDa
/ 3% BE

53 Âą 74

CTA
/ 70 kDa
/ 0.1% NaPP

52

0.01% NaPP
0.1% TW 80
0.001%
antifoam A†

42
not
reported

centrifugal
UF

124

74

Table 2 (cont’d).
Exp.
#

VirusA

Water
typeB

Sample
amendm
entC

8

HNoV
(GII)

lake

none

Virus
concentrationD
(4.6±8.1)×103
GC/L

Sample
volume,
L
10

Filter
/ MWCOE
/TreatmentF

EluentG

REXEED25S (29 kDa) 0.01% TW 80†
/ none

Primary Secondary
Total
Ref.
8
recovery , concentra- recoveryH
%
tion
,%
not
reported

PEG
precipitatio
n

2

167

Notes for Table 2:
A

HNoV (GII): human norovirus genogroup II; MNoV-1: murine norovirus-1

B

DI: deionized water; DI+tap: mixture of equal volume of DI and tap; Gly: glycine

C

NaPP: sodium polyphosphate

D

GC/L: genome copy per liter; PFU/L: plaque forming unit per liter

E

PS: polysulfone; CTA: cellulose triacetate. The value in parentheses is the molecular weight cutoff (MWCO) of the membrane filter; NA: not
available; kDa: kilodalton

F

Filter pretreatment, if any; BSA: bovine serum albumin; BE: beef extract; NaPP: sodium polyphosphate

G

TW 80: Tween 80, NaPP: sodium polyphosphate. Additional features of the elution protocol:

H

╧

shaking 30 min

†

crossflow

Recoveries were rounded to nearest integer

53

Table 3. Adenovirus recovery by VIRADEL as a function of water matrix and VIRADEL process parameters.
Exp.
#

VirusA

1
2

Water
typeB
tap

HAdV
41

3

ground I

Sample
Virus
Sample
amendment concentration volume,
C
L
8.5 to 1700
10 to
GC/L
1500
none

ground II

4

16)×102

(8 to
GC/L

20

2.6×106 GC/L

10

Primary
recoveryG
,%

Secondary
concentrationH

HAdV
2

glass wool (+)
/ none

3% BE
0.5 M Gly
pH 9.5∆

not
reported

PEG
precipitation

glass wool (+)
/ none

DI+tap
50

7

MK (+)
/ none

8

sea

none

22

89

2 to 7
3% BE
0.05 M Gly
pH 9.5

~1
not
reported

3% BE
flocculation

none

HAdV
41

Total
Ref.
recoveryG
,%
28

pH 3.5

5.3×105 GC/L

9

EluentF

8

5
6

Filter (+/-)D
/TreatmentE

1.6×107 GC/L

40

NanoCeram (+)
/ none

10

54

1% BE
0.25 M Gly
pH 9.5
3% BE
0.1 M Gly
pH 9.5∆
3% BE
0.1 M Gly
pH 9.5†
3% BE
0.1 M Gly
0.1% TW 80
pH 9.5†

66

1 to 2
0.01 to
0.02

1

5Âą3

1Âą0

not
performed

same as
primary
recovery

83

Table 3 (cont’d).
Exp.
#

VirusA

Water
typeB

Sample
Virus
Sample
amendment concentration volume,
C
L
1.6×107

11

GC/L

12
sea
4.5×105 GC/L
13

HAdV
41

none

40

Filter (+/-)D
/TreatmentE

EluentF

3% BE
0.1 M Gly
0.01% TW 80
pH 9.5†
soy-based
eluent
pH 7.0†
soy-based
NanoCeram (+)
eluent
/ none
0.01% TW 80
pH 7.0†

14

Primary
recoveryG
,%

Secondary
concentrationH

Total
Ref.
recoveryG
,%

1Âą1

0.2 Âą 0.1

1Âą1

not
performed

same as
primary
recovery

83

3Âą2

15

surface

16

treatedsurface

3% BE
0.1 M Gly
pH 9.5†

2.5×107 GC/L

2Âą1
1Âą1

17

103 PFU/L

82 Âą 11

102 PFU/L

37 Âą 5

sea
18
19

20
21

HAdV
5

RO
treated
sea

sewage
effluent

none

103 PFU/L

1

nanoalumina
fiber (+)
/ none

3% BE
pH 6.0∆

not
reported

centrifugal
UF

91 Âą 12

102 PFU/L

64 Âą 8

103 PFU/L

86 Âą 8

55

75

Table 3 (cont’d).
Exp.
#

VirusA

Water
typeB

Sample
Virus
Sample
amendment concentration volume,
C
L

22

sewage
effluent

none

23

sea

pH 3.5

RO
treated
sea
sewage
effluent

24

25
26
27

28

HAdV
41

HAdV
2

29
30

tap

nanoalumina
fiber (+)
/ none

102 PFU/L

EluentF

103

PFU/L

Secondary
concentrationH

3% BE
pH 6.0∆

44 Âą 6

0.5 mM
H2SO4
 1 mM
NaOH╧

not
reported

centrifugal
UF

0.2
pH 3.5
2.5×106 GC/L

1

pH 4.5

none

5×106
TCID50/L

20

HAdV
41

tap

lake

75

90 Âą 8

60 Âą 8
cellulose esters
1.5% BE
(-)
0.75% glycerol
/ none
pH 9.0╧

NanoCeram (+)
/ none

1.0% NaPP
PB
0.05M Gly
pH 9.3╘

55 Âą 19
<5

39 Âą 13

evaporation
or centrifugal
UF

centrifugal
UF

not
reported

149

not
reported
14 Âą 4

84

19 Âą 2
none

102 IU/L

10

NanoCeram (+)
/ none

1.5% BE
pH 9.8

not
reported

1.5% BE,
0.1% FeCl3,
flocculation

river
HAdV
41

Total
Ref.
recoveryG
,%

66 Âą10

1
cellulose esters
(-)
/ none

Primary
recoveryG
,%

DI

31

32

source
water

5 mM AlCl3,
pH 3.5

Filter (+/-)D
/TreatmentE

21 Âą 3

166

19 Âą 3
pH
6.5 to 7.0

(5.9±3.4)×104
GC/L

10

glass wool (+)
/ none

56

3% BE
0.5 M Gly
pH 9.5

not
reported

PEG
precipitation

5

167

Table 3 (cont’d).

33

HAdV
41

lake

Sample
Virus
Sample
amendment concentration volume,
C
L

none

(5.9±3.4)×104
GC/L

10

Filter (+/-)D
/TreatmentE

EluentF

NanoCeram (+)
/ none

1.5% BE
0.05 M Gly
pH 9

none

1.5% BE
0.05 M Gly
pH 7.5

pH 5

1.5% BE
0.05 M Gly
pH 10

34

35

HAdV
40

36

tap

107 to108 GC/L

0.1

1MSD (+)
/ none

37

38

none

HAdV
41

39

40
sea

25 mM AlCl3

Primary
recoveryG
,%
not
reported

Secondary
concentrationH
Celite
adsorptionelution
 PEG
precipitation

105 PFU/L

1

cellulose esters
(-)
/ none

41

57

Total
Ref.
recoveryG
,%

0.02

167

24 Âą 8

52 Âą 22
not
reported

Celite
adsorptionelution

14 Âą 7

92

9Âą3

1.5% BE
0.05 M Gly
pH 10

pH 5

HAdV
7

1.5% BE
0.05 M Gly
pH 7.5

stirring#
30 min

Water
typeB

shaking╧
30 min

VirusA

0.5 mM H2SO4
 1 mM NaOH

Exp.
#

64 Âą 4
13 Âą 4

79 Âą 5
not
performed
54

same as
primary
recovery

160

Table 3 (cont’d).
Water
typeB

42

HAdV
7

sea

Sample
Virus
Sample
amendment concentration volume,
C
L

25 mM AlCl3

105 PFU/L

1

Filter (+/-)D
/TreatmentE

cellulose esters
(-)
/ none

EluentF

stirring# 45 min

VirusA

0.5 mM H2SO4
 1 mM NaOH

Exp.
#

Primary
recoveryG
,%

Secondary
concentrationH

52

not
performed

Total
Ref.
recoveryG
,%
same as
primary
recovery

160

Notes for Table 3:
A

HAdV: human adenovirus.

B

DI: deionized water; DI+tap: mixture of equal volumes of DI water and tap water; source water: water collected from Water and Sewer
Authority.

C

GC/L: genome copies per liter; PFU: plaque forming unit per liter; TCID50/L: 50% tissue culture infective dose per liter; IU/L: infectious unit
per liter.

D

(+): electropositive filter; (-): electronegative filter.

E

Filter pretreatment, if any.

F

BE: beef extract; Gly: glycine; TW 80: Tween 80; PB: phosphate buffer (3.8 mM Na2HPO4, 6.5 mM KH2PO4); The arrow () indicates
sequential application of eluents.
In most studies, the elution was performed by filtering the eluent. Additional features of the elution protocol:
∆

Filters were soaked in eluent first prior to filtration of eluent. See each study for specific contact time.

†

Eluent is recirculated.

╧

Filter was placed into a tube containing NaOH and vortexed. See each study for specific vortex times.

58

Table 3 (cont’d).
Notes for Table 3:
╘

Filter was immersed in the eluent. Filtration housing unit was inverted 10 times followed by 15 min of hold time. The procedure was
performed 3 times prior to elution.

#

Filter was soaked in the flask containing NaOH and stirred with a magnetic bar.

G

Recoveries were rounded to the nearest integer (except when recoveries are below 1%).

H

The arrow () indicates the sequence of secondary concentration steps.

59

Table 4. Adenovirus recovery by crossflow ultrafiltration as a function of water matrix and ultrafiltration process parameters.
Exp.
VirusA
#

Water
typeB

Sample
Virus
Sample
amendment concentration, volume,
C
GC/LD
L

1

HAdV
DI + tap
2

none

2.6×106

10

2

HAdV
41

lake

none

(5.9 ± 3.4)×104

10

3

HAdV
5

cell
culture

none

>1012

0.45

(2.5 ± 0.5)×103

100

4
5

tap
HAdV
41

river

6

DI

7

tap
HAdV
40

8

0.01%
NaPP

(2.6 Âą

0.3)×103

50

Filter (+/-)E
/ MWCO
/ TreatmentF
Infilco
Degremont
filter
/ NA
/ 0.1% BSA
REXEED-25S
/ 29 kD
/ none
PES
/ 750 kDa
/ none

Secondary
Total
Re
concentra- recoveryH f.
tion
,%

1 mM NaOH╧

not reported

centrifugal
UF

3 to 6

66

0.01% TW 80†

not reported

PEG
precipitation

1

167

none

~75

not
performed

same as
primary
recovery

Celite
adsorptionelution

68 Âą 14

0.01%NaPP
REXEED-25S
0.01%
TW 80
/ 30 kDa
0.001% antifoam
/ none
A†

none

~109

0.01% NaPP
0.01% TW 80†

1

DI

tap

60

69 Âą 12
56 Âą 8

123

120

30 Âą 19

pre: 54 Âą 6
post: 99 Âą 8

PES
/ 30 kDa)
/ 5% CS

PES (30 kDa)
/ PEM
9

Primary
recoveryH,
%

EluentG

pre: 38 Âą 9
post: ~90
pre: 75 Âą 10
post:100 Âą 7
pre: 41 Âą 10
post: ~ 90

not
performed

same as
primary
recovery

121

Table 4 (cont’d).
Exp.
VirusA
#
10

Water
typeB

Filter (+/-)E
/ MWCO
/ TreatmentF

EluentG

lake I

11

lake II

12

lake II
HAdV
40

13

Sample
Virus
Sample
amendment concentration, volume,
C
GC/LD
L

PES
/ 30 kDa
/ 5% CS

0.01% NaPP
0.01% TW 80†

0.01% NaPP
0.01% TW 80
0.01% EDTA†
none

~109

lake II

15

lake II

PES
/ 30 kDa
/ PEM

0.01% NaPP
0.01% TW 80†

0.01% NaPP
0.01% TW 80
0.01% EDTA†

Notes for Table 4:
A

HAdV : human adenovirus

B

DI: deionized water; DI+tap: mixture of equal volume of DI and tap;

C

NaPP: sodium polyphosphate

D

GC/L: genome copy per liter

61

Secondary
Total
Re
H
concentra- recovery
f.
tion
,%

pre: ~20
post: 35 Âą 10

pre: ~20
post: ~84

1

lake I

14

Primary
recoveryH,
%
pre: ~40
post: 61 Âą 3

pre: ~40
post: 62 Âą 2
pre: ~20
post: 42 Âą 2

pre: ~20
post: ~84

not
performed

same as
primary
recovery

121

Table 4 (cont’d).
Notes for Table 4:
E

PES: polyethersulfone; The value in parentheses is the molecular weight cutoff (MWCO) of the membrane filter; NA: not available; kDa:
kilodalton

F

Filter pretreatment, if any; BSA: bovine serum albumin; CS: calf serum; PEM: polyelectrolyte multilayer

G

TW 80: Tween 80, NaPP: sodium polyphosphate; EDTA: Ethylenediaminetetraacetic acid. Additional features of the elution protocol:

H

╧

shaking for 30 min

†

crossflow

Recoveries were rounded to the nearest integer; pre: pre-elution recovery; post: post-elution recovery

62

Table 5. Primary recovery of bacteriophage MS2 by VIRADEL as a function of water matrix and VIRADEL process parameters.
Exp.
#

Water
typeA

1

Sample
Virus
Sample
amendment concentration, volume,
PFU/LB
L
pH 3.5

not reported

tap

10
0.1 M MnCl2,
pH 3.5

2

Filter (+/-)C
/TreatmentD

filterite (-)
/ none

not reported

3

4

5
tap

pH 6.5 to 7

106

1-MDS (+)
/ none

1

6

EluentE

5% BE
0.1% TW 80
pH 7

1.5% BE
0.05M Gly
pH 8∆
1.5% BE
0.05M Gly
pH 9∆
1.5% BE
0.05 M Gly
0.01% TW 80
pH 8∆
1.5% BE
0.05 M Gly
0.01% TW 80
0.1% NaPP
pH 8₸
1.5% BE
0.05 M Gly
0.01% TW 80
0.1% NaPP
pH 8∆

7

63

Primary
recoveryF,

Ref.

%
67Âą11
56

79Âą14

31 Âą 12

24 Âą 5

31 Âą 4
128

92 Âą 10

89 Âą 10

Table 5 (cont’d).
Exp.
#

Water
typeA

Sample
Virus
Sample
amendment concentration, volume,
PFU/LB
L
103

8
tap

1
1-MDS (+)
/ none

pH 6.5 to 7
5×104

9

Filter (+/-)C
/TreatmentD

20

EluentE
1.5% BE
0.05 M Gly
0.01% TW 80
0.1% NaPP
pH 8∆
1.5% BE
0.05 M Gly
0.01% TW 80
0.1% NaPP
pH 8▼

Primary
recoveryF,
%
44 Âą 9

128

32 Âą 13

10

1.5% BE
pH 9*

35

11

0.5 mM H2SO4
 1.5% BE
pH 9*

20

1mM NaOH*

16

12
sea G

none

(1.5 to 4.6)×105

0.1

cellulose
ester (-)
/ none

79

0.5 mM H2SO4
 1.5% BE
pH 9

9

14

1mM NaOH

6

15

0.5mM H2SO4
 1mM NaOH

2

13

64

Ref.

Table 5 (cont’d).
Exp.
#

Water
typeA

Sample
Virus
Sample
amendment concentration, volume,
PFU/LB
L

NanoCeram
(+)
/ none

16
DI
none

17

5×104

20

1-MDS (+)
/ none
NanoCeram
(+)
/ none
nitrocellulose
ester(-)
/ none

18
sea
19

Filter (+/-)C
/TreatmentD

0.1M MgCl2

21

22
tap
23

none

20
~5×106

24

NanoCeram
(+)
/ none

65

Ref.

65 Âą 23
1.5% BE
0.05M Gly
0.01% TW 80
pH 9.5∆

30 Âą 10
74

63 Âą 13

0.5mM H2SO4
 1mM NaOH

15 Âą 5

86 Âą 9
57 Âą 3

1.0% NaPP
PB
0.05 M Gly
pH 7.5╘

26 Âą 4

3% BE
pH 9.3╘

34 Âą 18

0.05 M Gly
pH 9.3╘

0.4 Âą 0.5

3% BE
0.05 M Gly
pH 9.3╘

25

Primary
recoveryF,
%

1.0% NaPP
PB
0.05 M Gly
pH 9.3╘

~5×102

20

EluentE

84

12 Âą 1

Table 5 (cont’d).
Exp.
#

Water
typeA

Sample
Virus
Sample
amendment concentration, volume,
PFU/LB
L

Filter (+/-)C
/TreatmentD

PB
0.05 M Gly
pH 9.3╘

26

27

28
tap

none

~5×106

20

NanoCeram
(+)
/ none

29

30

31

0.05 M MgCl2

107

river+DI
32

EluentE

0.1

0.05 M MgCl2
pH 3.5

66

cellulose
ester (-)
/ none

PB
0.05 M Gly
pH 7.5╘
PB
0.05 M Gly
0.3% TW 80
pH 9.3╘
0.1% NaPP
PB
0.05 M Gly
pH 9.3╘
0.6 M NaI
PB
0.05 M Gly
pH 9.3╘
3% BE
0.5 M NaCl
pH 9╧
0.5 mM H2SO4
0.05 M KH2PO4
0.1 M NaCl
0.1% TritonX-100
pH 9.2╧

Primary
recoveryF,

Ref.

%
26 Âą 5

24 Âą 7

37 Âą 2
84

40 Âą 7

3Âą2

52 Âą 6

169

16 Âą 3

Table 5 (cont’d).
Exp.
#

Water
typeA

Sample
Virus
Sample
amendment concentration, volume,
PFU/LB
L

Filter (+/-)C
/TreatmentD

2.9% TPB
6%Gly
pH 9╧

33
none
34

EluentE

107

river+DI

35

0.1

pH 5.5 to 6

Zeta Plus
60S (+)
/ none

Primary
recoveryF,

Ref.

%
1Âą0

PB
1% NaPP
0.05 M Gly
pH 9.3╧

2Âą0

0.05 M arginine
1% BE

0.04 Âą 0.05

169

pH 9╧

Notes on Table 5:
A

DI: deionized water; river+DI: river water diluted with sterile water

B

PFU/L: plaque forming units per liter

C

(+): electropositive filter; (-): electronegative filter

D

Filter pretreatment, if any

E

BE: beef extract; TW 80: Tween 80; Gly: glycine; NaPP: sodium polyphosphate; PB: phosphate buffer (3.8 mM Na2HPO4, 6.5 mM KH2PO4);
TPB: tryptose phosphate buffer; The arrow () indicates sequential application of eluents. In most studies, elution was performed by
filtering the eluent, but the elution protocol could include additional features. These features are marked as follows:
∆

Filters were soaked in eluent first prior to filtration of eluent. See each study for specific contact time

₸

Eluent heated to 37 ̊ C

▼

Eluent was pumped in the direction opposite to that of the water sample flow, then filter was soaked for 10 min in the eluent prior to
eluent collection

67

Table 5 (cont’d).
Notes on table 5:
* Sample prefiltered through PVDF membrane and drop-by-drop elution was performed
╘

Filter was immersed in the eluent. Filtration housing unit was inverted 10 times followed by 15 min hold time. Such procedure was
performed 3 times prior to filtering the eluent

╧

Filter was placed feed side down in a Petri dish containing eluent

F

Recoveries were rounded to the nearest integer (except for recoveries < 1%)

G

This work was performed with an indigenous coliphage (and not MS2 specifically). The study is included in the Table as illustrative of the
effect of the eluent on coliphage recovery.

68

Table 6. Primary recovery of bacteriophage MS2 by crossflow ultrafiltration as a function of water matrix and ultrafiltration process
parameters.
Exp.
#

Water
typeA

Sample
Virus
amendm concentration,
entB
PFU/LC

Sample
volume,
L

1

2

none

PS
(15 to 20 kDa)
/ 5% FBS
(1 hour)
PS
(15 to 20 kDa)
/ 5% FBS
(overnight)

3

4
~105

tap

5

10

EluentF

0.1%
NaPP

PS
(15 to 20 kDa)
/ 5% CS
PS
(15 to 20 kDa)
/ 0.1% NaPP
PS
(15 to 20 kDa)
/ 0.01% NaPP

7

0.01%
NaPP

69

Primary
recoveryG,

Ref.

%
44 Âą 30

51 Âą 19

50 Âą 14
none

108 Âą 16
104

PS
(15 to 20 kDa)
/ 5% FBS
(1 hour)

6

8

Filter
(MWCO/size)D
/TreatmentE
PS
(15 to 20 kDa)/
none
PS
(15 to 20 kDa)
/ 5% FBS
(overnight)

71 Âą 11

pre: 84 Âą 13
0.1% NaPP‡

post: 89 Âą 15
pre: 71 Âą 25
post: 82 Âą 25

0.01% NaPP‡

pre: 86 Âą 20
post: 96 Âą 21

Table 6 (cont’d).
Exp.
#

Water
typeA

9

Sample
Virus
amendmentB concentration,
PFU/LC

Sample
volume,
L

EluentF

Primary
recoveryG,

Ref.

% Âą 23
pre: 105

0.01% TW 80

PS
(15 to 20 kDa)
/ 0.01% NaPP

0.002% TW
80

10

Filter
(MWCO/size)D
/TreatmentE

0.01% NaPP‡

post:106 Âą
2370
pre:
post: 73

11

12

PS
(15 to 20 kDa)
/ none

none

~105

tap

10

PS
(15 to 20 kDa)
/ 0.01% NaPP

13
0.01% NaPP

14

15

tap

0.01% NaPP

720 Âą 240

100

16

tap

none

~104

100

70

PS
(30 kDa)
/ 5% CS
PS
(30 kDa)
/ 0.1% NaPP

tap‡

34Âą28

0.01% NaPP‡

59 Âą 10

0.01% NaPP
0.01% TW 80
0.01% TW 20
0.001%
antifoam A‡

65 Âą 35

0.01% NaPP
0.5% to 1%
Tween 80
0.001%
antifoam A ‡
0.01% NaPP
0.01% TW 80
0.001%
antifoam Y-30†
0.01% NaPP
0.5% TW 80
0.001%
antifoam A‡

104

91 Âą 33

120 Âą 22

105

52 Âą 34

106*

Table 6 (cont’d).
Exp.
#

Water
typeA

Sample
Virus
amendmentB concentration,
PFU/LC

Sample
volume,
L

Filter
(MWCO/size)D
/TreatmentE

EluentF

tap

0.01% NaPP

~104

100

PS
(30 kDa)
/ 0.1% NaPP

19

20

Ref.

%
84 Âą 12

17

18

Primary
recoveryG,

tap

none

104

100

none

5×104

20

22

23
DI

none

2×109

5

24

71

94 Âą 11

0.01% NaPP
0.1% TW 80
0.001%
antifoam A†

53 Âą 13

PS & CTA
‡
(20 to 70 kDa) 0.001% TW 80
/0.5% CS

21

river

0.01% NaPP
0.1% TW 80
0.001%
antifoam A‡

PES
(30 kDa) /
none

PS
(15 to 20 kDa)
/ none
PS
(15 to 20 kDa)
/ 3% BE

1%BE, 0.4%
Gly
pH 9.5†
MilliQ water
(1#)†

1%BE
0.4% Gly
pH 9.5 (2#)†

64 Âą 48

106*

163

73 to 84

117

42 to 64

30 Âą 7
107

none
54 Âą 24

Table 6 (cont’d).
Exp.
#

Water
typeA

25

DI +
0.05 M Gly
0.14 M
NaCl

26

Sample
Virus
amendmentB concentration,
PFU/LC

none
27

2×109

Sample
volume,
L

5

DI+
100 mM
Tris-HCl

(4.9±1.4)×102

29
0.01% NaPP

100
(7.4±6.8)×102

30

REXEED-25S
(30 kDa)
/ none

treated
wastewater

none

630Âą340

10

33

72

PS
(30 kDa)
/ none
PS
(65 kDa)
/ 5% CS

Primary
recoveryG,

Ref.

%
34 Âą 16

33 Âą 19
107

none
18 Âą 10

29 Âą 7

65 Âą 5a
127

none
99 Âą 5b

PS
(30 kDa)
/ 5% CS

31

32

EluentF

PS
(15 to 20 kDa)
/ 3% BE

28

tap

Filter
(MWCO/size)D
/TreatmentE
PS
(15 to 20 kDa)
/ none
PS
(15 to 20 kDa)
/ 3% BE
PS
(15 to 20 kDa)
/ none

84 Âą 2

none

79 Âą 18

80 Âą 6

118

Table 6 (cont’d).
Exp.
#

Water
typeA

Sample
Virus
amendmentB concentration,
PFU/LC

Sample
volume,
L

34

35

36

37

treated
wastewater

none

630Âą340

10

38

Filter
(MWCO/size)D
/TreatmentE
PS
(65 kDa)
/ none
CTA
(70 kDa)
/ 5% CS
CTA
(70 kDa)
/ none
PS
(30 kDa)
/ none
PS
(30 kDa)
/ none
CTA
(70 kDa)
/ none

39

40

41

tap

none

9×103 to 2×107

10

42

lake

none

(4.6±3.1)×104

10

73

EluentF

Primary
recoveryG,

Ref.

%
42 Âą 12

none

87 Âą 10

88 Âą 15

none

110 Âą 18
118

0.01% NaPP,
0.01% TW 80,
0.001%
antifoam Y-30†

120 Âą 20

none

80 Âą 14

CTA
(70 kDa)
/ none

0.01% NaPP,
0.01% TW 80,
0.001%
antifoam Y-30†

130 Âą 10

PES
(20 nm)
/ none

water†

31 Âą 8

122

0.01% TW 80†

68

167

REXEED-25S
(29 kDa)
/ none

Table 6 (cont’d).
Exp.
#

Water
typeA

Sample
Virus
amendmentB concentration,
PFU/LC

Sample
volume,
L

Filter
(MWCO/size)D
/TreatmentE

43

river

%
91 Âą 38

44

lake I

65 Âą 33

PS
(30 kDa)
/ 5% CS

EluentF

0.01% NaPP
0.01% TW 80
0.001%
antifoam Y-30†

Primary
recoveryG,

45

lake II

46

ground I

85 Âą 23

47

ground II

77 Âą 8

0.01% NaPP

(1.2±0.5)×102

50

53 Âą 19

Ref.

119

Notes on Table 6:
A

DI: deionized water; Gly: glycine

B

NaPP: sodium polyphosphate; TW 80: Tween 80

C

PFU/L: plaque forming units per liter

D

PS: polysulfone; CTA: cellulose triacetate; PS & CTA: recovery averaged with polysulfone membrane and cellulose triacetate membrane; PES:
polyethersulfone; The value in parentheses is the molecular weight cutoff (MWCO) or pore size of the membrane filter

E

Filter pretreatment, if any; FBS: fetal bovine serum; CS: calf serum; NaPP: sodium polyphosphate; BE: beef extract

74

Table 6 (cont’d).
Notes on table 6:
F

G

*

NaPP: sodium polyphosphate; TW 80: Tween 80; TW 20: Tween 20; BE: beef extract; Gly: glycine; The arrow () indicates sequential
application of eluents. Additional features of the elution protocol are marked as follows:
†

crossflow

‡

backflush

#

Number of rinses (1#: rinsed once; 2#: rinsed twice)

Recoveries were rounded to the nearest integer; pre: pre-elution recovery; post: post-elution recovery
a

recovery at a high filtration rate (2500 mL/min)

b

recovery at a low filtration rate (1750 mL/min)

In this paper, Polaczyk et al. performed secondary concentration as well and reported only one recovery value: “ultrafiltration recovery
efficiency”. It is this value that is included herein

75

REFERENCES

76

REFERENCES

1.

Waterborne zoonoses: Identification, causes and control, World Health
Organization, Geneva, Switzerland, 2004.

2.

H. Leclerc, L. Schwartzbrod and E. Dei-Cas, Crit. Rev. Microbiol., 2002, 28, 371409.

3.

J. Theron and T. E. Cloete, Crit. Rev. Microbiol., 2002, 28, 1-26.

4.

Surveillance for waterborne disease outbreaks associated with drinking water United States, 2001-2002, Morbidity and Mortality Weekly Report. Centers for
Disease Control and Prevention, 2004.

5.

Surveillance for waterborne disease and outbreaks associated with drinking
water and water not intended for drinking - United States, 2003-2004, Morbidity
and Mortality Weekly Report. Centers for Disease Control and Prevention, 2006.

6.

Surveillance for waterborne disease and outbreaks associated with drinking
water and water not intended for drinking - United States, 2005–2006, Morbidity
and Mortality Weekly Report. Centers for Disease Control and Prevention, 2008.

7.

Surveillance for waterborne disease outbreaks associated with drinking water United States, 2007-2008, Morbidity and Mortality Weekly Report. Centers for
Disease Control and Prevention, 2011.

8.

Surveillance for waterborne disease outbreaks associated with drinking water
and other nonrecreational water - United States, 2009-2010, Morbidity and
Mortality Weekly Report. Centers for Disease Control and Prevention, 2013.

9.

Surveillance for waterborne disease outbreaks associated with drinking water United States, 2011-2012, Morbidity and Mortality Weekly Report. Centers for
Disease Control and Prevention, 2015.

10.

N. I. o. Health, Biological sciences curriculum study. NIH Curriculum Supplement
Series. National Institutes of Health, Bethesda, MD, 2007.

11.

C. W. LeBaron, N. P. Furutan, J. F. Lew, J. R. Allen, V. Gouvea, C. Moe and S.
S. Monroe, MMWR. Recommendations and reports : Morbidity and mortality
weekly report. Recommendations and reports, 1990, 39, 1-24.

12.

S. M. Karst, Viruses, 2010, 2, 748-781.

13.

E. Papafragkou, J. Hewitt, G. W. Park, G. Greening and J. Vinje, Plos One, 2013,
8, e63485.
77

14.

S. E. Hrudey and E. J. Hrudey, Safe drinking water: Lessons from recent
outbreaks in affluent nations, IWA Publishing, London, 2004.

15.

J. L. Cashdollar and L. Wymer, J. Appl. Microbiol., 2013, 115, 1-11.

16.

S. R. Naik, R. Aggarwal, P. N. Salunke and N. N. Mehrotra, Bulletin of the World
Health Organization, 1992, 70, 597-604.

17.

Outbreak of aseptic meningitis associated with multiple enterovirus subtypesRomania, 1999, Morbidity and Mortality Weekly Report. Centers for Disease
Control and Prevention, 2000.

18.

A. L. Freeland, G. H. Vaughan Jr. and S. N. Banerjee, Acute Gastroenteritis on
Cruise Ships - United States, 2008-2014, Morbidity and Mortality Weekly Report.
Centers for Disease Control and Prevention, 2016.

19.

Investigation Update on the Coral Princess
(https://www.cdc.gov/nceh/vsp/surv/outbreak/2017/march18_coral_princess.html;
accessed on May 5, 2017).

20.

T. T. Fong, L. S. Mansfield, D. L. Wilson, D. J. Schwab, S. L. Molloy and J. B.
Rose, Environ. Health Perspect., 2007, 115, 856-864.

21.

C. E. O'Reilly, A. B. Bowen, N. E. Perez, J. P. Sarisky, C. A. Shepherd, M. D.
Miller, B. C. Hubbard, M. Herring, S. D. Buchanan, C. C. Fitzgerald, V. Hill, M. J.
Arrowood, L. X. Xiao, R. M. Hoekstra, E. D. Mintz and M. F. Lynch, Clin. Infect.
Dis., 2007, 44, 506-512.

22.

M. Abbaszadegan, M. LeChevallier and C. Gerba, J. Am. Water Works Assn.,
2003, 95, 107-120.

23.

Surveillance for waterborne disease outbreaks associated with recreational water
- United States, 2001-2002, Morbidity and Mortality Weekly Report. Centers for
Disease Control and Prevention, 2004.

24.

Surveillance for waterborne disease outbreaks associated with recreational water
- United States, 2003-2004, Morbidity and Mortality Weekly Report. Centers for
Disease Control and Prevention, 2006.

25.

Surveillance for waterborne disease and outbreaks associated with recreational
water use and other aquatic facility-associated health events - United States,
2005–2006 Morbidity and Mortality Weekly Report. Centers for Disease Control
and Prevention, 2008.

26.

Surveillance for waterborne disease outbreaks and other health events
associated with recreational water - United States, 2007-2008, Morbidity and
Mortality Weekly Report. Centers for Disease Control and Prevention, 2011.

78

27.

Recreational water associated disease outbreaks-United States, 2009–2010,
Morbidity and Mortality Weekly Report. Centers for Disease Control and
Prevention, 2014.

28.

Outbreaks of illness associated with recreational water - United States, 20112012, Morbidity and Mortality Weekly Report. Centers for Disease Control and
Prevention, 2015.

29.

R. G. Sinclair, E. L. Jones and C. P. Gerba, J. Appl. Microbiol., 2009, 107, 17691780.

30.

M. H. Nellor, R. B. Baird and J. R. Smyth, J. Am. Water Works Assn., 1985, 77,
88-96.

31.

Issues in potable reuse: The viability of augmenting drinking water supplies with
reclaimed water National Research Council, Washington, DC, 1998.

32.

A. P. Wyn-Jones and J. Sellwood, J. Appl. Microbiol., 2001, 91, 945-962.

33.

S. Choi and S. C. Jiang, Appl. Environ. Microbiol., 2005, 71, 7426-7433.

34.

I. Xagoraraki, D. H. Kuo, K. Wong, M. Wong and J. B. Rose, Appl. Environ.
Microbiol., 2007, 73, 7874-7881.

35.

J. W. He and S. Jiang, Appl. Environ. Microbiol., 2005, 71, 2250-2255.

36.

C. P. Gerba, J. B. Rose, C. N. Haas and K. D. Crabtree, Water Res., 1996, 30,
2929-2940.

37.

M. D. Sobsey, in Strategies for Drinking Water Contaminants, Commission on
Geosciences, Environment and Resources, National Academy Press,
Washington, DC, 1999, pp. 173-205.

38.

K. E. Gibson, K. J. Schwab, S. K. Spencer and M. A. Borchardt, Water Res.,
2012, 46, 4281-4291.

39.

S. Di Pasquale, M. Paniconi, B. Auricchio, L. Orefice, A. C. Schultz and D. De
Medici, J. Virol. Methods, 2010, 165, 57-63.

40.

T. M. Fumian, J. P. G. Leite, A. A. Castello, A. Gaggero, M. S. L. de Caillou and
M. P. Miagostovich, J. Virol. Methods, 2010, 170, 42-46.

41.

U. L. Tsai, M. D. Sobsey, L. R. Sangermano and C. J. Palmer, Appl. Environ.
Microbiol., 1993, 59, 3488-3491.

42.

W. Kang and J. L. Cannon, PLOS ONE, 2015, 10.

43.

J. Zhang, M. Mahalanabis, L. Liu, J. Chang, N. Pollock and C. Klapperich,
Diagnostics, 2013, 3, 155.
79

44.

J. Y. Liu, Q. P. Wu and X. X. Kou, J. Microbiol., 2007, 45, 48-52.

45.

T. M. Straub and D. P. Chandler, J. Microbiol. Methods, 2003, 53, 185-197.

46.

S. R. Farrah, S. M. Goyal, C. P. Gerba, C. Wailis and P. T. B. Shaffer, Water
Res. , 1976, 10, 897-901.

47.

S. R. Farrah, C. Wallis, P. T. B. Shaffer and J. L. Melnick, Appl. Microbiol., 1976,
32, 654-658.

48.

M. D. Sobsey, C. Wallis, M. Henderson and J. L. Melnick, Appl. Microbiol. , 1978,
26, 529-534.

49.

C. P. Gerba, Adv. Appl. Microbiol., 1984, 30, 133-168.

50.

J. L. Melnick, R. Safferman, V. C. Rao, S. Goyal, G. Berg, D. R. Dahling, B. A.
Wright, E. Akin, R. Stetler, C. Sorber, B. Moore, M. D. Sobsey, R. Moore, A. L.
Lewis and F. M. Wellings, Appl. Environ. Microbiol., 1984, 47, 144-150.

51.

M. D. Sobsey and J. S. Glass, Appl. Environ. Microbiol., 1984, 47, 956-960.

52.

N. Guttmanbass and J. Catalanosherman, Appl. Environ. Microbiol., 1985, 49,
1260-1264.

53.

M. D. Sobsey, T. Cromeans, A. R. Hickey and J. S. Glass, Water Sci. Technol.,
1985, 17, 665-679.

54.

J. Lukasik, T. M. Scott, D. Andryshak and S. R. Farrah, Appl. Environ. Microbiol.,
2000, 66, 2914-2920.

55.

M. Victoria, F. Guimaraes, T. Fumian, F. Ferreira, C. Vieira, J. P. Leite and M.
Miagostovich, J. Virol. Methods, 2009, 156, 73-76.

56.

T. M. Scott, J. Lukasik and S. R. Farrah, Can. J. Microbiol., 2002, 48, 305-310.

57.

L. A. Ikner, C. P. Gerba and K. R. Bright, Food Environ. Virol., 2012, 4, 41-67.

58.

K. E. Gibson and M. A. Borchardt, in Viruses in Foods, eds. S. M. Goyal and J. L.
Cannon, Springer, 2016, pp. 277-301.

59.

M. D. Sobsey and B. L. Jones, Appl. Environ. Microbiol., 1979, 37, 588-595.

60.

J. B. Rose, S. N. Singh, C. P. Gerba and L. M. Kelley, Appl. Environ. Microbiol.,
1984, 47, 989-992.

61.

J. J. Borrego, R. Cornax, D. R. Preston, S. R. Farrah, B. McElhaney and G.
Bitton, Appl. Environ. Microbiol., 1991, 57, 1218-1222.

80

62.

J. F. Ma, J. Naranjo and C. P. Gerba, Appl. Environ. Microbiol., 1994, 60, 19741977.

63.

D. C. Roepke, D. A. Halvorson, S. M. Goyal and C. J. Kelleher, Avian Diseases,
1989, 33, 649-653.

64.

Manual of Methods for Virology. U.S. Environmental Protection Agency, Report
600/4-84/013, Washington D.C., 2001.

65.

J. Brassard, K. Seyer, A. Houde, C. Simard and Y. L. Trottier, J. Virol. Methods,
2005, 123, 163-169.

66.

N. Albinana-Gimenez, P. Clemente-Casares, B. Calgua, J. M. Huguet, S.
Courtois and R. Girones, J. Virol. Methods, 2009, 158, 104-109.

67.

E. Haramoto, H. Katayama, E. Utagawa and S. Ohgaki, J. Virol. Methods, 2009,
160, 206-209.

68.

M. R. Karim, E. R. Rhodes, N. Brinkman, L. Wymer and G. S. Fout, Appl.
Environ. Microbiol., 2009, 75, 2393-2399.

69.

S. M. Goyal and C. P. Gerba, J. Virol. Methods, 1983, 7, 279-285.

70.

H. Katayama, A. Shimasaki and S. Ohgaki, Appl. Environ. Microbiol., 2002, 68,
1033-1039.

71.

T. G. Metcalf, Appl. Microbiol., 1961, 9, 376-379.

72.

C. Wallis and J. L. Melnick, J. Virol., 1967, 1, 472-477.

73.

W. Jakubowski, W. F. Hill and N. A. Clarke, Appl. Microbiol., 1975, 30, 58-65.

74.

H. B. Bennett, H. D. O'Dell, G. Norton, G. Shin, F. C. Hsu and J. S. Meschke,
Water Sci. Technol., 2010, 61, 317-322.

75.

D. Li, H. C. Shi and S. C. Jiang, J. Microbiol. Methods, 2010, 81, 33-38.

76.

A. Hata, K. Matsumori, M. Kitajima and H. Katayama, Food Environ. Virol., 2015,
7, 7-13.

77.

M. D. Sobsey and J. S. Glass, Appl. Environ. Microbiol., 1980, 40, 201-210.

78.

S. R. Farrah, C. P. Gerba, C. Wallis and J. L. Melnick, Appl. Environ. Microbiol.,
1976, 31, 221-226.

79.

A. M. Abdelzaher, H. M. Solo-Gabriele, M. E. Wright and C. J. Palmer, J.
Environ. Qual., 2008, 37, 1648-1655.

80.

C. M. Sabatino and S. Maier, Can. J. Microbiol., 1980, 26, 1403-1407.
81

81.

Manual of Methods for Virology. U.S. Environmental Protection Agency, Report
EPA/600/4-84/013 (originally published in 1984), Washington, DC., 2001.

82.

D. R. Preston, T. V. Vasudevan, G. Bitton, S. R. Farrah and J. L. Morel, Appl.
Environ. Microbiol., 1988, 54, 1325-1329.

83.

C. D. Gibbons, R. A. Rodriguez, L. Tallon and M. D. Sobsey, J. Appl. Microbiol.,
2010, 109, 635-641.

84.

L. A. Ikner, M. Soto-Beltran and K. R. Bright, Appl. Environ. Microbiol., 2011, 77,
3500-3506.

85.

N. Deboosere, S. V. Horm, A. Pinon, J. Gachet, C. Coldefy, P. Buchy and M.
Vialette, Appl. Environ. Microbiol., 2011, 77, 3802-3808.

86.

G. S. Fout, N. E. Brinkman, J. L. Cashdollar, S. M. Griffin, B. R. McMinn, E. R.
Rhodes, E. A. Varughese, M. R. Karim, A. C. Grimm, S. K. Spencer and M. A.
Borchardt, Method 1615. Measurement of Enterovirus and Norovirus
Occurrence in Water by Culture and RT-qPCR. U. S. Environmental Protection
Agency, Report EPA/600/R-10/18, 2010.

87.

L. T. Chang, S. R. Farrah and G. Bitton, Appl. Environ. Microbiol., 1981, 42, 921924.

88.

E. Haramoto and H. Katayama, Food Environ. Virol., 2013, 5, 77-80.

89.

E. Lambertini, S. K. Spencer, P. D. Bertz, F. J. Loge, B. A. Kieke and M. A.
Borchardt, Appl. Environ. Microbiol., 2008, 74, 2990-2996.

90.

S. R. Farrah, D. R. Preston, G. A. Toranzos, M. Girard, G. A. Erdos and V.
Vasuhdivan, Appl. Environ. Microbiol., 1991, 57, 2502-2506.

91.

P. Schaudies and D. A. Robinson, Literature review of molecular methods for
simulateneous detection of pathogens in water. U.S. Environmental Protection
Agency., Report 600/R-07/128 Washington D.C., 2007.

92.

B. R. McMinn, J. Virol. Methods, 2013, 193, 284-290.

93.

G. Belfort, Y. Rotem and E. Katzenelson, Water Res., 1975, 9, 79-85.

94.

G. Belfort, Y. Rotem and E. Katzenelson, Water Res., 1976, 10, 279-284.

95.

D. Berman, M. E. Rohr and R. S. Safferman, Appl. Environ. Microbiol., 1980, 40,
426- 428.

96.

D. M. Dziewulski and G. Belfort, Water Sci. Technol., 1983, 15, 75-89.

97.

H. V. Adikane, S. N. Nene, S. S. Kulkarni, P. U. Baxi, D. S. Khatpe and P. A.
Aphale, J. Membr. Sci., 1997, 132, 91-96.
82

98.

C. Chamorro, J. C. Espinoza, K. Soto and J. Kuznar, Archivos De Medicina
Veterinaria, 2006, 38, 77-82.

99.

V. R. Hill, A. L. Polaczyk, A. M. Kahler, T. L. Cromeans, D. Hahn and J. E.
Amburgey, J. Environ. Qual., 2009, 38, 822-825.

100.

P. F. Clark, R. C. Ainsworth and L. G. Kindschi, Exp. Biol. Med., 1933, 31, 255259.

101.

B. H. Sweet, J. S. McHale, K. J. Hardy, F. Morton, J. K. Smith and E. Klein,
Water Res., 1971, 5, 823-&.

102.

A. W. Morrow, J. Appl. Chem. Biotechnol., 1972, 22, 501-505.

103.

H. A. Morales-Morales, G. Vidal, J. Olszewski, C. M. Rock, D. Dasgupta, K. H.
Oshima and G. B. Smith, Appl. Environ. Microbiol., 2003, 69, 4098-4102.

104.

V. R. Hill, A. L. Polaczyk, D. Hahn, J. Narayanan, T. L. Cromeans, J. M. Roberts
and J. E. Amburgey, Appl. Environ. Microbiol., 2005, 71, 6878-6884.

105.

V. R. Hill, A. M. Kahler, N. Jothikumar, T. B. Johnson, D. Hahn and T. L.
Cromeans, Appl. Environ. Microbiol., 2007, 73, 4218-4225.

106.

A. L. Polaczyk, J. Narayanan, T. L. Cromeans, D. Hahn, J. M. Roberts, J. E.
Amburgey and V. R. Hill, J. Microbiol. Methods, 2008, 73, 92-99.

107.

H. Kim, H. Park and G. Ko, J. Food Protect., 2009, 72, 2547-2552.

108.

K. E. Gibson and K. J. Schwab, Appl. Environ. Microbiol., 2011, 77, 355-362.

109.

P. S. K. Knappett, A. Layton, L. D. McKay, D. Williams, B. J. Mailloux, M. R. Huq,
M. J. Alam, K. M. Ahmed, Y. Akita, M. L. Serre, G. S. Sayler and A. van Geen,
Groundwater, 2011, 49, 53-65.

110.

J. Rexroad, R. K. Evans and C. R. Middaugh, Journal of pharmaceutical
sciences, 2006, 95, 237-247.

111.

Y. Y. Feng, S. L. Ong, J. Y. Hu, X. L. Tan and W. J. Ng, Journal of industrial
microbiology & biotechnology, 2003, 30, 549-552.

112.

I. Samandoulgou, R. Hammami, R. M. Rayas, I. Fliss and J. Jean, Applied and
Environmental Microbiology, 2015, 81, 7680-7686.

113.

J. J. Hansen, P. S. Warden and A. B. Margolin, International journal of
environmental research and public health, 2007, 4, 61-67.

114.

R. Nims and M. Plavsic, Pharmaceuticals (Basel, Switzerland), 2013, 6, 358-392.

115.

A. L. Favier, W. P. Burmeister and J. Chroboczek, Virology, 2004, 322, 93-104.
83

116.

E. V. Pasco, H. Shi, I. Xagoraraki, S. A. Hashsham, K. N. Parent, M. L. Bruening
and V. V. Tarabara, J. Membr. Sci., 2014, 469, 140-150.

117.

S. Skraber, C. Gantzer, K. Helmi, L. Hoffmann and H. M. Cauchie, Food Environ.
Virol., 2009, 1, 66-76.

118.

P. B. Liu, V. R. Hill, D. Hahn, T. B. Johnson, Y. Pan, N. Jothikumar and C. L.
Moe, J. Microbiol. Methods, 2012, 88, 155-161.

119.

A. M. Kahler, T. B. Johnson, D. Hahn, J. Narayanan, G. Derado and V. R. Hill,
Water, 2015, 7, 1202-1216.

120.

E. R. Rhodes, E. M. Huff, D. W. Hamilton and J. L. Jones, J. Virol. Methods,
2016, 228, 31-38.

121.

H. Shi, I. Xagoraraki, K. N. Parent, M. L. Bruening and V. V. Tarabara, Appl.
Environ. Microbiol., 2016, 82, 4982-4993.

122.

L. Pei, M. Rieger, S. Lengger, S. Ott, C. Zawadsky, N. M. Hartmann, H. C.
Selinka, A. Tiehm, R. Niessner and M. Seidel, Environ. Sci. Technol., 2012, 46,
10073-10080.

123.

P. Nestola, D. L. Martins, C. Peixoto, S. Roederstein, T. Schleuss, P. M. Alves, J.
P. B. Mota and M. J. T. Carrondo, PLOS ONE, 2014, 9.

124.

K. E. Gibson and K. J. Schwab, Appl. Environ. Microbiol., 2011, 77, 385-391.

125.

J. Olszewski, L. Winona and K. H. Oshima, Can. J. Microbiol., 2005, 51, 295303.

126.

V. B. Rajal, B. S. McSwain, D. E. Thompson, C. M. Leutenegger, B. J. Kildare
and S. Wuertz, Water Res., 2007, 41, 1411-1422.

127.

E. R. Rhodes, D. W. Hamilton, M. J. See and L. Wymer, J. Virol. Methods, 2011,
176, 38-45.

128.

A. L. Polaczyk, J. M. Roberts and V. R. Hill, J. Microbiol. Methods, 2007, 68, 260266.

129.

D. B. Burns and A. L. Zydney, Biotechnol. Bioeng., 1999, 64, 27-37.

130.

J. A. Koehler, M. Ulbricht and G. Belfort, Langmuir, 1997, 13, 4162-4171.

131.

J. F. Hester, P. Banerjee and A. M. Mayes, Macromolecules, 1999, 32, 16431650.

132.

R. Attinti, J. Wei, K. Kniel, J. T. Sims and Y. Jin, Environ. Sci. Technol., 2010, 44,
2426-2432.
84

133.

P. A. Shields and S. R. Farrah, Appl. Environ. Microbiol., 2002, 68, 3965-3968.

134.

L. Gutierrez, S. E. Mylon, B. Nash and T. H. Nguyen, Environ. Sci. Technol.,
2010, 44, 4552-4557.

135.

K. Wong, B. Mukherjee, A. M. Kahler, R. Zepp and M. Molina, Environ. Sci.
Technol., 2012, 46, 11145-11153.

136.

K. S. Zerda, C. P. Gerba, K. C. Hou and S. M. Goyal, Appl. Environ. Microbiol.,
1985, 49, 91-95.

137.

L. J. Winona, A. W. Ommani, J. Olszewski, J. B. Nuzzo and K. H. Oshima, Can.
J. Microbiol. , 2001, 47, 1033-1041.

138.

D. O. Cliver, Appl. Microbiol., 1965, 13, 417-425.

139.

C. A. Sorber, B. P. Sagik and J. F. Malina, Water Res., 1972, 6, 1377-1388.

140.

A. Furiga, G. Pierre, M. Glories, P. Aimar, C. Roques, C. Causserand and M.
Berge, Appl. Environ. Microbiol., 2011, 77, 229-236.

141.

G. l. Pierre, A. l. Furiga, M. Berge, C. Roques, P. Aimar and C. Causserand, J.
Membr. Sci., 2011, 381, 41-49.

142.

R. Michalsky, P. H. Pfromm, P. Czermak, C. M. Sorensen and A. L. Passarelli, J.
Virol. Methods, 2008, 153, 90-96.

143.

M. Jin, X. Guo, X. W. Wang, D. Yang, Z. Q. Shen, Z. G. Qiu, Z. L. Chen and J.
W. Li, Environ. Sci. Technol., 2014, 48, 6947-6956.

144.

M. D. Sobsey and A. R. Hickey, Appl. Environ. Microbiol., 1985, 49, 259-264.

145.

N. Guttmanbass and J. Catalanosherman, Appl. Environ. Microbiol., 1986, 52,
556-561.

146.

A. De Keuckelaere, L. Baert, A. Duarte, A. Stals and M. Uyttendaele, J Virol.
Methods, 2013, 187, 294-303.

147.

S. Abd-Elmaksoud, S. K. Spencer, C. P. Gerba, A. H. Tamimi, W. E. Jokela and
M. A. Borchardt, Food Environ. Virol., 2014, 6, 253-259.

148.

J. L. Cashdollar, N. E. Brinkman, S. M. Griffin, B. R. McMinn, E. R. Rhodes, E. A.
Varughese, A. C. Grimm, S. U. Parshionikar, L. Wymer and G. S. Fout, Appl.
Environ. Microbiol., 2013, 79, 215-223.

149.

J. Wu, R. A. Rodriguez, J. R. Stewart and M. D. Sobsey, J. Appl. Microbiol.,
2011, 110, 1332-1340.

85

150.

S. E. Mylon, C. I. Rinciog, N. Schmidt, L. Gutierrez, G. C. L. Wong and T. H.
Nguyen, Langmuir, 2010, 26, 1035-1042.

151.

J. Heffron and B. K. Mayer, Environ. Sci.: Water Res. Technol., 2016, 2, 443459.

152.

R. A. Guy, P. Payment, U. J. Krull and P. A. Horgen, Appl. Environ. Microbiol.,
2003, 69, 5178-5185.

153.

M. Soto-Beltran, L. A. Ikner and K. R. Bright, Food Environ. Virol., 2013, 5, 9196.

154.

M. Abbaszadegan, M. S. Huber, C. P. Gerba and I. L. Pepper, Appl. Environ.
Microbiol., 1993, 59, 1318-1324.

155.

K. J. Schwab, R. Deleon and M. D. Sobsey, Appl. Environ. Microbiol., 1995, 61,
531-537.

156.

S. R. Farrah and G. Bitton, Can. J. Microbiol., 1979, 25, 1045-1051.

157.

S. R. Farrah, P. R. Scheuerman and G. Bitton, Appl. Environ. Microbiol., 1981,
41, 455-458.

158.

M. D. Sobsey, J. S. Glass, R. J. Carrick, R. R. Jacobs and W. A. Rutala, J. Am.
Water Works Assn., 1980, 72, 292-299.

159.

M. D. Sobsey, J. S. Glass, R. R. Jacobs and W. A. Rutala, J. Am. Water Works
Assn., 1980, 72, 350-355.

160.

S. Sun, Y. L. Shi, H. I. Tong, W. Kang, Z. Wang, E. Allmann and Y. A. Lu, J.
Virol. Methods, 2016, 229, 78-85.

161.

S. Yoon, C. Lee, K. Kim and A. G. Fane, Water Res., 1998, 32, 2180-2186.

162.

H. Lee, M. Kim, S. Y. Paik, C. H. Lee, W. H. Jheong, J. Kim and G. Ko, J. Water
Health, 2011, 9, 27-36.

163.

P. M. Holowecky, R. R. James, D. P. Lorch, S. E. Straka and H. D. A. Lindquist,
J. Appl. Microbiol., 2009, 106, 738-747.

164.

R. Lu, Q. Li, Z. Yin, I. Xagoraraki, V. V. Tarabara and T. H. Nguyen, J. Membr.
Sci., 2016, 497, 120-127.

165.

S. Petterson, R. Grøndahl-Rosado, V. Nilsen, M. Myrmel and L. J. Robertson,
Water Res., 2015, 87, 79-86.

166.

X. L. Pang, B. E. Lee, K. Pabbaraju, S. Gabos, S. Craik, P. Payment and N.
Neumann, J. Virol. Methods, 2012, 184, 77-83.
86

167.

D. S. Francy, E. A. Stelzer, A. M. Brady, C. Huitger, R. N. Bushon, H. S. Ip, M.
W. Ware, E. N. Villegas, V. Gallardo and H. D. Lindquist, Appl. Environ.
Microbiol., 2013, 79, 1342-1352.

168.

W. M. El-Senousy, M. I. Costafreda, R. M. Pinto and A. Bosch, Int. J. Food
Microbiol., 2013, 167, 74-79.

169.

T. H. Jones, V. Muehlhauser and G. ThĂŠriault, J. Virol. Methods, 2014, 206, 5-11.

87

CHAPTER THREE
The choice of virus propagation and purification methods affects
results of virus size and surface charge measurements

Abstract

Two virus propagation methods (in broth and double agar overlay) and three purification
procedures (PEG precipitation, centrifugal diafiltration, CsCl density gradient
centrifugation) were employed to grow and purify bacteriophages MS2 and P22. The
prepared suspensions were characterized in terms of virus size, charge, and
hydrophobicity using dynamic light scattering, phase analysis light scattering, and
contact angle measurements, respectively. The results of physicochemical
characterization of purified bacteriophages were found to depend on the choice of
propagation and purification methods. Regardless of the purification method applied,
virus propagation in broth showed advantage over that in double agar overlay, since the
latter introduced difficult-to-remove impurities attributable to residual agar. Of the three
purification methods evaluated with MS2 and P22 propagated in broth, CsCl density
gradient centrifugation gave the highest quality bacteriophage suspensions with a
narrow size distribution. The impurities remaining in the virus stock after PEG
precipitation and centrifugal diafiltration broadened the size distribution and interfered
with either charge measurements or hydrophobic characterization or both.
Electrophoretic mobilities determined after centrifugal ultrafiltration were similar to those
with CsCl density gradient centrifugation, but hydrophobicity characterization indicated
88

impurities. The isoelectric points of MS2 and P22 after PEG precipitation produced a
shift; the shift can be attributed to the change of virus hydrodynamic permeability in the
presence of residual PEG. Although time-consuming, propagation in broth combined
with CsCl density gradient centrifugation gives highest quality virus suspensions.

3.1 Introduction

Viral contamination of the water supply is a common cause of waterborne diseases
worldwide 1-3. Combined and sanitary sewer overflow, illicit discharge to storm water
systems and septic system failure can lead to a release of viruses into environment,
posing a threat to public health 4, 5.

Mechanistic studies of virus adsorption on soils and aqueous colloids have been
conducted by many researchers. In water and wastewater treatment, complete virus
removal by filtration can be difficult due to their small size 6, 7. In addition to virus
removal, filtration was also widely used to concentrate virus from samples prior to
detection. Low concentrations of viruses in natural waters and treatment systems
highlight the importance of sample concentration for reliable virus detection 8. Several
research efforts have been made to study the factors governing the virus fate during
filtration which can guide the design of virus removal 9-11 or concentration 8, 12, 13
processes for effective removal or recovery. Comprehensive studies of viruses’
interaction with various environmental and engineered surfaces are essential to
understanding mechanisms that control transport of viruses in different systems and

89

further prevent outbreak of viruses. Numerous studies on virus-surface interaction have
demonstrated the significant role of surface properties of viruses. Scrupulous
physicochemical characterization of viruses needs to be performed and thus, enables
rigorous modeling to understand the energetics of interactions. However, viruses
directly harvested from host bacteria or cell lines are stored in growth media containing
a large fraction of impurities such as incomplete virions, cell debris, microbial product
(protein, DNA, etc.). Experiments on virus physicochemical interaction and
characterization require that virions be separated from those impurities as the essential
step. Virus purification methods include density gradients centrifugation 8, 14-16,
polyethylene glycol (PEG) precipitation 16-19, direct diafiltration 20, 21 or dialysis 22, 23,
ultracentrifugation 24 and chromatography 25, 26.

Density gradient centrifugation can be of two general types: rate-zonal centrifugation
and isopycnic centrifugation. Sucrose and CsCl are commonly-used to prepare density
gradients in both methods. Rate-zonal centrifugation separates viruses primarily based
on differences in size and mass, which result in different sedimentation rates. In ratezonal separation, a virus sample is layered as a narrow zone on the top of continuous
density gradient. In contrast, isopycnic centrifugation separates viruses based solely on
differences in the buoyant density rather than size. In this method, a virus sample could
be either overlaid on or placed under prepared density gradient (either continuous or
discontinuous). Alternatively, to ease the sample handling, some gradient materials
(e.g. CsCl, Cs2SO4, iodixanol) can be directly dissolved into virus sample and gradient
can be self-generated during ultracentrifugation. Both rate-zonal and isopycnic

90

centrifugation methods could effectively separate viruses from other impurities by
forming a band that contains only the target virus. Unfortunately, some viruses (e.g.
herpesvirus 27, rotavirus 28, human respiratory syncytial virus 29) may lose infectivity
during density gradient centrifugation.

Virus precipitation using PEG relies on PEG acting as an “inert solvent sponge” that
sterically excludes viruses from the solvent and causes viruses precipitate from the
growth medium 30, 31. Other growth medium components including metals

32,

proteins 32

and DNA 33 can be co-precipitated. However, PEG removal by dialysis is not easy and
could be time-consuming due to the low diffusivity of this relatively large (typically 6 to 8
kDa) molecule 34, 35.

Centrifugal diafiltration, similar to dialysis is a technique that uses ultrafiltration
membranes to separates virions from other components in the growth medium based
on their size. With pressure applied across a membrane, virus purification using
diafiltration could be performed much faster than with dialysis, which is driven by a
concentration gradient and requires frequent changes of the buffer solution. In addition,
diafiltration concentrates and purifies viruses in one batch process minimizing virus loss
whereas traditional dialysis involving multiple changes of the dialysis bag only purifies,
but not concentrates virus. Two major drawbacks common to dialysis and diafiltration
are potential virus aggregation 36 and retention of impurities with the molecular weight
larger than the pore size of the dialysis or diafiltration membrane.

91

Chromatographic technique separates the target virus based on its interactions with
chromatographic resin. The interactions depend on the choice of the resin and virus
properties such as size, charge, hydrophobicity and ligand specificity, all of which
determine separation efficiency. Drawbacks of the method include possible virus
degradation due to harsh desorption condition and virus aggregation

8, 37, 38.

Differential ultracentrifugation involves the pelleting of viruses from growth medium
followed by resuspension of virus pellets with desired buffer. However, cost of
ultracentrifugation machine is high and ultracentrifugation is a time-consuming process,
especially for small viruses. Impurities with small molecular weight present at the bottom
in homogeneous sample could also cross-contaminate the virus pellets. Furthermore,
differential pelleting of viruses could possibly damage and aggregate some viruses,
leading to low recoveries. For example, bovine rotavirus recovery was significantly
reduced with ultracentrifugation (7%) relative to that with PEG precipitation (64%).

The same virus, but purified with different methods could result in distinct adsorption
behavior 39, aggregation dynamic 17, 36 and electrokinetics 36. Armanious et al. observed
that MS2 purified with PEG precipitation leads to less adsorption on poly-L-lysine
surface in QCM-D study than that purified with diafiltration 39. In a study by Dika et al.,
MS2 and MS2 like particle were prepared with three purification methods (dialysis, PEG
precipitation and density gradient) and for both particles, size as function of pH and
electrophoresis as function of ionic strength varied with different purification protocols
Poorly purified virus stock may disrupt physicochemical characterization in two ways:

92

36.

(1) measurement results were not only contributed by virions, but also adversely
affected by impurities present in the solution, leading to measurement artifact; (2)
impurities were adsorbed to viral capsid, masking the surface properties of virions

40.

These studies indicated that purity of virus stock is an important factor to consider when
evaluating the physicochemical interactions of virus in aqueous environments. Dika et
al. showed effects of virus purification on size and electrokinetics, but such effects
remain unknown on other important physicochemical properties of the virus such as its
isoelectric point (pI) and surface tension.

In this study, we purified viruses with three commonly used methods (CsCl density
gradient, PEG precipitation and centrifugal diafiltration) and evaluated effects of those
methods on determination of hydrodynamic size and electrophoresis as function of pH,
IEP and surface tensions. Two bacteriophages were used in this study: one is MS2 with
ssRNA and the other one is P22 with dsDNA. Both were typical surrogates for human
enteric viruses 12, 41-43.

The premise of the present work is that the propagation and purification procedures
should be selected together as an appropriate sample preparation method. While
human viruses can only be grown in an established cell line that is maintained in culture
vessel with growth medium, bacteriophages can be propagated using host bacteria
either cultivated on agar plate or suspended in liquid broth. The choice of the growth
medium determines which impurities can be introduced to the virus suspension during
the growth stage and what purification method is best suited for removing these

93

contaminants while avoiding virus loss. The optimal growth-purification sequence
should yield high titer purity virus stock while preserving virus infectivity. For this
purpose, we also compared infectious recoveries by different purification methods
tested.

3.2 Material and Methods

Table 7 summarizes the propagation and purification methods evaluated in this study.
The double layer agar method was not studied for the P22 phage.

Table 7. The propagation and purification methods evaluated in this study.
Preparation

Phage

Propagation medium

“MS2-agar-PEG”

Purification method
PEG precipitation

“MS2-agar-UF”

Double layer agar

“MS2-agar-CsCl”

Centrifugal diafiltration
CsCl 𝜌-gradient centrifugation

MS2
“MS2-broth-PEG”

PEG precipitation

“MS2-broth-UF”

E. coli broth

Centrifugal diafiltration

“MS2-broth-CsCl”

CsCl 𝜌-gradient centrifugation

“P22-broth-PEG”

PEG precipitation

“P22-broth-UF”

P22

Lysogeny broth

“P22-broth-CsCl”

Centrifugal diafiltration
CsCl 𝜌-gradient centrifugation

94

3.2.1 Virus propagation

Bacteriophage MS2 was purchased from ATCC (ATCC 15597-B1) and propagated in
both double agar overlay (recommended by ATCC) and liquid broth with Escherichia
medium (ATCC medium 271) and Escherichia coli strain C3000 (ATCC 15597) as the
host. Bacteriophage P22 and its host, Salmonella enterica subsp. enterica serovar
Typhimurium strain LT2, were provided by Dr. Kristin Parent (Michigan State
University). In contrast to MS2, P22 was propagated only in Lysogeny broth (LB). In this
study, virus propagated in double agar overlay is referred to as agar-propagated virus,
while virus propagated in suspended broth is referred to as broth-propagated virus.

3.2.1.1 Virus propagation in double agar overlay

For phage propagation in double layer agar, fresh colonies of host bacteria were first
inoculated into broth and incubated at 37 oC for approximately 2 h to reach the log
phase. Melted nutrient soft agar (8 mL, 0.5% w/v) was poured into each tube and kept
at 48 oC. A suspension of host bacteria in log phase (300 ÂľL) and diluted stock of
phages, (200 µL, 5×105 phage/mL) were then inoculated into soft agar sequentially.
After inoculation, soft agar was immediately overlaid on a nutrient agar plate (1.5% w/v).
Plates were incubated at 37 oC overnight. To harvest viruses, 5 mL of broth was added
on each plate and plates were placed again in 37 oC for 3 h. Then, to separate virus
from soft agar and host bacteria, soft agar was scraped off the surface of agar plates,
placed into a centrifuge tube and centrifuged at 8,000 g and 4 oC; after 15 min of

95

centrifugation the supernatant was collected, filtered through 0.22 Âľm filter and stored at
4 oC until purification.

3.2.1.2 Virus propagation in broth

For phage propagation in broth, a fresh colony of host bacteria was inoculated to 10 mL
broth one day prior to phage propagation and was incubated overnight. Next day,
overnight culture was added into fresh broth and the bacterial culture was shaken at
200 rpm at 37 oC. When optical density at wavelength 600 nm (OD600) reached 0.1,
phage stock was inoculated into bacterial culture with multiplicity of infection (MOI) of
0.1. Bacterial culture was further incubated at 37 oC with shaking (at 200 rpm for P22
and at 100 rpm for MS2) for another 8 h. Prior to virus harvest, several drops of
chloroform were added to culture. Culture was then rested for 10 to 15 min and
transferred to a centrifuge tube. Bacteria debris were removed via centrifugation at
8,000 g for 10 min at 4 oC. The supernatant with the phage was filtered through a 0.22
Âľm filter and stored at 4 oC until purification.

3.2.2 Virus purification

Figure 1 showed schematic graph of purification methods. The detailed procedure for
each purification method were discussed below.

96

3.2.2.1 Virus purification by CsCl density gradient centrifugation

Phages were pre-concentrated prior to purification due to volume limitation of
ultracentrifuge tubes used. For P22, phages were pelleted from stock via centrifugation
at 28,000 g for 90 min at 4 oC. After centrifugation, supernatant was discarded. Pellet
was resuspended in 2 mL of the pH 7.6 buffer (10 mM Tris, 10 mM MgCl2; later in the
text the buffer is denoted Tris-MgCl2) by overnight nutation at 4 oC. The small size of
MS2 virions makes pelleting this phage difficult and MS2 stock was concentrated down
to 2 mL using Amicon Ultra-4 centrifugal filters instead.

CsCl solutions with different densities were prepared for isopycnic centrifugation. Each
CsCl solution was layer-by-layer loaded into a tube with the highest density layer at the
bottom and the lowest density layer at the top. CsCl solutions were then capped with a
25% sucrose solution as a cushion. Density steps for MS2 were 1.50 g/cm 3 CsCl, 1.35
g/cm3 CsCl, 25% sucrose cushion; For P22, density steps were 1.60 g/cm3 CsCl, 1.40
g/cm3 CsCl, 25% sucrose cushion.

Phage resuspension was applied onto sucrose cushion and a virus band was obtained
via ultracentrifugation at 30,000 rpm at 18 oC for 3 h. The virus band was collected with
a syringe by puncturing the tube. CsCl in purified virus stock was then removed via
dialysis against 1 mM NaCl (pH 5.7, unadjusted) and the purified stock was stored at 4
oC.

97

3.2.2.2 Virus purification by PEG precipitation

PEG 6,000 was added into phage stock with gentle stirring to reach the final
concentration of 10% (w/v), followed by addition of NaCl with the final concentration of
0.5 M. Upon addition of PEG and NaCl, virus stock became turbid. Virus stock was
stored in the dark at 4 oC with gentle stirring overnight. The stock was then centrifuged
at 10,000 g for 30 min at 4 oC. The supernatant was discarded and the pellet was
resuspended via overnight nutation with 1 mM NaCl (pH 5.7, unadjusted) which was
filtered through a 0.22 Âľm pore size membrane prior to use.

The solution with the resuspended virus was treated by adding chloroform (1:1 v/v) to
remove remnant PEG. Upon addition of chloroform, resuspension was vigorously
vortexed for 30 s and then centrifuged at 1,700 g for 30 min. The aqueous fraction
above the white layer was carefully aspirated into tube as stock without disrupting the
white layer. Finally, the virus stock was dialyzed against 1 mM NaCl (pH 5.7,
unadjusted) and stored at 4 oC.

3.2.2.3 Virus purification by centrifugal diafitration

Amicon Ultra-4 centrifugal filters (MWCO 100 KDa) were used to purify viruses by
centrifugal diafiltration. Phage stock was loaded into Amicon Ultra filter unit and
centrifuged at 1,500 g to bring the concentrate volume to ~0.5 mL. Pre-filtered 1 mM
NaCl solution (pH 5.7, unadjusted) was then added to the concentrate to fill up the

98

centrifuge tube and the suspension was centrifuged again at 1,500 g until 0.5 mL
sample remained in the concentrate. This washing step was repeated at least 10 times
to remove broth remnants and to complete storage solution exchange. Finally, the
concentrate with was collected and used as the purified stock.

99

Figure 1. Schematic graph of purification with CsCl density gradient, polyethylene glycol precipitation and centrifugal diafiltration.

100

3.2.3 Hydrodynamic diameter and electrophoretic mobility measurements

Malvern Zetasizer Nano-ZS was used to measure both the hydrodynamic diameter and
the electrophoretic mobility of virions. Hydrodynamic diameter was measured by
dynamic light scattering while electrophoretic mobility was determined by phase
analysis light scattering. Zeta potential was then calculated from electrophoretic
mobilities. Prior to virus size and charge characterization, purified stocks were filtered
using 0.22 Âľm filter for P22 and 0.1 Âľm filter for MS2 to remove any aggregates that
could have formed after purification. Hydrodynamic diameter and electrophoretic
mobility were measured as functions of pH. Purified virus stocks were diluted with
background solutions with varied pH adjusted using NaOH and HCl. 1 mM NaCl
solution was used as background electrolyte for all pH except pH 7.6. The pH of 1 mM
NaCl solution could not be stabilized at pH 7.6 by adding NaOH or HCl only. Instead,
Tris-MgCl2 buffer was used as the background solution at this pH.

3.2.4 Surface tension and hydrophobicity determination

Surface tension of viruses purified by different methods was determined using the
protocol described in our previous work 8, 12. Briefly, to prepare a virus lawn, purified
virus stock was filtered through a 30 KDa ultrafiltration membrane. The membrane
coated with a multilayer cake of virus particles was dried at room temperature until the
water contact angle on the membrane reached a plateau. Contact angles of three probe
liquids (water, glycerol and diiodomethane) on the virus lawn were then measured using

101

sessile drop method. Surface tension components of each virus (𝛾𝑣𝐿𝑊 , 𝛾𝑣+ , 𝛾𝑣− ) were
obtained by substituting measured contact angles and known surface tensions of probe
liquids into the Young-DuprĂŠ equation:
(1 + cos 𝜃)𝛾𝑙𝑇𝑂𝑇 = 2 (√𝛾𝑣𝐿𝑊 𝛾𝑙𝐿𝑊 + √𝛾𝑣+ 𝛾𝑙− + √𝛾𝑣− 𝛾𝑙+ )

(3-1)

where 𝜃 is the contact angle of the probe liquid on the virus lawn, 𝛾 𝑇𝑂𝑇 is the total
surface energy, 𝛾 𝐿𝑊 is the LW component of surface energy, 𝛾 + is the electron-acceptor
parameter of surface energy, and 𝛾 − is the electron-donor parameter of surface energy.
Subscripts 𝑙 and 𝑣 refer to probe liquid and virus respectively. The free energy of
interfacial interaction between two virions immersed in water (Δ𝐺𝑣𝑤𝑣 ) was calculated
using equation (3-2):
Δ𝐺𝑣𝑤𝑣 =

−2 (√𝛾𝑣𝐿𝑊

−

2
𝐿𝑊
√𝛾𝑤 )

− 4 (√𝛾𝑣+ 𝛾𝑣− + √𝛾𝑤+ 𝛾𝑤− − √𝛾𝑣+ 𝛾𝑤− − √𝛾𝑣− 𝛾𝑤+ )

(3-2)

and used as a quantitative measure of hydrophobicity of the virus in question. A virus is
hydrophobic when Δ𝐺𝑣𝑤𝑣 < 0. The absolute value of Δ𝐺𝑣𝑤𝑣 indicates the degree of
hydrophobicity (or hydrophilicity, when Δ𝐺𝑣𝑤𝑣 > 0) of the virus.

102

3.3 Results and Discussion

3.3.1 Effects of the virus propagation method on the hydrodynamic size and
surface charge of MS2 purified by three different procedures

3.3.1.1 Effect on MS2 hydrodynamic size

MS2 propagated either on agar or in E. coli culture broth was purified by one of the
three methods: PEG precipitation, centrifugal diafiltration, and CsCl ρ-gradient
centrifugation (Table 7). To evaluate the effect of the propagation protocol on virus
physicochemical properties, we only compared results obtained with phages from
stocks purified by the same method, but propagated differently. Figure 2 shows
intensity-based size distribution in 1 mM NaCl, unadjusted pH 5.7 (storage solution for
purified virus) for agar-propagated MS2 and broth-propagated MS2 with each
purification applied. The average hydrodynamic diameter (𝑑ℎ ) and polydispersitivity
index in storage solution in each case were summarized in Table 8. 𝑑ℎ for MS2-brothCsCl batch in the storage solution was 30.2 ± 0.3 nm, which was similar to the individual
virion size (27 nm 44) obtained by TEM. The single peak at 30 nm with a narrow halfwidth (PDI ~0.12) suggests minimal aggregation. The average hydrodynamic diameter,
𝑑ℎ , was higher (37.8 ± 0.2 nm) in the MS2-agar-CsCl batch. An additional small peak
(18.3% of the intensity) at ~400 to 500 nm was observed in this condition with high PDI
(~0.32). Because the purification method was the same (CsCl density gradient), the
broadening of the PSD is an effect of the propagation method.
103

Figure 2. Size distribution of MS2 propagated by two different methods (in broth and on double
layer agar overlay) and purified by CsCl density gradient centrifugation (a), PEG precipitation (b)
and centrifugal ultrafiltration (c). Solution: 1mM NaCl, pH 5.7.

104

Average hydrodynamic diameter of MS2 in MS2-agar-PEG was 57 Âą 2 nm, significantly
larger than in the MS2-broth-PEG (38 Âą 2 nm). Similar trends were observed when MS2
stock was purified by centrifugal diafiltration, which gave 𝑑ℎ of 61 ± 3 nm in MS2-agarUF preparation and 39 ± 2 nm and in MS2-broth-UF preparation). For both purification
methods, the PDI values for agar-propagated MS2 (PDI of ~0.44 and ~0.66 for MS2agar-PEG and MS2-agar-UF) were higher than for broth-propagated MS2 (PDI of ~0.23
and ~0.29 for MS2-broth-PEG and MS2-broth-UF respectively). Regardless of the
propagation method used, a bimodal size distribution was obtained.

Table 8. Summary of average hydrodynamic diameter and polydispersity index for MS2 in
storage solution recorded for the six different preparations evaluated in this study.

Preparation

Average size, 𝑑ℎ (nm)

Polydispersity index, PDI

“MS2-agar-PEG”

57 Âą 2

0.66

“MS2-agar-UF”

61 Âą 3

0.44

“MS2-agar-CsCl”

38 Âą 0

0.32

“MS2-broth-PEG”

38 Âą 2

0.23

“MS2-broth-UF”

39 Âą 2

0.29

“MS2-broth-CsCl”

30 Âą 0

0.12

Regardless of the purification method applied, higher 𝑑ℎ , broadening of the main peak
and an additional peak with higher intensity were measured for agar-propagated MS2
indicating MS2 aggregation. We tentatively attribute this behavior to heteroaggregation
of MS2 virions with impurities introduced during MS2 propagation by the double agar
overlay method.
105

If a bimodal size distribution was obtained, an additional size measurement was
conducted with the same virus stock filtered again through a 0.1 Îźm pore size
membrane. This additional filtration step eliminated the peak corresponding to large
particles for all broth-propagated MS2 (regardless of the purification method) and for the
MS2-agar-CsCl preparation. Surprisingly, in MS2-agar-UF and MS2-agar-PEG, the
peak corresponding to large particles (~ 0.4 to 0.5 Îźm) remained even after multiple
filtrations through the 0.1 Îźm filter. We attribute the peak to the components of the agar
as agar is the only additional substance used in the double agar overlay procedure in
comparison with the broth-based growth.

Although agar gel had been removed from the virus harvest by centrifugation and
filtration (see Section 2.1.1), small agar molecules (MW range from <20 kDa to >150
kDa 45) or aggregates could remain in the stock. Virus pellets produced in the MS2agar-PEG process were gelatinous and gel aggregates were observed in the UF
concentrate during MS2-agar-UF. The gel could be agar aggregates formed from
smaller (≲ 0.1 μm) agar molecules concentrated along with the virus during purification.
Indeed it has been reported that agar molecules in filtrate can form double helix
structures and re-aggregate 45. We did observe an increase in the intensity of the size
distribution peak in the 400 to 500 nm range for post-filtration samples.

𝑑ℎ of MS2 as a function of pH for agar-propagated MS2 and broth-propagated MS2 with
each purification method was shown in Figure S1. Regardless of the purification

106

method, very large aggregates of 3 to 7 Îźm were observed in broth-propagated MS2
samples at pH 3.5 and lower. For agar-propagated MS2, such severe aggregation was
not observed when virus were purified with PEG precipitation or centrifugal diafiltration.
In those cases, 𝑑ℎ only increased to ~70 to 80 nm with a decrease in pH to 2.5. In MS2agar-CsCl samples, aggregates of ~3 μm were observed at pH 3.0.

3.3.1.2 Effect on MS2 charge

In a separate set of tests, the electrophoretic mobility, 𝜇, of MS2 propagated and
purified by six different protocols (Table 7) was measured. Eectrophoretic mobility and
calculated zeta potential as a function of pH are shown in Figure 3. The mobility of MS2agar-CsCl samples in the storage solution (pH 5.7, unadjusted) and 1 mM NaCl was 3.2 ¹ 0.2 ¾m cm¡s-1¡V-1. For any of purification methods applied, negative
electrophoretic mobility for MS2 propagated in double agar overlay at pH 5.7 was
always higher in magnitude than that propagated in broth. Depending on the purification
method applied, the pI of broth-propagated MS2 was either ~ 3.1 (for CsCl density
gradient purification and centrifugal diafiltration purification) or in the range between 2.2
and 2.5 (for PEG precipitation purification). In MS2-agar-PEG and MS2-agar-UF
samples, was not observed for the phage was found to be negative charged over the
entire pH range investigated (pH 2 to 10). This may explain the absence of large MS2
aggregates at low pH with these two purification methods as discussed earlier (see
section 3.3.1.1). Agar-based impurities such as agaropectin could potentially coat the
virus to impart a strong negative charge to it, even at very low pH. Agaropectin, a non-

107

gelling fraction of agar, is a highly negatively charged polysaccharide with acidic sidegroups such as ester sulfate, D-glucuronic acid, and pyruvic acid 46. Negative surface
charge and steric hindrance on MS2 caused by adsorption of agaropectin could
potentially explain the very moderate increase in 𝑑ℎ at low pH with these two
preparation methods. pI was observed (between pH 2.5 and 2.7) for MS2-agar-CsCl
samples; the value was lower than pH ~3.1 measured for the phage in MS2-broth-CsCl
samples.

108

Figure 3. Electrophoretic mobility and zeta potential as function of pH for MS2 propagated by
two different methods (in broth and on double layer agar overlay) and purified by CsCl density
gradient centrifugation (a), PEG precipitation (b) and centrifugal ultrafiltration (c).

The presented data and analysis clearly indicate that broth-based propagation
introduces less impurities that interfere with size and charge measurements of
bacteriophages. Of the three purification methods tested, CsCl density gradient
109

centrifugation was the only technique that could remove agar-derived impurities
introduced at the propagation stage. CsCl density gradient centrifugation, however, is a
labor-intensive and time-consuming process. This conclusion prompted a closer
inspection of the impact that each of the three purification methods has on results of the
physicochemical characterization of the virus.

3.3.2 Effects of the virus purification method on virus size and charge
measurements

To solely evaluate the effect of the purification method, P22 and MS2 were used as
model viruses. Both phages were propagated in liquid broth to avoid the interference of
impurities from agar (see section 3.3.1). Figure 4 shows the intensity based size
distribution in storage solution and electrophoretic mobility as a function of pH for both
MS2 and P22 with each purification method. Results for broth-propagated MS2 with
each purification method having been shown in Figure 2 (size measurement) and Figure
3 (electrophoretic mobility measurement) were regenerated in Figure 4(a) and Figure
4(b) here to better evaluate. those for broth-propagated P22 were shown in Figure 4 (a)
and Figure 4 (b).

110

Figure 4. Size distribution (a, c) and electrophoretic mobility (b, d) of broth-propagated MS2 (a, b) and P22 (c, d) phages purified by
CsCl density gradient centrifugation, PEG precipitation and centrifugal ultrafiltration. Solution used in size measurements: 1mM
NaCl, pH 5.7. Data for MS2 overlaps with parts of Figures 2 and 3 and are shown here for easier comparisons.

111

As mentioned in Section 3.3.1, MS2 size in MS2-broth-CsCl samples in the storage
solution was measured to be 30.2 Âą 0.3 nm with low PDI (~0.12) indicating a stable
suspension of monodisperse MS2 virions. In the MS2-broth-UF samples at the same
pH, 𝑑ℎ was ~38 nm with a higher PDI (~0.29) and a bimodal size distribution (main peak
at 32 nm and an additional lower band at ~400 nm) suggesting that aggregates form
during diafiltration. Compared with diafiltration purified MS2, PEG precipitation purified
MS2 gave similar 𝑑ℎ (~38 nm), but lower PDI (~0.23) with main peak at 32 nm and
negligible aggregation ranging from 1 to 5 μm. 𝑑ℎ for PEG purified MS2 in this study
was in agreement with those from other studies with the same purification method: 40
nm 17 and 40 to 60 nm 36. Since in MS2-broth-PEG aggregation was negligible (~4% of
the population), the higher 𝑑ℎ may have stemmed from PEG adsorbed to the viral
capsid; such sorption would decrease particle diffusivity and increase the hydrodynamic
size 36. This assumption is consistent with the finding by Gao et al. who reported that
conjugation of 5 kDa PEG to the adenovirus capsid (10% modification ratio) increased
virion size from 113 to 122 nm 47.

Figure 4 (c) illustrates the effects of the three purification methods on the hydrodynamic
diameter of bacteriophage P22. In the storage solution (1 mM NaCl, pH 5.7), the P22broth-CsCl was characterized by 𝑑ℎ of 69.4 ± 1.2 nm and a single peak at 68 nm with
the PDI of 0.04. The individual P22 virion size determined by TEM imaging was ~54
nm12. The difference could stem from the tail structure of P22, which impacts its
diffusivity 12, 48. As was the case with MS2, higher 𝑑ℎ (~72 nm) was measured for P22 in
P22-broth-PEG than in P22-broth-CsCl corroborating the hypothesis of PEG adsorption
112

on the capsid surface. Centrifugal diafiltration method resulted in bimodal size
distribution for P22, though 1 Îźm to 5 Îźm aggregates contributing to the secondary peak
in the size distribution were very minor (together corresponding to 2.8% of population).
𝑑ℎ increased to 76 nm with PDI at 0.18 in this case.

For both MS2 and P22 in the storage solution, the narrowest siz distributions were
always after purification by the CsCl density gradient. Comparing sizing data for MS2
and P22, one can see that each P22 suspension was more narrowly distributed than the
corresponding MS2 suspension purified by the same method. 𝑑ℎ as a function of pH for
both phages with each purification method were shown in Figure S2. As pH decreased,
severe aggregation of MS2 was observed in all sample preparations (Figure S2 (a)).
Larger aggregates (up to ~7 Îźm) were observed after PEG precipitation and diafiltration,
while CsCl density gradient resulted in relatively small aggregates (< 3 Îźm). In
comparison with MS2, the largest P22 aggregates were detected as pH decreased were
~600 nm with CsCl density gradient purification (Figure S2(b)). This suggests that, in
comparison with MS2, P22 virions could be better separated from the components of
the growth medium (broth) and are more stable in the 1 mM NaCl background solution.
For measurements at pH 7.6 Tris-MgCl2 buffer was used as the background solution. At
this pH, the sis distribution of P22 had a single narrow peak at ~70 nm, close to the size
measured in 1 mM NaCl at pH 5.7. In contrast, 𝑑ℎ for MS2 depended on the choice of
the purification method. MS2-broth-CsCl preparations had 𝑑ℎ of 29 nm, which was very
close to the size obtained in TEM. MS2-broth-UF and MS2-broth-PEG preparations,
however, had 𝑑ℎ values of 45 nm and 59 nm, respectively. These larger averages
113

could possibly be explained by impurities that may aggregate in higher IS (~25 mM) of
the Tris-MgCl2 buffer. The difference further underscores the advantage for CsCl
density gradient centrifugation as the purification method.

Figure 4 (c) and (d) presents electrophoretic mobility and corresponding zeta potential
values as functions of pH. In the storage solution, MS2 has electrophoretic mobility of 2.5 ¹ 0.3 ¾m¡s-1¡V-1¡cm when purified with diafiltration. Dika et al. 49, 50 and Langlet et al.
23

reported similar MS2 electrophoretic mobility in NaNO3 and at the same pH of 5.6 for

MS2 purified by dialysis, a method closely related to diafiltration. Regardless of the
purification method, the smaller magnitude of electrophoretic mobility was observed in
the Tris-MgCl2 buffer (pH 7.6) relative to 1mM NaCl, which can be attributed to electric
double layer compression at the higher ionic strength of buffer. The pI of MS2 ranges
from 2.2 to 4.0 depending on the solution chemistry 40. In our study, pI of MS2 was ~ 3.1
in both MS2-broth-UF and MS2-broth-CsCl preparations in 1 mM NaCl; this value is in
agreement with the results by Langlet et al, who reported pI of 3.1 for dialysis-purified
MS2 in 1 mM NaNO3 23. No significant differences in electrophoretic mobility were
observed for pH > 5.6 among the three purification methods. For pH <3.5, however.
MS2 phages in MS2-broth-PEG samples showed lower value of electrophoretic mobility
than MS2 purified with the other two methods. which lead to a decreased pI ranging
from 2.2 to 2.5 for PEG precipitation method.

Regardless of the purification method, the electrophoretic mobility for P22 measured in
Tris-MgCl2 (pH 7.6) was smaller in magnitude than that in 1 mM NaCl, pH 5.7, same

114

trend as MS2. For P22-broth-UF and P22-broth-CsCl in 1 mM NaCl, the pI of P22 was
~3.7, similar to pI of 3.4 reported by Fidalgo de Cortalezzi et al. for diafiltration-purified
P22 in 15 mM NaCl 51. The theoretical pI of P22 calculated by ProtParam, an analysis
software that calculates various physicochemical parameters for proteins was higher at
~5.0 52. The difference, also observed for other viruses could also reflect the effect of
the viral genome on the pI.

Similar to MS2, pI shift (reduced to ~3) was also observed for P22 when it is purified by
PEG precipitation method.

The pI shift for both phages may stem from interaction of PEG with virion though PEG is
neural polymer. A virus could be regarded as a permeable soft particle53, with the
surface charge partly dependent on the inner structure of virus (genome or protein
inside capsid). PEG-virion interaction could impact the electrophoretic mobility and pI in
a manner that depends on the structure of a PEG layer on the virion surface. PEG can
impact the electrokinetic properties of the virus also through its propensity to induce
virion aggregation at lower pH (see Figure S2) wherein the aggregate structure affects
both hydrodynamic and electrostatic properties of aggregated virions

53.

In our case, phage permeability may increase when aggregates occurred in presence of
PEG. Such increase in virion permeability may highlight the effect of internal genome
structure (RNA for MS2 with pI as low as ~2 36) on its electrophoretic mobility, thus
lower the pI of phage as a whole particle. Moreover, as mentioned in introduction, PEG

115

precipitates both viruses and, potentially, impurities (e.g. DNA, proteins). The impurities
co-precipitated with viruses may also be responsible for the pI shift observed for PEGpurified batches.

3.3.3 Effects of virus purification methods on surface tension parameters and
hydrophobicity of bacteriophages

We also evaluated the effect of the purification method on the values of surface tension
parameters calculated based on contact angle measurements. For P22 purified by CsCl
density gradient centrifugation, surface tension parameters and the interfacial free
energy had been reported in our previous work and were used here for comparison 12.
For all P22 preparations, apolar surface energy components (𝛾 𝐿𝑊 ) are between 42 and
44 mJ/m2 (Table 9), which is a range typical for biological materials 54. The purification
procedure did have an effect on the polar surface energy components. For P22-brothCsCl, the electron donor (𝛾 − ) and electron acceptor (𝛾 + ) parameters were 25.8 mJ/m2
and 0.06 mJ/m2, respectively, translating into the overall polar component, 𝛾 𝐴𝐵 of 2.4
mJ/m2. For P22-broth-PEG and P22-broth-UF preparations, we could not calculate the
appropriate virus surface tension components after substitution of the known probe
liquids surface tension and corresponding contact angles into Young-DuprĂŠ equation
(equation (3-1)).

116

Table 9. Measured contact angles, calculated surface energy parameters and free energy of interfacial interaction when immersed in
water for broth-propagated P22 bacteriophage as a function of the purification method.

Purification method

#

Contact angle

Surface energy parameters

Free energy of interfacial

with probe liquid (0)

(mJ/m2)

interaction in water,

H2O

Glycerol

DIM#

𝜸𝑳𝑾

𝜸+

𝜸−

𝜸𝑨𝑩

𝜸𝑻𝑶𝑻

𝚫𝑮𝒗𝒘𝒗 (mJ/m2)

CsCl density gradient

55 Âą 2

56 Âą 6

34 Âą 3

42.5

0.06

25.8

2.4

44.9

-6.3

PEG precipitation

44 Âą 5

65Âą 4

34 Âą 6

42.5

-

52.9

-

42.5

+38.1

Centrifugal diafiltraiton

49 Âą 8

75 Âą 6

30 Âą 2

44.2

-

57.3

-

44.2

+43.0

DIM = diiodomethane

117

Slightly negative values for √𝛾 + were obtained (~-0.9 for P22-broth-PEG and ~-1.8 for
P22-broth-UF). This may be explained by the presence of impurities that PEG
precipitation and diafiltration failed to remove (e.g. DNA, protein, other cellular lysis
debris and PEG). These impurities could be concentrated with the virus and nonuniformly distributed within the virus lawn. Each apparent contact angle on such surface
only depends on local structures instead of overall surface, which can bring error to our
measurements. This interpretation is consistent with the lower average water contact
angle on P22 lawn formed from P22-broth-CsCl than from P22-broth-CsCl and P22broth-CsCl preparations. Indeed, some of the potential impurities - DNA, RNA, and
PEG - are hydrophilic substances but with a negligible 𝛾 +

54.

Based on the surface

tension data reported in our previous work 8 and other studies 54, a very small value of
𝛾 + is also a common characteristic of non-enveloped viruses, which suggests that a
small error in contact angle measurements could lead to √𝛾 + <0.

We further calculated the interfacial free energy (Δ𝐺𝑣𝑤𝑣 ) for each purification method
using eqn (2). For PEG precipitation and diafitration methods, the calculation was made
with the assumption that 𝛾 + was negligible (~0). For P22-broth-CsCl, P22 was found to
be hydrophobic (Δ𝐺𝑣𝑤𝑣 <0) whereas for the other two methods Δ𝐺𝑣𝑤𝑣 was calculated to
be positive indicating that P22 is hydrophilic. Our results indicated that purification
methods could significantly impact the virus polarity characterization.

118

3.4 Summary

Two virus propagation methods (in broth and double agar overlay) and three purification
procedures (PEG precipitation, centrifugal diafiltration, CsCl density gradient
centrifugation) were employed to grow and purify bacteriophages MS2 and P22. The
prepared suspensions were characterized in terms of virus size, charge, and
hydrophobicity using dynamic light scattering, phase analysis light scattering, and
contact angle measurements, respectively. The physicochemical properties of purified
bacteriophages were found to depend on the choice of propagation and purification
methods.

Hydrodynamic size and electrophoretic mobility data for MS2 indicated that double agar
overlay growth introduces difficult-to-remove impurities that confound results of
physicochemical characterization of the virus. Specifically, agar-propagated MS2 stock
purified by PEG precipitation or centrifugal diafiltration included large (~ 0.4 to 0.5 Îźm)
“unfilterable” aggregates detected after repeated post-filtration with a 0.1 μm nominal
pore size membrane. Further, the isoelectric point of MS2 in this preparation could not
be determined as the bacteriophage remained electronegative even at the pH as low as
1.4. CsCl density gradient centrifugation could remove most impurities but still gave a
relatively broad MS2 size distribution. Such interferences were not observed for MS2
propagated in E. coli broth regardless of the purification method employed.
The degree of bacteriophage purity achievable with each of the three purification
methods was compared based on the data for MS2 and P22 propagated in broth media.

119

The impacts of the virus purification protocol on size and charge characteristics of P22
and MS2 were similar. The highest purity was achieved with CsCl density gradient
centrifugation, which consistently gave bacteriophage suspensions with a narrow size
distribution. PEG precipitation method produced broader size distributions and ionic
strength-dependent pI values. The dependency on ionic strength was interpreted to
result from PEG adsorption on the virus capsid rendering a “soft” aggregate of PEG and
one or more virions. Centrifugal diafiltration gave size and charge data of quality close
to that obtained with CsCl density gradient centrifugation; however, contact angle
measurements of probe liquids on lawns of diafiltration-purified viruses indicated
impurities that could not be discerned in particle size data.
PEG precipitation may be suitable for applications where the remnant PEG is not a
concern (e.g. due to its biological inertness) but should be avoided if impurity-free
virions are needed.

3.5 Conclusions

The performance of virus purification procedures depends on the choice of the
propagation method used to grow the virus. Agar-based propagation introduced
impurities that neither of the three purification procedures could remove fully. For
viruses propagated in broth media, the highest quality virus stock was obtained after
purification by CsCl density gradient centrifugation. The other two purification methods
are less costly and time-consuming and can be adequate choices for certain
applications. In general, the choice of the optimal purification procedure is a matter of

120

tradeoff where the demands for product quantity and quality (e.g. virus concentration
and infectivity, amount of residual impurities) should be weighed against issues of
practicability such as the time and cost of virus purification.

121

APPENDIX

122

APPENDIX

Figure 5. Average hydrodynamic diameter as function of pH for MS2 purified by CsCl density gradient (a), PEG precipitation (b) and
centrifugal ultrafiltration (c).

123

Figure 6. (a) Average hydrodynamic diameter as function of pH for MS2 purified by CsCl
density gradient, PEG precipitation and centrifugal ultrafiltration.

124

REFERENCES

125

REFERENCES

1.

T. T. Fong and E. K. Lipp, Microbiology and molecular biology reviews : MMBR,
2005, 69, 357-371.

2.

R. G. Sinclair, E. L. Jones and C. P. Gerba, Journal of applied microbiology,
2009, 107, 1769-1780.

3.

A. M. Gall, B. J. Marinas, Y. Lu and J. L. Shisler, PLoS pathogens, 2015, 11,
e1004867.

4.

USEPA, Journal, 2001.

5.

R. D. Arnone and J. P. Walling, Journal of water and health, 2007, 5, 149-162.

6.

Z. Altintas, M. Gittens, J. Pocock and I. E. Tothill, Biochimie, 2015, 115, 144-154.

7.

M. D. Sobsey, Managing water in the home: accelerated health gains from
improved water supply, World Health Organization Geneva, 2002.

8.

H. Shi, I. Xagoraraki, K. N. Parent, M. L. Bruening and V. V. Tarabara, Appl.
Environ. Microbiol., 2016, 4982-4993.

9.

R. M. Chaudhry, R. W. Holloway, T. Y. Cath and K. L. Nelson, Water research,
2015, 84, 144-152.

10.

J. Langlet, L. Ogorzaly, J.-C. Schrotter, C. Machinal, F. Gaboriaud, J. F. L. Duval
and C. Gantzer, J. Membr. Sci., 2009, 326, 111-116.

11.

H. Huang, T. A. Young, K. J. Schwab and J. G. Jacangelo, J. Membr. Sci., 2012,
409, 1-8.

12.

E. V. Pasco, H. Shi, I. Xagoraraki, S. A. Hashsham, K. N. Parent, M. L. Bruening
and V. V. Tarabara, J. Membr. Sci., 2014, 469, 140-150.

13.

A. L. Polaczyk, J. M. Roberts and V. R. Hill, Journal of Microbiological Methods,
2007, 68, 260-266.

14.

K. Wong, B. Mukherjee, A. M. Kahler, R. Zepp and M. Molina, Environmental
science & technology, 2012, 46, 11145-11153.

15.

J. F. Torres-Salgado, M. Comas-Garcia, M. V. Villagrana-Escareno, A. L. DuranMeza, J. Ruiz-Garcia and R. D. Cadena-Nava, Journal of Physical Chemistry B,
2016, 120, 5864-5873.

126

16.

T. L. Cromeans, A. M. Kahler and V. R. Hill, Applied and Environmental
Microbiology, 2010, 76, 1028-1033.

17.

T. H. Nguyen, N. Easter, L. Gutierrez, L. Huyett, E. Defnet, S. E. Mylon, J. K.
Ferri and N. A. Viet, Soft Matter, 2011, 7, 10449-10456.

18.

M. Tong, Y. Shen, H. Yang and H. Kim, Colloids and surfaces. B, Biointerfaces,
2012, 92, 340-347.

19.

C. T. Tanneru, J. D. Rimer and S. Chellam, Environmental science & technology,
2013, 47, 4612-4618.

20.

A. Armanious, M. Munch, T. Kohn and M. Sander, Environmental science &
technology, 2016, 50, 3597-3606.

21.

L. Gutierrez, X. Li, J. Wang, G. Nangmenyi, J. Economy, T. B. Kuhlenschmidt, M.
S. Kuhlenschmidt and T. H. Nguyen, Water research, 2009, 43, 5198-5208.

22.

B. Michen, F. Meder, A. Rust, J. Fritsch, C. Aneziris and T. Graule,
Environmental science & technology, 2012, 46, 1170-1177.

23.

J. Langlet, F. Gaboriaud, J. F. Duval and C. Gantzer, Water research, 2008, 42,
2769-2777.

24.

R. Attinti, J. Wei, K. Kniel, J. T. Sims and Y. Jin, Environmental science &
technology, 2010, 44, 2426-2432.

25.

R. Monjezi, B. T. Tey, C. C. Sieo and W. S. Tan, Journal of Chromatography B,
2010, 878, 1855-1859.

26.

K. Farkas, A. Varsani and L. P. Pang, Food and Environmental Virology, 2015, 7,
261-268.

27.

H. Pertoft, Virology, 1970, 41, 368-372.

28.

D. Y. Chen and R. F. Ramig, Virology, 1992, 186, 228-237.

29.

E. Gias, S. U. Nielsen, L. A. F. Morgan and G. L. Toms, Journal of Virological
Methods, 2008, 147, 328-332.

30.

G. D. Lewis and T. G. Metcalf, Appl. Environ. Microbiol., 1988, 54, 1983-1988.

31.

D. H. Atha and K. C. Ingham, J. Biol. Chem., 1981, 256, 12108-12117.

32.

A. J. Prussin, L. C. Marr and K. J. Bibby, FEMS Microbiol. Lett., 2014, 357, 1-9.

33.

N. Hammerschmidt, S. Hobiger and A. Jungbauer, Proc. Biochem., 2016, 51,
325-332.
127

34.

C. R. Cantor and R. K. Scopes, Protein Purification: Principles and Practice,
Springer, New York, 1994.

35.

M. Kleiner, L. V. Hooper and B. A. Duerkop, BMC Genomics, 2015, 16, 7.

36.

C. Dika, C. Gantzer, A. Perrin and J. F. Duval, Phys. Chem. Chem. Phys., 2013,
15, 5691-5700.

37.

K. S. Vijayaragavan, Doctor of Philosophy in Chemical Engineering Michigan
Technological University, 2014.

38.

V. Mautner, Methods Mol. Med., 2007, 130, 145-156.

39.

A. Armanious, M. Aeppli, R. Jacak, D. Refardt, T. Sigstam, T. Kohn and M.
Sander, Environ. Sci. Technol., 2016, 50, 732-743.

40.

B. Michen and T. Graule, J. Appl. Microbiol., 2010, 109, 388-397.

41.

J. Bae and K. J. Schwab, Appl. Environ. Microbiol., 2008, 74, 477-484.

42.

D. J. Dawson, A. Paish, L. M. Staffell, I. J. Seymour and H. Appleton, J. Appl.
Microbiol., 2005, 98, 203-209.

43.

Y. Masago, T. Shibata and J. B. Rose, Appl. Environ. Microbiol., 2008, 74, 58385840.

44.

J. H. Strauss, Jr. and R. L. Sinsheimer, Journal of molecular biology, 1963, 7, 4354.

45.

R. ArmisĂŠn and F. Gaiatas, in Handbook of Hydrocolloids (Second edition),
Woodhead Publishing, 2009, DOI: http://dx.doi.org/10.1533/9781845695873.82,
pp. 82-107.

46.

A. Nussinovitch, in Hydrocolloid Applications: Gum technology in the food and
other industries, Springer US, Boston, MA, 1997, DOI: 10.1007/978-1-46156385-3_1, pp. 1-18.

47.

J.-Q. Gao, Y. Eto, Y. Yoshioka, F. Sekiguchi, S. Kurachi, T. Morishige, X. Yao, H.
Watanabe, R. Asavatanabodee, F. Sakurai, H. Mizuguchi, Y. Okada, Y. Mukai,
Y. Tsutsumi, T. Mayumi, N. Okada and S. Nakagawa, Journal of Controlled
Release, 2007, 122, 102-110.

48.

R. E. Baltus, A. R. Badireddy, A. Delavari and S. Chellam, Environmental
science & technology, 2016, DOI: 10.1021/acs.est.6b05323.

49.

C. Dika, J. F. Duval, H. M. Ly-Chatain, C. Merlin and C. Gantzer, Appl Environ
Microbiol, 2011, 77, 4939-4948.

128

50.

C. Dika, M. H. Ly-Chatain, G. Francius, J. F. L. Duval and C. Gantzer, Colloids
and Surfaces A: Physicochemical and Engineering Aspects, 2013, 435, 178-187.

51.

M. M. Fidalgo de Cortalezzi, M. V. Gallardo, F. Yrazu, G. J. Gentile, O. Opezzo,
R. Pizarro, H. R. Poma and V. B. Rajal, Journal of Environmental Chemical
Engineering, 2014, 2, 1831-1840.

52.

K. N. Parent, E. B. Gilcrease, S. R. Casjens and T. S. Baker, Virology, 2012,
427, 177-188.

53.

J. Langlet, F. Gaboriaud, C. Gantzer and J. F. Duval, Biophysical journal, 2008,
94, 3293-3312.

54.

C. J. Van Oss, Interfacial forces in aqueous media, Taylor & Francis, Boca
Raton, Fla, 2006.

129

CHAPTER FOUR
Elution is a critical step for human adenovirus 40 recovery from tap
and surface water by crossflow ultrafiltration

Abstract

This paper examines the recovery of enteric adenovirus HAdV 40 by crossflow
ultrafiltration and interprets recovery values in terms of physicochemical interactions of
virions during sample concentration. Prior to ultrafiltration, membranes were either
blocked by 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 prediction by the extended Derjaguin-Landau-Verwey-Overbeek
theory, pre-elution recovery of HAdV, 𝑟𝑝𝑟𝑒 , from deionized water was higher with the
𝑃𝐸𝑀
𝐶𝑆
PEM-coated membranes (𝑟𝑝𝑟𝑒
= 74.8 ± 9.7%) than with CS-blocked membranes (𝑟𝑝𝑟𝑒
=

54.1 ± 6.2%). With either membrane type, the total virion recovery after elution, 𝑟𝑝𝑜𝑠𝑡 ,
𝐶𝑆
𝑃𝐸𝑀
was high in both deionized water (𝑟𝑝𝑜𝑠𝑡
= 99.5 ± 6.6%; 𝑟𝑝𝑜𝑠𝑡
= 98.8 Âą 7.7%) and tap
𝐶𝑆
𝑃𝐸𝑀
water (𝑟𝑝𝑜𝑠𝑡
= 89 ± 15% and 𝑟𝑝𝑜𝑠𝑡
= 93.7 Âą 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 virion and the
membrane. Addition of ethylenediamine tetraacetic acid (EDTA) to the eluent greatly
𝐶𝑆
𝑃𝐸𝑀
improved the elution efficacy (𝑟𝑝𝑜𝑠𝑡
= 88.6 ± 4.3%; 𝑟𝑝𝑜𝑠𝑡
= 87.0 Âą 6.9%) from surface

water even when organic carbon concentration in the water is high (9.4 Âą 0.1 mg/L).

130

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

This study presents results of the first comprehensive physicochemical characterization
of HAdV 40, an important human pathogen. The obtained data on HAdV 40 surface
properties enabled rigorous modeling to understand energetics of virion-virion and
virion-filter interactions. This is the first evaluation of crossflow filtration for concentration
and recovery of HAdV 40 and the first study where HAdV 40 is recovered by crossflow
filtration from practical water matrices. We demonstrate high HAdV 40 post-elution
recoveries from ultrapure water (99%), tap water (~91%), and high carbon content
surface water (~84%). These results are significant because of the very low adenovirus
recoveries that have been reported to date for other methods. We interpret recovery
data in terms of specific interactions and design the eluent composition accordingly to
maximize HAdV 40 recovery.

4.1 Introduction

Pathogenic 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,

131

coxsackievirus, 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 secondleading cause of childhood 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, 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 treatment systems, and eventually to prevent HAdV
outbreaks. However, the low concentrations of HAdV in various natural waters,101 to
104 genome copies (GC) per liter in river and lake water 8-10 and 103 to 105 GC/L in
wastewater treatment effluent 11, 12, make detection challenging and highlight the
importance of sample concentration for rapid and reliable virus detection.

VIRADEL (virus adsorption-elution), the standard method for virus concentration,
includes adsorption of virions on either electropositive or electronegative filters and
subsequent elution of 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 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 concentrating viruses from sea water using a
NanoCeram filter and 3%wt 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 of adenovirus may occur

132

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 relatively high
recovery of HAdV 40 with 1MDS filter (VIRADEL method) but the standard deviation
was also high (49 Âą 32%) 15.

Another virus concentration method, crossflow ultrafiltration (also called tangential flow
filtration), relies on size exclusion, rather than adsorption, to concentrate virions during
filtration and might give higher recoveries than the VIRADEL method. An ultrafiltration
(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 crossflow should minimize membrane fouling and virion deposition onto
the membrane. Yet, significant virion deposition on membranes in crossflow
ultrafiltration system is still observed 16-19. Similar to VIRADEL, elution of virions that
adsorb to the UF membrane may be accomplished 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, 𝜑X174 20, 23, P22 19) and human
pathogenic viruses (e.g. poliovirus 16, 24, echovirus 18, and norovirus 25).

Crossflow filtration has three major advantages over VIRADEL: (1) lower cost of UF
membranes compared to VIRADEL filters, (2) simultaneous concentration of multiple
types of pathogens, and (3) removal of small compounds that may inhibit qPCR
analysis. However, a fraction of the virions may still adsorb on the membrane to

133

decrease recoveries 26. To enhance recovery, UF membranes are typically coated with
proteinaceous solutions (e.g. calf serum) 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 will limit the application of calf serum
blocked (CS-blocked) membranes for rapid or field sampling 27. In a previous study, to
improve virion recovery during crossflow 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 blocking, the PEM-coated membranes showed higher recovery of bacteriophage
P22 from both DI water and a membrane bioreactor effluent

19.

Despite the need for improved HAdV concentration and recovery, crossflow UF has not
been comprehensively evaluated for this application. Pei et al. employed crossflow UF
to recover viruses including HAdV 2 from tap water, but they coupled this process with
VIRADEL. These authors reported 42.4% recovery of HAdV 2 by VIRADEL only,
whereas they did not determine the recovery of HAdV 2 in crossflow UF

22.

Skraber et

al. determined the recovery of MS2, 𝜑X174 and poliovirus in crossflow 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 pre-filtered virus stock after virus harvest 28. Rhodes et
al. examined recoveries of multiple viruses including enteroviruses, 𝜑X 174 and HAdV
41 from tap and river water by crossflow ultrafiltration; post-elution recoveries of HAdV
41 were 69.4 Âą 12.4% from tap water and 55.8 Âą 7.9% from river water while pre-elution

134

recoveries were not reported 29. The authors 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 of
adenoviral serotypes are available for HAdV 2 (PZC = 3.5 to 4.0

30)

and HAdV 5 (PZC ~

4.5 31) but not for the enteric HAdV 40 and HAdV 41. Moreover, no studies quantify 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 is to improve HAdV recovery from water. We build on our
earlier experimental and modeling study 19 to pursue two specific objectives: 1)
elucidate specific physicochemical mechanisms that control HAdV 40 adhesion to a
membrane surface and 2) evaluate crossflow UF for concentration and recovery of
HAdV 40 from several water matrices. We employ two types of membranes to examine
the effects of membrane properties on virion recovery. The approach includes
comprehensive physicochemical characterization of HAdV 40, evaluation of virion-virion
and virion-membrane interactions using the XDLVO, and crossflow filtration tests with
feed waters spiked with HAdV 40. The premise of the study is that understanding virion
interactions can guide process design for improving pre-elution and post-elution
recoveries.

135

4.2 Materials and Methods

4.2.1 Reagents and water samples

All reagents were of ACS analytical grade or higher purity. Heparin (sodium salt),
chitosan (medium molecular weight), calf serum, ethylenediaminetetraacetic acid
(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, and glycerol was purchased from J.T. Baker (Center Valley, PA). All
solutions were prepared using deionized water (DI) water with a resistivity of 18.2
MΩcm (Barnstead NANOpure System, Dubuque, IA).

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 Îźm
membranes (mixed cellulose esters, Millipore, Billerica, MA) immediately after
collection, stored at 4 oC and used within one 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 (OI Analytical model 1010 analyzer, OI
Analytical, College Station, TX). Concentrations of cations (Na+, K+, Mg2+, and Ca2+)

136

were determined by inductively coupled plasma mass spectrometry (iCAP Q quadrupole
ICP-MS, Thermo Scientific, Waltham, MA).

4.2.2 Human adenovirus 40: Propagation, purification and characterization

HAdV 40 (ATCC® VR-931™, Manassas, VA) was purchased from ATCC and
propagated in A549 cells. The concentration of propagated virus, as measured by
quantitative real-time polymerase chain reaction (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. The Supplemental
Material (APPENDIX) file gives 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 characterization tests. The
hydrodynamic diameter of the virion was determined by dynamic light scattering (DLS,
90 Plus, Brookhaven Instruments Corp., Holtsville, NY), and the 𝜁-potential of the virion
was calculated from phase analysis light scattering (ZetaPALS, Brookhaven
Instruments Corp., Holtsville, NY). To study the effect of pH on the aggregation state
and 𝜁-potential of HAdV 40 virion, 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

137

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). Sample was rested on the grid for 1 min and washed away by drops of water. 1%
uranyl formate was then used to stain the grid and excess stain was removed by filter
paper. Grid was dried prior to TEM imaging. Images were recorded using a JEM2200FS (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 protocol described previously
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. Contact
angles of three probe liquids (DI water, glycerol and diiodomethane) on the virion lawn
were then measured with a goniometer (model 250, ramĂŠ-hart, Succasunna, NJ) using
the sessile drop method to determine surface energy components of HAdV 40.

138

4.2.3 XDLVO energy of virion-virion and virion-membrane interactions

XDLVO energy profiles were calculated (see APPENDIX, section S6, for details) using
the surface properties (charge and energy) of HAdV 40 and membranes. Membranes
were characterized in our previous study 19.

4.2.4. Membrane preparation

Prior to modification, polyethersulfone membranes (Omega, Pall Corp., East Hills, NY)
with a molecular weight cutoff of 30 kDa were soaked in 0.1 M NaOH for 3 h and then in
DI water for 24 h at 4 oC with water exchange after 12 h, as recommended by the
manufacturer. To coat the membrane with a PEM, 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 oC 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. Alternating deposition of HE and CHI continued until 4.5
bilayers were obtained with HE as the outmost (terminal) layer. To block a membrane
with calf serum, the membrane was placed in a crossflow filtration cell, and 500 mL of
5% (v/v) calf serum was circulated over the membrane overnight at 1.0 L/min at room
temperature with no pressure applied across membrane. After blocking, the membrane
was rinsed with DI water twice at the same rate for 10 min. The PEM-coated and CS-

139

blocked membranes were characterized in terms of charge and surface energy as
described earlier 19.

4.2.5. Virus concentration and recovery tests

Virions were recovered from DI water, tap water and surface water. Prior to recovery
experiments, 1 L 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 hour of recirculation (approximate
time needed for filtration) and compared with that in the feed initially. No difference in
virus concentration was observed, indicating that no loss of virion to the filtration
apparatus. The APPENDIX shows the schematic of the crossflow filtration apparatus
(Figure 17). The membrane was placed in the crossflow cell (CF042P, Sterlitech, Kent,
WA) and compacted for 90 min at 40 psi (~ 276 kPa). A 1 L water sample was spiked
with 1 mL of HAdV 40 stock (resulting in the virion concentration of ~106 to 107 GC/mL
in the diluted solution) and 5 mL was collected to quantify virions in feed sample. 996
mL remaining sample was pressurized using compressed N2 gas and delivered to the
membrane cell using a peristaltic pump (model 621 CC, Watson Marlow, Wilmington,
MA) at 2.0 L/min. A digital flow meter (101-7, McMillan, Georgetown, TX) monitored the
crossflow rate. The transmembrane pressure was maintained at ~ 20 psi (~ 138 kPa)
and continually monitored by a pressure transducer (0.5-5.5 V output, Cole Parmer,
Vernon Hills, IL). The ratio of the crossflow flux to the initial permeate flux, 𝐽𝑐𝑓 /𝐽𝑝 ,

140

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 flow meter and digital balance were transmitted to the computer via a data
acquisition module (PCI-6023E/SC-2345, National Instruments, Austin, TX). Filtration
was stopped when 900 mL were filtered. The remaining 100 mL of the retentate were
analyzed for virus content so that pre-elution recovery could be calculated.

After filtration, virions were eluted off the membrane using 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 was performed at a crossflow rate of 2.0 L/min with no
transmembrane pressure applied. 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.

4.2.6 Virus quantification and recovery

DNA of HAdV 40 in water samples was extracted with a MagNA Pure Compact System
(Roche Diagnostics USA, Indianapolis, IN) immediately after sample collection. qPCR
was used to determine HAdV 40 concentration (see APPENDIX, section A.2, for a
description of qPCR procedures). Possible effects of organic compounds in the feed

141

water on virion quantification by qPCR were evaluated by performing inhibition tests.
Virion pre-elution recovery (𝑟𝑝𝑟𝑒 ), post-elution recovery (𝑟𝑝𝑜𝑠𝑡 ) and log removal (𝐿𝑅𝑉)
were calculated using the following equations:
𝐶 𝑉

(4-1)

𝑟𝑝𝑟𝑒 = 𝐶 𝑟𝑉𝑟 ,
𝑓 𝑓

𝑟𝑝𝑜𝑠𝑡 =

𝐶𝑟 𝑉𝑟 +𝐶𝑒 𝑉𝑒
𝐶𝑓 𝑉𝑓

𝐶𝑓 +𝐶𝑟

𝐿𝑅𝑉 = 𝑙𝑜𝑔10 (

(4-2)

,

2𝐶𝑝

(4-3)

).

where 𝐶𝑓 , 𝐶𝑟 , 𝐶𝑒 , and 𝐶𝑝 represent virion concentrations in feed, retentate, eluate and
permeate samples, respectively, and 𝑉𝑓 , 𝑉𝑟 , 𝑉𝑒 , 𝑉𝑝 are the volumes of these samples. In
calculating the 𝐿𝑅𝑉 (equation (4-3)), the virus concentration in feed was calculated by
averaging the initial concentration of HAdV 40 in feed prior to filtration and the
1

̅̅̅𝑓 = (𝐶𝑓 + 𝐶𝑟 ). Thus, the
concentration of the virus in retentate after filtration: 𝐶
2

reported virus removal (𝐿𝑅𝑉 ) is the value averaged over the entire filtration process.

4.3 Results and Discussion

4.3.1 Characteristics of source waters

Table 10 provides water quality parameters for the DI water, tap water, and Lake
Lansing water (sampled in spring and fall). The biggest differences among the waters
are 1) much lower values of TOC, conductivity and cation concentrations in DI water
and 2) high TOC and Ca2+ concentrations in the Lake Lansing water samples.

142

Table 10. Water quality characteristics.

Cation concentration (mg/L)

pH

Conductivity

TOC

range

(ÎźS/cm)

(mg/L)

Na+

K+

Mg2+

Ca2+

Deionized water

5.7 to 6.0

n/a#

n/a#

0.18 Âą 0.00

0.02 Âą 0.00

0.04 Âą 0.00

0.06 Âą 0.00

Tap water

7.5 to 8.0

319.5 Âą 27.6

1.1 Âą 0.1 22.18 Âą 2.33 9.32 Âą 0.32

1.86 Âą 0.09 24.37 Âą 1.83

Lake water (fall)

7.0 to 7.5

336.3 Âą 12.7

7.5 Âą 0.2 16.42 Âą 2.53 10.91 Âą 0.11

1.47 Âą 0.03 30.33 Âą 0.54

Lake water (spring) 7.5 to 7.8

369.3 Âą 32.0

9.4 Âą 0.1 16.46 Âą 2.60 11.54 Âą 0.20

1.50 Âą 0.03 41.09 Âą 0.72

Water source

# below

detection limit.

143

4.3.2 Physicochemical properties of HAdV 40: Size, 𝜻-potential and surface
charge density

Coliphages commonly serve as surrogates for human viruses

33.

virus removal and adsorption show that coliphages (e.g. MS2

34, 35,

However, studies on
𝜑X174 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
37,

36,

fastidiousness in cell culture 38-40 (though cultivation improvement was reported with

modified host cell line 41, 42), and difficulty of isolation 40, 43) make them different from
other HAdV serotypes.

Figure 7. Transmission electron microscopy image of HAdV 40.

This uniqueness makes detailed characterization of HAdV 40 and 41 necessary.
Aggregation state and 𝜁-potential as functions of pH were reported for HAdV serotype 2
30,

but not for HAdV 40. In this study, we measured HAdV 40 size by dynamic light
144

scattering and compared these data with the results of TEM imaging. We also
determined the 𝜁-potential and surface charge density of the virion as functions of pH.

Based on TEM images (e.g. Figure 7), the diameter of HAdV 40 is ~ 80 nm. Figure 8
shows the number-based hydrodynamic diameter distribution of HAdV 40 as a function
of pH, and Table 11 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 average
ℎ
hydrodynamic diameters of HAdV 40, 𝑑̅𝐻𝐴𝑑𝑉
, ranged from 94 Âą 3 nm to 102 Âą 9 nm,

which agrees with typical size values reported earlier for HAdV of different serotypes 4446.

The small increase in the hydrodynamic diameter relative to the diameter

determined by TEM could stem from fibers on the HAdV 40 surface 30, 47 that we do not
take into account when estimating HAdV 40 diameters from TEM images. Such fibers
should, however, decrease particle diffusivity in light-scattering experiments.
Additionally, shrinkage of HAdV 40 during negative staining may decrease the
diameters observed in TEM images 48. The polydispersity index of the particle
diameters in the HAdV 40 suspensions at pH 5.8, 6.7, and 7.6 was 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 up to 1 Âľm in
size appeared at pH 4.0 and 4.3 in several measurements. In contrast to aggregation
between pH 4.0 and 4.7, at pH 2.8 the virion hydrodynamic diameter is only 112 Âą 20
nm, suggesting minimal aggregation. This is consistent with studies showing that the
enteric adenovirus remains infective in the acidic conditions in the gut 49, 50. A high
positive 𝜁-potential at low pH (Figure 9) probably minimizes aggregation. At pH 9.7 the

145

average hydrodynamic diameter of HAdV 40 (80 Âą 5 nm) was significantly lower than
values measured at pH values from 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 stem from virus degradation. Size reduction at pH 10 also
occurred with HAdV 2 30.

146

Figure 8. Normalized number-based size distribution of HAdV 40 as a function of pH. Vertical
red dashed lines indicate the average modal diameter (99 nm, see APPENDIX, Table 15) in the
buffer recommended for the storage of purified HAdV 40 (10 mM Tris-HCl and 1 mM EDTA, pH
7.6). Vertical blue dash-dot lines denote the average HAdV diameter (~ 80 nm) as determined
by TEM. The APPENDIX shows the HAdV 40 size distribution in tap water (Figure 20).

147

Table 11. Parameters of HAdV 40 size distributions at different pH values a. Shaded (gray) areas denote pH values where significant
aggregation occurs.
Water matrix

1 mM NaCl

buffer b

tap water

1 mM NaCl

pH

2.8

4.0

4.3

4.7

5.8

6.7

7.6

7.5 to 8.0

9.7

̅ 𝒉𝑯𝑨𝒅𝑽 (nm)
𝒅

112 Âą 20

284 Âą 99

238 Âą 69

149 Âą 3

94 Âą 3

95 Âą 3

103 Âą 9

109 Âą 14

80 Âą 5

Polydispersity

0.22 Âą 0.06

0.31 Âą 0.04

0.31 Âą 0.01

0.10 Âą 0.01

0.06 Âą 0.01

0.07 Âą 0.02

0.07 Âą 0.01

0.11 Âą 0.01

0.12 Âą 0.03

Note:
a. Additional parameters - modal diameter and half width and half maximum – are given in the APPENDIX (Table 15)
b. 10 mM Tris-HCl and 1 mM EDTA (buffer recommended for the storage of purified HAdV 40).

148

Figure 9 present the virion 𝜁-potentials determined from electrophoretic mobilities (see
APPENDIX, Figure 23) using an approximate expression derived by Ohshima

53;

the

expression provides an accurate (<1% error) estimate for Îś-potentials for arbitrary
values of 𝜅𝑎, where 𝑎 is the radius of the virion and 𝜅 is the Debye-Hückel parameter
(see APPENDIX, section A.9). Due to the large virion diameters, Smoluchowski’s
expression for electrophoretic mobility, applicable for 𝜅𝑎 ≫ 1, is not accurate. Thus, we
used Ohshima’s expression.

Figure 9. 𝜁-potential of HAdV 40 as a function of pH.

The 𝜁-potential increased 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 Âą 0.5% of
the total population) appeared, possibly as an artifact of the dialysis process 32.
149

Relatively small absolute values of the 𝜁-potential between pH 4.0 and 4.7 probably
allow for the extensive aggregation described above.

As Figure 9 shows, the PZC of HAdV 40 particles is between pH 4.0 and 4.3. Favier et
al. predicted the theoretical PZC of each major capsid protein of HAdV 40 using their
primary sequences and showed that the PZC values of 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 of an intact virion is lower than the PZC values of its
structural protein components 54. PZC values were reported to be in the 3.5 to 4.0
range for HAdV 2 30 and 4.5 for HAdV 5 31, consistent with the low PZC of HAdV 40.

At pH values where virions aggregate, the measured 𝜁–potential is that of an aggregate
of virions, and not a single virion. In this case, we calculated virion 𝜁–potentials using an
expression developed by Makino and Ohshima

55;

the expression connects the surface

charge density of a particle with its 𝜁–potential, diameter, and Debye length (see
APPENDIX, section A.10). The calculation assumed 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 𝜁-potential of HAdV 40 in tap water (pH 7.5 to 8.0) was -17.7 mV, which is
much smaller than the 𝜁-potential (-30.5 mV) of HAdV 40 in a buffered solution with a

150

similar pH (pH 7.6). The difference could stem from specific adsorption of divalent or
multivalent ions onto virions in tap water.

4.3.3 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 12 shows,
the apolar surface energy component (𝛾 𝐿𝑊 ) of HAdV 40 was 41.6 mJ/m2, a value typical
for biological materials 56. The electron donor (𝛾 − ) and electron acceptor (𝛾 + )
components of surface energy were 14.7 mJ/m2 and 0.01 mJ/m2, respectively, making
the polar component of HAdV 40 surface energy, 𝛾 𝐴𝐵 = 2√𝛾 + 𝛾 − , equal to 0.84 mJ/m2.
Very small values of 𝛾 + were also reported for tobacco mosaic virus 56, bacteriophage
P22 19 and multiple proteins 56, suggesting that this is a common characteristic of
proteins and non-enveloped viruses.

Based on values of 𝛾 𝐿𝑊 , 𝛾 + and 𝛾 − for water and HAdV 40, the calculated polar
adhesion energy between HAdV 40 and water is 79.7 mJ/m 2 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.

151

Table 12. Contact angles, calculated surface energy parameters and the enerinterfacial
interaction when immersed in water for HAdV 40.
Contact angle

Surface energy parameters

Free energy of

with probe liquid (0)

(mJ/m2)

interfacial virion-virion

𝑳𝑾

H2O

Glycerol

DIM#

𝜸

68 Âą 2

64 Âą 1

36 Âą 2

41.6

+

−

𝜸

𝜸

0.01

14.7

interaction in water,
𝑨𝑩

𝑻𝑶𝑻

𝜸

𝜸

0.8

42.4

∆𝐺𝑣𝑤𝑣 (mJ/m2)
-30.4

4.3.4. XDLVO energy profile of virion-virion interfacial interaction in aqueous
media

Several studies suggest that the classic Derjaguin-Landau-Verwey-Overbeek theory
does not accurately predict the interaction between colloids and various surfaces in
aqueous 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 model to evaluate virion-virion and virion-membrane interfacial
interactions. Inputs to the model include the surface characteristics of HAdV 40 (see
sections 3.2 and 3.3) and membranes 19.

Figure 10 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 zeta potential values used in XDLVO
calculations were for virions in the recommended buffer (1 mM EDTA + 10mM Tris). For
pH values > 5.8, the height of the energy barrier increases with pH with the sole

152

exception of pH of 7.6. The deviation from the trend is due to the different background
solution used (10 mM Tris and 1 mM EDTA instead of 1 mM NaCl; see section 3.2).
Negligible aggregation of virions for pH ≥5.8 is due to the high energy barrier (from 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 𝜁-potential (29.4 mV) at pH 2.8, the energy
barrier is 20.7 kT and the virion suspension is stable.

Figure 10. XDLVO energy profile of virion-virion interfacial interactions in 1 mM NaCl a for
HAdV 40 at different pH values. (For pH 7.6, size and zeta potential values used in XDLVO
calculations were for virions in the recommended buffer (1 mM EDTA + 10mM Tris), and not in 1
mM NaCl.)

153

4.3.5 XDLVO energy of virion-membrane interfacial interaction in membrane feed
solutions: DI water and tap water.

XDLVO calculation of 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 only be measured in 1 mM NaCl electrolyte,
so this value was used to approximate the membrane charge in the HAdV-spiked DI
water. The approximation is reasonable because the electrophoretic mobility of virions
measured in DI water (-1.65 ± 0.19 µm∙S-1∙V-1∙cm; pH 5.8) and in 1 mM NaCl solution (1.72 ± 0.48 µm∙S-1∙V-1∙cm; pH 5.8) were not statistically different.

At separation distances > 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 APPENDIX, Figures 21 and 22). However, as the separation
distance increases, the magnitude of each interaction decreases. The magnitude of van
der Waals interaction energy, for example, decreases to less than 0.5 kT as the
separation distance increases to approximately 15 nm. Over a shorter range (from the
minimum equilibrium cut-off 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), the PEM-coated membrane
is negatively charged (𝜁 = -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

154

surface, the secondary minimum in the membrane-virion interaction energy profile is
shallow (-4.2 kT) (Figure 11A, solid line). The same membrane carries a positive
charge (𝜁 = 5.6 ± 0.4 mV) in tap water. The charge reversal could 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 in tap water, electrostatic

attraction coupled with van der Waals attraction at large separation distances and a
predominant repulsive acid-base interaction at small separation distances yields a -10.3
kT secondary minimum at the separation distance of 3.2 nm (Figure 11A, dashed line).
This secondary minimum may lead to reversible adsorption of HAdV 40 onto the
membrane surface during tap water filtration.

155

A

B

Figure 11. XDLVO energy of interfacial interaction of HAdV 40 in aqueous media with a PEM-coated membrane (A) and a CSblocked membrane (B). The energies were calculated from surface parameters determined at pH 5.8.

156

Even though the PZC of bovine serum albumin, a major component of calf serum, is in
the 4.7 to 5.6 range 61-63, CS-blocked membranes in 1 mM NaCl solution at pH=5.8
(unadjusted) carry a weak positive charge (𝜁 =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 from 7.5 to 8.3 64). The XDLVO energy profile for the interaction
of an HAdV 40 virion and a CS-blocked membrane shows a -48.2 kT secondary
minimum at a separation distance of 2.4 nm (Figure 11B, solid line), which suggests
deposition of a significant amount of virion.

In tap water, the 𝜁-potential of CS-blocked membranes was 2.8 ± 1.0 mV, similar to that
in 1 mM NaCl. Compared with experiments in 1 mM NaCl, the reduced negative charge
of HAdV 40 in tap water (see Figure 9) leads to lower 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 (Figure 11B, dashed line).

4.3.6 Virus recovery from DI water

Pre-elution virion recovery (see eq. (1)) after crossflow UF of DI water spiked with
virions was significantly greater with a PEM-coated membrane (74.8 Âą 9.7%) than with a
CS- blocked membrane (54.1 Âą 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

157

(Figure 11A, solid line). This adsorption may stem from microscale attraction 56, 65,
whereas XDLVO theory describes only macroscopic interactions. For example, in
experiments on B. cepacia adhesion Hwang et al. reported bacterial adhesion in the
absence of a secondary minimum in the XDLVO energy profile and suggested 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 (~ 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 APPENDIX, Table 14) are smaller
than the length 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 APPENDIX,
Table 14) still smaller than 30 nm HAdV fibers. Other possible explanations for the
incomplete pre-elution 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).

158

A

B

Figure 12. Recovery of HAdV 40 with CS-blocked and PEM-coated membranes from HAdV-spiked deionized water (A) and tap water
(B).

159

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
(Figure 11B, 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 pre-elution
recovery after crossflow UF is higher with PEM-coated membranes than with CSblocked membranes is consistent with our previous findings for crossflow filtration of
bacteriophage P22 in DI water, where PEM-coated membranes gave significantly
higher pre-elution recovery (80%) than CS-blocked membranes (30%) 19.

After filtration, we eluted virions to increase recovery. For both PEM-coated and CSblocked membranes, eluents containing 0.01% NaPP and 0.01% Tween 80 give high
post-elution recoveries (see eq. (2)) of 99.5 Âą 6.6% for PEM-coated membranes and
98.8 Âą 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 virion. Thus, both NaPP and Tween 80 could help
release the virion from the membrane.

The relatively high feed concentration of adenovirus (~106 to 107 GC/mL) in our study
may lead to a higher recovery and a smaller variance than 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

160

low concentration of virions in the permeate (and further quantify both virion removal
and the fraction of the virion adsorbed on the membrane). The HAdV 40 removal was
high with both CS-blocked membrane (𝐿𝑅𝑉 up to 2.17) and PEM-coated membrane
(𝐿𝑅𝑉 up to 4.64), so the permeate concentration was much lower than the feed
concentration.

4.3.7 Virus recovery from tap and surface water

We also evaluated the HAdV 40 recovery 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 APPENDIX,
section A.5 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 pre-elution virion
recovery after crossflow UF with PEM-coated membranes (40.5 Âą 9.9%) and with CSblocked membranes (38.3 Âą 9.3%). Based on XDLVO predictions for CS-blocked
membranes (Figure 11B), the pre-elution recovery from tap water should be higher than
from DI water. However, the opposite occurred – for CS-blocked membranes, the
average pre-elution recovery from tap water was ~ 16% lower than from DI water

161

(Figure 12). The discrepancy suggests that non-XDLVO 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 = 1.1Âą 0.1 mg/L) and on the membrane surface. Permeate flux declined ~
15% to 20% during crossflow UF of tap water (see APPENDIX, Figure 18) suggesting
the formation of a fouling layer on the membrane surface.

The pre-elution recovery from tap water after crossflow UF with the PEM-coated
membrane was ~ 34% lower than from DI water (Figure 12). In addition to the possible
effects described above for the CS-blocked membranes, charge reversal of the PEM
membrane in tap water may decrease the pre-elution recovery. Pre-elution recovery
from tap water has a larger variance, 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.

162

A

B

Figure 13. Recovery of HAdV 40 from surface water after crossflow UF with CS-blocked membranes (A) and PEM-coated
membranes (B).

163

Surface water is a complex matrix containing naturally occurring organic and inorganic
species, both dissolved and colloidal. Pre-elution recoveries from surface water
collected in the fall were ~ 40% with both PEM- and CS-coated membranes, which is
similar to recoveries from tap water, but significantly lower than recoveries from DI
water. The TOC in surface water collected in the fall was 7.5 Âą 0.2 mg/L. For surface
water, permeate flux declined more than 50% over 90 min of crossflow UF (see
APPENDIX, Figure 18) with membranes of each type. Foulants formed a cake layer
and masked the anti-adhesive 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, silica particles) and co-deposited
on the membrane. Post-elution recovery was higher for tap water (Figure 12B) than for
surface water suggesting that fouling also decreased the effectiveness of the elution
process.

Many studies showed that divalent cations such as Ca2+ (30.3 Âą 0.5 mg/L in fall surface
water) 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. Calcium bridging could also occur between carboxyl groups on
natural organic matter and on the HAdV 40 viral 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 in UF from

164

complex waters. This is consistent with the result that pre-elution recovery from surface
water was not statistically different for the two membrane types (Figure 13).

We also analyzed virion recovery from the surface water collected from the same lake in
early spring after ice melted. Flux decline also occurred for both membranes during
filtration of spring surface water (see APPENDIX, Figure 18), and pre-elution recoveries
with both membranes were ~20%. Lower recovery from spring surface water compared
to that from fall surface water could stem from higher concentrations of Ca2+ (41.1 Âą 0.7
mg/L) and TOC (9.4 Âą 0.1 mg/L), see Table 10.

Post-elution recovery with CS-blocked membranes was 61.0 Âą 2.8% and 34.9 Âą 10.1%
for water samples collected in fall and spring, respectively. For PEM-coated
membranes, post-elution yielded 62.4 Âą 2.2% and 41.6 Âą 2.0% recoveries for fall and
spring samples. Thus, virions adsorbed from surface water on both membranes were
not eluted as effectively (Figure 13) as those adsorbed from DI or tap water (Figure 12)
when using an eluent with NaPP and Tween 80 only.

To increase the elution efficiency, we added 0.01 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 post-elution recovery from
spring and fall surface water, averaged over all PEM-coated and CS-blocked
membranes, increased to 84.3 Âą 4.5%. The increased elution efficiency is consistent
with the hypothesis that calcium binding decreases virion recovery from surface water.

165

In a previous study we showed that a PEM consisting of a weak polycation and a strong
polyanion disassembles as low pH 79. Because HAdV 40 tolerates acidic conditions
such a PEM could enable a simple additional processing step to improve the overall
recovery: disassembling the PEM after elution could release trapped virions to
potentially achieve 100% recovery.

4.3.8 Virus removal

As Figure 14 shows, in tests with HAdV 40-spiked waters of all types, virion removal
(see eq. (3)) by PEM-coated membranes was typically an order of magnitude higher
than removal by CS-blocked membranes. However, the loss of 1% of virions to the
permeate by CS-blocked membranes should not significantly decrease recoveries.

Figure 14. Removal of HAdV 40 by crossflow ultrafiltration with calf-serum blocked membranes
and membranes coated by a polyelectrolyte multilayer.

166

The low 𝐿𝑅𝑉 by CS-blocked membranes might result from an uneven calf serum
distribution on the membrane surface and incomplete blockage. The blocking
procedure involves recirculating 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 simply correlate with the
extent of fouling alone. Pore blockage by humic acid could enhance the 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 and DI water.

4.4 Conclusions

Physicochemical characterization of HAdV 40 showed that this relatively large virus
(hydrodynamic diameter of 102.5 Âą 8.6 at pH 7.6 in 10 mM Tris and 1 mM EDTA) is
hydrophobic (∆𝐺𝑣𝑤𝑣 = -30.4 mJ/m2) with a point of zero charge between pH 4.0 and pH
4.3. Extended Derjaguin-Landau-Verwey-Overbeek theory predicted that the
hydrophilic and negatively charged (𝜁 = -7.0 ± 3.0 mV at pH 5.8 in 1 mM NaCl) PEMcoated membranes should be advantageous for recovering HAdV from DI water in

167

comparison with positively charged CS-coated membranes (𝜁 = 3.0 ± 2.0 mV at pH 5.8
in 1 mM NaCl).

We tested the validity of the XDLVO prediction in filtration experiments. Pre-elution
recovery of HAdV from DI water (ionic strength of ~0.2 mM when spiked with HAdV
𝑃𝐸𝑀
stock) after crossflow UF was higher with PEM-coated membranes (𝑟𝑝𝑟𝑒
= 74.8 Âą 9.7%)
𝐶𝑆
than with CS-blocked membranes (𝑟𝑝𝑟𝑒
= 54.1 Âą 6.2%). Although pre-elution recovery

was lower from tap water, for both PEM- and CS-coated membranes virion elution using
an aqueous solution of sodium polyphosphate and TWEEN 80 was effective for both DI
𝐶𝑆
𝑃𝐸𝑀
𝑃𝐸𝑀
water (𝑟𝑝𝑜𝑠𝑡
= 99.5 ± 6.6%; 𝑟𝑝𝑜𝑠𝑡
= 98.8 ± 7.7%) and tap water (𝑟𝑝𝑜𝑠𝑡
= 88.8 Âą 15.3% and
𝐶𝑆
𝑟𝑝𝑜𝑠𝑡
= 93.7 Âą 6.9%). The near 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. Pre𝐶𝑆
𝑃𝐸𝑀
and post-elution recoveries from surface waters were significantly lower (𝑟𝑝𝑟𝑒
and 𝑟𝑝𝑟𝑒
𝐶𝑆
𝑃𝐸𝑀
as low as ~ 21%; 𝑟𝑝𝑜𝑠𝑡
and 𝑟𝑝𝑜𝑠𝑡
< 65%) and showed no statistically significant

difference between the two membrane types. However, addition of EDTA to the eluent
𝐶𝑆
𝑃𝐸𝑀
greatly increased the elution efficacy (𝑟𝑝𝑜𝑠𝑡
= 88.6 ± 4.3%; 𝑟𝑝𝑜𝑠𝑡
= 87.0 Âą 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. In tap water, post-elution virion recovery is nearly 100% for both PEM- and
CS-coated membranes despite the higher removal of virions with PEM-coated
168

membranes (𝐿𝑅𝑉 𝑃𝐸𝑀 = 2.82 ± 0.32 and 𝐿𝑅𝑉 𝐶𝑆 = 1.96 ± 0.53 in tap water). For more
complex water matrices such as surface water, the composition of the eluent is the most
important 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), and not on virion-membrane
interactions. An eluent that includes a polyanion (sodium polyphosphate), a non-ionic
surfactant (Tween 80) and a chelating agent (EDTA) recovers HAdV effectively (~ 88%)
even from high TOC (9.4 Âą 0.1 mg/L) surface water.

169

APPENDIX

170

APPENDIX

A.1 HAdV 40 propagation procedures

The HAdV 40 propagation procedure was adapted from protocols by Wong et al.

30

and

Mautner 32. A549 cells, which are susceptible to a wide range of enteric viruses, were
used as the HAdV 40 host. These cells were cultured in 50 mL of cell growth medium in
a culture flask with a surface area of 150 cm2 until 90% confluence was achieved.
Medium in the culture flasks was decanted, and cell cultures were inoculated with 2 mL
of virus stock. The flask was incubated at 37 oC for 60 min and rocked gently at 15 min
intervals. After 60 min, 2 mL of viral inoculums was discarded and ~25 to 30 mL of cell
maintenance medium was added to the flask. The tightly sealed flask was incubated at
37 oC and the cytopathic effect (CPE) was checked daily. As HAdV 40 is a fastidious
virus, 4 days are usually needed to observe a good CPE. Viruses should be harvested
when cells round up and detach from the flask surface. Harvesting too early leads to low
virus production, and harvesting too late may lead to virus release to medium. To
harvest viruses, the remaining cells were gently pipetted off the flask surface. Cell
suspension was centrifuged at 900 g for 10 min at 4 oC, and supernant was then
discarded. Cell pellets were resuspended in maintenance medium and then subjected
to a freeze/thaw procedure (-70 oC / 37 oC) to release virions from cells. After 3
freeze/thaw cycles, cell resuspension was centrifuged at 12,000 g for 10 min at 4 oC.
Supernatant containing viruses was collected, aliquoted, and stored at -70 oC for further
use.

171

Viruses were further purified using a CsCl density step gradient (density steps from the
bottom up: 1.45 g/cm3 CsCl, 1.38 g/cm3 CsCl, 1.32 g/cm3 CsCl, 40% glycerol cushion).
HAdV 40 stock was then added to the top and a virus band was obtained by
ultracentrifugation at 80,000 g (AH-650 rotor, Sorvall, Thermo Scientific, Waltham, MA)
and 4 oC for 3 h. The band was collected by side puncture of the tube with a syringe.
CsCl was removed by dialysis against the desired buffer at 4 oC with gentle stirring.

Figure 15. HAdV 40 band obtained by CsCl density gradient centrifugation.

172

Figure 16. Transmission electron microscopy image of HAdV 40 viruses.

A.2 HAdV 40 quantification by qPCR

A MagNA Pure Compact System (Roche Diagnostics USA, Indianapolis, IN) was used
to extract HAdV 40 DNA from 390 ÎźL samples collected immediately after filtration
experiments. 10 ÎźL of carrier RNA (1 mg/mL) was added to increase DNA recovery
during DNA extraction. Extracted DNA was stored in 100 ÎźL eluates and at -20 oC prior
to qPCR analysis. Each extract was quantified by qPCR in triplicate, and qPCR was
performed using the LightCycler 1.5 system (Roche Diagnostics USA, Indianapolis, IN)
with the primer and probe sequences described in Table 13. The volume of each PCR
reaction mixture is 20 ÎźL containing 5 ÎźL of extracted DNA sample, 0.8 ÎźL of forward
primer (10 mM), 0.4 ÎźL of each of two reverse primers (10 mM), 0.6 ÎźL of Taqman
Probe (10 mM), 10 ÎźL of probe master mix (2X, Light Cycler 480 Probes, Roche) and
2.8 ΟL of PCR-grade water. qPCR mixtures were heated at 95°C for 15 min for DNA
denaturation prior to 45 cycles of DNA amplification (95°C for 10 s, 60°C for 30 s and
72°C for 12 s). Finally, mixtures were cooled for 30 s at 40°C.
173

To prepare a standard curve (qPCR crossing-point (CP) values versus number of HAdV
40 DNA copies), plasmid DNA carrying cloned HAdV 40 hexon gene served as
standards and was prepared following the method described earlier 9.

Table 13. Primer and probe sequences used in qPCR for HAdV 40 detection
Forward Primer (for
serotype 40 and 41)
Reverse primer 1
(for serotype 40)
Reverse primer 2
(for serotype 41)

HAdV-F4041-hex157f

ACC-CAC-GAT-GTA-ACC-ACA-GAC

HAdV-F40-hex245r

ACT-TTG-TAA-GAG-TAG-GCG-GTT-TC

HAdV-F41-hex246r

CAC-TTT-GTA-AGA-ATA-AGC-GGT-GTC

HAV-F4041hex214rprobe

6-FAM-CGA-CKG-GCA-CGA-AKC-GCA-GCG-T-BHQ1

Taqman probe

A.3 Crossflow ultrafiltration (UF) apparatus

membrane cell
peristaltic pump

permeate

feed
pressure sensor

feed
gas

retentate
retentate

flowmeter

N2

feed tank

Figure 17. Crossflow filtration apparatus.

174

A.4 Permeate flux in HAdV 40 recovery tests.

Figure 18 shows the normalized permeate flux as a function of time during virus
recovery from different water matrices. Flux values were normalized by the flux value at
t = 0.

Figure 18. Normalized permeate flux as function of filtration time for CS-blocked membranes (A)
and PEM-coated membranes (B).

175

A.5 Testing of potential inhibition of qPCR in complex water matrices.

We evaluated whether chemicals present in complex water matrices inhibit DNA
extraction and qPCR detection. One liter sample of DI water, tap water and surface
water were spiked with 1 mL of HAdV 40 stock. Log concentrations of DNA from
complex water matrices (tap water and surface water) as well as DI water were
calculated based on cross point (CP) values of DNA extracts obtained by qPCR.
Similar experiments were conducted to evaluate the potential inhibition in the retentate
produced by filtering tap water and surface water samples. One liter samples were
filtered using the protocol for virus recovery tests, but no virus was introduced into the
feed. 100 mL retentate samples were collected and spiked with 1 mL of HAdV 40 stock.
The same procedure was followed for DI water samples. Log concentration of DNA
from complex water matrices (tap water and surface water) as well as DI water were
calculated. Our results (Figure 19) show no significant difference among the DNA
concentrations determined in all of the feed and retentate matrices. This indicates there
is either no inhibition or the same inhibition level for all water types in both feed and
retentate.

176

Log concentration

10
feed

8

retentate
6
4
2
0
DI

Tap

Lake
(Fall)

Lake
(Spring)

Figure 19. Log concentration of DNA detected by qPCR in feed and retentate water matrices
spiked with the same concentrations of virions.

A.6 Calculation of the virus-virus interaction energy and virus-membrane
interaction energy using the extended Derjaguin-Landau-Verwey-Overbeek
(XDLVO) theory
The total XDLVO interfacial interaction energy, 𝑈 𝑋𝐷𝐿𝑉𝑂 , is given by:
𝑈 𝑋𝐷𝐿𝑉𝑂 = 𝑈 𝐿𝑊 + 𝑈 𝐸𝐿 + 𝑈 𝐴𝐵

(A-1)

where 𝑈 𝐿𝑊 , 𝑈 𝐸𝐿 , and 𝑈 𝐴𝐵 correspond to Lifshitz - van der Waals, electrical double
layer, and Lewis acid-base interactions, respectively. Equations for each interaction as
a function of separation distance are introduced below.

177

A.6.1 The case of virus-virus (sphere-sphere) interaction in water (𝑼𝒗𝒘𝒗 )

𝑈𝑣𝑤𝑣 is the total interfacial interaction energy between two virions (𝑣) immersed in water
𝐿𝑊
𝐸𝐿
𝐴𝐵
(𝑤). As stated by eq. (A-1), 𝑈𝑣𝑤𝑣 includes three components: 𝑈𝑣𝑤𝑣
, 𝑈𝑣𝑤𝑣
, and 𝑈𝑣𝑤𝑣
.

A.6.1.1 Lifshitz - van der Waals (LW) interaction energy, 𝑼𝑳𝑾
𝒗𝒘𝒗

𝐿𝑊
𝑈𝑣𝑤𝑣
=

−𝐴
2𝑎2
2𝑎2
ℓ(4𝑎 + ℓ)
[
+
+ ln
]
2
(ℓ + 2𝑎)2
6 ℓ(4𝑎 + ℓ) (ℓ + 2𝑎)

(A-2)

where 𝑎 is the radius of the virion, ℓ is the separation distance and 𝐴 is the Hamaker
constant:
𝐴 = 12𝜋ℓ0 2 𝛥𝐺ℓ𝐿𝑊
0

(A-3)

where ℓ0 is the minimum equilibrium cut-off distance, which is estimated as 0.158 nm 5658

and Δ𝐺ℓ𝐿𝑊
is the LW adhesion energy per unit area at ℓ0 :
0
Δ𝐺ℓ𝐿𝑊
= −2(√𝛾𝑣𝐿𝑊 − √𝛾𝑤𝐿𝑊 )2
0

(A-4)

where 𝛾𝑣𝐿𝑊 and 𝛾𝑤𝐿𝑊 are LW (apolar) components of surface energy for virus and water
respectively.

A.6.1.2 Electrical double layer (EL) interaction energy, 𝑼𝑬𝑳
𝒗𝒘𝒗

𝐸𝐿
𝑈𝑣𝑤𝑣

𝑘𝑇 2 2
= 32𝜋𝜀𝑟 𝜀0 𝑎 ( ) 𝛾 exp(−𝜅ℓ)
𝑧𝑒

(A-5)

where

178

(A-6)

2𝑁𝐴 𝑒 2 𝐼
𝜅=√
𝜀0 𝜀𝑟 𝑘𝑇

where 𝑁𝐴 is Avogadro’s constant and 𝐼 is ionic strength. The dimensionless 𝛾 is given
by:
𝑧𝑒𝜓0
𝛾 = 𝑡𝑎𝑛ℎ (
)
4𝑘𝑇

(A-7)

where 𝜓0 is the surface potetntial of the virion, which can be estimated from the 𝜁–
potential of the virion, 𝜁𝑣 :
𝑑
𝜓0 = 𝜁𝑣 (1 + ) exp(𝜅𝑑)
𝑎

(A-8)

𝑑 is the distance from the particle’s surface to the slipping plane and is usually 0.3 to 0.5
nm. We used 𝑑 of 0.5 nm. The difference in exp(𝜅𝑑) due to the choice of 𝑑 = 0.3 nm
rather than 𝑑 = 0.5 nm, is at most 7% (seen at pH 7.6).

ℎ
Eq. (A-5) is valid for 𝜅𝑎 > 5 and ℓ < 𝑎 81. In our study, 0.5𝜅𝑑̅𝐻𝐴𝑑𝑉 ≥ 5.8 in 1mM NaCl

(Table 14) making eq. (A-5) applicable.

Equation (A-8) is valid for 𝜁–potential values up to ~50 mV 57. In our study, the absolute
̅
| 𝒊𝒔 ≤ 45.1 (Table 14), making eq. (A-8) applicable.
value of |𝜁𝐻𝐴𝑑𝑉
A.6.1.3 Lewis acid-base (AB) interaction energy, 𝑼𝑨𝑩
𝒗𝒘𝒗

ℓ0 − ℓ
𝐴𝐵
𝑈𝑣𝑤𝑣
= 𝑎𝜋𝜆Δ𝐺ℓ𝐴𝐵
𝑒𝑥𝑝 (
)
0
𝜆

(A-9)

179

where 𝜆 is the decay length for water. The commonly used value of 𝜆 is 0.6 nm 58.
Δ𝐺ℓ𝐴𝐵
is the AB adhesion energy per unit area at ℓ0 :
0
(A-10)

Δ𝐺ℓ𝐴𝐵
= −4 (√𝛾𝑣+ − √𝛾𝑤+ ) (√𝛾𝑣− − √𝛾𝑤− )
0

where 𝛾𝑣+ and 𝛾𝑤+ are electron-acceptor parameters of surface energy for virion and
water respectively, 𝛾𝑣− and 𝛾𝑤− are electron-donor parameters of surface energy for virion
and water respectively.

A.6.2 The case of virion-membrane (sphere-plate) interaction (𝑼𝒗𝒘𝒎 )

𝑈𝑣𝑤𝑚 is the total interfacial interaction energy between virions (𝑣) and membrane (𝑚)
immersed in water (𝑤).

A.6.2.1 Lifshitz - van der Waals (LW) interaction energy, 𝑼𝑳𝑾
𝒗𝒘𝒎

For the sphere-plate case, Lifshitz – van der Waals (LW) interaction energy is given by
82

𝐴 𝑎
𝑎
ℓ
𝐿𝑊
𝑈𝑣𝑤𝑚
=− [ +
+ ln
]
6 ℓ ℓ + 2𝑎
ℓ + 2𝑎

(A-11)

For the virus-membrane interaction, Δ𝐺ℓ𝐿𝑊
can be calculated as:
0
𝐿𝑊
Δ𝐺ℓ𝐿𝑊
= 2 (√𝛾𝑤𝐿𝑊 − √𝛾𝑚
) (√𝛾𝑣𝐿𝑊 − √𝛾𝑤𝐿𝑊 )
0
𝐿𝑊
where 𝛾𝑚
is the LW component of surface energy for the membrane.

180

(A-12)

A.6.2.2 Electrical double layer (EL) interaction energy, 𝑼𝑬𝑳
𝒗𝒘𝒎

𝐸𝐿
𝑈𝑣𝑤𝑚

1 + 𝑒 −𝜅ℓ
2)
= 𝜋𝜀𝑟 𝜀0 𝑎 [2𝜁𝑣 𝜁𝑚 ln (
) + (𝜁𝑣2 + 𝜁𝑚
ln(1 − 𝑒 −2𝜅ℓ )]
1 − 𝑒 −𝜅ℓ

(A-13)

where 𝜁𝑚 is the 𝜁–potential of the membrane.

A.6.2.3 Lewis acid-base (AB) interaction energy, 𝑼𝑨𝑩
𝒗𝒘𝒎
ℓ0 − ℓ
𝐴𝐵
𝑈𝑣𝑤𝑚
= 2𝜋𝑎𝜆Δ𝐺ℓ𝐴𝐵
exp
(
)
0
𝜆

(A-14)

For the virus-membrane interaction, Δ𝐺ℓ𝐴𝐵
is calculated as:
0
− + 𝛾 − − 𝛾 − ) + 2 𝛾 − (√𝛾 + + √𝛾 + − √𝛾 + )
Δ𝐺ℓ𝐴𝐵
= 2√𝛾𝑤+ (√𝛾𝑚
√ 𝑣 √ 𝑤
√ 𝑤
𝑚
𝑣
𝑤
0

(A-15)

+ −
−𝛾 +)
− 2 (√𝛾𝑚
𝛾𝑣 + √𝛾𝑚
𝑣
+
−
𝛾𝑚
and 𝛾𝑚
are electron-acceptor and electron-donor parameters of surface energy for

membrane.

To calculate each component of surface energy for the membrane and the virus, we
measured contact angles of DI water, glycerol and diiodomethane on both solid
surfaces (membrane surface and virus lawn) using a goniometer. We obtained the
components of surface energy for the solids by substituting contact angles and surface
energy components of probe liquids into eq. (A-16):
(1 + cos 𝜃)𝛾𝑙𝑇𝑂𝑇 = 2 (√𝛾𝑠𝐿𝑊 𝛾𝑙𝐿𝑊 + √𝛾𝑠+ 𝛾𝑙− + √𝛾𝑠− 𝛾𝑙+ )

181

(A-16)

𝛾 𝐴𝐵 = 2√𝛾 + 𝛾 −

(A-17)

𝛾 𝑇𝑂𝑇 = 𝛾 𝐿𝑊 + 𝛾 𝐴𝐵

(A-18)

where 𝜃 is the contact angle of the probe liquid on the solid, 𝛾 𝑇𝑂𝑇 is total surface energy,
𝛾 𝐿𝑊 is the LW component of surface energy,𝛾 + is the electron-acceptor parameter of
surface energy, and 𝛾 − is the electron-donor parameter of surface energy. Subscripts 𝑙
and 𝑠 refer to probe liquid and solid surface respectively.

The free energy of interfacial interaction between two virions when immersed in water
(Δ𝐺𝑣𝑤𝑣 ) can be determined using the following equation

83:

2

Δ𝐺𝑣𝑤𝑣

= −2 (√𝛾𝑣𝐿𝑊 − √𝛾𝑤𝐿𝑊 ) − 4 (√𝛾𝑣+ 𝛾𝑣− + √𝛾𝑤+ 𝛾𝑤− − √𝛾𝑣+ 𝛾𝑤− − √𝛾𝑣− 𝛾𝑤+ )

(A-19)

A.7 HAdV 40 size distribution in tap water.

Figure 20. Normalized number-based size distribution of HAdV 40 in tap water (pH = 7.5 to
8.0). The vertical dashed red line indicates the average modal diameter (98.5 nm) in the buffer
recommended for the storage of purified HAdV 40 (10 mM Tris-HCl and 1 mM EDTA, pH 7.6).
The vertical dash-dot blue line denotes the average HAdV diameter (~ 80 nm) determined from
TEM.

182

A.8 XDLVO energy of virus-membrane interfacial interaction: Relative contributions from van der Waals,
electrostatic and acid-base interactions.

Figure 21. XDLVO energies of interfacial interaction of HAdV 40 with CS-blocked membranes in (A) DI water (𝐼 ≅ 0.2 mM after
spiking with HAdV 40) and (B) tap water.

183

Figure 22. XDLVO energies of interfacial interaction of HAdV 40 with PEM-coated membranes in (A) DI water (𝐼 ≅ 0.2 mM after
spiking with HAdV 40) and (B) tap water.

184

A.9 On the determination of ζ–potential based on measured electrophoretic
mobility

Figure 23 present the results of electrophoretic mobility measurements.

Figure 23. Electrophoretic mobility of HAdV 40 as a function of pH.

To convert measured mobilities into ζ–potential values, we used an expression derived
by Ohshima 53:

𝜇=

2 𝜀𝑟 𝜀0
1
𝜁 1+
3
3 𝜂
2.5
2 [1 +
]
[
𝜅𝑎[1 + 2𝑒 −𝜅𝑎 ] ]

(A-20)

185

where 𝜅 (nm-1) is the Hückel parameter (𝜅 −1 (nm) is the Debye length), 𝜂 is the
viscosity of the electrolyte, in which particles are suspended, and 𝑎 is the radius of the
particle. Eq. (A-20) provides an accurate (<1% error) estimate of ζ–potential for any
value of 𝜅𝑎. This is in contrast to Smoluchowski’s (eq. (A-21)) and Hückel’s (eq. (A-22))
expressions for electrophoretic mobility, which are applicable for 𝜅𝑎 ≫ 1 and 𝜅𝑎 ≪ 1,
respectively.
𝜇=

𝜀𝑟 𝜀0
𝜁
𝜂

(A-21)

𝜇=

3 𝜀𝑟 𝜀0
𝜁
2 𝜂

(A-22)

A.10 On the determination of 𝜻–potential of virus aggregates.

At pH values where virions aggregate, the 𝜁–potential measured is that of an aggregate
of virions, and not a single virion. We calculated the 𝜁–potential of individual (nonaggregated) virions using eq. (A-23), which connects the surface charge density of a
particle (𝜎) with its 𝜁 –potential, diameter (𝑑𝑝 ) and Debye length (𝜅 −1) 55:
𝑧𝑒𝜁
8 𝑙𝑛 [𝑐𝑜𝑠ℎ (
)]
2𝜀𝑟 𝜀0 𝜅𝑘𝑇
𝑧𝑒𝜁
1
2
1
4𝑘𝑇
𝜎=
𝑠𝑖𝑛ℎ (
) √1 +
+
2
𝑧𝑒𝜁
𝑧𝑒𝜁
𝑑𝑝
𝑧𝑒
2𝑘𝑇
𝑠𝑖𝑛ℎ2 (
)
𝜅 2 𝑐𝑜𝑠ℎ2 (4𝑘𝑇) (𝜅 𝑑𝑝 )
2𝑘𝑇
2

(A-23)

This calculation assumes that 𝜎 is an intensive property and does not depend on the
aggregation state of the virion. To determine the 𝜁–potential of an individual virion from
the 𝜁–potential measured for virion aggregates, we first used eq. (A-23) to calculate 𝜎
based on the measured 𝜁–potential and average size of virus aggregates. With this
186

value of 𝜎, we then used eq. (A-23) again with the hydrodynamic diameter of an
individual virion to calculate the 𝜁–potential of an individual virus (Table 14). Because 𝜁–
potential measurements involved adding acid or base to adjust pH, the ionic strength
was different at different pH values. To account for these differences we estimated the
ionic strength at each pH from measured values of conductivity.

At pH values where virus aggregation occurred, the difference between 𝜁-potentials of
aggregated virions (measured 𝜁-potential) and that of an individual virion (calculated 𝜁potential) was relatively minor (Table 14).

187

Table 14. Charge characteristics of HAdV 40 at different pH values. Shaded (gray) areas denote pH values where significant
𝑎𝑔𝑔
aggregation occurs and 𝜁𝐻𝐴𝑑𝑉 ≠ 𝜁𝐻𝐴𝑑𝑉 .
Water matrix

1 mM NaCl

pH

buffer a

tap water

1 mM NaCl

DI water e

2.8

4.0

4.3

4.7

5.8

6.7

7.6

7.5 to 8.0

9.7

5.8

7.5
6.49

1.8
0.77

1.7
-0.89

1.6
-1.84

1.5
-3.02

1.6
-3.97

12.1
-9.63

5.1
-4.06

1.9
-5.88

0.2
n/af

𝒂𝒈𝒈
𝜻̅𝑯𝑨𝒅𝑽 (mV) c

n/ad

6.8

-7.8

-14.9

n/ad

n/ad

n/ad

n/ad

n/ad

n/af

𝜻̅𝑯𝑨𝒅𝑽 (mV)

29.4

7.5

-8.8

-17.8

-28.0

-34.9

-34.0

-20.3

-45.1

n/af

Debye length, 𝜿−𝟏 (nm)

3.5

7.2

7.4

7.5

7.8

7.5

2.8

3.8

6.9

23.3

Hückel parameter, 𝜿 (nm-1)

0.285

0.139

0.136

0.133

0.128

0.133

0.361

0.263

0.145

0.043

16.0

19.8

16.1

9.9

6.0

6.3

18.6

14.3

5.8

2.2

Ionic strength (mM)
𝝈𝑯𝑨𝒅𝑽 (mC/m2)

𝒉

̅𝑯𝑨𝒅𝑽
𝟎. 𝟓𝜿𝒅

b

Note:
a 10 mM Tris-HCl and 1 mM EDTA (buffer recommended for the storage of purified HAdV 40).
b Ionic strength at each pH is estimated based on electrical conductance measurements.
c Values of 𝜁 𝑎𝑔𝑔 are also reported in Figure 9. The values are included here for the ease of comparison with 𝜁
𝐻𝐴𝑑𝑉 .
𝐻𝐴𝑑𝑉
d There is no detectable aggregation of HAdV 40 at this pH.
e After spiking DI water with HAdV 40 stock
f Not measured

188

Table 15. Additional parameters of HAdV 40 size distribution at different pH values. Shaded (gray) areas denote pH values where
significant aggregation occurs.
Water
pH
matrix
HWHM

1 mM NaCl

buffer a

tap water

1 mM NaCl

2.8

4.0

4.3

4.7

5.8

6.7

7.6

7.5 to 8.0

9.7

11.8 Âą 5.7

31.3 Âą 17.3

22.0 Âą 1.4

18.8 Âą 7.5

13.3 Âą 2.4

7.6 Âą 0.7

15.0 Âą 3.1

6.9 Âą 2.0

5.8 Âą 1.8

98.5 Âą 8.7

101.3 Âą 5.0

76.4 Âą 5.7

𝒎𝒐𝒅
101.7 Âą 7.9
203.4 Âą
233.2 Âą
149.8 Âą 8.1 94.7 Âą 4.0
91.5 Âą 2.7
(nm)
Note:
81.4
89.7
(nm)
a 10
mM Tris-HCl and 1 mM EDTA (buffer recommended for the storage of purified HAdV 40).
b Half width at half maximum.
b𝒅𝑯𝑨𝒅𝑽

189

REFERENCES

190

REFERENCES

1.

http://www.who.int/water_sanitation_health/diseases/burden/en/).

2.

K. D. Crabtree, C. P. Gerba, J. B. Rose and C. N. Haas, Water Sci. Technol.,
1997, 35, 1-6.

3.

K. Mena and C. Gerba, in Rev. Environ. Contam. Toxicol., ed. D. Whitacre,
Springer New York, 2009, vol. 198, ch. 4, pp. 133-167.

4.

G. Fongaro, M. A. do Nascimento, C. Rigotto, G. Ritterbusch, A. D. da Silva, P.
A. Esteves and C. R. M. Barardi, Virol. J., 2013, 10.

5.

S. C. Jiang, Environ. Sci. Technol., 2006, 40, 7132-7140.

6.

C. E. Enriquez, C. J. Hurst and C. P. Gerba, Water Res., 1995, 29, 2548-2553.

7.

J. A. Thurston-Enriquez, C. N. Haas, J. Jacangelo, K. Riley and C. P. Gerba,
Appl. Environ. Microbiol., 2003, 69, 577-582.

8.

S. Choi and S. C. Jiang, Appl. Environ. Microbiol., 2005, 71, 7426-7433.

9.

I. Xagoraraki, D. H. W. Kuo, K. Wong, M. Wong and J. B. Rose, Appl. Environ.
Microbiol., 2007, 73, 7874-7881.

10.

N. Albinana-Gimenez, M. P. Miagostovich, B. Calqua, J. M. Huguet, L. Matia and
R. Girones, Water Res., 2009, 43, 2011-2019.

11.

T. T. Fong, M. S. Phanikumar, I. Xagoraraki and J. B. Rose, Appl. Environ.
Microbiol., 2010, 76, 715-723.

12.

O. Olayinka and O. Anthony, Virol. J., 2015, 12, 98.

13.

C. D. Gibbons, R. A. Rodriguez, L. Tallon and M. D. Sobsey, J. Appl. Microbiol.,
2010, 109, 635-641.

14.

L. A. Ikner, M. Soto-Beltran and K. R. Bright, Appl. Environ. Microbiol., 2011, 77,
3500-3506.

15.

B. R. McMinn, J. Virol. Methods, 2013, 193, 284-290.

16.

D. Berman, M. E. Rohr and R. S. Safferman, Appl. Environ. Microbiol., 1980, 40,
426-428.

17.

H. A. Morales-Morales, G. Vidal, J. Olszewski, C. M. Rock, D. Dasgupta, K. H.
Oshima and G. B. Smith, Appl. Environ. Microbiol., 2003, 69, 4098-4102.
191

18.

A. L. Polaczyk, J. Narayanan, T. L. Cromeans, D. Hahn, J. M. Roberts, J. E.
Amburgey and V. R. Hill, J. Microbiol. Methods, 2008, 73, 92-99.

19.

E. V. Pasco, H. Shi, I. Xagoraraki, S. A. Hashsham, K. N. Parent, M. L. Bruening
and V. V. Tarabara, J. Membr. Sci., 2014, 469, 140-150.

20.

V. R. Hill, A. M. Kahler, N. Jothikumar, T. B. Johnson, D. Hahn and T. L.
Cromeans, Appl. Environ. Microbiol., 2007, 73, 4218-4225.

21.

P. B. Liu, V. R. Hill, D. Hahn, T. B. Johnson, Y. Pan, N. Jothikumar and C. L.
Moe, J. Microbiol. Methods, 2012, 88, 155-161.

22.

L. Pei, M. Rieger, S. Lengger, S. Ott, C. Zawadsky, N. M. Hartmann, H. C.
Selinka, A. Tiehm, R. Niessner and M. Seidel, Environ. Sci. Technol., 2012, 46,
10073-10080.

23.

S. Skraber, C. Gantzer, K. Helmi, L. Hoffmann and H. M. Cauchie, Food Environ.
Virol., 2009, 1, 66-76.

24.

G. Belfort, Y. Rotem and E. Katzenelson, Water Res., 1975, 9, 79-85.

25.

D. S. Francy, E. A. Stelzer, A. M. Brady, C. Huitger, R. N. Bushon, H. S. Ip, M.
W. Ware, E. N. Villegas, V. Gallardo and H. D. Lindquist, Appl. Environ.
Microbiol., 2013, 79, 1342-1352.

26.

L. A. Ikner, C. P. Gerba and K. R. Bright, Food Environ. Virol., 2012, 4, 41-67.

27.

V. R. Hill, A. L. Polaczyk, D. Hahn, J. Narayanan, T. L. Cromeans, J. M. Roberts
and J. E. Amburgey, Appl. Environ. Microbiol., 2005, 71, 6878-6884.

28.

P. Nestola, D. L. Martins, C. Peixoto, S. Roederstein, T. Schleuss, P. M. Alves, J.
P. B. Mota and M. J. T. Carrondo, Plos One, 2014, 9.

29.

E. R. Rhodes, E. M. Huff, D. W. Hamilton and J. L. Jones, J. Virol. Methods,
2016, 228, 31-38.

30.

K. Wong, B. Mukherjee, A. M. Kahler, R. Zepp and M. Molina, Environ. Sci.
Technol., 2012, 46, 11145-11153.

31.

E. I. Trilisky and A. M. Lenhoff, J. Chromatogr. A, 2007, 1142, 2-12.

32.

V. Mautner, in Adenovirus Methods and Protocols, eds. W. M. Wold and A.
Tollefson, Humana Press, 2007, vol. 130, ch. 11, pp. 145-156.

33.

M. M. F. Mesquita and M. B. Emelko, in Bacteriophages, ed. I. Kurtboke, Intech,
2014.

34.

M. Abbaszadegan, B. K. Mayer, H. Ryu and N. Nwachuku, Environ. Sci.
Technol., 2007, 41, 971-977.
192

35.

M. I. Bellou, V. I. Syngouna, M. A. Tselepi, P. A. Kokldnos, S. C.
Paparrodopoulos, A. Vantarakis and C. V. Chrysikopoulos, Sci. Total Environ.,
2015, 517, 86-95.

36.

A. H. Kidd, J. Chroboczek, S. Cusack and R. W. H. Ruigrok, Virology, 1993, 192,
73-84.

37.

J. D. Song, X. L. Liu, D. L. Chen, X. H. Zou, M. Wang, J. G. Qu, Z. Z. Lu and T.
Hung, Virology, 2012, 432, 336-342.

38.

G. W. Gary, Jr., J. C. Hierholzer and R. E. Black, J. Clin. Microbiol., 1979, 10, 96103.

39.

J. C. De Jong, J. G. Kapsenberg, C. J. Muzerie, A. G. Wermenbol, A. H. Kidd, G.
Wadell, R. G. Firtzlaff and R. Wigand, J. Med. Virol., 1983, 11, 215-231.

40.

J. C. Dejong, K. Bijlsma, A. G. Wermenbol, M. W. Verweijuijterwaal, H.
Vanderavoort, D. J. Wood, A. S. Bailey and A. Osterhaus, J. Clin. Microbiol.,
1993, 31, 1562-1569.

41.

V. Sherwood, H.-G. Burgert, Y.-H. Chen, S. Sanghera, S. Katafigiotis, R. E.
Randall, I. Connerton and K. H. Mellits, J. Gen. Virol., 2007, 88, 71-76.

42.

M. Kim, M. Y. Lim and G. Ko, Appl Environ Microbiol., 2010, 76, 2509-2516.

43.

C. T. Tiemessen and A. H. Kidd, J. Gen. Virol., 1995, 76, 481-497.

44.

I. Wilhelmi, E. Roman and A. Sanchez-Fauquier, Clin. Microbiol. Infect., 2003, 9,
247-262.

45.

M. A. Kennedy and R. J. Parks, Mol. Ther., 2009, 17, 1664-1666.

46.

G. R. Nemerow, P. L. Stewart and V. S. Reddy, Curr. Opin. Virol., 2012, 2, 115121.

47.

L. Gutierrez, S. E. Mylon, B. Nash and T. H. Nguyen, Environ. Sci. Technol.,
2010, 44, 4552-4557.

48.

J. L. Putaux, A. Buleon, R. Borsali and H. Chanzy, Int. J. Biol. Macromol., 1999,
26, 145-150.

49.

A. L. Favier, W. P. Burmeister and J. Chroboczek, Virology, 2004, 322, 93-104.

50.

Z.-Z. Lu, X.-H. Zou, L.-X. Dongi, J.-G. Qu, J.-D. Song, M. Wang, L. Guo and T.
Hung, J. Gene Med., 2009, 11, 128-138.

51.

J. J. Hansen, P. S. Warden and A. B. Margolin, Int. J. Environ. Res. Public
Health, 2007, 4, 61-67.
193

52.

C. L. Bean, J. J. Hansen, A. B. Margolin, H. Balkin, G. Batzer and G. Widmer, Int.
J. Environ. Res. Public Health, 2007, 4, 53-60.

53.

H. Ohshima, J. Colloid Interface Sci., 1994, 168, 269-271.

54.

E. Norrby, in Curr. Top. Microbiol. Immunol., eds. W. Arber, W. Braun, F.
Cramer, R. Haas, W. Henle, P. H. Hofschneider, N. K. Jerne, W. Kikuth, P.
Koldowsky, H. Koprowski, O. Maaløe, R. Rott, H. G. Schweiger, M. Sela, L.
Syruček, P. K. Vogt and E. Wecker, Springer Berlin Heidelberg, 1968, vol. 43, ch.
1, pp. 1-43.

55.

K. Makino and H. Ohshima, Langmuir, 2010, 26, 18016-18019.

56.

C. J. Van Oss, Interfacial forces in aqueous media - second edition, Taylor &
Francis Group, 2006.

57.

C. J. van Oss, J. Mol. Recognit., 2003, 16, 177-190.

58.

J. A. Brant and A. E. Childress, J. Membr. Sci., 2002, 203, 257-273.

59.

A. Martin-Molina, J. A. Maroto-Centeno, R. Hidalgo-Alvarez and M. QuesadaPerez, Colloids Surf., A, 2008, 319, 103-108.

60.

C. Schneider, M. Hanisch, B. Wedel, A. Jusufi and M. Ballauff, J. Colloid
Interface Sci., 2011, 358, 62-67.

61.

K. M. Kanal, G. D. Fullerton and I. L. Cameron, Biophys. J., 1994, 66, 153-160.

62.

S. R. Ge, K. Kojio, A. Takahara and T. Kajiyama, J. Biomater. Sci. Pol., 1998, 9,
131-150.

63.

L. R. S. Barbosa, M. G. Ortore, F. Spinozzi, P. Mariani, S. Bernstorff and R. Itri,
Biophys. J., 2010, 98, 147-157.

64.

R. V. Josephson, E. M. Mikolajick and D. P. Sinha, J. Dairy Sci., 1972, 55, 15081510.

65.

K. Hori and S. Matsumoto, Biochem. Eng. J., 2010, 48, 424-434.

66.

G. Hwang, S. Kang, M. Gamal El-Din and Y. Liu, Colloid Surface B, 2012, 91,
181-188.

67.

M. W. Hahn and C. R. O'Melia, Environ. Sci. Technol., 2004, 38, 210-220.

68.

D. Sutlovic, M. D. Gojanovic, S. Andelinovic, D. Gugic and D. Primorac, Croat.
Med. J., 2005, 46, 556-562.

69.

C. N. Albers, A. Jensen, J. Baelum and C. S. Jacobsen, Geomicrobiol. J., 2013,
30, 675-681.
194

70.

B. L. Yuan, M. Pham and T. H. Nguyen, Environ. Sci. Technol., 2008, 42, 76287633.

71.

J. Kim, W. Q. Shan, S. H. R. Davies, M. J. Baumann, S. J. Masten and V. V.
Tarabara, Environ. Sci. Technol., 2009, 43, 5488-5494.

72.

M. Pham, E. A. Mintz and T. H. Nguyen, J. Colloid Interface Sci., 2009, 338, 1-9.

73.

L. Gutierrez and T. H. Nguyen, Environ. Sci. Technol., 2012, 46, 8705-8713.

74.

M. P. Tong, Y. Shen, H. Y. Yang and H. Kim, Colloids Surf., B, 2012, 92, 340347.

75.

A. L. Rose and T. D. Waite, Marine Chem., 2003, 84, 85-103.

76.

T. Karlsson, P. Persson and U. Skyllberg, Environ. Sci. Technol., 2006, 40, 26232628.

77.

X.-H. Guan, C. Shang and G.-H. Chen, Chemosphere, 2006, 65, 2074-2081.

78.

Q. L. Li and M. Elimelech, Environ. Sci. Technol., 2004, 38, 4683-4693.

79.

P. Ahmadiannamini, M. L. Bruening and V. V. Tarabara, J. Membr. Sci., 2015,
491, 149-158.

80.

Z. Yin, V. V. Tarabara and I. Xagoraraki, J. Membr. Sci., 2015, 482, 120-127.

81.

M. Elimelech, X. Jia, J. Gregory and R. Williams, Particle Deposition and
Aggregation: Measurement, Modelling and Simulation, Elsevier Science, 1998.

82.

J. Gregory, J. Colloid Interface Sci., 1981, 83, 138-145.

83.

C. J. van Oss, Interfacial forces in aqeuous media - second edition, Taylor &
Francis Group, 2006.

195

CHAPTER FIVE
Conclusions

Conclusions on individual projects described in the dissertation are provided at the end
of corresponding Chapters. This Chapter lists several overarching conclusions drawn
based on the entirety of the work:

The optimal choice of a sample purification method depends on the intended application
and is most often a matter of tradeoff. One needs to consider the relevance of
“endpoints” (e.g. desired virus purity, efficiency of recovery, effects on virus infectivity)
for the application to choose an appropriate method. Practical considerations such as
time and cost often factor into the decision.

Understanding virus interactions with dissolved species in the sample at the molecular
level is important for increasing virus recovery from complex matrices such as crude
stock or natural water. Relatively little is known definitively about such interactions now.
Virus fate during sample concentration is governed by physicochemical interactions
between the virus and the filter. Primary concentration methods, currently the bottleneck
in the overall concentration process, need to be designed to harness these interactions
to achieve higher and more reproducible recoveries.

Interlaboratory and interstudy comparisons of virus recovery data must be drawn with
extreme caution due to the variability of various parameters used in different studies.
196

The effects of concentration process parameters are complex and, in many cases,
appear to be interrelated. Understanding the virus-filter interactions and the effects individual and combined - of various parameters is important for the design of an
effective membrane-based concentration process. Systematic study to identify general
trends surrogate microorganisms and water matrices would be an improvement over the
trial and error approach used now.

Due to the challenge in virus recovery from environmental waters, public health
protection could not rely on detection alone; instead it should be accompanied by
regulatory measures to control the public access to potentially contaminated water
sources. To protect public health most effectively, the continued development of the
detection technology should be accompanied by regulatory work and environmental
policy decision making.

197