Tải bản đầy đủ - 0trang
IV. Protein Sorption and Denaturation
JIN AND FLURY
Figure 4 Schematic view of protein interaction with a solid surface. Adapted with permission
from Andrade, J. D. and Hlady, V. (1986). Protein adsorption and material biocompatibility: A tutorial
review and suggested hypothesis. Adv. Polym. Sci. 79, 1–63. Copyright Springer-Verlag GmbH &
more sorption sites become active during unfolding of the protein, and the desorption becomes less likely (Fig. 5).
During adsorption, various parts of the protein molecule attach to the solid
surface. To desorb the molecule from the surface, all of these attachment sites
have to be released simultaneously, which requires a considerable amount of energy
(Norde, 1995). Diluting the aqueous phase as in the standard desorption procedure
usually does not desorb proteins from a surface very well. Desorption may be
increased by changing pH, ionic strength of the solution, or by adding compounds,
such as other proteins, that compete for sorption sites. It is also likely that in some
cases the desorbed molecules differ from their original structure, and it has been
found that in many cases the adsorption is not completely reversible (Norde, 1995).
It is also known that in the case of irreversible adsorption, the amount of proteins
adsorbed increases with increasing bulk concentration (Ramsden, 1995b). There
is a controversy in the literature about the explanation for this ﬁnding. Hypotheses
are that (i) proteins adsorb in two or more distinct orientations at the surface,
SUBSURFACE VIRUS FATE AND TRANSPORT
Figure 5 Schematic view of sorption of a soft protein. (1) Adsorption/desorption of native protein.
(2) Structural rearrangement at the surface interface. (3) Adsorption/desorption of conformationally
changed molecule. (4) Transformation to native state. Adapted from Kleijn and Norde (1995). The
adsorption of proteins from aqueous solution on solid surfaces. Heterog. Chem. Rev. 2, 157–172.
Copyright C John Wiley & Sons Limited. Reproduced with permission.
(ii) proteins form an ordered two-dimensional crystal on the surface, and (iii)
proteins denature at the surface (Ramsden, 1995b).
As discussed earlier, several types of interactions (e.g., electrostatic, hydrophobic, and hydrogen-bonding) are involved in the sorption of protein and other
colloidal particles. The magnitudes of the various interactions determine the thermodynamic aspects of the sorption behavior in colloid and protein systems. Strong
interactions between particles and the surface, often leading to irreversible sorption, have been observed with polymer microspheres (Johnson and Elimelech,
1995; Johnson and Lenhoff, 1996; Semmler et al., 1998) and proteins (Andrade
and Hlady, 1986; Kondo and Mihara, 1996; Norde and Lyklema, 1978; Soderquist
and Walton, 1980). In many protein systems, however, protein–surface interactions
are relatively weak and are affected by changing solution conditions (Yuan et al.,
2000). In general, the extent and mechanisms of protein sorption depend upon
the balance between particle–surface attraction and particle–particle repulsion. In
their review, Yuan et al. (2000) outlined the following possible types of protein
1. Irreversible sorption. When the attraction between a charged particle and an
oppositely charged surface is very strong, the particle is likely to maintain a
stable position at the surface. Such an irreversible sorption behavior cannot
be captured accurately with equilibrium models. Rather, random sequential
adsorption (RSA)-type models have been found to be more appropriate.
2. Reversible sorption at low surface coverage. The particles may not bind permanently to any speciﬁc segment of the solid surface when particle–surface
JIN AND FLURY
interaction is relatively weak. Instead, they may attach to various parts of the
surface or undergo desorption, in which case equilibrium sorption models can
be used to describe such behavior.
Whether an irreversible or a reversible sorption reaction occurs is determined
by the balance between particle–surface and particle–particle interactions, both
of which are inﬂuenced by solution pH and ionic strength as well as particle size
(Oberholzer and Lenhoff, 1999; Yuan et al., 2000). Based on energetic considerations, irreversible sorption is more frequent for large particles, whereas reversible
sorption is more frequent for smaller particles (Yuan et al., 2000).
Protein sorption has usually been described with the Langmuir isotherm. The
assumptions made for the Langmuir isotherm to be applicable are that (i) adsorption occurs in a monolayer, (ii) the surface is homogeneous, (iii) adsorption does
not affect the energy status of other molecules, (iv) only one adsorbing species
is present, and (v) sorption is reversible. Recent evidence, however, has shown
mechanisms different from those of Langmuir-type behavior for protein sorption.
Alternative approaches to describe protein sorption kinetics are discussed in the
Many different approaches have been proposed to describe kinetics of protein
sorption. The basic model is the kinetic Langmuir model,
= κ1 φC − κ2 S,
φ = Smax − S.
In this model, the sorption process is completely reversible. The parameter φ represents the number of available sorption sites. For equilibrium conditions, Eqs. (26)
and (27) reduce to the Langmuir isotherm (Eq. 1).
To account for the speciﬁc characteristics of protein sorption, such as the timedependent conformational changes, the basic Langmuir model has been extended
by adding additional compartments. The reactions between the compartments are
assumed to be reversible, irreversible, or time-dependent. Andrade and Hlady
(1986) give an overview over these types of models and propose a general model
as shown in Fig. 6. This model assumes that the desorption is time dependent and
SUBSURFACE VIRUS FATE AND TRANSPORT
Figure 6 A general kinetic model for protein sorption in case of reversible denaturation. Adapted
with permission from Andrade and Hlady (1986). Protein adsorption and material biocompatibility: A
tutorial review and suggested hypothesis. Adv. Polym. Sci. 79, 1–63. Copyright Springer-Verlag GmbH
& Co. KG.
that denaturated proteins quickly regain their native state. The return to the native
state is justiﬁed when no covalent bonds have been formed during the denaturation
process (Andrade and Hlady, 1986). This general model differs slightly from the
concept shown in Fig. 5, where denaturated proteins may sorb reversibly and
denaturation is considered to be more likely irreversible. Speciﬁcs on how to
model time-dependent reactions are given in Soderquist and Walton (1980). These
authors proposed that adsorption and desorption rates can be written as
κ1,i C(ti ) −
κ2,i S(ti ),
where the rate constants κ 1,i and κ 2,i change with time.
The kinetic Langmuir model implies that sorption is reversible and takes place
at deﬁnite sites, and the sorbed molecules do not interfere with each other. However, at most surfaces sorption takes place at randomly selected positions. For
large molecules, this random sorption on the surface may lead to gaps between
sorbed molecules that cannot be ﬁlled anymore. It has also been observed that the
adsorption is in many cases irreversible.
The random adsorption leads to inaccessible sorption sites, and the available fraction of sites φ is actually smaller than Smax − S (Widom, 1966). This
surface-exclusion effect has been formulated mathematically by sequentially placing particles of given size randomly on a line or plane and is often referred to as
the “parking problem”(Feder and Giaever, 1980; Feder, 1980). Figure 7 shows a
one-dimensional schematic of a Langmuir and parking problem process. Schaaf
and Talbot (1989a, 1989b) presented the ﬁrst kinetic analysis of the surfaceexclusion process, and they write the kinetic expression in a generalized Langmuir
= κ1 φ( )C − κ2 ,
JIN AND FLURY
Schematic of a one-dimensional Langmuir sorption process and “parking problem.”
where is the surface coverage given by = S/Smax, and the surface exclusion
effect is represented by the function φ( ). This surface exclusion can be formulated
in different ways, depending on the processes occurring during sorption.
Random sequential adsorption theory (RSA) was developed to model the irreversible deposition of colloidal particles. In the RSA process it is assumed that
(i) the particles cannot overlap on the surface, (ii) adsorption is irreversible (κ 2 = 0
in Eq. 29), and (iii) sorbed particles do not diffuse on the surface. In this case, and
assuming spherical particles, the function φ( ) is given by (Schaaf and Talbot,
6 3 2
+ O( 4 ).
φ =1−4 +
The RSA model is a two-dimensional simulation of the adsorption process in which
a disk is placed at a random location on a planar surface. The placement continues
until no more disks can ﬁt on the surface without overlap with others previously
placed on the surface. The jamming limit, given as the fraction of the total surface
area covered by disks, is found empirically to be max = 0.547. Equation (30) is
valid for < 0.3. At higher surface coverage, the kinetics can be approximated
with (Schaaf and Talbot, 1989b)
φ = 8.98(
− )3 .
Only steric interactions between adsorbed particles are accounted for in the basic
RSA model (Yuan et al., 2000). Deviations from RSA deposition have been found
in several experimental studies of charged particle adsorption where the maximum
surface coverage was dependent upon ionic strength (Johnson and Elimelech, 1995;
Johnson and Lenhoff, 1996; Semmler et al., 1998). The results of these studies
indicate that the effective jamming limit was signiﬁcantly lower than the 54.7%
predicted by the basic RSA model at low ionic strength and increased to approach
the 54.7% maximum coverage with increasing ionic strength. Therefore, a modiﬁed version of the RSA model has been developed to include particle–particle
electrostatic repulsion. Such modiﬁcation requires the adsorbing particle to avoid
overlap with particles already adsorbed as well as to avoid adsorbing in close
SUBSURFACE VIRUS FATE AND TRANSPORT
proximity to each other when electrostatic repulsion is large. Taking into account
the effect of the electrostatic particle–particle repulsion, the maximum coverage
can be expressed as (Semmler et al., 1998)
= 0.547(a/aeff )2 ,
where a is the actual radius of a charged particle and aeff is the radius of the effective
area excluded by the charged particle.
A variety of other extensions and generalizations of the basic RSA model have
been developed in recent years. Among these are the consideration of the relative
importance of diffusion and sedimentation prior to particle deposition (Lavalle
et al., 1999) and the application of RSA to particle deposition during ﬂow
in porous media (Ko and Elimelech, 2000; Ko et al., 2000). During transport
of charged particles in porous media, hydrodynamic as well as electrostatic
interactions cause deviations from standard RSA behavior, and a modiﬁed
blocking function taking into account particle size, ﬂow velocity, and electrolyte
concentrations has been developed (Ko et al., 2000).
Figure 8 shows the effect of the dynamic blocking function on the rate of adsorption in a ﬂowthrough reactor. We consider a ﬂowthrough reactor where the
concentration in the aqueous phase is constant, and plot the rate of adsorption versus
the amount sorbed on the sorbent. The solid line denotes an irreversible Langmuir
kinetics (Eqs. 26, 27 with κ 2 = 0), the dashed line represents RSA (Eqs. 29–31).
The parameter κ 1 has been chosen as 1. Langmuir kinetics shows a linear dependence, whereas in RSA the rate of adsorption is initially faster and decreases with
increasing surface coverage.
Figure 8 Rate of surface coverage as a function of surface coverage for inrreversible Langmuir
and RSA type of kinetics in a ﬂowthrough reactor.
JIN AND FLURY
The RSA kinetics has been experimentally veriﬁed in some cases and was found
to be superior to the Langmuir kinetics (Adamczyk et al., 1992; Kurrat et al.,
1994; Ramsden, 1995a). Many other theoretical approaches of the RSA type have
been developed (Adamczyk et al., 1992, 1999; Adamczyk, 2000; Evans, 1993;
Kurrat et al., 1994; Oberholzer et al., 1997), but most of them have not yet been
V. VIRUS SURVIVAL
A. FACTORS INFLUENCING VIRUS SURVIVAL
Many factors may inﬂuence virus survival in the subsurface environment
(Table I). Virus inactivation is deﬁned as a loss of viral titer with time due to
disruption of coat proteins and degradation of nucleic acid (Gerba, 1984). A large
body of evidence indicates that temperature is the primary inﬂuential factor in the
survival of viruses in soil and groundwater (Straub et al., 1992; Yahya et al., 1992;
Yates et al., 1985). Leﬂer and Kott (1974) found that it took 42 days for 99% inactivation of poliovirus in sand at 25◦ C, whereas for the same degree of inactivation
more than 175 days were required at 18◦ C. Poliovirus was found to persist for
more than 180 days in saturated sand and sandy loam soils at 4◦ C, whereas no
viruses could be recovered from the soils incubated at 37◦ C after 12 days (Yeager
and O’Brien, 1979). Hurst et al. (1980) studied the survival of poliovirus at three
temperatures: 1, 23, and 37◦ C in a loamy sand. They found that the inactivation rate
was signiﬁcantly correlated with incubation temperature, noting faster inactivation
rates at the highest temperatures.
Generally, viruses seem to survive longer under moist as compared to dry conditions. Bagdasaryan (1964) observed that several enteroviruses, including poliovirus 1, coxsackievirus B3, and echoviruses 7 and 9, could survive for 60 to
90 days in soil with 10% moisture as compared with only 15 to 25 days in airdried soils. Ninety-nine percent inactivation of poliovirus occurred in 1 week as
the soil moisture content was reduced from 13 to 0.6%; however, 7 to 8 and 10
to 11 weeks were required for the same amount of inactivation in soils with 25
and 15% moisture content, respectively (Sagik et al., 1978). Hurst et al. (1980)
found that the poliovirus inactivation rate increased as the moisture content was
increased from 5 to 15%, then decreased as more liquid was added.
When viruses are sorbed to a solid surface, such as a clay particle, they are
generally protected from inactivation. Sobsey et al. (1986) compared the rate of
inactivation of hepatitis A virus in ﬁve different soil types including a clay soil, a
clay loam, a loamy sand, a sand, and an organic muck. Survival was the greatest in
the clay soils, in which at least 8 weeks were required to inactivate 99% of the
infectious viruses. This may be due to virus sorption to the soils. Reductions in virus
SUBSURFACE VIRUS FATE AND TRANSPORT
inactivation rates have been reported for virus adsorbed to soil (Hurst et al., 1980),
to clay minerals (Babich and Stotzky, 1980; Lipson and Stotzky, 1984a; Taylor
et al., 1980), to estuarine sediments (Bitton, 1974; Gerba and Schaiberger, 1975;
Liew and Gerba, 1980; Smith et al., 1978), and to sewage solids (Gerba et al., 1978).
The protective effect of virus association with particulate matter or other surfaces
includes protection from proteolytic enzymes or other substances which inactivate
viruses, increased stability of the viral capsid, prevention of virus aggregation
formation, and blocking of ultraviolet radiation (Gerba, 1984). Protection of viruses
against thermoinactivation has been suggested from the results of Bitton et al.
(1976), Liew and Gerba (1980), and Stotzky et al. (1981).
Although virus association with solids in natural environments has generally
been observed to be protective, the association with metal and metal oxide surfaces has been found to enhance virus inactivation. Bacteriophages MS-2 and f2
were inactivated in ﬂuids that had been in contact with Al3+, Zn2+, and Mg2+
(Yamamoto et al., 1964). Atherton and Bell (1983) observed degradation of MS-2
into small fragments after trichloroethylene elution from magnetite at pH 10. Poliovirus readily adsorbed to several metal oxide surfaces and was inactivated at the
surface of MnO2 and CuO (Murray and Laband, 1979). Studies also have shown
that Cu2+ and Fe3+ can cause virus inactivation, and viruses that are enveloped or
contain RNA are more sensitive to inactivation to Cu2+ than those that are nonenveloped or contain DNA (Sagripanti 1992; Sagripanti et al., 1993). The presence
of iron oxides (mainly goethite) was found to be responsible for the sorption and
inactivation of MS-2; however, φX174 was not affected by the metal oxides (Chu
et al., 2000). These results indicate that the degree of inactivation due to metals
and metal oxides is virus speciﬁc.
Other factors such as UV radiation (Battigelli et al., 1993), dissolved oxygen
level (Scheuerman et al., 1991), organic matter concentration and pH (Campos
et al., 2000), certain cations (Quignon et al., 1998; Sagripanti et al., 1993; Siegl
et al., 1984; Wallis et al., 1962), and the presence of other microorganisms (Hurst,
1988; Yates et al., 1990) have also been found to inﬂuence virus survival in the
subsurface environment. Because viruses can be quite persistent and mobile in
soil and groundwater, there is a need to better understand and quantify the factors
that inﬂuence virus survival behavior. A more detailed discussion on the factors
affecting virus inactivation can be found in Schijven and Hassanizadeh (2000).
B. MODELING OF VIRUS INACTIVATION
Virus inactivation is often assumed to follow ﬁrst-order kinetics (Sim and
Chrysikopoulos, 1996; Yates and Yates, 1988), although deviations from this rate
law have been noted for many years (Chang, 1966). In particular, the interplay of
different factors affecting virus survival will likely cause deviations from simple
JIN AND FLURY
Several types of regression equations have been used by Hurst et al. (1992)
to relate virus inactivation to environmental parameters. Among eight equations
= log10 β0 + β1 log10 X 1 + · · · + βn log10 X n + βt log10 t
was found to be most suitable (Hurst et al., 1992), where N0 is the titer at time zero,
Nt is the titer at some subsequent observation time t, and X1 through Xn represent
independent variables, i.e., water conductivity, water turbidity, incubation time,
incubation temperature, and ability to support bacterial growth. The values β 0
through β t represent the respective coefﬁcients assigned to the independent variables X1 through Xn.
Other possible causes of non-ﬁrst-order inactivation include natural variations
in the lability of individual virus particles (virions) in a viral population (Pollard
and Solosko, 1971; Yamagishi and Ozeki, 1972), the association of viruses with
nonviral colloidal material (Babich and Stotzky, 1980; Bitton, 1974; Gerba and
Schaiberger, 1975), or the presence of viral aggregates in the inactivating suspension (Sobsey, 1966). Grant (1995) developed a kinetic model for the inactivation
of viruses that exhibit a range of initial aggregate sizes. Virus aggregation is a
concern mainly because it affects the biological activity of viruses in the following
ways: (i) if viral aggregates consist of many virions that are individually infective,
the overall level of infectivity of an aggregated suspension is reduced relative to
the dispersed state; and (ii) aggregates may be more resistant to inactivation than
single virions (Floyd and Sharp, 1978). In his model, Grant (1995) assumed that
viral aggregates are more resistant to inactivation than a single virion because all
virions within an aggregate must be inactivated before the aggregate as a whole
is considered inactive, and undamaged components of inactive virions within an
aggregate may recombine to cause host–cell infection (multiplicity reactivation
(MR)). Grant proposed three different inactivation laws for three different systems: (i) monodispersed and completely homogeneous, (ii) aggregated viruses
without, and (iii) aggregated viruses with MR. Survival curves calculated from the
model compared well with experimental inactivation data.
There is evidence that virus inactivation rate coefﬁcients may change over time
(Grant et al., 1993; Hurst et al., 1980; Parkinson and Huskey, 1971; Shah and
McCamish, 1972). When a virus population consists of different subpopulations
with different resistances against inactivation, the overall inactivation rate coefﬁcients will decrease with time (Grant et al., 1993; Parkinson and Huskey, 1971).
The time dependence of the inactivation rate coefﬁcient can be modeled either
explicitly as a function of time (Sim and Chrysikopoulos, 1996) or as a function
of residence time (Flury and Jury, 1999). In batch systems, the two approaches are
identical; however, in a ﬂow system, time and residence-time-dependent processes
are different depending on the boundary conditions (Flury and Jury, 1999).
SUBSURFACE VIRUS FATE AND TRANSPORT
VI. THE ROLE OF THE GAS–LIQUID INTERFACE
IN PROTEIN/VIRUS INACTIVATION
The effects of the air–water interface (AWI) on virus survival were ﬁrst encountered by investigators studying the fate of airborne viruses in aerosols. It was
suggested that virus adsorption and inactivation at the AWI may be very similar
to those for individual proteins (Adams, 1948; Trouwborst et al., 1974). Shaklee
and Meltzer (1909) observed that shaking a pepsin solution in the presence of an
AWI resulted in degradation (i.e., loss of biological activity) of pepsin, whereas in
the absence of the AWI, i.e., when the bottles were completely ﬁlled with liquid,
no degradation of pepsin occurred. They also observed that bubbling air or CO2
through the solution results in denaturation of pepsin and that the denaturation rate
increased with decreasing pH of the solution. These early observations have been
conﬁrmed and extended in many other studies (Table V).
The mechanisms of denaturation or inactivation is closely related to sorption
of viruses and proteins at the AWI. Studies have shown that proteins in solution
diffuse to the AWI and adsorb, and subsequently may undergo conformational
changes from their native state to a state of minimal interfacial tension (van del
Vegt et al., 1996). Such a conformational change causes unfolding of a protein’s
three-dimensional conﬁguration and results in its denaturation (Adams, 1948;
Augenstine and Ray, 1957; Donaldson et al., 1980; Tronin et al., 1996). The unfolding of the protein molecular structure is an endothermic process (Zittle, 1953).
Whether or not denaturation of a protein occurs at an AWI is partially determined
by its structural characteristics, which are the cases for the distinction between
“hard” and “soft” proteins. Hard proteins are structurally stable, possibly due to
the presence of greater numbers of intrachain disulﬁde bonds, and generally resist
denaturation upon adsorption (Arai and Norde, 1990; Norde and Favier, 1992;
Tripp et al., 1995). Soft proteins are much less structurally stable and are more
likely to undergo conformational change and lose their biological activities (Tripp
et al., 1995). When proteins desorb from the interface, their native state generally
cannot be regained. Due to the signiﬁcant change in physicochemical properties of
the molecule, protein sorption at the AWI has been considered to be an irreversible
process (MacRitchie, 1987; Neurath and Bull, 1938; Quinn and Dawson, 1970).
Although a single protein is much less complex, it is reasonable to consider a
virus as a large conglomerate of distinct proteins. The hydrophobic regions of a
virus protein partition to the gas phase at an AWI via conformational change
and may render the virus inactive (Trouwborst et al., 1974). However, studies
have shown that the extent of inactivation by the air–water interfacial forces is a
virus-dependent phenomenon. Trouwborst et al. (1974) demonstrated that during
shaking, the EMC virus was not inactivated at the AWI, while phages T3 and T5
were partially inactivated, and T1, MS-2, and Semliki Forest viruses were rapidly
Virus Inactivation and Protein Denaturation at the Gas–Liquid Interface
T1, . . .,T7
Shaking results in ﬁrst-order denaturation; no denaturation
in absence of interface; denaturation increases with
increasing acidity of solution; no denaturation in full
bottles; in absence of air–water interface; denaturation
not caused by oxidation; different degree of denaturation
for different proteins.
Vigorous shaking results in zero-order denaturation.
Protein unfold at interface and form monomolecular ﬁlm;
stirring creates new interface area.
Spreading of protein at gas–liquid interface results in
denaturation of protein; proteins become insoluble.
Proteins form surface ﬁlm; coagulum formed when
solution is shaken; rate of denaturation dependent on
size of bottle, shaking intensity but not concentration.
Inactivation by bubbling air laden with certain vapors
through virus solution.
Inactivation by shaking in buffered saline solution and by
bubbling gas through solution; inactivation increased as
pH reduced from 7 to 5.
Inactivation by shaking and bubbling air through solution;
inactivation dependent on pH; presence of gelatin
Unfolding of macromolecules largely determined by
interfacial energy; sorption follows Langmuir isotherm.
Irreversible binding of proteins to interface.
EMC virus not affected, bacteriophages T3,T5 only little
affected, T1, MS-2, and Semliki Forest virus inactivated
by bubbling air or nitrogen gas through solution;
inactivation prevented by adding peptone and apolar
carboxylic acids; rate of inactivation dependent on salt
concentration, more sorption at higher salt concentration.
Inactivation of MS-2 at air–water interface.
Denaturation at interface depends on gas–liquid contact
time and surface regeneration rate; denaturation reduced
in presence of surfactants.
Denaturation at interface due to unfolding; insoluble
coagulum formed when solution is shaken.
Denaturation of T7 at the interface due to partial unfolding
of protein structure in aeration and shaking experiments;
P22H5 much more stable than T7.
Denaturation at the interface due to partial unfolding of
Inactivation in polypropylene bottles, but not in glass
bottles; inactivation occurs at the triple-phase interface
Shaklee and Meltzer (1909)
Langmuir and Waugh (1938)
Neurath and Bull (1938)
Grubb et al. (1947)
James and Augenstein (1966)
Quinn and Dawson (1970)
Trouwborst et al. (1974)
Trouwborst and de Jong (1973)
Donaldson et al. (1980)
ˇ sko (1992)
Bricelj and Siˇ
Tronin et al. (1996)
Thompson et al. (1998)