- Email: [email protected]
Adsorption of a non-enveloped mammalian virus to functionalized nanofibers
Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
Contents lists available at ScienceDirect
Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb
Adsorption of a non-enveloped mammalian virus to functionalized nanofibers Xue Mi, Caryn L. Heldt ∗ Department of Chemical Engineering, Michigan Technological University, 1400 Townsend Dr., Houghton, MI, USA
a r t i c l e
i n f o
Article history: Received 11 April 2014 Received in revised form 29 May 2014 Accepted 3 June 2014 Available online 11 June 2014 Keywords: Virus removal Microfiltration Kinetics Thermodynamics DLVO theory
a b s t r a c t In the pursuit of finding superior methods to remove pathogens from drinking water, this study examines the adsorption of a non-enveloped, mammalian virus to highly charged nanofibers. N-[(2-Hydroxyl3-trimethylammonium) propyl] chitosan (HTCC) nanofibers were synthesized by the addition of a quaternary amine to chitosan. HTCC was blended with polyvinyl alcohol (PVA) to produce nanofibers by electrospinning. The nanofibers were stabilized against water by crosslinking with glutaraldehyde. When studied in the range of 100–200 nm in diameter, larger fibers were able to adsorb about 90% more virus than smaller fibers. The kinetics of the adsorption was modeled with pseudo-first order kinetics and equilibrium was achieved in as little as 10 min. Equilibrium adsorption was modeled with the Freundlich isotherm with a Freundlich constant of 1.4. When the Freundlich constant deviates from 1, this demonstrates that there is heterogeneity at the adsorption surface. The heterogeneity likely occurs at the nanofiber surface since a polymeric blend of two polymers was used to electrospin the nanofibers. The model mammalian virus, porcine parvovirus (PPV), has a fairly homogeneous, icosahedral protein capsid available for adsorption. The fast adsorption kinetics and high capacity of the nanofibers make HTCC/PVA a potential filter material for the removal of pathogens from drinking water. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Diarrheal diseases caused by pathogens in drinking water lead to the death of over a million people each year [1]. These pathogens are typically bacteria or viruses. Bacterial pathogens can be removed by disinfection with UV light [2] or removal by ultrafiltration [3]. However, viral pathogens are more difficult to remove. The small size of many viruses makes nanofiltration a requirement for sizedbased removal [4] and leads to high transmembrane pressures. Viruses can be removed by chemical treatment, the most common is chlorine. However, the interaction of chlorine with natural organic matter (NOM) is suspected to create carcinogenic byproducts [5]. Chlorination also requires a coordinated supply system, which is often lacking in underdeveloped countries. Due to the many limitations of providing clean drinking water in underdeveloped countries, we are interested in engineering methods of virus removal to purify drinking water that do not rely on chemical disinfection or size-based removal. Adsorption of virus to a solid surface is another option to purify drinking water. An electrochemical carbon nanotube filter has been
∗ Corresponding author. Tel.: +1 906 487 1134; fax: +1 906 487 3213. E-mail address: [email protected] (C.L. Heldt). http://dx.doi.org/10.1016/j.colsurfb.2014.06.007 0927-7765/© 2014 Elsevier B.V. All rights reserved.
shown to both adsorb and inactivate MS2 bacteriophage particles [6]. Although the power requirements for this filter are low, many places in underdeveloped countries do not have adequate access to electricity to continually run a filter of this type. The adsorption of viruses to alumina [7], iron [8], and clays [9] has shown that many different surfaces can adsorb viruses and there is a complex relationship between electrostatic and hydrophobic interactions. The hydrophobic interaction of super-powdered activated carbon (S-PAC) and MS2 bacteriophage achieved a 4 log reduction value (LRV), equivalent to 99.99% removal, after contact for 8 h [10]. In general, chemicals with low-solubility are well adsorbed by activated carbon [11]. Up to this point, activated carbon adsorption has not been considered a typical virus removal step. The hydrophobic and polycationic coating of N,N-dodecyl,methyl-polyethylenimine (PEI) was able to quickly and efficiently disinfect aqueous solutions containing the non-enveloped poliovirus and rotavirus [12] and the enveloped influenza virus [13]. A novel ion adsorber was capable of greater than 4 LRV of mouse minute virus (MMV), xenotropic murine leukemia virus (MLV) and simian virus 40 (SV40) [14]. The adsorber contained eight layers of hydrophilic PVDF base membrane, derivatized with a quaternary amine ligand providing an anion exchange surface [14]. Also explored were trimeric peptide ligands (WRW and KYY), which removed all detectable porcine parvovirus (PPV) from solutions by a minimum of 4.5 LRV [15]. The
320
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
Fig. 1. Synthesis of HTCC from chitosan using GTMAC.
small peptides had two hydrophobic and one positively charged amino acid [15]. However, for negatively charged adsorbers, low virus removal was found. A 0.22 m pore size, modified PVDF membrane filter with a negatively-charged surface was reported to remove less than 0.5 LRV MS2 bacteriophage [3]. One negativelycharged commercial GS9034 microfilter was reported to remove only 1 LRV of MS2 bacteriophage from water [16]. The major disadvantages of ion absorbers are low pathogen capacity and long contact times. To increase the adsorption capacity and decrease the needed contact time, we are exploring high surface area nanofibers with a high charge density. Highly charged chitosan derivates that can be electrospun into nanofibers have been our material of choice for virus removal. Chitosan derivatives that contain a high positive charge are well-known for their antimicrobial properties [17–19]. We have shown that the chitosan derivative, N-[(2-Hydroxyl-3trimethylammonium) propyl] chitosan (HTCC), also has the ability to reduce the infectivity of mammalian viruses [20,21]. In order to create a device to purify water, we have electrospun HTCC into nanofibers. Electrospinning is a common method of creating nanofibers for applications ranging from tissue engineering [22], electronics [23], and membrane materials [24]. Two main attributes of electrospun mats are the ease of fabrication and the high surface to volume ratio, which facilitates adsorption. Additionally, the pore size of the created filters was in the range of several microns. This allows for a high water flux with low transmembrane pressures. Ma et al. demonstrated that increasing the hydrophilicity of an adsorption filter also increased the water flux while maintain high bacteria and virus removal [24]. In the present study, we expanded on our past work that looked at the creation of water-stable, highly charged nanofibers for virus removal [20]. We evaluated the adsorption of the non-enveloped, mammalian virus PPV. PPV is a chemically stable and small virus, with a diameter of 18–26 nm [25]. This makes the virus difficult to remove either chemically or by size, which is why it was chosen as a model virus for removal. 2. Experimental 2.1. Synthesis and characterization of HTCC HTCC was synthesized as described earlier [20,21], and is shown in Fig. 1. Briefly, chitosan (75–85% deacetylated, MW = 190,000–310,000 Da) and glycidyltrimethylammonium chloride (GTMAC) (≥90%) were mixed in Nanopure water (resistance >18 M) at 85 ◦ C for 10 h. The resulting HTCC mixture was dialyzed to remove the excess GTMAC and then filtered to remove the excess chitosan. The final product was purified with cold acetone precipitation and dried. The resulting product was characterized with FTIR in a PerkinElmer FT-IR Spectrum One Spectrometer (Shelton, CT). The degree of quaternization (DQ) was measured by the titration of chloride using silver nitrate, as has been described previously [21]. The DQ was determined to be 76.4 ± 4.3%.
2.2. HTCC nanofiber formation and characterization Blends of the synthesized HTCC and purchased polyvinyl alcohol (PVA) (99% hydrolyzed, MW = 89,000–98,000 Da) were created at a 10% (w/w) polymer and a ratio of 4:6 HTCC:PVA in Nanopure water. Nanofibers were created with a homemade electrospinning apparatus [21] consisting of a multi speed syringe pump (Braintree Scientific Inc., Braintree, MA), a Glassman positive DC high voltage power supply (High Bridge, NJ), capable of generating voltages in the range of 0–30 kV, and a rotating drum collector covered with aluminum foil run by an Electro Craft Torque power pump (Gallipolis, OH). The needle was 5 cm from the rotating drum collector, a 20 kV voltage was applied, the syringe pump was run at 4.5 ml/h, and a rotation speed of 1500 rpm was used for the drum collector. The voltage and pump speed were varied to create fibers of different diameters. The fibers were collected on Whatman filter paper that was attached to the collector. HTCC nanofibers were crosslinked with 30% glutaraldehyde vapor at 37 ◦ C for 4 h, as described earlier [20], to impart water stability to the nanofibers. The nanofibers were imaged with a Hitachi S-4700 cold-field emission scanning electron microscope (FE-SEM) (Tustin, CA) after sputter coating with 5 nm of platinum/palladium (Hummer Sputtering System, Union City, CA). The accelerating voltage for the FE-SEM was 5 kV, and the magnification was from 1000× to 80,000×. To determine the fiber diameter, 50 random fibers from three SEM-micrographs were calculated with Nano Measurer and OriginLab software. 2.3. Virus removal Virus removal was accomplished with the model porcine parvovirus (PPV) (strain NADL-2) that was propagated and titrated on porcine kidney cells (PK-13), which has been described previously [26,27]. The virus was titrated with the cell viability assay, MTT. The log removal value (LRV) was used to determine the amount of virus that adsorbed to the nanofibers, LRV = −log10
c f
co
(1)
where cf is the infectious virus concentration after adsorption and c0 is the initial infectious virus concentration, both in MTT50 /ml. For virus adsorption studies, nanofibers on a filter paper support were punched to create a 0.50 cm2 circle. The nanofibers were incubated with 0.5 ml of virus of different concentrations and different times. For the kinetic studies, the initial virus concentration was 7 log10 (MTT50 /ml) in Nanopure water. For the equilibrium studies, the incubation time was 10 min with various starting concentrations diluted in Nanopure water. All studies had end-over-end rotation with a Roto-shake Genie rocker (Scientific Industries Inc., Bohemia, NY). The amount adsorbed to the nanofibers was calculated with Eq. (2): qi =
(co − ci )V M
(2)
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
321
Fig. 2. FTIR of chitosan and HTCC.
where qi is the amount of adsorb virus at any time in MTT50 /mg, ci is the concentration of virus in solution at any time in MTT50 /ml, V is the volume of solution added to the fibers, which was kept constant at 0.5 ml, and M is the amount of nanofibers in mg. 3. Results and discussion 3.1. Characterization of HTCC nanofibers In order to create nanofibers that adsorb a model mammalian virus, we first synthesized HTCC from chitosan. The FTIR spectra are shown in Fig. 2. The primary amine peak found at 1590 cm−1 in chitosan is reduced in the HTCC. The new methyl groups create a C–H bending peak at 1474 cm−1 in HTCC. These are the same results as found by others [17]. We have created a highly charged nanofiber made of HTCC and PVA by electrospinning. Our past work has shown that crosslinking the fibers with glutaraldehyde will impart water stability without reducing the adsorption of an enveloped and non-enveloped virus [20]. As shown in Fig. 3, glutaraldehyde crosslinked fibers are water stable after 60 min of incubation in water. The positively charged surface of the HTCC is likely the most prominent adsorption mechanism. It has been shown when positively-charged chitosan and neutrally charged PVA were electrospun, a core-shell fiber was formed where the charged chitosan was the shell material [28]. This leads us to believe that the HTCC is highly concentrated to the outside of the fiber, and we plan to explore this in the future.
3.2. Nanofiber diameter and density We hypothesized that the nanofiber diameter would affect the adsorption of virus to the nanofibers. This is due to the aspect ratio of the virus to the fibers. We explored fibers that ranged from 100 to 200 nm in diameter. The virus used in this work, PPV, is 18–26 nm in diameter [25]. With this aspect ratio, the virus will likely experience a difference in the curvature of the fibers. We changed the electrospinning voltage and feed rate to create different diameter fibers. This has been a documented method of controlling nanofiber diameter in electrospinning [20,29]. As shown in Fig. 4A, as the diameter increased, there was an increase in virus adsorption. In particular, an increase from 100 to 200 nm produce a 1 LRV increase in virus removal, which is equivalent to a 90% increase in virus removal. This was contrary to the expected results. We expected that the smaller diameter fibers, which have a higher surface area, would have a higher LRV. We wanted to confirm that the different electrospinning conditions used to change the fiber diameter did not change the amount of nanofibers collected. In Fig. 4B, it shows that there was no correlation between the density of the nanofibers and the nanofibers diameter. Therefore, we conclude that the positive relationship between the nanofibers and virus adsorption is due to the diameter and not a change in the density due to the change in electrospinning conditions. As more nanostructures are being created and used in biological systems, the adsorption of molecules as a function of curvature
Fig. 3. SEM of HTCC-PVA nanofibers. (A) Freshly spun fibers, (B) fibers after glutaraldehyde crosslinking, and (C) crosslinked fibers after 60 min immersion in water.
322
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
Fig. 5. LRV of virus as a function of time. The error bars are the standard deviation of three virus removal experiments.
removal, the effect of curvature on adsorption should continue to be explored. 3.3. Kinetics of virus adsorption
Fig. 4. Fiber diameter effect on virus removal. Electrospinning applied voltage and feed rate were varied to create fibers of various diameter. (A) Fiber diameter as a function of virus removal in LRV. The error bars are the standard deviation of three virus removal experiments. (B) Relationship between the density of electrospun nanofibers and the diameter.
is being explored. Tethered enzymes are more stable and have a higher activity when attached to a high curvature surface [30]. The high curvature (defined as 1/radius of the nanostructure), separated the enzymes on the surface. Exploring how adsorption is affected by curvature, the adsorption of methane was studied ab initio to graphene (a flat surface) and the inside and outside of a carbon nanotube. The adsorption energy was the lowest for the internal area of the nanotube, followed by flat graphene and the highest for the external area of the nanotube [31]. Experimentally, the mass adsorption of laminin, a protein that has a flat structure, increased with decreasing curvature [32]. Although the geometry of small molecule adsorption and protein adsorption is not the same as an icosahedral non-enveloped virus adsorption to nanofibers, there are few studies currently available for comparison. Therefore, our results follow the current trends of other adsorption studies in that flatter, or less curved surfaces have higher adsorption. A better geometric approximation of the interaction of the virus with nanofibers is to use the interaction of a sphere and a cylinder, which represent the virus and the nanofibers, respectively. The Derjaguin–Landau–Verwy–Overbeek (DLVO) theory was used to characterize the hydrophobic van der Waals interactions and the electrostatic double-layer interactions between a sphere and a cylinder [33]. It was determined that as Rc /Rs became greater than 10, where Rc is the radius of the cylinder and Rs is the radius of the sphere, the DLVO theory predicted that the interaction was equal to the interaction of a sphere and a flat plate. Using this ratio, we predict that 200 nm diameter nanofibers have the same interactions with a spherical virus of 20 nm as does a flat surface. A flat surface has the maximum interaction energy as compared to a curved surface and a sphere. Therefore, 200 nm is likely the diameter for maximum virus-nanofibers interaction. As the nanofiber diameter increases past 200 nm, the high surface to volume ratio of nanofibers may become hindered. We were not able to experimentally confirm this since constraints in electrospinning only allowed us to study the diameter range of 100–200 nm fibers. As more adsorptive nanostructures are created for virus and protein
The removal of PPV with time is shown in Fig. 5. The LRV of the virus quickly arrived at equilibrium after 10 min of adsorption. This is much faster than the binding of heavy metal ions, including nickel, cadmium, lead and copper, to electrospun chitosan/PEO nanofibers, which took 120 min for equilibrium [34]. It would be expected that metals would adsorb faster than virus if this were only a diffusion-limited process. This highlights that the adsorption of virus to HTCC nanofibers is kinetically favorable and fast. The virus adsorption kinetic data was fit to a pseudo-zero order, pseudo-first order and pseudo-second order rate law [34,35], as shown in Eqs. (3)–(5), respectively: qi = ko t
(3)
In(qe − qi ) = In qe − k1 t
(4)
t 1 1 = + t qi qe k2 q2e
(5)
where k0 is the zero order rate constant, k1 is the first order rate constant, k2 is the second order rate constant, qe is the amount of adsorb virus at equilibrium in MTT/mg and t is the adsorption time. From the fitted data in Fig. 6, it can be seen that the pseudo-zero order and the pseudo-first order rate law fit the data best. This is also shown in Table 1, which contains the rate constants, the correlation coefficients (R2 ), and the p-values as determined by an F-test. Although the p-value for the pseudo zero-order model is low, there is no significance to this model. A zero-order rate law assumes that regardless of concentration, the nanofibers will continue to adsorb more virus. Realistically, we know that there is a finite amount of virus that can be adsorbed to a solid surface and therefore this cannot be a realistic model of virus adsorption. Although the p-value for the pseudo first-order model is high, it is the most realistic of the models. The model allows for a finite adsorption. The p-value is likely high due to the low number of data points that were taken in the kinetic range of the model (see Fig. 6B). Due to the difficulty of the experiments, more data points were not feasible. Due to the low R2 value and high p-value, we discard the pseudo second-order model. We conclude that with the limited data we have, that the pseudo first-order model best describes the data. The pseudo-first order rate constant is higher by 5-fold than the calculated rate constant for metal binding to nanofibers [34]. Although we measured the infectious adsorption of the virus, parvoviruses are known to have many non-infectious particles, up to 1 infectious particle to every 102 –104 virus particles [36], making it likely that several orders of magnitude more virus are adsorbing to the nanofibers than calculated here.
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
323
Table 1 Kinetic rate constants. Rate model
Pseudo-zero order Pseudo-first order Pseudo-second order *
Experimental
Rate constants
qe (MTT50 mg−1 )
qe (MTT50 mg−1 )
ki
R2
p-value*
na 55,730 55,730
na 65,180 108,700
5630 MTT mg−1 min−1 0.12 min−1 3.8 × 10−7 (MTT/mg)−1 min−1
0.96 0.99 0.50
0.02 0.07 0.30
p-value determined from an F-test.
3.4. Adsorption isotherms Different starting concentrations of PPV were incubated for 10 min with the crosslinked HTCC nanofibers. From the kinetic trials, we determined that 10 min was long enough to reach equilibrium, as shown by the LRV vs time plot in Fig. 5. The virus removal was fit to many different isotherm models [35,37]. We first fit the data to the linear portion of the Langmuir isotherm with the following model: q = qm kL C
(6)
and the fit is shown in Fig. 7A. This model has the same assumptions as the Langmuir isotherm except that qm is not achieved in the experiment and therefore only the product qm KL can be calculated. Virus production methods make it difficult to obtain high virus concentrations and viruses tend to aggregation at high concentrations. The linear Langmuir isotherm allows similar systems
Fig. 7. Isotherm modeling of equilibrium virus removal. (A) Linearized fit of the linear Langmuir isotherm. (B) Linearized fit of the Freundlich isotherm.
to be compared that likely have the same qm , and has been used to describe virus adsorption [38]. The adsorption of PPV to nanofibers does not show a good correlation for the linear adsorption model, as shown in Table 2. The best fitting isotherm was the Freundlich isotherm, shown in Fig. 7B. This is a commonly found isotherm in biological adsorption [9]. The Freundlich isotherm is an empirical adsorption model which assumes a heterogeneous surface with different adsorption sites [37]. The linear form of the Freundlich isotherm is expressed as: log q = log kf + n log C
(7)
where Kf is the Freundlich isotherm constant related to adsorption capacity and n is the Freundlich isotherm constant related to the interaction. When n is 1, the idealized Langmuir isotherm is obtained. The Langmuir isotherm assumes a homogeneous surface and that the adsorption energy is independent of surface coverage. As shown in Table 2, the Freundlich constant for the adsorption of PPV to nanofibers was 1.4. This describes a heterogeneity that is present in the adsorption process. It has three surface proteins, Table 2 Isotherm constants. Isotherm model
Isotherm constants Ki
Fig. 6. Kinetic modeling of virus removal. (A) Linearized fit of the pseudo-zero order reaction. (B) Linearized fit of the pseudo-first order reaction. (C) Fit of the kinetic models to the data. Experimental data (diamonds), pseudo-zero order fit (dotted line), pseudo-first order fit (solid line) and pseudo-second order fit (dashed line).
Freundlich model Linear Langmuir model *
−1 n
0.12 (ml mg 4.8 ml mg−1
p-value determined from an F-test.
)
n
R2
p-value*
1.4
0.91 0.74
0.05 0.12
324
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
VP1, VP2 and VP3. 80% of the virus surface is VP2 and the other two proteins are cleavage products of VP2 [39], meaning that they have the same amino acid sequence minus the amino acids that were cleaved. Since PPV is fairly homogeneous, it is likely that the heterogeneity occurs at the nanofiber surface. The nanofibers are formed from a blend of HTCC and PVA. There could be either HTCC or PVA or both present on the surface of the nanofibers. Another possibility is that the polymer blend could form core-shell fibers, as has been shown for other electrolytic polymers electrospun with non-ionogenic partners [28]. If HTCC is on the surface of the nanofibers, there is likely heterogeneity in the HTCC polymer. We began with chitosan that was 75–85% deacetylated. Of the deacetylated amines, we functionalized almost all of the amines, because the HTCC was measured to contain 76% quaternary amines. Therefore, the HTCC was a mixture of quaternary amines, acetylated amines, and possible a few remaining primary amines. A third possibility for the presence of heterogeneity is the multi-layer binding of the virus on the nanofiber surface. Since virus tends to agglomerate, it is likely that virus will bind to one another and this can also cause the Freundlich constant to be greater than 1. This virus–virus interaction may also be the cause of the higher p-values in the kinetic modeling. The adsorption of bacteriophages to two different types of clays was modeled with the Freundlich isotherm [9]. It was found that n was close to 1 for the bacteriophage-clay adsorption. The clays used were negatively charged at the pH studied. In contrast, the nanofibers studied here were positively charged and attracted the virus due to an electrostatic interaction. Virus removal is often studied for positively charged surfaces [40,41]. The difference between adsorption of a virus to a positively and negatively charge surface could explain the difference in the value of n, as well as the extreme geometric differences between the surfaces. The value of Kf is slightly smaller for the nanofibers studied here compared to clays that bound the least bacteriophage, the STx-1b clay, which were reported as 0.27 (ml mg−1 )n for X174 and 0.76 (ml mg−1 )n for MS2 [9]. 4. Conclusions In this work, we have studied the adsorption of a non-enveloped, mammalian virus to positively-charged HTCC nanofibers. The nanofibers were stabilized against water dissolution by crosslinking with glutaraldehyde. As nanofiber diameter increased from 100 to 200 nm, there was an increase of 1 LRV, which is equivalent to an increase in 90% of the virus removal. The adsorption was best described by a pseudo-first order rate model and the Freundlich isotherm. Equilibrium was achieved in 10 min, which is a fast adsorption. The Freundlich constant was 1.4, which demonstrates that there are likely heterogeneous PPV binding sites on the nanofibers. Due to the method of nanofiber manufacturing by electrospinning two polymers, it is likely that the heterogeneity is in the nanofiber or multilayer PPV adsorption. The characterized nanofiber mats have a high virus adsorption and the adsorption occurs quickly. These nanofiber mats have a high potential to become future water filtration membranes for the removal of pathogenic organisms. Acknowledgements We would like to thank Dr. Patricia Heiden for use of her electrospinning apparatus and fruitful discussions. We would also like
to thank Mr. K. Saagar Vijayaragavan for assistance with the SEM images. Funding for this project was received from the Department of Chemical Engineering at Michigan Tech, the Research Enhancement Fund at Michigan Tech and NSF (CBET-1125585). References [1] WHO, 2010. Available from: http://www.euro.who.int/en/what-we-do/healthtopics/environmental-health/water-and-sanitation/facts-and-figures (cited accessed: 22.11.10). [2] M.B. Fisher, M. Iriarte, K.L. Nelson, Water Res. 46 (6) (2012) 1745–1754. [3] B.T. Zhu, D.A. Clifford, S. Chellam, Water Res. 39 (20) (2005) 5153–5161. [4] A.H. Koenderman, H.G. ter Hart, I.M. Prins-de Nijs, J. Bloem, S. Stoffers, A. Kempers, G.J. Derksen, B. Al, L. Dekker, J. Over, Biologicals 40 (6) (2012) 473–481. [5] D. Liviac, A. Creus, R. Marcos, Environ. Mol. Mutagen. 50 (5) (2009) 413–420. [6] M.S. Rahaman, C.D. Vecitis, M. Elimelech, Environ. Sci. Technol. 46 (3) (2012) 1556–1564. [7] F. Meder, J. Wehling, A. Fink, B. Piel, K. Li, K. Frank, A. Rosenauer, L. Treccani, S. Koeppen, A. Dotzauer, K. Rezwan, Biomaterials 34 (17) (2013) 4203–4213. [8] C. Shi, J. Wei, Y. Jin, K.E. Kniel, P.C. Chiu, Sep. Purif. Technol. 84 (2012) 72–78. [9] C.V. Chrysikopoulos, V.I. Syngouna, Colloid Surf. B: Biointerfaces 92 (2012) 74–83. [10] T. Matsushita, H. Suzuki, N. Shirasaki, Y. Matsui, K. Ohno, Sep. Purif. Technol. 107 (2013) 79–84. [11] J.S. Yang, D.X. Yuan, T.P. Weng, Desalination 263 (1–3) (2010) 271–278. [12] A.M. Larson, B.B. Hsu, D. Rautaray, J. Haldar, J.Z. Chen, A.M. Klibanov, Biotechnol. Bioeng. 108 (3) (2011) 720–723. [13] B.B. Hsu, S.Y. Wong, P.T. Hammond, J.Z. Chen, A.M. Klibanov, Proc. Natl. Acad. Sci. U.S.A. 108 (1) (2011) 61–66. [14] M. Phillips, J. Cormier, J. Ferrence, C. Dowd, R. Kiss, H. Lutz, J. Carter, J. Chromatogr. A 1078 (1–2) (2005) 74–82. [15] C.L. Heldt, P.V. Gurgel, L.A. Jaykus, R.G. Carbonell, Biotechnol. Prog. 24 (3) (2008) 554–560. [16] H.Y. Ma, C. Burger, B.S. Hsiao, B. Chu, J. Mater. Chem. 21 (21) (2011) 7507–7510. [17] S.H. Lim, S.M. Hudson, Carbohydr. Res. 339 (2) (2004) 313–319. [18] M. Ignatova, N. Manolova, I. Rashkov, Eur. Polym. J. 43 (4) (2007) 1112–1122. [19] Q.L. Li, S. Mahendra, D.Y. Lyon, L. Brunet, M.V. Liga, D. Li, P.J.J. Alvarez, Water Res. 42 (18) (2008) 4591–4602. [20] X. Mi, K.S. Vijayaragavan, C.L. Heldt, Carbohydr. Res. 387 (2014) 24–29. [21] B. Bai, X. Mi, X. Xiang, P.A. Heiden, C.L. Heldt, Carbohydr. Res. 380 (2013) 137–142. [22] Y. Li, F. Chen, J. Nie, D. Yang, Carbohydr. Polym. 90 (4) (2012) 1445–1451. [23] L.Q. Mai, L. Xu, C.H. Han, X. Xu, Y.Z. Luo, S.Y. Zhao, Y.L. Zhao, Nano Lett. 10 (11) (2010) 4750–4755. [24] H.Y. Ma, B.S. Hsiao, B. Chu, J. Membr. Sci. 452 (2014) 446–452. [25] S. Halder, R. Ng, M. Agbandje-McKenna, Fut. Virol. 7 (3) (2012) 253–278. [26] C.L. Heldt, R. Hernandez, U. Mudiganti, P.V. Gurgel, D.T. Brown, R.G. Carbonell, J. Virol. Methods 135 (1) (2006) 56–65. [27] M.F. Tafur, K.S. Vijayaragavan, C.L. Heldt, Antivir. Res. 99 (1) (2013) 27–33. [28] X. Mu, Y. Liu, D. Fang, Z. Wang, J. Nie, G. Ma, Carbohydr. Polym. 90 (4) (2012) 1582–1586. [29] X. Geng, O.-H. Kwon, J. Jang, Biomaterials 26 (2005) 5427–5432. [30] P. Asuri, S.S. Karajanagi, H. Yang, T.J. Yim, R.S. Kane, J.S. Dordick, Langmuir 22 (2006) 5833–5836. [31] D.G.A. Smith, K. Patkowski, J. Phys. Chem. C 118 (1) (2014) 544–550. [32] N. Giamblanco, E. Martines, G. Marletta, Langmuir 29 (26) (2013) 8335–8342. [33] K. Li, Y. Chen, Colloid Surf. A: Physicochem. Eng. Asp. 415 (0) (2012) 218–229. [34] M. Aliabadi, M. Irani, J. Ismaeili, H. Piri, M.J. Parnian, Chem. Eng. J. 220 (2013) 237–243. [35] D. Vu, Z. Li, H. Zhang, W. Wang, Z. Wang, X. Xu, B. Dong, C. Wang, J. Colloid Interface Sci. 367 (1) (2012) 429–435. [36] S.I. Lang, S. Boelz, A.Y. Stroh-Dege, J. Rommelaere, C. Dinsart, J.J. Cornelis, J. Virol. 79 (1) (2005) 289–298. [37] S. Ozdemir, F.M. Bekler, V. Okumus, A. Dundar, E. Kilinc, Environ. Prog. Sustain. Energy 31 (4) (2012) 544–552. [38] C.L. Heldt, P.V. Gurgel, L.A. Jaykus, R.G. Carbonell, Biotechnol. Prog. 25 (5) (2009) 1411–1418. [39] A.A. Simpson, B. Herbert, G.M. Sullivan, C.R. Parrish, Z. Zadori, P. Tijssen, M.G. Rossmann, J. Mol. Biol. 315 (2002) 1189–1198. [40] J. Weaver, S.M. Husson, L. Murphy, R. Wickramasinghe, Biotechnol. Bioeng. 110 (2013) 491–499. [41] W. Riordan, S. Heilmann, K. Brorson, K. Seshadri, Y. He, M. Etzel, Biotechnol. Bioeng. 103 (5) (2009) 920–929.
Contents lists available at ScienceDirect
Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb
Adsorption of a non-enveloped mammalian virus to functionalized nanofibers Xue Mi, Caryn L. Heldt ∗ Department of Chemical Engineering, Michigan Technological University, 1400 Townsend Dr., Houghton, MI, USA
a r t i c l e
i n f o
Article history: Received 11 April 2014 Received in revised form 29 May 2014 Accepted 3 June 2014 Available online 11 June 2014 Keywords: Virus removal Microfiltration Kinetics Thermodynamics DLVO theory
a b s t r a c t In the pursuit of finding superior methods to remove pathogens from drinking water, this study examines the adsorption of a non-enveloped, mammalian virus to highly charged nanofibers. N-[(2-Hydroxyl3-trimethylammonium) propyl] chitosan (HTCC) nanofibers were synthesized by the addition of a quaternary amine to chitosan. HTCC was blended with polyvinyl alcohol (PVA) to produce nanofibers by electrospinning. The nanofibers were stabilized against water by crosslinking with glutaraldehyde. When studied in the range of 100–200 nm in diameter, larger fibers were able to adsorb about 90% more virus than smaller fibers. The kinetics of the adsorption was modeled with pseudo-first order kinetics and equilibrium was achieved in as little as 10 min. Equilibrium adsorption was modeled with the Freundlich isotherm with a Freundlich constant of 1.4. When the Freundlich constant deviates from 1, this demonstrates that there is heterogeneity at the adsorption surface. The heterogeneity likely occurs at the nanofiber surface since a polymeric blend of two polymers was used to electrospin the nanofibers. The model mammalian virus, porcine parvovirus (PPV), has a fairly homogeneous, icosahedral protein capsid available for adsorption. The fast adsorption kinetics and high capacity of the nanofibers make HTCC/PVA a potential filter material for the removal of pathogens from drinking water. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Diarrheal diseases caused by pathogens in drinking water lead to the death of over a million people each year [1]. These pathogens are typically bacteria or viruses. Bacterial pathogens can be removed by disinfection with UV light [2] or removal by ultrafiltration [3]. However, viral pathogens are more difficult to remove. The small size of many viruses makes nanofiltration a requirement for sizedbased removal [4] and leads to high transmembrane pressures. Viruses can be removed by chemical treatment, the most common is chlorine. However, the interaction of chlorine with natural organic matter (NOM) is suspected to create carcinogenic byproducts [5]. Chlorination also requires a coordinated supply system, which is often lacking in underdeveloped countries. Due to the many limitations of providing clean drinking water in underdeveloped countries, we are interested in engineering methods of virus removal to purify drinking water that do not rely on chemical disinfection or size-based removal. Adsorption of virus to a solid surface is another option to purify drinking water. An electrochemical carbon nanotube filter has been
∗ Corresponding author. Tel.: +1 906 487 1134; fax: +1 906 487 3213. E-mail address: [email protected] (C.L. Heldt). http://dx.doi.org/10.1016/j.colsurfb.2014.06.007 0927-7765/© 2014 Elsevier B.V. All rights reserved.
shown to both adsorb and inactivate MS2 bacteriophage particles [6]. Although the power requirements for this filter are low, many places in underdeveloped countries do not have adequate access to electricity to continually run a filter of this type. The adsorption of viruses to alumina [7], iron [8], and clays [9] has shown that many different surfaces can adsorb viruses and there is a complex relationship between electrostatic and hydrophobic interactions. The hydrophobic interaction of super-powdered activated carbon (S-PAC) and MS2 bacteriophage achieved a 4 log reduction value (LRV), equivalent to 99.99% removal, after contact for 8 h [10]. In general, chemicals with low-solubility are well adsorbed by activated carbon [11]. Up to this point, activated carbon adsorption has not been considered a typical virus removal step. The hydrophobic and polycationic coating of N,N-dodecyl,methyl-polyethylenimine (PEI) was able to quickly and efficiently disinfect aqueous solutions containing the non-enveloped poliovirus and rotavirus [12] and the enveloped influenza virus [13]. A novel ion adsorber was capable of greater than 4 LRV of mouse minute virus (MMV), xenotropic murine leukemia virus (MLV) and simian virus 40 (SV40) [14]. The adsorber contained eight layers of hydrophilic PVDF base membrane, derivatized with a quaternary amine ligand providing an anion exchange surface [14]. Also explored were trimeric peptide ligands (WRW and KYY), which removed all detectable porcine parvovirus (PPV) from solutions by a minimum of 4.5 LRV [15]. The
320
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
Fig. 1. Synthesis of HTCC from chitosan using GTMAC.
small peptides had two hydrophobic and one positively charged amino acid [15]. However, for negatively charged adsorbers, low virus removal was found. A 0.22 m pore size, modified PVDF membrane filter with a negatively-charged surface was reported to remove less than 0.5 LRV MS2 bacteriophage [3]. One negativelycharged commercial GS9034 microfilter was reported to remove only 1 LRV of MS2 bacteriophage from water [16]. The major disadvantages of ion absorbers are low pathogen capacity and long contact times. To increase the adsorption capacity and decrease the needed contact time, we are exploring high surface area nanofibers with a high charge density. Highly charged chitosan derivates that can be electrospun into nanofibers have been our material of choice for virus removal. Chitosan derivatives that contain a high positive charge are well-known for their antimicrobial properties [17–19]. We have shown that the chitosan derivative, N-[(2-Hydroxyl-3trimethylammonium) propyl] chitosan (HTCC), also has the ability to reduce the infectivity of mammalian viruses [20,21]. In order to create a device to purify water, we have electrospun HTCC into nanofibers. Electrospinning is a common method of creating nanofibers for applications ranging from tissue engineering [22], electronics [23], and membrane materials [24]. Two main attributes of electrospun mats are the ease of fabrication and the high surface to volume ratio, which facilitates adsorption. Additionally, the pore size of the created filters was in the range of several microns. This allows for a high water flux with low transmembrane pressures. Ma et al. demonstrated that increasing the hydrophilicity of an adsorption filter also increased the water flux while maintain high bacteria and virus removal [24]. In the present study, we expanded on our past work that looked at the creation of water-stable, highly charged nanofibers for virus removal [20]. We evaluated the adsorption of the non-enveloped, mammalian virus PPV. PPV is a chemically stable and small virus, with a diameter of 18–26 nm [25]. This makes the virus difficult to remove either chemically or by size, which is why it was chosen as a model virus for removal. 2. Experimental 2.1. Synthesis and characterization of HTCC HTCC was synthesized as described earlier [20,21], and is shown in Fig. 1. Briefly, chitosan (75–85% deacetylated, MW = 190,000–310,000 Da) and glycidyltrimethylammonium chloride (GTMAC) (≥90%) were mixed in Nanopure water (resistance >18 M) at 85 ◦ C for 10 h. The resulting HTCC mixture was dialyzed to remove the excess GTMAC and then filtered to remove the excess chitosan. The final product was purified with cold acetone precipitation and dried. The resulting product was characterized with FTIR in a PerkinElmer FT-IR Spectrum One Spectrometer (Shelton, CT). The degree of quaternization (DQ) was measured by the titration of chloride using silver nitrate, as has been described previously [21]. The DQ was determined to be 76.4 ± 4.3%.
2.2. HTCC nanofiber formation and characterization Blends of the synthesized HTCC and purchased polyvinyl alcohol (PVA) (99% hydrolyzed, MW = 89,000–98,000 Da) were created at a 10% (w/w) polymer and a ratio of 4:6 HTCC:PVA in Nanopure water. Nanofibers were created with a homemade electrospinning apparatus [21] consisting of a multi speed syringe pump (Braintree Scientific Inc., Braintree, MA), a Glassman positive DC high voltage power supply (High Bridge, NJ), capable of generating voltages in the range of 0–30 kV, and a rotating drum collector covered with aluminum foil run by an Electro Craft Torque power pump (Gallipolis, OH). The needle was 5 cm from the rotating drum collector, a 20 kV voltage was applied, the syringe pump was run at 4.5 ml/h, and a rotation speed of 1500 rpm was used for the drum collector. The voltage and pump speed were varied to create fibers of different diameters. The fibers were collected on Whatman filter paper that was attached to the collector. HTCC nanofibers were crosslinked with 30% glutaraldehyde vapor at 37 ◦ C for 4 h, as described earlier [20], to impart water stability to the nanofibers. The nanofibers were imaged with a Hitachi S-4700 cold-field emission scanning electron microscope (FE-SEM) (Tustin, CA) after sputter coating with 5 nm of platinum/palladium (Hummer Sputtering System, Union City, CA). The accelerating voltage for the FE-SEM was 5 kV, and the magnification was from 1000× to 80,000×. To determine the fiber diameter, 50 random fibers from three SEM-micrographs were calculated with Nano Measurer and OriginLab software. 2.3. Virus removal Virus removal was accomplished with the model porcine parvovirus (PPV) (strain NADL-2) that was propagated and titrated on porcine kidney cells (PK-13), which has been described previously [26,27]. The virus was titrated with the cell viability assay, MTT. The log removal value (LRV) was used to determine the amount of virus that adsorbed to the nanofibers, LRV = −log10
c f
co
(1)
where cf is the infectious virus concentration after adsorption and c0 is the initial infectious virus concentration, both in MTT50 /ml. For virus adsorption studies, nanofibers on a filter paper support were punched to create a 0.50 cm2 circle. The nanofibers were incubated with 0.5 ml of virus of different concentrations and different times. For the kinetic studies, the initial virus concentration was 7 log10 (MTT50 /ml) in Nanopure water. For the equilibrium studies, the incubation time was 10 min with various starting concentrations diluted in Nanopure water. All studies had end-over-end rotation with a Roto-shake Genie rocker (Scientific Industries Inc., Bohemia, NY). The amount adsorbed to the nanofibers was calculated with Eq. (2): qi =
(co − ci )V M
(2)
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
321
Fig. 2. FTIR of chitosan and HTCC.
where qi is the amount of adsorb virus at any time in MTT50 /mg, ci is the concentration of virus in solution at any time in MTT50 /ml, V is the volume of solution added to the fibers, which was kept constant at 0.5 ml, and M is the amount of nanofibers in mg. 3. Results and discussion 3.1. Characterization of HTCC nanofibers In order to create nanofibers that adsorb a model mammalian virus, we first synthesized HTCC from chitosan. The FTIR spectra are shown in Fig. 2. The primary amine peak found at 1590 cm−1 in chitosan is reduced in the HTCC. The new methyl groups create a C–H bending peak at 1474 cm−1 in HTCC. These are the same results as found by others [17]. We have created a highly charged nanofiber made of HTCC and PVA by electrospinning. Our past work has shown that crosslinking the fibers with glutaraldehyde will impart water stability without reducing the adsorption of an enveloped and non-enveloped virus [20]. As shown in Fig. 3, glutaraldehyde crosslinked fibers are water stable after 60 min of incubation in water. The positively charged surface of the HTCC is likely the most prominent adsorption mechanism. It has been shown when positively-charged chitosan and neutrally charged PVA were electrospun, a core-shell fiber was formed where the charged chitosan was the shell material [28]. This leads us to believe that the HTCC is highly concentrated to the outside of the fiber, and we plan to explore this in the future.
3.2. Nanofiber diameter and density We hypothesized that the nanofiber diameter would affect the adsorption of virus to the nanofibers. This is due to the aspect ratio of the virus to the fibers. We explored fibers that ranged from 100 to 200 nm in diameter. The virus used in this work, PPV, is 18–26 nm in diameter [25]. With this aspect ratio, the virus will likely experience a difference in the curvature of the fibers. We changed the electrospinning voltage and feed rate to create different diameter fibers. This has been a documented method of controlling nanofiber diameter in electrospinning [20,29]. As shown in Fig. 4A, as the diameter increased, there was an increase in virus adsorption. In particular, an increase from 100 to 200 nm produce a 1 LRV increase in virus removal, which is equivalent to a 90% increase in virus removal. This was contrary to the expected results. We expected that the smaller diameter fibers, which have a higher surface area, would have a higher LRV. We wanted to confirm that the different electrospinning conditions used to change the fiber diameter did not change the amount of nanofibers collected. In Fig. 4B, it shows that there was no correlation between the density of the nanofibers and the nanofibers diameter. Therefore, we conclude that the positive relationship between the nanofibers and virus adsorption is due to the diameter and not a change in the density due to the change in electrospinning conditions. As more nanostructures are being created and used in biological systems, the adsorption of molecules as a function of curvature
Fig. 3. SEM of HTCC-PVA nanofibers. (A) Freshly spun fibers, (B) fibers after glutaraldehyde crosslinking, and (C) crosslinked fibers after 60 min immersion in water.
322
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
Fig. 5. LRV of virus as a function of time. The error bars are the standard deviation of three virus removal experiments.
removal, the effect of curvature on adsorption should continue to be explored. 3.3. Kinetics of virus adsorption
Fig. 4. Fiber diameter effect on virus removal. Electrospinning applied voltage and feed rate were varied to create fibers of various diameter. (A) Fiber diameter as a function of virus removal in LRV. The error bars are the standard deviation of three virus removal experiments. (B) Relationship between the density of electrospun nanofibers and the diameter.
is being explored. Tethered enzymes are more stable and have a higher activity when attached to a high curvature surface [30]. The high curvature (defined as 1/radius of the nanostructure), separated the enzymes on the surface. Exploring how adsorption is affected by curvature, the adsorption of methane was studied ab initio to graphene (a flat surface) and the inside and outside of a carbon nanotube. The adsorption energy was the lowest for the internal area of the nanotube, followed by flat graphene and the highest for the external area of the nanotube [31]. Experimentally, the mass adsorption of laminin, a protein that has a flat structure, increased with decreasing curvature [32]. Although the geometry of small molecule adsorption and protein adsorption is not the same as an icosahedral non-enveloped virus adsorption to nanofibers, there are few studies currently available for comparison. Therefore, our results follow the current trends of other adsorption studies in that flatter, or less curved surfaces have higher adsorption. A better geometric approximation of the interaction of the virus with nanofibers is to use the interaction of a sphere and a cylinder, which represent the virus and the nanofibers, respectively. The Derjaguin–Landau–Verwy–Overbeek (DLVO) theory was used to characterize the hydrophobic van der Waals interactions and the electrostatic double-layer interactions between a sphere and a cylinder [33]. It was determined that as Rc /Rs became greater than 10, where Rc is the radius of the cylinder and Rs is the radius of the sphere, the DLVO theory predicted that the interaction was equal to the interaction of a sphere and a flat plate. Using this ratio, we predict that 200 nm diameter nanofibers have the same interactions with a spherical virus of 20 nm as does a flat surface. A flat surface has the maximum interaction energy as compared to a curved surface and a sphere. Therefore, 200 nm is likely the diameter for maximum virus-nanofibers interaction. As the nanofiber diameter increases past 200 nm, the high surface to volume ratio of nanofibers may become hindered. We were not able to experimentally confirm this since constraints in electrospinning only allowed us to study the diameter range of 100–200 nm fibers. As more adsorptive nanostructures are created for virus and protein
The removal of PPV with time is shown in Fig. 5. The LRV of the virus quickly arrived at equilibrium after 10 min of adsorption. This is much faster than the binding of heavy metal ions, including nickel, cadmium, lead and copper, to electrospun chitosan/PEO nanofibers, which took 120 min for equilibrium [34]. It would be expected that metals would adsorb faster than virus if this were only a diffusion-limited process. This highlights that the adsorption of virus to HTCC nanofibers is kinetically favorable and fast. The virus adsorption kinetic data was fit to a pseudo-zero order, pseudo-first order and pseudo-second order rate law [34,35], as shown in Eqs. (3)–(5), respectively: qi = ko t
(3)
In(qe − qi ) = In qe − k1 t
(4)
t 1 1 = + t qi qe k2 q2e
(5)
where k0 is the zero order rate constant, k1 is the first order rate constant, k2 is the second order rate constant, qe is the amount of adsorb virus at equilibrium in MTT/mg and t is the adsorption time. From the fitted data in Fig. 6, it can be seen that the pseudo-zero order and the pseudo-first order rate law fit the data best. This is also shown in Table 1, which contains the rate constants, the correlation coefficients (R2 ), and the p-values as determined by an F-test. Although the p-value for the pseudo zero-order model is low, there is no significance to this model. A zero-order rate law assumes that regardless of concentration, the nanofibers will continue to adsorb more virus. Realistically, we know that there is a finite amount of virus that can be adsorbed to a solid surface and therefore this cannot be a realistic model of virus adsorption. Although the p-value for the pseudo first-order model is high, it is the most realistic of the models. The model allows for a finite adsorption. The p-value is likely high due to the low number of data points that were taken in the kinetic range of the model (see Fig. 6B). Due to the difficulty of the experiments, more data points were not feasible. Due to the low R2 value and high p-value, we discard the pseudo second-order model. We conclude that with the limited data we have, that the pseudo first-order model best describes the data. The pseudo-first order rate constant is higher by 5-fold than the calculated rate constant for metal binding to nanofibers [34]. Although we measured the infectious adsorption of the virus, parvoviruses are known to have many non-infectious particles, up to 1 infectious particle to every 102 –104 virus particles [36], making it likely that several orders of magnitude more virus are adsorbing to the nanofibers than calculated here.
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
323
Table 1 Kinetic rate constants. Rate model
Pseudo-zero order Pseudo-first order Pseudo-second order *
Experimental
Rate constants
qe (MTT50 mg−1 )
qe (MTT50 mg−1 )
ki
R2
p-value*
na 55,730 55,730
na 65,180 108,700
5630 MTT mg−1 min−1 0.12 min−1 3.8 × 10−7 (MTT/mg)−1 min−1
0.96 0.99 0.50
0.02 0.07 0.30
p-value determined from an F-test.
3.4. Adsorption isotherms Different starting concentrations of PPV were incubated for 10 min with the crosslinked HTCC nanofibers. From the kinetic trials, we determined that 10 min was long enough to reach equilibrium, as shown by the LRV vs time plot in Fig. 5. The virus removal was fit to many different isotherm models [35,37]. We first fit the data to the linear portion of the Langmuir isotherm with the following model: q = qm kL C
(6)
and the fit is shown in Fig. 7A. This model has the same assumptions as the Langmuir isotherm except that qm is not achieved in the experiment and therefore only the product qm KL can be calculated. Virus production methods make it difficult to obtain high virus concentrations and viruses tend to aggregation at high concentrations. The linear Langmuir isotherm allows similar systems
Fig. 7. Isotherm modeling of equilibrium virus removal. (A) Linearized fit of the linear Langmuir isotherm. (B) Linearized fit of the Freundlich isotherm.
to be compared that likely have the same qm , and has been used to describe virus adsorption [38]. The adsorption of PPV to nanofibers does not show a good correlation for the linear adsorption model, as shown in Table 2. The best fitting isotherm was the Freundlich isotherm, shown in Fig. 7B. This is a commonly found isotherm in biological adsorption [9]. The Freundlich isotherm is an empirical adsorption model which assumes a heterogeneous surface with different adsorption sites [37]. The linear form of the Freundlich isotherm is expressed as: log q = log kf + n log C
(7)
where Kf is the Freundlich isotherm constant related to adsorption capacity and n is the Freundlich isotherm constant related to the interaction. When n is 1, the idealized Langmuir isotherm is obtained. The Langmuir isotherm assumes a homogeneous surface and that the adsorption energy is independent of surface coverage. As shown in Table 2, the Freundlich constant for the adsorption of PPV to nanofibers was 1.4. This describes a heterogeneity that is present in the adsorption process. It has three surface proteins, Table 2 Isotherm constants. Isotherm model
Isotherm constants Ki
Fig. 6. Kinetic modeling of virus removal. (A) Linearized fit of the pseudo-zero order reaction. (B) Linearized fit of the pseudo-first order reaction. (C) Fit of the kinetic models to the data. Experimental data (diamonds), pseudo-zero order fit (dotted line), pseudo-first order fit (solid line) and pseudo-second order fit (dashed line).
Freundlich model Linear Langmuir model *
−1 n
0.12 (ml mg 4.8 ml mg−1
p-value determined from an F-test.
)
n
R2
p-value*
1.4
0.91 0.74
0.05 0.12
324
X. Mi, C.L. Heldt / Colloids and Surfaces B: Biointerfaces 121 (2014) 319–324
VP1, VP2 and VP3. 80% of the virus surface is VP2 and the other two proteins are cleavage products of VP2 [39], meaning that they have the same amino acid sequence minus the amino acids that were cleaved. Since PPV is fairly homogeneous, it is likely that the heterogeneity occurs at the nanofiber surface. The nanofibers are formed from a blend of HTCC and PVA. There could be either HTCC or PVA or both present on the surface of the nanofibers. Another possibility is that the polymer blend could form core-shell fibers, as has been shown for other electrolytic polymers electrospun with non-ionogenic partners [28]. If HTCC is on the surface of the nanofibers, there is likely heterogeneity in the HTCC polymer. We began with chitosan that was 75–85% deacetylated. Of the deacetylated amines, we functionalized almost all of the amines, because the HTCC was measured to contain 76% quaternary amines. Therefore, the HTCC was a mixture of quaternary amines, acetylated amines, and possible a few remaining primary amines. A third possibility for the presence of heterogeneity is the multi-layer binding of the virus on the nanofiber surface. Since virus tends to agglomerate, it is likely that virus will bind to one another and this can also cause the Freundlich constant to be greater than 1. This virus–virus interaction may also be the cause of the higher p-values in the kinetic modeling. The adsorption of bacteriophages to two different types of clays was modeled with the Freundlich isotherm [9]. It was found that n was close to 1 for the bacteriophage-clay adsorption. The clays used were negatively charged at the pH studied. In contrast, the nanofibers studied here were positively charged and attracted the virus due to an electrostatic interaction. Virus removal is often studied for positively charged surfaces [40,41]. The difference between adsorption of a virus to a positively and negatively charge surface could explain the difference in the value of n, as well as the extreme geometric differences between the surfaces. The value of Kf is slightly smaller for the nanofibers studied here compared to clays that bound the least bacteriophage, the STx-1b clay, which were reported as 0.27 (ml mg−1 )n for X174 and 0.76 (ml mg−1 )n for MS2 [9]. 4. Conclusions In this work, we have studied the adsorption of a non-enveloped, mammalian virus to positively-charged HTCC nanofibers. The nanofibers were stabilized against water dissolution by crosslinking with glutaraldehyde. As nanofiber diameter increased from 100 to 200 nm, there was an increase of 1 LRV, which is equivalent to an increase in 90% of the virus removal. The adsorption was best described by a pseudo-first order rate model and the Freundlich isotherm. Equilibrium was achieved in 10 min, which is a fast adsorption. The Freundlich constant was 1.4, which demonstrates that there are likely heterogeneous PPV binding sites on the nanofibers. Due to the method of nanofiber manufacturing by electrospinning two polymers, it is likely that the heterogeneity is in the nanofiber or multilayer PPV adsorption. The characterized nanofiber mats have a high virus adsorption and the adsorption occurs quickly. These nanofiber mats have a high potential to become future water filtration membranes for the removal of pathogenic organisms. Acknowledgements We would like to thank Dr. Patricia Heiden for use of her electrospinning apparatus and fruitful discussions. We would also like
to thank Mr. K. Saagar Vijayaragavan for assistance with the SEM images. Funding for this project was received from the Department of Chemical Engineering at Michigan Tech, the Research Enhancement Fund at Michigan Tech and NSF (CBET-1125585). References [1] WHO, 2010. Available from: http://www.euro.who.int/en/what-we-do/healthtopics/environmental-health/water-and-sanitation/facts-and-figures (cited accessed: 22.11.10). [2] M.B. Fisher, M. Iriarte, K.L. Nelson, Water Res. 46 (6) (2012) 1745–1754. [3] B.T. Zhu, D.A. Clifford, S. Chellam, Water Res. 39 (20) (2005) 5153–5161. [4] A.H. Koenderman, H.G. ter Hart, I.M. Prins-de Nijs, J. Bloem, S. Stoffers, A. Kempers, G.J. Derksen, B. Al, L. Dekker, J. Over, Biologicals 40 (6) (2012) 473–481. [5] D. Liviac, A. Creus, R. Marcos, Environ. Mol. Mutagen. 50 (5) (2009) 413–420. [6] M.S. Rahaman, C.D. Vecitis, M. Elimelech, Environ. Sci. Technol. 46 (3) (2012) 1556–1564. [7] F. Meder, J. Wehling, A. Fink, B. Piel, K. Li, K. Frank, A. Rosenauer, L. Treccani, S. Koeppen, A. Dotzauer, K. Rezwan, Biomaterials 34 (17) (2013) 4203–4213. [8] C. Shi, J. Wei, Y. Jin, K.E. Kniel, P.C. Chiu, Sep. Purif. Technol. 84 (2012) 72–78. [9] C.V. Chrysikopoulos, V.I. Syngouna, Colloid Surf. B: Biointerfaces 92 (2012) 74–83. [10] T. Matsushita, H. Suzuki, N. Shirasaki, Y. Matsui, K. Ohno, Sep. Purif. Technol. 107 (2013) 79–84. [11] J.S. Yang, D.X. Yuan, T.P. Weng, Desalination 263 (1–3) (2010) 271–278. [12] A.M. Larson, B.B. Hsu, D. Rautaray, J. Haldar, J.Z. Chen, A.M. Klibanov, Biotechnol. Bioeng. 108 (3) (2011) 720–723. [13] B.B. Hsu, S.Y. Wong, P.T. Hammond, J.Z. Chen, A.M. Klibanov, Proc. Natl. Acad. Sci. U.S.A. 108 (1) (2011) 61–66. [14] M. Phillips, J. Cormier, J. Ferrence, C. Dowd, R. Kiss, H. Lutz, J. Carter, J. Chromatogr. A 1078 (1–2) (2005) 74–82. [15] C.L. Heldt, P.V. Gurgel, L.A. Jaykus, R.G. Carbonell, Biotechnol. Prog. 24 (3) (2008) 554–560. [16] H.Y. Ma, C. Burger, B.S. Hsiao, B. Chu, J. Mater. Chem. 21 (21) (2011) 7507–7510. [17] S.H. Lim, S.M. Hudson, Carbohydr. Res. 339 (2) (2004) 313–319. [18] M. Ignatova, N. Manolova, I. Rashkov, Eur. Polym. J. 43 (4) (2007) 1112–1122. [19] Q.L. Li, S. Mahendra, D.Y. Lyon, L. Brunet, M.V. Liga, D. Li, P.J.J. Alvarez, Water Res. 42 (18) (2008) 4591–4602. [20] X. Mi, K.S. Vijayaragavan, C.L. Heldt, Carbohydr. Res. 387 (2014) 24–29. [21] B. Bai, X. Mi, X. Xiang, P.A. Heiden, C.L. Heldt, Carbohydr. Res. 380 (2013) 137–142. [22] Y. Li, F. Chen, J. Nie, D. Yang, Carbohydr. Polym. 90 (4) (2012) 1445–1451. [23] L.Q. Mai, L. Xu, C.H. Han, X. Xu, Y.Z. Luo, S.Y. Zhao, Y.L. Zhao, Nano Lett. 10 (11) (2010) 4750–4755. [24] H.Y. Ma, B.S. Hsiao, B. Chu, J. Membr. Sci. 452 (2014) 446–452. [25] S. Halder, R. Ng, M. Agbandje-McKenna, Fut. Virol. 7 (3) (2012) 253–278. [26] C.L. Heldt, R. Hernandez, U. Mudiganti, P.V. Gurgel, D.T. Brown, R.G. Carbonell, J. Virol. Methods 135 (1) (2006) 56–65. [27] M.F. Tafur, K.S. Vijayaragavan, C.L. Heldt, Antivir. Res. 99 (1) (2013) 27–33. [28] X. Mu, Y. Liu, D. Fang, Z. Wang, J. Nie, G. Ma, Carbohydr. Polym. 90 (4) (2012) 1582–1586. [29] X. Geng, O.-H. Kwon, J. Jang, Biomaterials 26 (2005) 5427–5432. [30] P. Asuri, S.S. Karajanagi, H. Yang, T.J. Yim, R.S. Kane, J.S. Dordick, Langmuir 22 (2006) 5833–5836. [31] D.G.A. Smith, K. Patkowski, J. Phys. Chem. C 118 (1) (2014) 544–550. [32] N. Giamblanco, E. Martines, G. Marletta, Langmuir 29 (26) (2013) 8335–8342. [33] K. Li, Y. Chen, Colloid Surf. A: Physicochem. Eng. Asp. 415 (0) (2012) 218–229. [34] M. Aliabadi, M. Irani, J. Ismaeili, H. Piri, M.J. Parnian, Chem. Eng. J. 220 (2013) 237–243. [35] D. Vu, Z. Li, H. Zhang, W. Wang, Z. Wang, X. Xu, B. Dong, C. Wang, J. Colloid Interface Sci. 367 (1) (2012) 429–435. [36] S.I. Lang, S. Boelz, A.Y. Stroh-Dege, J. Rommelaere, C. Dinsart, J.J. Cornelis, J. Virol. 79 (1) (2005) 289–298. [37] S. Ozdemir, F.M. Bekler, V. Okumus, A. Dundar, E. Kilinc, Environ. Prog. Sustain. Energy 31 (4) (2012) 544–552. [38] C.L. Heldt, P.V. Gurgel, L.A. Jaykus, R.G. Carbonell, Biotechnol. Prog. 25 (5) (2009) 1411–1418. [39] A.A. Simpson, B. Herbert, G.M. Sullivan, C.R. Parrish, Z. Zadori, P. Tijssen, M.G. Rossmann, J. Mol. Biol. 315 (2002) 1189–1198. [40] J. Weaver, S.M. Husson, L. Murphy, R. Wickramasinghe, Biotechnol. Bioeng. 110 (2013) 491–499. [41] W. Riordan, S. Heilmann, K. Brorson, K. Seshadri, Y. He, M. Etzel, Biotechnol. Bioeng. 103 (5) (2009) 920–929.