Earlier studies have shown that 3-layer-modified cellulose fibers with poly(acrylic acid) (PAA) as the middle layer between two cationic polyelectrolyte polyvinylamine (PVAm) layers have strong antibacterial efficacy in terms of both bacteria adsorption and bacterial growth inhibition. In the present work, the fossil-based PAA middle layer was replaced by sustainable wood-based cellulose nano-fibrils (CNF), i. e., the fibers were modified by a 3-layer PVAm/CNF/PVAm system. Interestingly, the antibacterial efficacy of this system was greater than that of the previous PVAm/PAA/PVAm system. A higher salt concentration and lower assembly pH in the multilayer build-up resulted in better bacterial reduction. As the surface of a cellulose fiber is heterogeneous, making it difficult to characterize and visualize at high resolution, more homogeneous cellulose model surfaces were prepared by spin coating the dissolved cellulose fiber onto a silica surface to model the fiber surface. With increasing ionic strength, more aggregated and heterogeneous structures can be observed on the PVAm/CNF/PVAm modified model surfaces. The adsorbed bacteria distributed on the structured surfaces were clearly seen under fluorescence microscopy. Adsorbed amounts of bacteria on either aggregate or flat regions were quantified by scanning electron microscopy (SEM). More adsorbed bacteria were clearly seen on aggregates than on the flat regions at the surfaces. Degrees of bacteria deformation and cell damage were also seen under SEM. The surface roughness of the modified model surfaces was examined by atomic force microscopy (AFM), and a positive correlation was found between the surface roughness and the bacterial adhesion. Thus, an additional factor that controls adhesion, in addition to the surface charge, which is probably the most dominant factor affecting the bacteria adhesion, is the surface structures, such as roughness.
Bacterial infections were a deadly enemy to humans until the first antibiotic penicillin was discovered. After many years of efforts, more advanced antibiotics have been developed and a lot of lives have therefore been saved. However, it is also a double-edged weapon since the over development and abuse of antibiotics creates drug resistance in bacteria which can reduce the effectiveness of antibiotics so that bacteria continue to multiply. This has now become a global concern. New antibiotics and biocides are being developed to diminish the new emerging resistant bacteria, but bacteria always find a way to survive and it seems that humans can never catch up with the pace of bacteria evolution (Andersson and Hughes 2012, Lukačišinová and Bollenbach 2017).
Bacteria can be classified as Gram-positive and Gram-negative based on the structural characteristics of the bacteria cell wall. Many Gram-positive bacteria have a thicker peptidoglycan that is densely functionalized with teichoic acids which are anionic glycopolymers, whereas the outer membrane of the Gram-negative bacteria cell wall is mainly composed of anionic lipopolysaccharides. Most bacteria therefore carry a net negative surface charge. A potential antibacterial solution is thus to target the cell wall instead of the enzyme inside bacteria cell, and to develop materials with an antibacterial surface immobilized with a cationic polymer that can act on the cell membrane by physical adsorption, damaging the bacteria cell wall integrity and reducing its viability (Milović et al. 2005). If the positive charge density is high enough a surface embedded with a positive charge should be able to adsorb bacteria through electrostatic interactions and destabilize the membrane integrity by an ion-exchange process (Kugler et al. 2005, Lichter and Rubner 2009, Murata et al. 2007). It has also been claimed that polycations with hydrophobic modifications interact with the hydrophobic bacteria cell wall causing bacteria lysis (Lewis and Klibanov 2005, Tiller et al. 2001). According to this concept, Illergård et al. modified the cellulosic fibers with a cationic polyelectrolyte polyvinylamine (PVAm) applied by the layer-by-layer (LbL) technique, and it was found that a 3-layer build-up of PVAm together with polyacrylic acid (PAA) as the anionic middle layer achieved the best antibacterial effect (Illergård et al. 2012). It has also been shown that PVAm is a better option for making antibacterial fiber than the other commonly used cationic polyelectrolytes poly(diallyl dimethyl ammonium chloride) (PDADMAC) and poly(allylamine hydrochloride) (PAH) due to a higher charge density and better incorporation with PAA (Chen et al. 2017).
In order to enhance the antibacterial efficacy in terms of bacterial adsorption as well as bacterial growth inhibition, and to make an all bio-based antibacterial material without fossil-based polymers, the PAA middle layer was replaced by the more sustainable wood-based cellulose nano-fibrils (CNF), i. e. the fibers were modified by a 3-layer PVAm/CNF/PVAm system. Earlier work by Wågberg et al. explored the build-up of a polyelectrolyte multilayer (PEM) with CNF and have shown a good control of layer thickness with either different polyelectrolytes such as poly(ethylene imine) PEI, PDADMAC etc., or different salt concentrations (Wågberg et al. 2008). The colloidal stability of a CNF dispersion in aqueous solution was investigated (Fall et al. 2011), who explained the effect of pH and salt concentration on the aggregation of CNF, showing that the CNF tends to aggregate at lower pH and higher salt concentrations. In the present study, CNF as the middle layer was deposited at different pH (3.5 and 7.5) and different salt concentrations (1, 10, 100 mM) in order to evaluate the optimal combination and explore the factors that could affect the interaction between bacteria and modified cellulose fibers. A model cellulose film was also prepared by spin-coating dissolved cellulose fibers onto silicon wafers to simulate the fiber. In order to study the interactions in detail in a way that was not possible by studying fibers (Gunnars et al. 2002), i. e. roughness measurements by atomic force microscopy (AFM), and fluorescence microscopy without cellulose fiber auto-fluorescing problems etc.
Materials and methods
Chemicals and substrates
PVAm was supplied by BASF SE (Ludwigshafen, Germany) with a molecular weight of 340 kDa. PAA was obtained from Sigma-Aldrich (Stockholm, Sweden) with a molecular weight of 240 kDa. All polymers were dialysed and freeze-dried before use. Disintegrated bleached chemical softwood kraft pulp was supplied by ESSITY (Mölndal, Sweden). The fibers were first washed by a standard procedure (Wågberg and Björklund 1993) to remove metal ions, dissolved polymers and colloids. The pH of the fiber suspension was adjusted to 2 with 1 M HCl and the suspension was stored for 30 min, and thereafter washed with deionized water until the conductivity of the filtrate was below 5 µS/cm. 0.1 M NaHCO3 was then added to the suspension, the pH was adjusted to 9 with 1 M NaOH, and the fibers were converted to their sodium form. A nanofibrillated carboxymethylated cellulose stock sample (Generation 2) was obtained from RISE AB (Stockholm, Sweden) as a gel-like pulp containing 2.5 % of fibers in MilliQ water (MQ) (Millipore, Solna, Sweden), and a dispersion of this CNF stock was prepared (Wågberg et al. 2008).
Preparation of cellulose model surfaces
Cellulose model surfaces were prepared, according to Gunnars et al., on silicon wafers (single-side polished, p-type) from Siltronic AG (Munchen, Germany). The wafer was washed with a sequence of MilliQ water, ethanol, and MilliQ water, dried with N2 gas, and then oxidized at ambient atmospheric pressure in an oven at 1000 °C for 30 min. The oxidized wafer was then cut into 10 × 10 mm pieces, and hydrophilized by dipping it in a 10 wt % NaOH solution for 30 s followed by rinsing and drying. Prior to applying regenerated cellulose to the wafers, the wafer was dipped into a PVAm solution (0.1 g L−1, pH 7.5) for 15 min, rinsed with MilliQ water and dried with N2, in order to provide an anchoring layer. The same cellulose fibers as those used in fiber modification were dissolved in NMMO (50 wt % in water, Sigma Aldrich) and DMSO (99 %, Sigma Aldrich), and cellulose II surfaces were regenerated onto PVAm-treated silicon wafers by spinning at 1500 rpm for 15 s followed by 3500 rpm for 30 s on a spincoater (KW-4A-2, Chemat Technologt, Northridge, CA, US). The cellulose film was precipitated and washed by immersing the coated wafers in MilliQ water for 1 hour and leaving them for another hour in fresh MilliQ water, followed by drying with N2 and leaving in oven at 105 °C for 6 hours to enhance the surfaces stability (Gunnars et al. 2002).
5 g cellulose pulp fibers was added to deionised water in which 2 M NaCl was added to adjust the salt concentration to 100 mM at pH 9.5. Thereafter, the first layer of polymer was deposited by adding 10 g L−1 PVAm to the fiber suspension at a concentration of 0.1 g L−1 and 0.5 % w/w fiber consistency with constant stirring for 15 min. A second layer was then deposited with 0.1 g L−1 concentration of PAA at a salt concentration of 100 mM and pH adjusted to 3.5 in PVAm/PAA/PVAm modified sample and 0.1 g L−1 CNF at the pH and salt combinations shown in Table 1 for the PVAm/CNF/PVAm modified samples. The third layer, the outermost layer of PVAm, was deposited at the same pH and salt concentration as the first layer. The pulp was washed thoroughly with deionised water after each deposition step until the conductivity of filtrate was below 5 µS/cm. The LbL deposition on the model surface followed the same schemes as the modification of cellulose fibers, except that the layer assembly was achieved by dipping and drying in nitrogen instead of stirring and freeze-drying. Modified silica wafers were kept in a vacuum desiccator.
Nitrogen analysis was carried out using an ANTEK MultiTek instrument (PAC, Houston, TX, USA) to assess the amount of adsorbed cationic polymer (the only nitrogen-containing source) on the cellulose fiber. When fiber samples were burned in a low oxygen content, the nitrogen-containing compounds were oxidized and excited to NO2, and the light emission was registered and presented as N-counts. 15 mg of each fiber sample was tested in triplicate, and the amount of polymer adsorbed was calculated from a calibration curve. The results are presented as mean values together with the 95 % confidence interval.
|pH 3.5||pH 3.5||pH 3.5|
|1 mM NaCl||10 mM NaCl||100 mM NaCl|
|pH 7.5||pH 7.5||pH 7.5|
|1 mM NaCl||10 mM NaCl||100 mM NaCl|
Topographies of the CNF/PVAm-modified cellulose model surfaces were examined using an upright bright field microscope (ICS standard 25 ZEISS, NY, US). Images were captured using a 10X air objective with the same exposure.
Quartz crystal microbalance with dissipation (QCM-D)
A QCM-D E4 instrument (Bioline Scientific AB, Göteborg, Sweden) was used to obtain an insight of layer assembling of CNF/PVAm on cellulose model surface. A detailed description of the working protocol can be found elsewhere. The flow sequence of adsorptions on QSX-334 cellulose-coated silica quartz crystals (Q-sense) were: Rinsing-PVAm-rinsing-CNF-rinsing-PVAm solution. The concentrations of PVAm and CNF were 0.1 g L−1, which was same as in the modification of the model surface. Rinsing solutions were MilliQ water adjusted to different pH and salt concentrations, depending on the conditions of assembly of the next layer. For a rigid film, the adsorbed mass can be determined from the change in resonance frequency (∆f) of the crystal according to the Sauerbrey equation:
where ∆m is the adsorbed mass, C is the crystal sensitivity factor, and n is the overtone number. Using the QCM-D instrument, both the frequency and the dissipation were monitored at the same time. The energy dissipation relates to the rigidity of the film, the dissipation factor D being defined as
where is the energy dissipated during one oscillation period, and is the stored energy in the system (Marx 2003).
Gram-negative Escherichia coli (E. coli) K-12 (BioRad, Solna, Sweden) were incubated in a nutrient broth (BD Difco, Stockholm, Sweden) at 37 °C with continuous shaking. The bacterial cells were harvested after 18 hours of incubation by centrifugation at 5000 rpm for 5 min using a bench-top centrifuge (VWR, Stockholm, Sweden) at room temperature. The cells were washed twice with 1/4 strength Ringer’s solution (Merck, Stockholm, Sweden), and redispersed in 1/4 Ringer’s solution. 1 ml of approx. 10−7 CFU mL−1 of the bacterial solution was then added to 9 ml of fiber suspension at 1 % fiber consistency, the pH was controlled by adding 0.1 ml 0.1 M TRIS buffer (pH 7.5) (Sigma-Aldrich, Stockholm, Sweden), and the mixed suspension was incubated for 6 hours at 37 °C with continuous shaking. The exact numbers of initially added bacterial cells and of those left in the suspension after incubation for 6 hours were determined by making a 10-fold dilution series followed by counting the colonies on Petrifilm™ (3M, Sollentuna, Sweden). The counting was done manually, with less than 100 bacteria, or by particle analyser by means of ImageJ.
Bacterial growth inhibition test
In according with the protocol described by Illergård et al., bacterial growth was monitored by adding 10 % nutrient broth to each fiber-bacteria suspension after the bacterial removal tests. The suspensions were incubated with nutrient for 18 hours at 37 °C with continuous shaking, and the extent of bacterial growth inhibition was determined from the optical density (OD) at λ = 620 nm on a MultiSkan FC microplate spectrophotometer (Thermo Scientific, Stockholm, Sweden). When the concentration of bacteria was below the detection limit (OD 0.036) of the instrument, the bacteria concentration was determined by the above-mentioned bacterial counting on Petrifilm (Illergård et al. 2012).
A fluorescence microscope Nikon Eclipse Ti-U (BergmanLabora AB, Danderyd, Sweden) was used to study the quantities and distributions of the bacteria adsorbed on cellulose model surfaces. The bacteria E. coli was transformed in PGreen plasmid according to a calcium-chloride-heat-shock-transformation protocol (Mandel and Higa 1970), which could express green fluorescent protein. Modified and unmodified cellulose model surfaces were submerged in 107 CFU mL−1E. coli PGreen in Ringer’s solution for 18 hours with continuous shaking, followed by a gentle washing with Ringer’s solution. The surfaces were placed upside down on a glass slide and observed under 20× magnification. Pictures were captured with the same exposure time and configuration as in manual mode.
Bacterial cells were fixed on model surfaces prior to the AFM and SEM tests. The specimens were submerged in 2.5 % glutaraldehyde in 0.1 M phosphate buffer (pH 7.4) overnight at 4 °C to kill the bacteria and cross-link their proteins. They were then briefly rinsed in deionised water followed by solvent exchange in 70 % ethanol for 15 min, 90 % ethanol for 15 min, 95 % ethanol for 15 min, absolute ethanol for 15 min, all at 4 °C, and finally in acetone (Mustafa et al. 1998).
Atomic force microscopy (AFM)
An Atomic force microscope was used for surface characterization of the cellulose model surfaces both with and without adsorbed bacteria. The roughness of the cellulose film on the silica wafer in fluid was measured using a SCANASYST mode with AFM cantilevers of SCANASYST-FLUID+ on an AFM MultiMode 8 (Bruker, Santa Barbara, CA), with a scanning area of 5 × 5 μm2, which was a suitable scale for creating less variations during roughness measurements. Five test points were measured in triplicate at random positions of CNF aggregates formed on modified model surfaces and of the flat regions without aggregation. To obtain a clear and high-resolution 3D image of how bacteria attached on model surfaces, the bacterial cells adsorbed were fixed as mentioned above, and the surfaces were scanned in MilliQ-water with the scanning mode of SCANASYST in fluid. The morphology of the bacterium could thus be seen.
Scanning electron microscopy (SEM)
Model surfaces with adsorbed bacteria were dried in air after AFM measurements and then coated with 10 nm of Pt in a 208HR high-resolution sputter coater (Cressington, Watford, UK) before being studied in SEM using S-4800 field emission scanning electron microscope (Hitachi, Tokyo, Japan). The images were captured under 18.0 k and 1.50 k magnification at 5.0 kV.
Results and discussion
Modification of cellulose fiber
Cellulose pulp fibers were modified by a layer-by-layer deposition of PVAm/CNF/PVAm. CNF as the middle layer, was adsorbed at varied pH and salt concentration, respectively. One pulp specimen was modified by PVAm/PAA/PVAm as a comparison. All the modified fibers were characterized by the amount of cationic polyelectrolyte (PVAm) adsorbed determined by Antek nitrogen analysis. From calibration based on the same polymer solution, the amounts of PVAm adsorbed shown in Figure 1 were calculated.
The differences in adsorption were due to the differences in the third layer build-up of PVAm since all the samples had the same adsorption of PVAm in the first layer. The PAA/PVAm modified cellulose fiber had a much higher amount of PVAm adsorbed than all the other CNF/PVAm samples. Among the CNF/PVAm samples, the total adsorbed amount of PVAm was higher when the CNF layer was deposited at pH 3.5 than at pH 7.5, except for the CNF layer assembly at 10 mM salt concentration.
Modification of cellulose model surface
Table 2 shown multiple stereo microscopic images of model surfaces of cellulose coated silica wafer modified by PVAm/PAA/PVAm and PVAm/CNF/PVAm. When CNF was used as the middle layer in this 3-layers system, and the assembling conditions i. e. pH and NaCl concentration were varied, which were as the same conditions as fiber modification, different surface topographies can be observed. In the images c, e, g and h that the CNF aggregated and formed “clusters”, in which the CNF was adsorbed at pH 3.5 and 100 mM salt concentration, and much less aggregates were observed at pH 7.5 at 10 mM salt concentration, whereas no CNF aggregates were observed at pH 7.5 at 1 mM as well as in PVAm/PAA/PVAm one.
Polyelectrolyte adsorption is commonly referred to as an electrostatically driven process, but the underlying thermodynamic driving force is the entropy gain when the counter-ions to the charges of the solids and the polyelectrolytes are released upon adsorption (Swerin and Wågberg 1994). The properties of the modified substrates depend on the choice of polymer and on the process parameters, such as salt concentration and pH. Adsorption of PVAm onto a fiber surface can be induced by increasing the salt concentration, since this decreases the near-neighbour repulsion of the adsorbed charge (Xie et al. 2013) and the pH is also crucial since PVAm is a weak polyelectrolyte. Previously, Illergård et al. have studied the adsorption of PVAm onto a fiber and layer-assembly with PAA, and then found an optimized adsorption condition at pH 9.5 with a 100 mM NaCl salt concentration (Illergård et al. 2010). The adsorption of CNF was affected by the colloidal stability of the aqueous CNF dispersion, which depended on the pH and salt concentration respectively. According to Fall et al., CNF aggregation can be observed at low pH as well as an elevated salt concentration (Fall et al. 2011). When the electrostatic repulsion between fibrils decreases, at low pH, the increased protons close to the CNF fibril lead to the protonation of carboxyl groups and reduces the surface charge. As shown in Table 2, at a salt concentration of 1 mM and 10 mM, the aggregation was observed on surfaces at lower pH value (3.5). When salt was introduced, it enhanced the ion concentration led to a reduction of the thickness of the diffuse layer near the fibril and resulted in a weakened long-range repulsion between the fibers. Thus, it explained why the large amount of aggregation was only found on CNF deposited in the salt concentration as high as 100 mM at a pH of 7.5. The total amount of PVAm adsorbed onto the fiber in the first and third layers adsorbed on fiber was determined by Antek nitrogen analysis. Since all the pulp samples started with the same adsorption of PVAm as the first layer, the difference in the adsorbed amounts presented in Figure 1 can be interpreted as differences in PVAm adsorption in the third layer. As shown in Figure 1, highest adsorbed amount of PVAm in the third layer was presented when using PAA as the middle layer. PAA and PVAm are weak polyelectrolytes, their charge densities can be altered at different pH. During the adsorption of PAA, the pH was 3.5 and salt concentration was 100 mM, the polymer was more coiled, and more was adsorbed due to a lower charge. By increasing pH to 9.5, charge of PAA was increased and the polymer chain became more extended, whereas PVAm will lose charge and become more coiled. Thus, at a salt concentration of 100 mM NaCl, although there was a larger charge screen effect, higher amount of coiled PVAm with lower charge density were adsorbed on a higher charged PAA, and this recharging adsorption kinetics was reported by Illergård et al. (2010). However, this effect was not so obvious in CNF sample at the same condition, since at this high salt concentration, because the CNF at pH 3.5 was preferable to form larger aggregates and not as easy to alter the conformation as polymer, the higher adsorbed amount of PVAm could be due to a higher surface area. To confirm the multilayer formation and to understand the adsorption behaviour under different conditions, the consecutive depositions of PVAm and CNF under different adsorption conditions were monitored by a quartz crystal microbalance with dissipation (QCM-D), in which the change in oscillation frequency and energy dissipation were recorded in situ as a function of time. The results are shown in Figure 2. The depositions shown in the figures began with rinsing and CNF after the deposition of the first PVAm layer, so that all four samples began with the same adsorption. Different deposition behaviours of CNF can be observed at different pHs.
Two adsorption patterns can be observed qualitatively from Figures 2 (a), (c) and (b), (d). The change of pH for CNF adsorption made this difference while the change of salt from 1 mM to 10 mM NaCl did not change the adsorption pattern but affect the behaviours. The first main difference between (a) and (b) is the strong decay of the total frequency shown in (a) during the initial rinsing step using 1 mM NaCl and pH 3.5, which was due to the expansion of PVAm polymer chain at lower salt and low pH, in which condition, the amine groups on the polymer chain were protonated and revealed more positive charge that increase the repulsion polymer chain as well as hydrophilicity led to more bounded H2O molecules. This behaviour was confirmed by the great increase of dissipation during the rinsing, and the adsorption layer of PVAm became much softer. This behaviour was not observed in (b) since the pH drop from 9.5 to 7.5 and salt decreasing from 100 mM to 1 mM after rinsing hardly made a difference of adsorbed PVAm polymer chain. Second big difference between (a) and (b) was the CNF adsorption at both pH conditions. As presented in (a), more rigid layer was formed after a slower deposition of CNF since the dissipation shown no sign of change, whereas in (b), faster adsorption and higher frequency change was observed, a softer layer of swollen CNF was deposited according to the increase of dissipation. The adsorption kinetics of (a) relied on the charge of PVAm, on which bound water amount and the charge of CNF, while (b) had less bound water on PVAm, the adsorption kinetics relied more on the charge of CNF. Thus in (a) less charged CNF with less hydrophilicity at pH 3.5 shown a slower and less adsorption because it is harder to overcome the barrier of bound water. In (b), CNF with higher charge and higher swellability adsorbed faster and more swollen CNF layer was deposited. This also caused a desorption of CNF when upcoming increasing of pH, weakly adsorbed CNF was partially rinsed away probably due to the repulsion within the CNF aggregates. The increasing salt concentration of (c) and (d) allowed to a varied adsorption kinetics but not yet alter the adsorption pattern. It could be observed in (c) that, the decay of frequency after the rinsing was not as high as (a), meant the increasing of salt decrease the diffuse ion layer (screen the charge) and weaken the water binding ability of PVAm polymer chain. In addition, during the CNF adsorption, a noticeable drop of dissipation was shown in (c) instead of (a), and the adsorption rate of CNF was much faster in (c) with the fast decay of dissipation, indicated the bound water on PVAm layer was excluded and formed compressed layer. This difference of (a) and (c) could be explained by larger CNF aggregates were much easier to overcome the barrier of less bound water molecule and exclude them to adsorb on the PVAm layer due to the increase of ionic strength, but less CNF was adsorbed due to a weakened electrostatic interaction. In Figures 2 (b) and (d), since the CNF adsorption exclude the effect of bound water layer around PVAm, the adsorption kinetics of CNF were more dominant by its charge and conformation. A much higher frequency change shown in (b) and (d) due to the higher charged swollen CNF at pH 7.5 adsorbed on rigid PVAm layers compared with (a) and (c), and the increase of dissipation confirmed the layer became softer after adsorption. However, the main difference could be observed between (b) and (d) was adsorbed amount and adsorption rate. It was the result of CNF that was more aggregated, but the electrostatic force was weakened at higher salt concentration. Bigger CNF aggregates could carry with more bound water but took longer to be adsorbed. By using QCM with dissipation, the adsorption of PVAm/CNF layer assembly in detail was revealed and helped to find better conditions for LbL modification.
Surface roughness of modified surfaces
Using a model surface for the study made it possible not only to obtain a better visualization, but also to study the surface properties using AFM. In this case, the surface roughness of the model surfaces was determined in fluid and is presented in Figure 3. The roughness of the aggregate regions is clearly higher than that of the flat regions of the CNF/PVAm modified model surfaces.
Bacterial assays on modified cellulose fibers
Bacterial removal by adsorption as well as bacterial growth inhibition tests were firstly carried out on CNF/PVAm modified cellulose fibers for comparison with the previously well-described PAA/PVAm system (Illergård et al. 2011). In Figure 4, it can be seen that when PAA was replaced with CNF, at a higher salt concentration and lower assembly pH, the pulp samples had greater bacterial removal abilities. After the bacterial removal test, 10 % of nutrient broth was added to each batch which was then incubated for 18 hours under continuous shaking. The optical density (OD) was then determined, a greater capability of inhibited bacteria growth being indicated by a lower OD value. The results are presented in Figure 5. The inhibition abilities of CNF/PVAm modified fibers were close to those of PAA/PVAm fibers and the growth of bacteria was dramatically less than in the ‘+’ control, in which no modified fiber was added. It can also be seen that the inhibition ability of each CNF/PVAm sample followed the same trend as in the bacterial removal test, so that the greater the bacterial removal ability the greater is also the inhibition ability.
Visualization of bacteria on modified model surfaces
To visualize the interaction between bacteria and CNF/PVAm modified cellulose fiber, above mentioned cellulose model surfaces were used instead of pulp fiber. Since the model surface is much more homogeneous and flat, which made it possible to easily examine them under the microscope and obtain more valuable information. Figure 6 shown multiple images of fluorescence microscopy images of bacteria adsorbed on modified surfaces. When CNF was used as the middle layer in this 3-layers system, and the assembling conditions i. e. pH and NaCl concentration were varied, different surface topographies were formed. It can be seen in the blue images from Table 2 that the CNF aggregated and formed “clusters” in some of the samples. The bacteria were also accumulated in these aggregate areas as shown in the green images of c, e, g and h in Figure 6. On the rather flat surfaces, b, d and f, bacteria were adsorbed evenly. Adsorbed densities of bacteria on aggregate regions and flat regions can be seen are shown in Figure 7.
Figure 8 shows the SEM images of samples with CNF aggregates. From, to the left, the model surface that modified by CNF as a middle layer at pH 3.5 and 10 mM salt and, on the right, the flat region of the same surface. More bacteria were adsorbed on the CNF aggregate region than on the flat region. All other samples with aggregates showed the same behaviour and the images are shown the supporting information.
Figure 9 shown the morphology of E. coli adsorbed on a CNF/PVAm model surface. Bacteria that were deformed, elongated and burst can be seen in image a, Image b, at a greater magnification, shows that the cells in the middle up and bottom left were completely torn up. The fimbriae of the bacteria cell were attached firmly to PVAm adsorbed cellulose nano-fibrils. The 10-fold bacteria fixed on the CNF/PVAm modified surfaces were also examined under AFM in fluid, and cell damage and elongation can be seen in Figures 9 c, d.
Bacterial interaction on modified surfaces
The antibacterial properties of the modified fibers were evaluated mainly by their ability to remove bacteria from the solution and how well they could inhibit bacterial growth after adding nutrient by counting the bacteria remaining or by monitoring the OD. Bacterial removal was the result of bacterial adsorption onto the modified surface, probably driven mainly by electrostatic interaction between the cationic outer-layer of modified fiber and the anionic bacteria surface (Illergård et al. 2012, Terada et al. 2005). E. coli is a Gram-negative bacterium, which has an outer covering of phospholipids and lipopolysaccharides, and the lipopolysaccharides impart a strongly negative charge to the surface. In present study, the growth inhibition properties presented in Figure 5 show that the effect is related to the bacterial removal ability; the better the removal the better the growth inhibition. This result had also shown in earlier studies (Illergård et al. 2015, Chen et al. 2017). The high bacterial adsorption capacity of the cationic polyelectrolyte modified fiber could interfere with a part of the inhibition test since the test was designed by monitoring bacteria left in the solution, but bacterial growth was inhibited due to the existence of PVAm, which could cause by a lower bacterial viability and damage to the bacteria cell wall.
PVAm with an accessible positive charge was considered as the main reason for the adsorption of bacteria, in this study, it was found that the adsorption was further enhanced by introducing a CNF as the middle layer during layer-by-layer deposition. Adsorbed amount of PVAm on outmost layer was higher in PVAm/PAA/PVAm modified fibers than that in PVAm/CNF/PVAm ones as shown in Antek results. If we assumed higher PVAm adsorption led to a higher charge, there should be more bacteria adsorbed on PAA one, on contrary, lower adsorbed amount of PVAm on the CNF/PVAm modified samples had the same or even stronger bacterial adhesion ability. The introduction of CNF to the system was most probably the reason for this phenomenon. This contradictory result suggested that, either PVAm adsorbed on CNF had higher accessible charge, or apart from electrostatic interaction, some other factor also influenced the bacterial adhesion. Similar results were reported by Henschen et al., who found the cellulose substrate with structures modified by 1 and 3 layers PVAm comprising PAA as middle layer had a better bacterial adhesion activity, even though they had a lower total adsorbed amount of PVAm than the ones without structures. A possible explanation is that, at a lower pH and higher salt concentration, CNF tended to form aggregates and to create a structured surface, which could enhance the interaction with the bacteria cell (Henschen et al. 2017). It is difficult to study this at a fundamental level on wood fibers since they are too porous and heterogeneous. However, applying the dissolved cellulose fiber as a coating on the surface of a silica wafer solved these difficulties. These cellulose model surfaces were much more homogeneous, and the aggregates formed during the multilayer build-up was easily seen. At a lower pH and higher salt concentration, aggregates were formed during the adsorption of CNF, and CNF adsorption on fiber samples under these conditions also showed a better bacterial adhesion. SEM images in Figure 8 had also showed more bacteria adsorption in the aggregate region. This behaviour could be accumulated and resulted in a higher adsorption of bacteria on the CNF/PVAm modified fiber, while the existence of CNF as the middle layer created surfaces with a different topography. The aggregation of CNF made the 2D surfaces into a 3D structure that increased the contact area as well as the roughness. Regions consisting of CNF aggregates that had a greater roughness also adsorbed more bacteria than the flat ones. By comparing the bacteria removal (bacteria adsorption capability) on the fiber with the roughness it was found that the surfaces with aggregates had the higher roughness, and also had a larger number of bacteria adsorbed than the lower roughness PAA/PVAm. The bacterial adsorption on the fibers correlated with the surface roughness. Similar results were also reported by Ji et al., who described that a structured surface with higher roughness allows more bacterial adsorption (Ji et al. 2015). This could possibly explain why CNF/PVAm modified pulps had comparable and even better antibacterial properties than the sample modified by PAA/PVAm, even with a lower total adsorption of PVAm than that of PAA/PVAm. The accessible positive charge could be another reason which will be discussed in our future work.
Better antibacterial features were achieved on a cationic PVAm layer-by-layer modified cellulose fibers by replacing the anionic middle layer of petroleum-based PAA with wood-fiber-based cellulose nanofibrills. The three-layer assembly of CNF/PVAm on cellulose fibers had a lower total adsorption of the antibacterial agent PVAm than the PAA/PVAm but had a better interaction with bacteria. More bacterial adhesion occurred on CNF aggregates, and this correlated with a higher surface roughness according to AFM. Thus, the higher bacterial adhesion on CNF/PVAm modified fiber was probably due to the greater roughness caused by CNF aggregates. Bacterial adhesion on surfaces without aggregates also correlated positively with the surface roughness. The improvement of bacterial adsorption was observed possibly due to the increase in surface roughness resulting from the introduction of CNF to the system. This work has shown the potential to make a more bio-based antibacterial alternative by replacing PAA with CNF in layer-by-layer assembly. Future work should be focused on some mechanisms studies to investigate the effects of different factors; therefore, we can have a better control of them to achieve a better antibacterial property especially combining with a more sustainable, natural-resource-based modification.
Funding source: China Scholarship Council
Award Identifier / Grant number: 201407930001
Funding statement: We thank the Chinese Scholarship Council (201407930001) for financial support and RISE Bioeconomy for technical support with the nitrogen analysis.
We are grateful to Prof. Lars Wågberg and Dr. Josefin Illergård for helpful discussions during these studies.
Conflict of interest: The authors declare no conflicts of interest.
Chen, C., Illergård, J., Wågberg, L., Ek, M. (2017) Effect of cationic polyelectrolytes in contact-active antibacterial layer-by-layer functionalization. Holzforschung 71:649–658.10.1515/hf-2016-0184Search in Google Scholar
Fall, A.B., Lindström, S.B., Sundman, O., Ödberg, L., Wågberg, L. (2011) Colloidal stability of aqueous nanofibrillated cellulose dispersions. Langmuir 27:11332–11338.10.1021/la201947xSearch in Google Scholar
Henschen, J., Larsson, P.A., Illergård, J., Ek, M., Wågberg, L. (2017) Bacterial adhesion to polyvinylamine-modified nanocellulose films. Colloids Surf. B, Biointerfaces 151:224–231.10.1016/j.colsurfb.2016.12.018Search in Google Scholar
Illergård, J., Enarsson, L.E., Wågberg, L., Ek, M. (2010) Interactions of hydrophobically modified polyvinylamines: adsorption behavior at charged surfaces and the formation of polyelectrolyte multilayers with polyacrylic acid. ACS Appl. Mater. Interfaces 2:425–433.10.1021/am9006879Search in Google Scholar
Illergård, J., Römling, U., Wågberg, L., Ek, M. (2012) Biointeractive antibacterial fibres using polyelectrolyte multilayer modification. Cellulose 19:1731–1741.10.1007/s10570-012-9742-0Search in Google Scholar
Illergård, J., Wågberg, L., Ek, M. (2011) Bacterial-growth inhibiting properties of multilayers formed with modified polyvinylamine. Colloids Surf. B, Biointerfaces 88:115–120.10.1016/j.colsurfb.2011.06.023Search in Google Scholar
Illergård, J., Wågberg, L., Ek, M. (2015) Contact-active antibacterial multilayers on fibres: a step towards understanding the antibacterial mechanism by increasing the fibre charge. Cellulose 22:2023–2034.10.1007/s10570-015-0629-8Search in Google Scholar
Ji, Y.W., Cho, Y.J., Lee, C.H., Hong, S.H., Chung, D.Y., Kim, E.K., Lee, H.K. (2015) Comparison of Surface Roughness and Bacterial Adhesion Between Cosmetic Contact Lenses and Conventional Contact Lenses. Eye Contact Lens 41:25–33.10.1097/ICL.0000000000000054Search in Google Scholar
Kugler, R., Bouloussa, O., Rondelez, F. (2005) Evidence of a charge-density threshold for optimum efficiency of biocidal cationic surfaces. Microbiology 151:1341–1348.10.1099/mic.0.27526-0Search in Google Scholar
Lichter, J.A., Rubner, M.F. (2009) Polyelectrolyte multilayers with intrinsic antimicrobial functionality: the importance of mobile polycations. Langmuir 25:7686–7694.10.1021/la900349cSearch in Google Scholar
Milović, N.M., Wang, J., Lewis, K., Klibanov, A.M. (2005) Immobilized N-alkylated polyethylenimine avidly kills bacteria by rupturing cell membranes with no resistance developed. Biotechnol. Bioeng. 90:715–722.10.1002/bit.20454Search in Google Scholar
Murata, H., Koepsel, R.R., Matyjaszewski, K., Russell, A.J. (2007) Permanent, non-leaching antibacterial surface–2: how high density cationic surfaces kill bacterial cells. Biomaterials 28:4870–4879.10.1016/j.biomaterials.2007.06.012Search in Google Scholar
Mustafa, K., Lopez, B.S., Hultenby, K., Wennergberg, A., Arvidson, K. (1998) Attachment and proliferation of human oral fibroblasts to titanium surfaces blasted with TiO2 particles. A scanning electron microscopic and histomorphometric analysis. Clin. Oral Implants Res. 9:195–207.10.1034/j.1600-0501.1998.090307.xSearch in Google Scholar
Swerin, A., Wågberg, L. (1994) Size-exclusion chromatography for characterization of cationic polyelectrolytes used in papermaking. Nord. Pulp Pap. Res. J. 9(1):18.10.3183/npprj-1994-09-01-p016-017Search in Google Scholar
Terada, A., Yuasa, A., Tsuneda, S., Hirata, A., Katakai, A., Tamada, M. (2005) Elucidation of dominant effect on initial bacterial adhesion onto polymer surfaces prepared by radiation-induced graft polymerization. Colloids Surf. B, Biointerfaces 43:99–107.10.1016/j.colsurfb.2005.03.016Search in Google Scholar
Wågberg, L., Decher, G., Norgen, M., Lindström, T., Ankerfors, M., Axnäs, K. (2008) The build-up of polyelectrolyte multilayers of microfibrillated cellulose and cationic polyelectrolytes. Langmuir 24:784–795.10.1021/la702481vSearch in Google Scholar
Xie, F., Nylandeer, T., Piculell, L., Utsel, S., Wågberg, L., Åkesson, T., Forsman, J. (2013) Polyelectrolyte Adsorption on Solid Surfaces: Theoretical Predictions and Experimental Measurements. Langmuir 29:12421–12431.10.1021/la4020702Search in Google Scholar
© 2018 Walter de Gruyter GmbH, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 Public License.