Three-Dimensional Organization of Self-Encapsulating Gluconobacter oxydans Bacterial Cells.

Self-organized bacteria have been the subject of interest for a number of applications, including the construction of microbial fuel cells. In this paper, we describe the formation of a self-organized, three-dimensional network that is constructed using Gluconobacter oxydans B-1280 cells in a hydrogel consisting of poly(vinyl alcohol) (PVA) with N-vinyl pyrrolidone (VP) as a cross-linker, in which the bacterial cells are organized in a particular side-by-side alignment. We demonstrated that nonmotile G. oxydans cells are able to reorganize themselves, transforming and utilizing PVA-VP polymeric networks through the molecular interactions of bacterial extracellular polysaccharide (EPS) components such as acetan, cellulose, dextran, and levan. Molecular dynamics simulations of the G. oxydans EPS components interacting with the hydrogel polymeric network showed that the solvent-exposed loops of PVA-VP extended and engaged in bacterial self-encapsulation.

Introduction

Despite recent advances and technological innovations in microbial electrochemical technologies, including the bioproduction of biofuel, waste treatment, and biosensing,[1−4] poor electron transfer between microbes and electrodes is a major concern and a limiting factor for large-scale applications of the technology.[2−4] In light of this drawback, intensive research has concentrated on understanding the mechanisms responsible for the extracellular electron transfer (EET) process and on methods by which electrocatalytic bacteria communicate.[2,4−9] It has been discovered that bacteria are able to form “pili-like” nanowires that facilitate an EET process by linking the respiratory chains located on membranes of bacterial cells to adjacent external surfaces, such as oxidized metals in the natural environment, or to engineered electrodes in renewable energy devices.[4,6,8,9] Bacteria utilize mainly (1) indirect pathways involving redox-active mediators, (2) primary metabolites or other intermediates, or (3) direct modes, involving physical contact in which naturally occurring outer-membrane c-type cytochromes shuttle electrons for the reduction or oxidation of electrodes.[10,11] Understanding the electron transfer mechanism is important in allowing the optimization of the performance of microbial fuel cells.

Geobacter sulfurreducens bacterial cells have been reported to form pili that serve as biological nanowires that allow the transfer of electrons from bacterial cell surfaces to the surface of Fe(III) oxides.[6,7] Another metal-reducing bacteria, Shewanella oneidensis MR-1,[8,9] was observed to form nanowires from the extension of their bacterial outer membrane and periplasm. It was also reported that S. oneidensis MR-1 cells that had been treated with cisplatin produced elongated cells, which showed approximately a five-fold improvement in current densities compared to normal, untreated cells.[12] Furthermore, filamentous bacteria of the family Desulfobulbaceae have been found to conduct electricity over centimeter-long distances, thereby coupling the processes of oxygen reduction at marine sediment surfaces with sulphite oxidation in subsurface layers.[5,13]

Gluconobacter oxydans cells require redox shuttles to assist the transfer of electrons from the active sites of cells to electrodes.[1,14−17] This Gram-negative, nonmotile bacterium produces periplasmic membrane-bound pyrroloquinoline quinone (PQQ), which contains complex enzymes that allow the efficient oxidation of a number of substrates.[1,14−16,18−20]G. oxydans is a promising candidate for the construction of efficient microbial fuel cells because of the periplasmic localization of active sites of PQQ-dependent redox enzymes.[1,14−17,20] In particular, G. oxydans encapsulated in a confined space showed enhanced levels of electric power generation.[14] It remains unclear, however, how the G. oxydans cells were encapsulated within polymeric hydrogels. In this paper, we report on the three-dimensional (3D) organization of G. oxydans subsp. industrius B-1280 cells that are encapsulated within the confined environment of a hydrogel system constructed from poly(vinyl alcohol) cross-linked with N-vinyl pyrrolidone (VP).[16] The proposed mechanism is supported by experimental evidence in addition to molecular dynamics simulation for four exopolysaccharides (acetan, cellulose, dextran, and levan) present on the outer cell wall of the G. oxydans cells.[15,18] Synthetic and analytical chemistry techniques, including ultra-small-angle neutron scattering (USANS), biological assays and computational modeling were used to demonstrate that G. oxydans cells were engaged in self-encapsulation and were able to construct a network by utilizing the molecular components of the PVA–VP hydrogel. Specifically, we showed that interactions taking place between bacterial extracellular polysaccharide components and surface loops of the PVA–VP hydrogel had the greatest influence over the structure of resultant hydrogel and were able to transform its original three-dimensional structure into a well-organized biological network.

Results and Discussion

Three-Dimensional Organization of the PVA–VP Hydrogels

To understand the way in which the G. oxydans cells are encapsulated within the hydrogel, an understanding of the structure of original hydrogel polymeric network is essential. Therefore, the hydrogel was synthesized through the free radical polymerization of linear PVA using VP as a cross-linker.[14] Ceric ammonium nitrate (CAN) was used to initiate the chain reaction, causing the formation of the free radical species shown in Figure A. The free oxygen radicals of PVA were then cross-linked with double bonds present on VP monomers, leading to the formation of a three-dimensional porous polymeric network, as confirmed by scanning electron microscopy (SEM) analysis of freeze-fractured hydrogel samples (Figure B). X-ray diffraction was previously used to obtain an insight into the process by which PVA hydrogel porous structures were formed.[21] The results of this study indicated the presence of entangled swollen amorphous and crystalline domains of PVA that acted as “bulges” within the gel network. Raman microspectroscopic imaging of samples also confirmed the formation of a PVA–VP porous structure within hydrogel films in their hydrated state (Figure B and Figure S1A,B, Supporting Information). The estimated pore diameter distribution of the resultant hydrogels is presented in Figure C. It was found that PVA–VP hydrogels exhibited a pore diameter of 0.92 ± 0.38 μm and a swelling ratio of 410% (Figure S1C, Supporting Information), most likely due to the low proportion of crystalline segments being present in PVA–VP.[22,23]

Figure 1

Synthetic pathway and physical characteristics of porous poly(vinyl alcohol) (PVA) of molecular weight 67 kDa cross-linked with VP. (A) Schematic diagram showing the radical polymerization between linear polymeric PVA and VP monomer to form PVA–VP hydrogel. (B) Porous structure of the hydrogel visualized in its freeze-dried and hydrated states using SEM (scale bar 5 μm) and Raman microspectroscopy (scanning areas of 5 μm × 5 μm). (C) Pore diameter distribution of PVA–VP estimated using ImageJ.

G. oxydans Bacterial Cell Network Formation

The three-dimensional organization of G. oxydans cells in the PVA–VP hydrogels was visualized using SEM, confocal laser scanning microscopy (CLSM), and Raman microspectroscopy (RM) (Figure ). The examination of the freeze-fractured hydrogels containing the bacterial cells showed that the original three-dimensional structure of hydrogels had been completely altered. In the hydrogel environment, G. oxydans cells were found to have self-organized into a network (Figure A), in contrast to the structure adopted in the absence of PVA–VP (Figure ). CLSM and Raman microspectroscopy image analyses confirmed that the G. oxydans cells had formed a network in the hydrogel (Figure B,C). The average biological cluster lengths were found to be 10.5 ± 6.4 μm, as inferred from the ImageJ analysis data (Figure D).

Figure 2

Self-assembly of G. oxydans bacterial cells into the biological network in the PVA–VP hydrogel system. (A) SEM showing the reorganization of G. oxydans cells into clusters wrapped with the polymer (blue highlight indicates the continuous chains of cells). (B) Confocal laser scanning microscopy images, with viable cells stained green with SYTO 9 and nonviable cells stained red with propidium iodide and (C) Raman microspectroscopy images (scanning areas of approximately 5 × 5 μm2) showing G. oxydans cell clusters. (D) Cluster length distributions of G. oxydans bacterial cells showing that cells formed clusters in 67 kDa PVA–VP hydrogel. The kinetics of self-organization of G. oxydans cells in PVA–VP hydrogel was investigated using time-lapsed microscopy (E) CLSM and (F) USANS. (G) Schematic diagram illustrating the re-arrangement of bacteria disrupting the polymeric network.

The dynamic formation of the network by G. oxydans cells in the PVA–VP hydrogel system was further investigated using time-lapsed CLSM and USANS (Figure E,F). Over the period of an hour, short clusters of 4–5 cells were found to have formed (Movie S1, Supporting Information). A further elongation of the short clusters continued after this time for the next 24 h, resulting in the formation of larger self-assembled clusters (Figure F, Figure S2A and Movie S2, Supporting Information). The analysis of the USANS spectra showed that the increased intensity of scattering lengths (q) ranged from 0.5 × 10–4 to 1.0 × 10–4 Å–1 over the 24 h period, indicating that the G. oxydans cell clusters had become larger. Because of the complexity and polydispersity of the globular knot complex structures of PVA–VP together with G. oxydans, further resolving the scattering patterns of USANS was not feasible. However, the change in scattering intensity gradients of the neutron beam clearly indicated the cluster formation by G. oxydans cells. A similar approach was previously used to identify the formation of the protein clusters, such as those of monoclonal antibodies.[24]

To understand the way in which the G. oxydans cells interact with the PVA–VP hydrogel system, the hydrogel was labeled with polyethylene glycol (PEG)-functionalized fluorescent silica nanoparticles (SiO2 NPs) as described elsewhere[25,26] and in Supporting Information S1.1. The NPs were bound to the OH groups present on the PVA–VP (Figure A). The location of the binding sites of the SiO2 NPs on the PVA–VP was visualized using CLSM. The CLSM micrographs confirmed the presence of the original porous three-dimensional (3D) organization of the cells, as highlighted by the various positions of the SiO2 NPs (Figure A). After the addition of the G. oxydans cells, however, the SiO2 NPs were found to have completely dissociated from the PVA–VP hydrogel polymeric chains (Figure A). These results suggested that the G. oxydans cells replaced the SiO2 NPs to occupy the binding sites present in the PVA–VP (Figure B). Notably, after over 70 h of analysis, no statistically significant change was observed in the pH level of the PVA–VP hydrogel system (Figure S2B, Supporting Information), thus eliminating the possibility that an alteration in the 3D organization of the hydrogels with bacterial cells could have occurred as a consequence of changes in pH.

Figure 3

Competitive interactions of bacterial cells and PEG-functionalized SiO2 nanoparticles with the PVA–VP hydrogel. (A) CLSM micrographs showing that G. oxydans bacterial cells (right image, green color) completely disrupted the structure of original hydrogel (left image), fluorescent SiO2 nanoparticles (SiO2 NPs) (red color) labeled hydroxyl groups of the PVA–VP hydrogel. (B) Schematic diagram of the hypothetical mechanism by which hydrogel polymeric chains and G. oxydans cells interact. Disruption of hydrogen bond formation between PEG-functionalized SiO2 NPs and PVA–VP hydrogel and hydrogen bond formation between extracellular polysaccharides of G. oxydans cells and PVA–VP.

The examination of the CLSM micrographs revealed that the G. oxydans cells had secreted an extracellular polymeric material (Figure S1E, Supporting Information) and had adopted a characteristic but partial side-by-side orientation within the clusters being formed (Figure B and Figure S1F, Supporting Information). Markedly, when Escherichia coli K12 cells were introduced into the PVA–VP hydrogel, no self-organized clusters were detected (Figure S3, Supporting Information). To the best of our knowledge, a side-by-side alignment in a hydrogel matrix has only been reported for nonbiological, ellipsoidal microparticles.[27] Examples of self-oriented electroactive cells reported in the literature include the pili or pili-like nanowires formed by S. oneidensis(8,9,12,28) and filamentous bacteria of the family Desulfobulbaceae, which can undergo cable-like alignment.[5,13] It has also been reported that nonelectroactive bacteria such as Lactobacillus fermentum and Bifidobacterium breve have been able to align themselves in response to the application of an external magnetic field after being coated with maghemite nanoparticles.[29] In addition to these reports, some artificial systems have been developed to mediate the aggregation of nonelectroactive bacteria through the modification of the surfaces of bacterial cells.[30,31]

Biocatalytic Activity of G. oxydans Cell Clusters

The encapsulated G. oxydans cell clusters in PVA–VP hydrogels required only 20–30 min to generate 90.0 mV/mg, whereas freely suspended bacteria generated lower electricity (48.0 mV/mg) after 60–70 min (Table S1, Supporting Information). The microbial fuel cell containing the encapsulated G. oxydans cells was able to maintain the generated electrical potential over a period of 8 days (Figure S4, Supporting Information), whereas the electric potential generated by the suspended cells has been reported to be stable only for a 24 h period. The conductivity of the bacterial network was estimated to be 0.32 mS/m, which is much higher than that of freely suspended bacteria (0.11 mS/m) (Table S1, Supporting Information). Cyclic voltammograms (CVs) were recorded using bare graphite electrodes with freely suspended bacteria in the electrochemical cell and at the PVA–VP-encapsulated G. oxydans-modified electrode (Figure ). These voltammograms highlighted a significant difference between the two systems. It is suggested that the improved performance of the microbial fuel cells constructed using the encapsulated G. oxydans clusters resulted in the increased efficiency in the transfer of extracellular electrons between bacterial clusters.

Figure 4

Cyclic voltammograms (CV) of suspended and encapsulated G. oxydans. The cyclic voltammogram (CV) of a (A) bare graphite electrode with free-suspended bacteria (30 mg) and 2,6-DCPIP (0.08 mM) at the electrochemical cell, (B) bare graphite electrode with free-suspended bacteria (30 mg), 2,6-DCPIP (0.08 mM) and glucose (0.01 M), (C) PVA–VP/G. oxydans-modified electrode bacteria (30 mg) with 2,6-DCPIP (0.08 mM), (D) PVA–VP/G. oxydans-modified electrode (30 mg) with 2,6-DCPIP (0.08 mM) and glucose (0.01 M), electrolyte: 30 mM phosphate buffer at pH 6.0.

Molecular Dynamics Simulation of G. oxydans Extracellular Polysaccharide Components Interacting with the Hydrogel Polymeric Network

The outer cell wall of the G. oxydans cells is surrounded by an extracellular polysaccharide (EPS) capsule, which typically contains several polysaccharide units, including acetan, cellulose, dextran, and levan.[15,18] Because the contact between the G. oxydans cells and the PVA–VP hydrogel occurs via the EPS, all-atom model heptamers of the four EPS components were constructed and solvated with explicit water, and their interaction with the model PVA–VP hydrogel was then simulated as a function of time (Figure A and Table S2, Supporting Information).

Figure 5

Modification of PVA–VP/water interfaces by different bacterial extracellular polysaccharide components (acetan, cellulose, dextran, and levan). (A) Model structures of four typical polysaccharide components (acetan, cellulose, dextran, and levan) interacting with hydrated PVA–VP polymeric networks (oxygen atoms are colored red, PVA–VP carbon atoms colored cyan, and acetan, cellulose, dextran, and levan carbon atoms colored blue, green, tan, and pink, respectively). (B) Atomic density profile of the PVA–VP/water interface showing the change in atomic density taking place when extracellular polysaccharide components (acetan, cellulose, dextran, and levan) are present compared to PVA–VP alone.

The mesoporous poly(vinyl alcohol)–N-vinylpyrrolidone (PVA–VP) hydrogel model was constructed from 5 atactic oligomers of 23 monomer units with a molar ratio of 22:1 VA/VP, which were randomly packed into a confined layer using the Theodorou–Suter algorithm.[32,33] Following packing, the oligomer termini were attached to the nearest adjacent terminus to create an infinite polymer surface. The system was then solvated with a water volume to reproduce the experimental concentration of polymer in water of 54.9 mg/mL and simulated using molecular dynamics (MD) for 200 ns to relax the system. The heptamers of acetan, bacterial cellulose, dextran, and levan were constructed, solvated, and simulated for at least 50 ns to relax the structures. Following relaxation of the polymer hydrogel and polysaccharide heptamers, each polysaccharide was positioned in a 40 × 50 × 70 Å3 unit cell containing the solvated polymer with a minimum distance of 10 Å to the polymer surface to examine the polymer–hydrogel interactions.

Of the four model EPS components, acetan was found to have the greatest influence over the structure of the hydrogel (Figure B). Specifically, the solvent-exposed loops of the PVA–VP (shown in light green in Figure B and Movie S3, Supporting Information) extended away from the hydrogel surface and bulk phase, and polymer–polymer interactions were replaced by polymer–EPS interactions. The mechanism of the PVA–VP deformation was characterized by the presence of initially metastable direct and water-mediated hydrogen bonds being formed between the acetan and a loop of the PVA–VP (Figure ). The formation of hydrogen bonds coincided with a breakage of the hydrogen bonds between the exposed polymer loops and the polymer surface, which was accompanied by an increase in the conformational entropy as the loop moved away from the bulk of the polymer. This was followed by additional rearrangements of the loop to maximize the short-range interactions with the acetan molecules. The strong specific interactions taking place between the acetan (and to a lesser extent the other three EPS components) and the loops of the PVA–VP through hydrogen bond formation and van der Waals forces (Figure and Movie S3, Supporting Information) may explain how the bacterial outer layer was able to interact with and modify the structure of the PVA–VP as well as displace the SiO2 NPs from the hydrogel.

Figure 6

Hydrogen bond counts between a loop of PVA–VP and the PVA–VP bulk, acetan, and water. (A) Hydrogen bond formation between the PVA–VP loops and acetan as a function of time. (B) Corresponding events depicting the hydrogen bond formation taking place between acetan and PVA–VP loops on a molecular scale (oxygen atoms are colored red, PVA–VP carbon atoms are colored green, acetan carbon atoms colored blue, and hydrogen bonding colored purple with dashed lines).

Of the four polysaccharides studied, acetan was found to perturb the structure of the PVA–VP hydrogel to the greatest extent, followed by cellulose (Figure , Table S2, Supporting Information). Levan exhibited the least effect on the hydrogel. Cellulose remained the most extended polysaccharide throughout the simulations, maintaining a radius of gyration (Rg) of 10.6 ± 0.3 Å, followed by acetan (7.4 ± 1.4 Å). Acetan, however, exhibited the greatest fraction of the contact surface area with the PVA–VP and provided the highest population of polysaccharide–polymer hydrogen bonds of any of the polysaccharides being studied (Table S2, Supporting Information). Interestingly, the interaction energy decomposition analysis indicated that only approximately 25% of the interaction enthalpy between the acetan and PVA–VP was due to the contribution from electrostatic interactions, whereas the remaining contribution was due to van der Waals interactions. We emphasize that this is a conceptual and representative model of the interactions between the polysaccharides present in the bacterial outer layer and the PVA–VP hydrogel, and all four polysaccharides studied exhibited similar behavior and are capable, in principle, of modifying the polymer structure to some extent. It is likely that all bacterial polysaccharides act cooperatively, causing the withdrawal of individual chains from the polymer bulk, effectively “melting” the polymer network and creating an arrangement favorable to the bacteria.

Conclusions

To the best of our knowledge this is the first study that reports the self-organization of G. oxydans cells into a biological network by actively utilizing the properties of the PVA–VP hydrogel system in which it is contained. Here, we demonstrated that the self-organization of the G. oxydans cells occurs through the formation of an extended network of bacterial cells aligned in a partial side-by-side configuration to form biological clusters. This particular orientation was most likely adopted to facilitate the efficient transfer of electrons, because it results in an increased interaction contact area. The alignment of the G. oxydans cells in the polymeric hydrogel allowed the microbial fuel cells containing them to generate a greater electric potential than that obtained by suspended G. oxydans cells in the absence of the hydrogel. The power generation of the PVA–VP/G. oxydans system was found to be stable for more than 8 days. A remarkable feature of the G. oxydans cells is their ability to utilize the PVA–VP hydrogel system to self-encapsulate and self-organize into an extended network via strong specific intermolecular interactions between the polysaccharides present on the exopolysaccharide capsules and the exposed loops of the PVA–VP polymer. This process has the potential to be exploited in the novel design of highly efficient electricity generation cells that utilize components of industrial organic waste.

Materials and Methods

Hydrogel Synthesis

Poly(vinyl alcohol) samples, with molecular weights of 67 kDa (Sigma-Aldrich), were used to form a hydrogel using VP as the cross-linking agent. More details regarding the synthetic procedure for these hydrogels is described in Supporting Information S1.1.

Poly(ethylene oxide)-functionalized fluorescent nanoparticles (of diameter approximately 15 nm) were prepared and incorporated into the hydrogel as detailed in Supporting Information S1.1. These were used to characterize the porous cellular microstructure of the hydrogels and to investigate the interactions taking place between the bacteria and the hydrogel.[26] The swelling behavior of the PVA–VP hydrogels in Milli-Q water was studied and reported in terms of the swelling ratio (Q), where Q = swollen mass/initial dry hydrogel mass.[26] The detailed procedure is described in Supporting Information S1.1. The synthesis of PVA–VP and nanoparticle-immobilized PVA–VP is described elsewhere[14,26] and detailed in Supporting Information S1.2.

Encapsulation of G. oxydans and E. coli Bacterial Cells in Hydrogels

G. oxydans subsp. industrius VKM B-1280 was obtained from Russian Culture Collection from Skryabin Institute of Biochemistry and Physiology of Microorganisms, Russian Academy of Science. The G. oxydans cells were grown at 30 °C in a medium of the following composition: 200.0 g/L d-sorbitol (Sigma-Aldrich) and 20.0 g/L yeast extract (BD, USA). E. coli K12, used for comparative purposes in CLSM studies, was grown in nutrient broth (Oxoid). The bacterial cells were stored in stock solutions prepared with the same medium containing 20% glycerol. The stock samples (1 mL tubes) were stored at −80 °C. The bacterial cultures were refreshed prior to use on agar plates incubated at 30 °C for 48 h.

Prior to the encapsulation process, cells were harvested after a 48 h incubation period and washed twice with 18.0 mM phosphate–citrate buffer (pH 6.0). Wet cell pellets were mixed with the hydrogel solutions to obtain a final concentration of 30 mg/mL. The mixture was stirred for 24 h, after which time the mixture was progressively diluted to obtain the final concentration of 6 mg/mL. Hydrogel films were prepared by casting approximately 300 μL of the diluted mixture onto a glass slide and allowing it to dry at 25 °C for 24 h. The same process of E. coli encapsulation was carried out with 10 mM PBS (pH 7.4). A functional analysis of the constructed microbial fuel cell (MFC) with the encapsulated G. oxydans is given in Supporting Information S1.3.

Scanning Electron Microscopy

The synthetic hydrogels prepared with and without the G. oxydans cells were characterized using SEM (ZEISS SUPRA 40VP, Oberkochen, BW, Germany) at 3 kV under 5000×, 10 000×, 50 000× magnification. Freeze-dried and fractured hydrogel films with and without bacteria were mounted on an L-shaped aluminum holder using a conductive tape and gold-sputtered as previously described.[26]

Raman Spectroscopy

Raman microspectroscopy of the PVA–VP hydrogels with and without G. oxydans was mapped using a Raman microspectrometer (WITec) with a 532 nm laser wavelength (hν = 2.33 eV). A 100× objective was used to characterize the hydrogel samples. A grid of 50 spectra × 50 spectra was acquired for a scanning area of 5 μm × 5 μm. The integration time for a single spectrum was 2.0 s. Scanning was repeated independently using three different samples. Visualization of the G. oxydans bacterial cell clusters in the PVA–VP was performed without using the freeze-drying preparation steps used for imaging via CLSM and Raman microspectroscopy.

Ultra-Small-Angle Neutron Scattering Spectroscopy

Ultra-small-angle neutron scattering (USANS) was performed using the Kookaburra beamline at the Australian Nuclear Science and Technology Organisation (ANSTO, NSW, Australia) to determine the microstructure of the PVA–VP hydrogels with and without the G. oxydans being present. Two perfect channel-cut crystals were used to measure the broadening of a monochromatic neutron beam with a wavelength of 4.74 Å, thus measuring the small-angle scattering from a sample placed between them, allowing a range of momentum transfer quantities to be utilized. The scattering length, q, was calculated according to the following equationwhere q (Å–1) is the scattering length; λ (nm) is the wavelength of incident neutrons; and θ (rad) is the scattering angle.

A q range of 3 × 10–5 to 3 × 10–3 Å–1 was used to investigate the self-assembled structure formed between the hydrogel and G. oxydans over a 36 h period. The samples were contained in a 2 mm path length fused quartz cell with an aperture of 35 mm × 35 mm. Data were reduced using standard procedures that involved the determination of the transmission with respect to an empty cell and normalized in addition to subtraction of the background scattering from the empty sample cell. Incoherent scattering from water, which was approximately 0.8 cm–1, was considered an insignificant contribution to the background spectra over the range being studied.

Confocal Laser Scanning Microscopy

Confocal laser scanning microscopy was used to visualize the proportion of live and dead cells using a LIVE/DEAD BacLight Bacterial Viability Kit, L7012, which contained a mixture of SYTO 9 and propidium iodide fluorescent dyes (Molecular Probes, Invitrogen, Grand Island, NY, USA). SYTO 9 permeated both intact and damaged membranes of the cells, binding to nucleic acids and fluorescing green when excited by a 485 nm wavelength laser. The propidium iodide entered only the cells that had sustained significant membrane damage, and hence considered to be nonviable, and binds with a higher affinity to the nucleic acids within the cells than does the SYTO 9. Bacterial suspensions were stained according to the method previously reported[26] and imaged using a FluoView FV10i inverted microscope (Olympus, Tokyo, Japan).

Molecular Dynamics (MD) Model Construction

The published parameters from the condensed-phase optimized molecular potentials for atomistic simulation studies (COMPASS) force field[35] were applied to all systems, with missing angles and dihedrals taken from the polymer consistent force field (PCFF).[36] We have used both COMPASS and PCFF previously to model solvated polymer interactions,[37] and this family of force fields has been shown to be reliable for modeling polymers and interfaces.[38−41] All simulations were run in the MD code LAMMPS[34] in the constant pressure and temperature (NPT) ensemble with the Nosé–Hoover thermostat and barostat to maintain a temperature of 298 K and a pressure of 1 atm and utilized a 1 fs timestep and 10 ps output frequency. Electrostatics beyond a 15.5 Å cutoff were treated with the PPPM summation method with an accuracy of 1 × 10–5 kcal/mol and van der Waals interactions with an atom-based summation utilizing a 15.5 Å cutoff and tail correction. All simulations were run for up to 300 ns, with at least three different initial velocity distributions for each system. Equilibration was defined as the point where the heavy-atom root mean squared deviation (rmsd) reached a steady value with a standard deviation of less than 1 Å. All analyses were performed on 10 ns of equilibrated trajectory. Hydrogen bonds were calculated with a maximum donor–acceptor distance of 2.5 Å and a minimum angle of 120°. Atomic density profiles were created using the VMD Density Profile Tool.[42]