Electrochemical scaffold generates localized, low concentration of hydrogen peroxide that inhibits bacterial pathogens and biofilms

We hypothesized that low concentrations of H2O2 could be generated through the electrochemical conversion of oxygen by applying an electric potential to a conductive scaffold and produce a low, but constant, concentration of H2O2 that would be sufficient to destroy biofilms. To test our hypothesis we used a multidrug-resistant Acinetobacter baumannii strain, because this species is often implicated in difficult-to-treat biofilm infections. We used conductive carbon fabric as the scaffold material (“e-scaffold”). In vitro experiments demonstrated the production of a maximum constant concentration of ~25 μM H2O2 near the e-scaffold surface. An e-scaffold was overlaid onto an existing A. baumannii biofilm, and within 24 h there was a ~4-log reduction in viable bacteria with an ~80% decrease in biofilm surface coverage. A similar procedure was used to overlay an e-scaffold onto an existing A. baumannii biofilm that was grown on a porcine explant. After 24 h, there was a ~3-log reduction in viable bacteria from the infected porcine explants with no observable damage to the underlying mammalian tissue based on a viability assay and histology. This research establishes a novel foundation for an alternative antibiotic-free wound dressing to eliminate biofilms.

in vitro, in vivo and animal model studies 13,14,16,17,19 . Despite the apparent effectiveness of DC, the mechanism by which ES improves wound healing remains unknown 18 . This is best illustrated by the examples summarized in Table 1, which report contradictory conclusions for DC applications.
Previous studies employed a range of DC voltage, current setting, polarity of the electrode placed on a wound infection, length of application time and other variables ( Table 1). As a consequence, it is difficult to draw conclusions about the general effectiveness of DC-based ES as a therapeutic treatment 18 . For instance, an electric current of 32 μ A/cm 2 applied through a copper mesh electrode with negative polarity for 2 h, three times a day, eliminated P. aeruginosa from infected skin ulcers 16 . In contrast, the application of a 52-μ A/cm 2 electric current through the same electrode material with negative polarity required 72 h of continuous application to eliminate P. aeruginosa from an infected wound model successfully 15 . Most investigators speculate that electrical current is responsible for antibacterial effects, but no mechanisms have been confirmed 15,16 . Others have applied DC voltage (3.5 V) to inhibit P. aeruginosa on an electrode surface and speculated that toxic compounds are responsible 20 , but this mechanism has not been confirmed either 13,14,20 . Thus, despite numerous hypotheses regarding the mechanism of action of ES, there is no unifying theory on which to standardize treatments to eliminate biofilm from wound infections or standardize investigations 18 . This lack of understanding likely derives, in part, from too little emphasis being placed on the role of electrochemical processes occurring at an electrode surface applied to a wound.
ES uses two inert electrodes to control and drive electrical current and control biofilm [21][22][23][24] . Until recently, however, the community has lacked the tools and methods to investigate the micro-environmental changes that are caused by electrochemical reactions 22 . Recently, our research group reported that continuous (40 h) electrochemical generation of low concentrations of H 2 O 2 was detected near a stainless steel electrode with negative polarity and that the H 2 O 2 appeared to delay biofilm development 22 . The electrochemical formation of H 2 O 2 results from the partial reduction of dissolved oxygen in an aqueous solution on an electrode as per equation (1) 22,25 . ( ) The reduction potential of H 2 O 2 is + 85 mV Ag/AgCl , but because of its high activation overpotential, H 2 O 2 production usually requires a negative polarization potential 26 . When an electrode in a wound environment is polarized below + 85 mV Ag/AgCl , oxygen will be reduced to generate H 2 O 2 , which can prevent/delay biofilm growth 22 . Depending on the concentration, the electrochemical generation of H 2 O 2 should be compatible with wound healing because a low concentration of H 2 O 2 is normally produced in wounds as a cellular inflammatory response and H 2 O 2 is required for healing 27 , probably through the stimulation of keratinocyte differentiation 28,29 . Clearly, only a low concentration of H 2 O 2 can be tolerated so as to avoid oxidative damage to tissue 30,31 . Also, such electrochemical generation of H 2 O 2 should be continuous over time. Thus, we hypothesized that an electrochemical scaffold capable of continuous controlled delivery of a low concentration of H 2 O 2 can function as an efficient antibiotic-free wound dressing to destroy biofilms. Our goals were to (1) develop an electrochemical scaffold ("e-scaffold") that would reduce dissolved oxygen to H 2 O 2 , (2) test its biocidal efficacy at eliminating A. baumannii biofilms grown in vitro and on porcine explant models, and (3) use a porcine explant model to determine whether the e-scaffold damages underlying tissue.

Results and Discussion
As an e-scaffold material, we chose a carbon-based conductive fabric because of its biological compatibility, flexibility and wide use as both electrode and wound dressing material [32][33][34] . We standardized the operating electric potential for maximum sustained H 2 O 2 concentration generated near the e-scaffold surface, based on microelectrode measurements 22 . Polarized e-scaffold was overlaid onto an existing A. baumannii biofilm for 24 h. Afterward, changes in biofilm surface coverage were quantified from biofilm images 6,35 and the colony-forming units were counted. To verify that the electrochemical generation of H 2 O 2 is the dominant mechanism for biofilm elimination, we neutralized the mechanism in situ by externally adding catalase to decompose the H 2 O 2 generated by an e-scaffold in an identical A. baumannii biofilm. This allowed us to determine whether an e-scaffold can operate without H 2 O 2 production. Similarly, in a separate experiment, we added H 2 O 2 to give concentrations as experimentally observed to be generated by the e-scaffold to confirm that H 2 O 2 is required to damage A. baumannii biofilms. The e-scaffold was further tested against infected porcine explants.

H 2 O 2 is generated at the e-scaffold surface. H 2 O 2 becomes detectable near the e-scaffold surface
when it is polarized at − 300 mV Ag/AgCl (Fig. 1A). This observation is consistent with the two-electron pathway (equation 1) previously described for oxygen reduction at carbon-based electrodes 32 . A maximum H 2 O 2 concentration was detected at − 600 mV Ag/AgCl . Consequently, for all subsequent experiments the e-scaffolds were polarized at − 600 mV Ag/AgCl . Depth profile measurements demonstrated that the H 2 O 2 concentration was ~25 μ M at the e-scaffold surface but declined to almost zero at 300 μ m from the e-scaffold surface (Fig. 1B). Such low concentrations of H 2 O 2 are sufficient to promote wound healing by eliminating biofilm without damaging mammalian tissue 22,30,31 . As expected, the non-polarized e-scaffolds produced no detectable H 2 O 2 (Fig. 1B). We also quantified pH profiles but observed no change in pH near the electrode surface 32 . This demonstrates that pH effects are unlikely to contribute to the e-scaffold antibacterial mechanism as had been suggested by previous papers 24,36 . In vitro biofilm control by e-scaffold. Figure 2A shows sample images of green fluorescent protein (GFP)-expressing A. baumannii biofilms initially and after 24 h for both non-polarized (control) and polarized e-scaffolds. The control biofilms continued to grow into larger biofilm clusters over the 24-h period, whereas the biofilms disappeared after 24 h of exposure to the e-scaffold. These observations are consistent with our previous study demonstrating that biofilm is eliminated from 316L stainless steel surface in the presence of electrochemically generated H 2 O 2 22 . After a 24-h application of the e-scaffold, the biofilm surface was reduced significantly, from 25.0 ± 2.0% to 7.0 ± 2.3% (P < 0.05, Student's t-test), whereas the biofilm coverage increased to 34.0 ± 3.5% for the control biofilms (Fig. 2B). The colony-forming units (log (CFU/cm 2 ) of e-scaffold-treated biofilms decreased to 4.35 ± 0.27; those of the control were 8.29 ± 0.05 (Fig. 2C). Interestingly, several researchers have reported that bacteria are eliminated near electrodes that have a negative DC polarity with a current density similar to that in our experiments (Table 1). Most investigators speculate that this antibacterial effect results from the negative electric current per se. Istanbullu et al. 22 recently reported that H 2 O 2 is generated by a polarized surface, and for the current study, we surmised that this is the mechanism responsible for the beneficial effects of an e-scaffold. As an independent test of this mechanism, we added exogenous H 2 O 2 to A. baumannii biofilms. When 45 mM H 2 O 2 was delivered in a single administration, there was a ~3-log reduction in A. baumannii CFU compared to the control without H 2 O 2 . This magnitude of reduction is similar to that of in situ biofilm reduction (Fig. 3A). However, the application of H 2 O 2 at a concentration > 100 μ M is not practical, as it is reported to be cytotoxic for mammalian cells [37][38][39] . Continuous delivery at lower concentrations over longer periods of time reportedly removes biofilm without damaging mammalian cells 22,30,31 . When an equivalent concentration of H 2 O 2 was delivered continuously over the course of 24 h, we observed only a ~2-log reduction in bacterial counts (Fig. 3B). While this latter experiment was expected to work as well as the e-scaffold, we surmise that the difference occurred because the e-scaffold provides a better distribution of H 2 O 2 by electrotaxis and thus achieves greater biofilm reduction. The two stabilizers used in these experiments (sodium sulfate and manganese phosphate) had no independent effect on the biofilms (Fig. 3A).
We estimated that 0.01 mg/ml catalase was needed for complete decomposition of the H 2 O 2 that was generated by the e-scaffold ( Supplementary Fig. S1B). When a 5-fold greater concentration of catalase was applied to the biofilms, the e-scaffold only produced a ~1-log reduction in the number of viable cells (Fig. 3B), consistent with H 2 O 2 being the principal bactericidal mechanism. The catalase itself had no significant effect on biofilm (Fig. 3B). Because the e-scaffold reduced the number of viable cells by ~4 log (Fig. 3A) and the addition of catalase blocked all but a ~1-log reduction (Fig. 3B), we concluded that the biocidal activity is due to electrochemically generated H 2 O 2 . The estimated 25% difference could be due to electrostatic or electrophoretic effects of negative electric current 40,41 (Fig. 3). Alternatively, it is possible that H 2 O 2 was not eliminated by the catalase enzyme rapidly enough and there was some loss of viable cells.
The e-scaffold eliminates biofilm from infected porcine explants. The efficacy of H 2 O 2 generated from the e-scaffold was tested against infected porcine explants. We observed a ~3-log reduction in colony-forming units (log (CFU/cm 2 ) for e-scaffold treated biofilms (Fig. 4). We also found that 95% of the tissue cells in the e-scaffold treated uninfected explants remained viable compared to the untreated control, as measured using a viability stain (n = 3, P = 0.85, one-way ANOVA). Blinded histological assessments showed that there was no significant damage to the underlying tissue given exposure to the potentiated e-scaffold (Supplementary, Fig. S2 and Fig. S3).
Collectively these results demonstrate that the e-scaffold can reduce a biofilm community by four orders of magnitude by generating reactive H 2 O 2 without apparent damage to the underlying tissue. Others have reported no tissue damage from the direct application of similar concentrations of H 2 O 2 10,30,42 . This work confirmed that the polarized e-scaffold successfully eliminates biofilms and that the electrochemical generation of H 2 O 2 is the dominant mechanism of action. This e-scaffold design requires oxygen to diffuse to the carbon fiber surfaces 22 . If needed, to increase H 2 O 2 production and concentration the e-scaffold could be exposed to an atmosphere enriched with oxygen [43][44][45] . In practice, the electrolyte medium could be replaced to prevent drying and to improve reaction rate while keeping the wound bed moist 9,16,29 .

Conclusions
Considerable attention is being focused on alternative, antibiotic-free biofilm removal strategies. Electrochemically generated H 2 O 2 near biofilm surfaces can eliminate the biofilm, but this requires continuous delivery at low concentrations in clinical settings. We have established the first step for an e-scaffold that continuously generates H 2 O 2 at an applied potential of − 600 mV Ag/AgCl over the course of 24 h. With a maximum sustained concentration of ~25 μ M H 2 O 2 generated at any given time in vitro, the e-scaffold reduced the number of A. baumannii in biofilms by ~4 log. When it was applied to an infected porcine explant there was a ~3-log reduction in the biofilm community without detectable damage to the underlying mammalian tissues. Thus, the e-scaffold eliminates A. baumannii biofilm through a defined mechanism, and deserves further investigation as a wound treatment.

Materials and Methods
Electrochemical scaffold (e-scaffold). The e-scaffold consisted of three electrodes. The working electrode held a negative polarity to reduce oxygen and generate H 2 O 2 . To complete the electrochemical cell, we used a counter electrode and a custom-made Ag/AgCl reference electrode. A custom-built e-scaffold was fabricated using carbon fabric (Panex 30 PW-06, Zoltek Companies Inc., St Louis, MO). The fabric was cut in a circular shape (6.42 cm 2 ) to serve as the e-scaffold, and a smaller circular carbon fabric "patch" (2.14 cm 2 ) was used as the counter electrode. The counter electrode was attached to the e-scaffold using a thin layer (~1 mm) of silicone rubber sealant (DAP Dynaflex 230 Premium Indoor/ Outdoor Sealant, catalog #18357) that provided insulation between the electrodes while still allowing oxygen to diffuse to the bottom surface of the e-scaffold for H 2 O 2 generation.
For controlled generation of H 2 O 2 , precise and accurate control of the potential of the e-scaffold is essential 22 . Therefore, the electric potential applied to the e-scaffold was controlled using a Gamry Series G 300 ™ potentiostat (Gamry Instruments, Warminster, PA, USA) against a saturated Ag/AgCl reference electrode 6 . Ti wires (0.025 Ti, Malin Co., Cleveland, OH, Lot #27567) were used to connect the electrode ends to the external cables leading to a potentiostat (Fig. 5). The connection resistance was consistently < 2 Ω. The e-scaffold was overlaid either onto biofilms grown in vitro on glass surface or onto infected or uninfected porcine explants (Fig. 5). This configuration allowed the ventral surface of the e-scaffold to be exposed directly to biofilms.   For these experiments the e-scaffold was polarized at − 600 mV Ag/AgCl and the average current density was − 56 μ A/cm 2 . The gray patterned bars represent the percent viability of cells in the fresh (t = 0 h), control (untreated, t = 24 h) and e-scaffold treated (t = 24 h) uninfected porcine explants. The data are means from nine three porcine explant each with triplicate measurements, and the error bars represent standard errors of the means calculated from triplicate measurements. No significant difference between control and e-scaffold treated samples was observed (P = 0.85, one-way ANOVA).

Quantifying H
(Zeiss Stemi 2000). The microelectrode tip was then moved down to within ~50 μ m of the e-scaffold surface (Supplementary, Fig. S4). At this position linear sweep voltammetry was initiated from + 400 mV Ag/AgCl to − 800 mV Ag/AgCl at 10 mV/s. The onset of H 2 O 2 production was measured from this voltammetry, and − 600 mV Ag/AgCl was selected as the optimum potential to produce H 2 O 2 near the surface. Following linear sweep voltammetry, the e-scaffold was polarized to − 600 mV Ag/AgCl and the current was allowed to reach a steady value. Starting at 1000 μ m from the e-scaffold surface, the microelectrode tip was stepped down in 5-μ m increments towards the e-scaffold surface. After each increment the H 2 O 2 concentration was measured to develop a depth-resolved concentration profile. The accumulation of H 2 O 2 at the e-scaffold surface and the penetration distance into the bulk were measured using these depth profiles. The depth profile of H 2 O 2 for a non-polarized e-scaffold surface was similarly measured as a control. The profiles were measured on one side since the other side was insulated by a layer of inert silicone rubber sealant and no reaction happens on that surface.

Growing in vitro biofilms. To test the efficacy of e-scaffold at destroying existing biofilm, we used
Acinetobacter baumannii biofilms in this study. An overnight culture of A. baumannii (ATCC #BAA-1605) was grown in full-strength Luria Broth (LB) medium (Sigma-Aldrich, catalog #L3522) and was resuspended in 5% LB medium (OD 600 ≈ 0.5). For imaging experiments, GFP expressing A. baumannii (ATCC #17978) was used and LB medium was supplemented with ampicillin (100 μ g/mL; Sigma-Aldrich, catalog #A5354-10ML). Sterile glass bottom petri dishes (MatTek Corporation, catalog #P35G-1.5-20-C) were used to grow and image the biofilms (Supplementary, Fig. S5A). After 2 h of initial attachment, the bacterial suspension was removed and the biofilms were washed twice to remove planktonic cells and then refreshed with 5% LB medium. Biofilms were allowed to develop for 24 h. Before the e-scaffolds were applied, the bulk liquid was refreshed.
Application of e-scaffold to in vitro biofilms. E-scaffolds were sterilized by autoclaving (121 °C, 15 min) and were saturated with sterilized liquid medium. Then the bulk liquid above the existing A. baumannii biofilms were removed carefully and the e-scaffolds were placed onto it. Fresh medium was added to the system. The scaffolds were then polarized at -600 mV Ag/AgCl for 24 h, after which biofilms were processed for the quantification of viable cells by scraping the e-scaffold and biofilms from the glass surfaces of the petri dish into 5 ml of LB medium (1 g/L). These suspensions were centrifuged, the resulting cell pellet was resuspended in 1 ml of LB medium (1 g/L), and serial dilutions were prepared. Colony-forming units (CFU) were counted using a drop-plate count method 46 . Biofilms exposed to non-polarized e-scaffolds were used as a control.
Biofilm imaging and analysis. An inverted epifluorescence microscope (Nikon Eclipse Ti-S inverted microscope) with a Nikon DS-Qi1Mc camera and a CFI Plan Fluor ELWD 40X objective (N.A. 0.60, W.D. 3.72.7 mm) was used to image the cells. Biofilms were imaged before exposure to the e-scaffold (t = 0 h) and after 24 h of exposure. To remove any planktonic cells, biofilms were washed twice and refreshed with 5% LB medium supplemented with ampicillin prior to imaging. Image Structure Analyzer (ISA) was used to calculate surface coverage by biomass from the digitized biofilm images automatically 6 . At least ten discrete images were taken each time to obtain statistically representative data 47 . Average values were calculated for these ten images. The average values of three biological replicates were used to calculate the means and standard errors. We used surface coverage, which is the ratio of the area of the biomass to the total area of the image, as the main indicator of biofilm structure. The higher the percentage surface coverage, the higher the coverage of the glass surface by biofilms. Figure 5. Schematic of experimental setup for treatment of biofilm exposed to e-scaffold with illustration of electrochemical H 2 O 2 production. Electrodes are connected to a potentiostat (not shown in figure). Microscope glass coverslips and porcine explants were used as biofilm growth surfaces for in vitro and infected porcine explant experiments, respectively.
Scientific RepoRts | 5:14908 | DOi: 10.1038/srep14908 External H 2 O 2 addition. To test whether H 2 O 2 by itself can remove biofilms in vitro and from infected porcine explants, we added exogenous H 2 O 2 (VWR, Catalog #RC3819-16, adjusted to a concentration similar to that produced by the e-scaffold) to A. baumannii biofilms. The total amount of H 2 O 2 generated from the e-scaffold was estimated by charge balance calculations from equation (1) 6 and the integration of current vs. time data observed from the potentiostat (Supplementary, Fig. S1A). First, we challenged the biofilms with the total calculated amount of H 2 O 2 (45 mM) in a single administration. Then, in separate experiments, we added H 2 O 2 continuously (similar to the e-scaffold) to biofilms at an average of 2 mM/h for 24 h. This allowed us to simulate the total amount of H 2 O 2 produced over time by the e-scaffold. To minimize the rapid oxidation of H 2 O 2 we added stabilizers (0.005% sodium sulfate and 0.003% manganese phosphate) to the solution. The stabilizers were also included in a separate control treatment.
We added catalase (Sigma-Aldrich, catalog #C1345) to A. baumannii biofilm and measured its protection against H 2 O 2 produced by the e-scaffold. Prior to the addition, the H 2 O 2 decomposition rate per unit of catalase was determined from H 2 O 2 microelectrode measurements. The total amount of catalase required per min for the complete decomposition of H 2 O 2 was calculated based on the rate of H 2 O 2 generation by the e-scaffold (Supplementary, Fig. S1B). To ensure complete H 2 O 2 decomposition, catalase was added in excess of the calculated value (5×≡0.05 mg/ml). In a separate experiment we tested the ability of this amount of catalase to inhibit biofilms.

Biofilm-infected porcine explants.
The e-scaffold was tested against biofilm-infected porcine explants. We followed previously published protocols to prepare infected explants 50,51 . Ear tissues were harvested from domestic pigs (obtained from C&L Lockers, Moscow, ID, USA), immediately cooled to 4 °C and kept for less than an hour at this temperature before being processed at the laboratory. No purpose-bred animals were used for these experiments. After the tissues were cleaned with 70% ethanol and the hair was removed using an electric razor, skin was excised with a scalpel. For the intact epidermis model the excised skin was sectioned at a thickness of approximately 500 μ m, using Padgett's dermatome, and punched into 12-mm-diameter discs, excluding skin with visible structural changes (scratches, erosion or scars). For the partial cutaneous wound model, mid-dermal sheets with a thickness of 500 μ m were harvested 52 . Skin punches with the dermal side down were used to seed polycarbonate transwell inserts (Greiner Bio-One North America, Inc., catalog #657641) with a 0.4-μ m pore size membrane separating each explant from the outer well prefilled with 2 ml of cell nutrient medium. These were maintained at 37 °C and 95% humidity in a 5% CO 2 environment. The nutrient medium consisted of serum-free Dulbecco's Modified Eagles Medium (DMEM) (Thermo Scientific, catalog #SH3024301) supplemented with L-glutamine (0.584 g/L), ampicillin (50 μ g/ml) and antifungal amphotericin B (0.4 μ g/ml). Biofilms were initiated by adding 5 μ l of overnight culture of A. baumannii (ATCC #17978, OD 600 ≈ 0.5) to the center of each explant surface. After 4 days the biofilm-infected porcine explants were ready to use.
Application of the e-scaffold to infected porcine explants. E-scaffolds were prepared as described above and overlaid onto A. baumannii biofilm-infected porcine explants. The inner well with the explant and e-scaffold was filled with 4 ml of sterile PBS as electrolyte (Supplementary, Fig. S6). Similar to the in vitro experiment, e-scaffold surface exposed to biofilm was polarized at − 600 mV Ag/AgCl . Biofilms exposed to non-polarized e-scaffolds were used as a control. After 24 h of polarization, the e-scaffolds as well as the explants with biofilms from both polarized and control wells were collected and processed for serial dilution and bacterial cell counts as described above.
Cytotoxicity test of the e-scaffold on porcine explants. The cell viability in the uninfected porcine explants with induced wounds was quantified to test whether the polarized e-scaffold treatment damaged the tissues 50 . After application of the polarized e-scaffold for 24 h, the porcine explant cell viability was quantified using PrestoBlue cell viability reagent (Life Technologies, catalog #A-13261) with the standard protocol (Life Technologies). Briefly, explants were incubated in 300 μ L of 10% PrestoBlue (in DMEM) for 3 h at 37 °C in an environment with 95% humidity and 5% CO 2 . The absorbance of the medium was then measured at 570 nm and 600 nm. The percent reduction of PrestoBlue was calculated from this absorbance and the molar extinction coefficient of oxidized and reduced PrestoBlue. The extent of PrestoBlue reduction calculated for explants exposed to polarized e-scaffolds was compared to that of the control, i.e. explants with non-polarized e-scaffolds. A normalized viability score of 100% was given to the explant showing the highest percent reduction of PrestoBlue for this control.

Histopathology.
Where indicated replicate tissue explants were fixed in 10% neutral buffered formalin and 5-μ m-thick sections were prepared and stained with matoxylin and eosin by the Washington Animal Disease Diagnostic Laboratory. The resulting light micrographs were subjected to a treatment-blind evaluation by a board-certified veterinary anatomic pathologist.
Scientific RepoRts | 5:14908 | DOi: 10.1038/srep14908 Statistical Analysis. Unless indicated otherwise, all experiments were performed using three independent replicates. The average results were expressed as mean ± standard deviation and analyzed using one-way ANOVA and Student's t-test to identify any significant difference between samples with and without e-scaffold application. Statistical analyses were performed using SigmaPlot (version 12.5).