Protein-Based Films as Antifouling and Drug-Eluting Antimicrobial Coatings for Medical Implants

Abstract

Nosocomial infections caused by bacterial contamination of medical devices and implants are a serious healthcare concern. We demonstrate here the use of fluorous-cured protein (FCP) nanofilm coatings for generating antimicrobial surfaces. In this approach, bacteria-repelling films are created by heat curing proteins in fluorous media. These films are then loaded with antibiotics, with release controlled via electrostatic interactions between therapeutic and protein film building blocks to provide bactericidal surfaces. This film fabrication process is additive-free, biocompatible, biodegradable, and can be used to provide antimicrobial coatings for both 2D and 3D objects for use in indwelling devices.

Graphical Abstract

INTRODUCTION

Infections caused by bacterial contamination of medical devices such as stainless-steel IV poles and implants pose a serious healthcare problem.^1,2 Nosocomial infections are caused by bacterial colonization on biomedical surfaces in healthcare settings.^3,4 In 2011, more than 700,000 nosocomial infections occurred in the United States, resulting in nearly 75,000 deaths.⁵ Antimicrobial surfaces have shown promise in the prevention of nosocomial infections.⁶

Antimicrobial coatings are used to prevent bacterial contamination of a wide range of surfaces,^{7, 8, 9, 10, 11} including medical devices^{12, 13} and implants.^{14, 15, 16} The general design of antimicrobial coatings are based on two main strategies: bacteria repulsion¹⁷ and release of antimicrobials.¹⁸ Tuning the chemical and morphological characteristics of a surface coating can prevent bacterial adhesion or eliminate bacteria upon contact.¹⁹ On the other hand, biocidal activity can be imparted to polymeric coatings by loading antibacterial agents, such as nanoparticles,²⁰ halogens,²¹ and antibiotics.²² The localized burst release of antimicrobials can efficiently reduce bacteria colonization on biomaterial surfaces as well as the possible risk of bacteria gaining resistance and tolerance to the antimicrobial agents.

Antimicrobials coatings must ensure sufficient loadability and localized burst release of antimicrobials to efficiently mitigate bacterial infections and minimize the risk of bacterial drug resistance.^{23, 24} At the same time, the biocompatibility and bioresorbability of coating materials is crucial to minimize immune reactions and cytotoxicity, especially for implants and catheters that are in direct contact with tissue.²⁵ Therefore, many efforts have focused on improving the biocompatibility of these coating scaffolds, while ensuring high drug loadability.²⁶ Proteins are excellent candidates for fabricating functional biomaterials such as antimicrobial coatings, due to their inherent biocompatibility, biodegradability and functional diversity.^{27, 28,29} However, most of the protein film fabrication strategies utilize cross-linkers to ensure aqueous stability, thereby adversely affecting their biocompatibility, degradability, and protein structure and function.^{30, 31} Alternatively, structural proteins that self-assemble into water-stable films such as silk, collagen etc. may be utilized. However, this limits the choice of proteins available and modification strategies must be employed to incorporate functionalities.^{32, 33} Recent studies have demonstrated fabrication of protein films through amyloid-like protein aggregation arising from rapid reduction of intramolecular disulfide bonds. This is a non-toxic and versatile strategy that has been utilized for antimicrobial applications.^34,35

Previously, we developed an additive-free methodology for fabricating protein films,³⁶ demonstrating that heating films in a fluorous environment (using perfluoroperhydrophenanthrene, PFHP) provided water-stable films (fluorous-cured proteins, FCPs) with minimal protein denaturation. This retention of protein structure concomitantly translated the properties of the proteins to the film surface, generating coatings that are hydrophilic and retained surface properties of native proteins such as charge and hydrophilicity.^{36, 37,38} Furthermore, PFHP leaves no residue on the protein coating and is regarded as safe per FDA guidelines,³⁹ providing an additive-free, biocompatible strategy for fabrication of protein films. We report here the fabrication of antimicrobial-loaded BSA-based protein coatings that serve as a reservoir for controlled-release of antibiotics, while simultaneously retaining their anti-fouling properties. The overall negative charge of the BSA protein film was utilized to imbibe films with cationic cargo, demonstrated through loading of charged fluorescent dyes. Furthermore, we demonstrate that release behavior is also dictated by electrostatic interactions and salt concentration. Lastly, antimicrobial activity was imparted by loading and release of a cationic antibiotic (colistin), providing surfaces featuring both resistance to bacterial colonization and controlled release of antimicrobials. These surfaces provide potential platforms for medical device and implant coatings.

RESULTS AND DISCUSSION

Our initial studies focused on bovine serum albumin (BSA) nanofilms. BSA (MW: 66.4 kDa, pI: 4.8) is inexpensive, readily available and considered a non-reactive protein that has often been used as a blocker in immunohistochemistry.^40,41 The negative surface charge and inherent zwitterionic property of BSA films prevents biomolecular and cellular adhesion, enabling the fabrication of anti-fouling coatings.⁴² BSA films were fabricated by spin-casting a 20% w/w BSA aqueous solution onto plasma-cleaned silicon wafers. These protein films were next stabilized by immersing protein-coated substrates into a preheated fluorous media - perfluoroperhydrophenanthrene (PFHP). After curing at 180 °C for 15 min in PFHP, water stable BSA films were generated (Figure S1). Films with different thickness were obtained by changing the concentration of precursor protein solution (Figure 1(a)). Heat-cured BSA films (treated in air at 180 °C) show significant loss of protein structure and surface charge,³⁶reducing their biocompatibility. Moreover,^. this loss of charge surfaces would be predicted to reduce or eliminate loading of the films based on electrostatic interactions.

Figure 1.

(a) Atomic force microscopic images and cross sections for scratched protein films prepared by 5%, 10% and 20% w/w BSA solution. (b) Films stability measured by the change of thickness (quantified using ellipsometry) after loading with dye, incubating in PBS, and treating with protease. (c) Loading capacity of protein films with different thickness (inset is the pictures of films taken under UV irradiation).

Cargo loadability was demonstrated first using cationic rhodamine 123 (R123) dye. Stabilized BSA films were incubated in 0.05 mM dye solution. The thickness of R123-BSA films was measured before loading, after loading, during release in PBS and post-treatment with protease (trypsin), as shown in Figure (1b). The thickness of BSA films remains constant during incubation in aqueous media for 7 days, indicating good water stability. Moreover, protein films fully degrade within 24 hr in the presence of protease (0.01M trypsin solution) indicating that stabilization and cargo-loading does not affect biodegradability of BSA films. The loading capacity of BSA films was quantified by measuring the fluorescent intensity of R123 released from the completely degraded films, normalized by the surface area of substrates (Figure 1c). The R123-loaded BSA films of different thicknesses were incubated in 1 mL of 0.01% trypsin for 24 hours, and the amount of R123 was determined by a calibration curve generated by measuring fluorescence intensity of different concentrations of R123 (Figure S2). The loading capacity increased proportionally to the film thickness, indicating that R123 was successfully incorporated within the BSA films and not simply adhered to the film surface. The results indicate the BSA films can be utilized as a reservoir for loading drugs.

The role of electrostatic interactions on the loading properties of protein films was quantified using oppositely charged dyes with similar size and hydrophobicity.⁴³ BSA films were incubated in 0.05 mM of anionic fluorescein (FL) and cationic R123 dyes for 1, 3, 6, and 24 hours. Despite the structural similarity of both dyes, the charge of the molecule significantly affects their loading into the film (Figure 2) regardless of the incubation time. There is almost no loading observed when the dye is negatively charged (FL), indicating that the electrostatic interaction between payload and film is the dominant factor for cargo loading. Further evidence for electrostatics driving incorporation was provided by cationic lysozyme films (Figure S3), where inverse behavior was observed. We next prepared R123 in a mildly acidic solution of pH = 4 to further demonstrate that the charge of the protein film modulates cargo loading. As the isoelectric point of BSA is pI = 4.8, BSA films are expected to be cationic at pH = 4, thereby hindering loading of R123. As expected, the loading of R123 in BSA films at pH 4 decreased substantially when both the cargo and film have the same charge (Figure 2).

Figure 2.

Loading capacity of BSA films prepared by incubating with Rhodamine 123 (R123), fluorescein (FL), and R123 in pH 4.

The release behaviour of the R123 films was monitored by measuring the fluorescence intensity of the supernatant at different time intervals. In Figure 3a, burst release was observed within 1hr of incubation in the case of R123. As expected, low loading of FL and R123 at pH = 4 resulted in low release. However, Figure S3(a) indicates that release rate is independent of type of dye loaded as well as pH of loading solution. Following these experiments, the role of ionic strength of the solution on release rate was investigated by varying the salt concentration. We hypothesized that the release behavior of R123 from BSA films would be affected by the salt concentration of the environment due to the change of electrostatic interactions between films and cargos. In low salt conditions (5 mM NaCl), stronger binding between R123 and BSA films was observed, leading to slow release (Figure 3b). On the other hand, in high salt solution (150 mM NaCl, PBS), the interaction became weaker and lead to an increased release, as seen in Figure 3b. Figure S3(b) shows release at different salt concentrations over 50 hr.

Figure 3.

(a) Release patterns of dye-loaded BSA films prepared by incubating with R123 and FL. (b) Cumulative release of R123 from BSA films in buffers prepared using different sodium chloride concentrations. Release rates were calculated as 170.8 %release/hr at 150 mM salt concentration, 134.8 %release/hr at 50 mM salt concentration and 49.5 %release/hr at 5 mM salt concentration.

The electrostatically regulated interaction of fluorous-cured protein films and cargo was next utilized to develop antifouling drug-eluting coatings for localized burst-release of cationic drugs. We hypothesized that incorporation of a cationic antibiotic into BSA-coated surgical screws would impart active protection from infections through increased antimicrobial activity in the vicinity of the implant, supplementing the passive protection provided by the anti-fouling behavior of the protein film.³⁶ We chose colistin as a the antimicrobial due to its polycationic moiety and efficacy against multidrug resistant strains.⁴⁴ Colistin was loaded into BSA films in a similar manner to R123 loading – films of thickness 85, 180, and 450 nm, generated by varying the concentration of BSA solution during spin coating, were first stabilized in PFHP. Films were then incubated in a 20 mg/ mL colistin solution overnight. The mount of colistin loaded was quantified using LDI-MS, as seen in Figure 4a (calculation described in Table S1). A significant difference in loading was observed between 85 and 180 nm films but not between 180 and 450 nm. This may be attributed to the larger size of the colistin molecule as compared to R123 that would be expected to affect dye penetrability. Next, the antimicrobial activity resulting from release of colistin was tested by the Kirby Bauer Diffusion Assay. Colistin-loaded films of varying thicknesses were placed in a bed of P. aeruginosa, prepared by seeding 10 μL of 10⁸ cfu/mL of P. aeruginosa solution on an agar plate. A clear inhibition zone (Figure 4b) was observed around the colistin-loaded films as well as the colistin-loaded control paper disk due to release of colistin into the agar in the vicinity of the substrate. This serves as a demonstration of localized release as high antibiotic concentrations are observed in the vicinity of the substrate. The normalized inhibition zone (calculations described in Figure S4) was calculated by dividing the total area of the inhibition zone by the area of the BSA film (silicon chips), as shown in Figure 4c. As expected, no inhibition zone was observed in the case of the negative controls: PBS-loaded disks and the BSA film without colistin.

Figure 4.

a) Colistin loading in BSA films of varying thicknesses as measured by LDI-MS. (b) Kirby Bauer Diffusion assay showing antimicrobial activity of colistin-loaded films, compared with positive and negative controls. (c) Normalized inhibition zones of Colistin-loaded BSA films calculated disk diffusion assay.

The zwitterionic nature and overall negative charge makes BSA ideal for developing antifouling coatings for medical. We hypothesized that colistin-loaded BSA films would provide dual protection from bacterial infections through passive resistance to bacterial fouling and active antibacterial activity. Therefore, we utilized surgical screws as a model medical implant to test both the anti-fouling and antimicrobial activity of colistin-loaded BSA coated surgical screws. Figure 5(a) illustrates the procedure for fabricating colistin-loaded BSA coatings on surgical screws. Plasma-treated clean surgical screws were coated by dipping in a 20% w/v BSA solution, followed by PFHP stabilization. Following this, screws were incubated a 20 mg/ mL colistin solution overnight. Red fluorescent protein (RFP) expressing E. coli was incubated with coated screws for 24 hours to evaluate anti-fouling behavior. Fluorescence microscopy images show a significant decrease in bacterial adhesion on BSA films both with and without loaded colistin, as compared to uncoated surgical screws (Figure 5b, c and d).

Figure 5.

(a) Schematic illustration of the fabrication of protein-coating on 3D surgical screw. (b-d) Fluorescence microscopy images of (b) bare, (c) BSA-coated and (d) colistin-loaded surgical screws incubated with red fluorescent protein (RFP) expressing E. coli for 24 hours. (e) Kirby-Bauer disc diffusion antibacterial activity assay for BSA-coated screws.

The biocidal efficacy of colistin-loaded BSA coatings was quantified by Kirby-Bauer diffusion assay. BSA coated screws with and without colistin were inserted into an agar plate seeded with 10 μL of 10⁸ cfu/mL of P. aeruginosa solution. A clear inhibition zone was observed around the colistin-loaded BSA coated screws, consistent with the results observed in Figure 4. As expected, pure BSA coatings as well as uncoated screws showed no antimicrobial activity (Figure 5e). Burst release of antibiotic is expected within two hours, based on the release rates observed in Figure 3(a). This rapid release is advantageous, as high concentrations of antibiotics can be released locally. High concentrations minimize the likelihood of development of drug resistance, while localized release is expected to minimize off-target effects. These results demonstrate that colistin-loaded BSA coatings are viable candidates for designing superior antibacterial coatings for medical implants due to their anti-fouling property and localized antibiotic release.

CONCLUSION

In summary, we developed a robust and efficient strategy for fabricating antimicrobial coatings on medical devices using naturally abundant proteins. Fluorous-curing film fabrication preserves the electrostatic properties of protein precursors enabling the fabrication of anionic BSA films that function as anti-fouling coatings. Further, the electrostatic interaction between cargo and protein films was harnessed for selective loading and release of oppositely charged cargos. Effective loading and localized release of cationic antibiotic imparted active protection against bacterial contamination by enhancing antimicrobial activity in the vicinity of the implant. Taken together, fluorous-cured antibiotic-loaded BSA coatings offer dual protection against bacterial contamination of implants through the passive anti-fouling behavior of BSA and the active burst-release of cationic antibiotic in the vicinity of the implant. The rate of drug release is expected to be affected by the type of protein as well as the biological fluid that interacts with the coated implant. Therefore, film surface charge and thickness may be modulated to control rate of release. Utilizing proteins to formulate coatings enables fabrication of biodegradable coatings in a sustainable manner. The choice of protein, in addition to dictating anti-fouling behavior, also ensures biocompatibility. This approach is a sustainable, biocompatible, and effective antimicrobial strategy for the prevention and treatment of implant-related nosocomial infections.

EXPERIMENTAL METHODS.

Materials:

Bovine serum albumin and lysozyme were purchased from Fisher Scientific and used without further purification. Silica wafers were purchased from WRS Materials. Perfluoroperhydrophenanthrene (PFHP), tetradecafluorohexane, fluorescein, rhodamine 123, and colistin sulfate were purchased from Millipore Sigma. MilliQ water was purified by using a Millipore water purification system. Titanium surgical screws screw was purchased from Alpha Bio Tec.

Film preparation:

5–20% w/v of protein (BSA or Lyso) solutions were prepared in MilliQ water and spin-coated at 3000 rpm for 25 seconds onto an oxygen-plasma cleaned silicon substrate to yield protein films. As prepared protein films were then incubated in pre-heated perfluoperhydrophenanthrene (PFHP) at 180 °C for 20 min, followed by washing with tetradecafluorohexane and drying with N₂ gas.

Surgical screws were cleaned in oxygen plasma before dip-coating with 20% w/v BSA solution. The screw was dried in a fume hood for 3 hours before heat-treating in PFHP at 180 °C for 20 min. After washing with tetradecafluorohexane, the screws were dried with nitrogen gas.

Dye and antibiotic loaded films:

Protein coatings were incubated in either 0.05 mM fluorescein or rhodamine 123 solutions (in PBS) for 24 h followed by with milliQ water. The procedure for antibiotic loading is the same except 20 mg/mL of colistin sulfate solution in PBS was used. For pH experiments, R123 was dissolved in a pH 4 PBS buffer, adjusted using HCl.

Control release experiment:

The dye-loaded protein films were incubated in 3 mL PBS and the release was monitored by measuring fluorescence signal of the supernatant at ex: 490nm, em: 515nm for FL and ex: 500 nm, em: 525 nm for R123 using a plate reader.

Colistin release study:

5, 10 and 20% w/v BSA solutions were used to prepare BSA films with different thicknesses as described above. Stabilized BSA films were then incubated in 20 mg/ mL colistin solution for 24 hr to enable colistin loading. Colistin-loaded BSA films were then incubated in 3 mL PBS for 15 hr to allow for the release of loaded colistin. After 15 hr, the supernatant from each sample was collected, spiked with 500 μg/ mL of Vancomycin solution and analysed using LDI-MS. LDI-MS signal was utilized to calculate the intensity ratio – intensity of colistin signal/ intensity of vancomycin signal. Thereafter, the concentration of vancomycin was utilized to calculate the concentration of colistin in the supernatant. Calculations are further elaborated in Table S1.

Kirby-Bauer disc diffusion:

P. aeruginosa were inoculated in 3 mL LB broth and grown to stationary phase at 37°C. The cultures were then harvested by centrifugation and washed three times with 0.85% sodium chloride solution. The concentration of the resuspended bacteria solution was determined by measuring the optical density at 600 nm. Seeding solution were prepared by diluting to 0.1 OD₆₀₀ (10⁸ colony forming units) in M9 minimal media. Agar gel plates were prepared by pouring a sterile solution of 6g Agar and 10g LB in 400 mL of water onto polystyrene petri dishes. 10 μL of the seeding solution was spread onto the agar plates using a sterile glass spreader. Colistin-loaded, BSA Coated as well as uncoated silicon chips were placed onto the agar plate and incubated overnight at 37°C. A paper disc loaded with 1 mg/ mL was utilized as a positive control. The inhibition zone is the area around antibiotic-loaded substrates where no bacterial colonies were observed. The normalized inhibition zone was calculated by dividing the area of inhibition by the area of the substrate (silicon/ paper) for each case.

In the case of the surgical screws, agar plates were prepared and seeded with bacteria using the same protocol. Following this colistin-loaded, BSA coated and bare screws were screwed into the agar to observe colistin release.

Bacterial adhesion experiment:

DsRed expressing E. coli were inoculated in 3 mL LB broth and grown to stationary phase at 37 °C. The cultures were then harvested by centrifugation and washed three times with 0.85% sodium chloride solution. The concentration of the resuspended bacteria solution was determined by measuring the optical density at 600 nm. Seeding solution were made by diluting to 0.1 OD₆₀₀ (10⁸ colony forming units/ mL) in M9 minimal media supplemented with 1mM IPTG (isopropyl β-D-1-thiogalactopyranoside). 2 ml of this solution was poured onto the bare, BSA-coated and Colistin-loaded screws and incubated for 24 hours in ambient condition to enable bacteria adhesion. The screws were then washed 3 times with PBS before imaging by a confocal microscope.

Microscopy:

Bacterial film coatings were analyzed by confocal scanning light microscopy (CLSM). All analysis was performed using the A1SP: Nikon A1 Spectral Detector Confocal.

Scheme 1.

Protein nanofilms fabricated by FCP method retain their surface properties and can be loaded with charged cargos via electrostatic interaction. Antimicrobial coatings are fabricated by loading negatively charged BSA nanofilms with cationic antibiotics.