Electrodeposited Assembly of Additive-Free Silk Fibroin Coating from Pre-Assembled Nanospheres for Drug Delivery.

Electrophoretically deposited (EPD) polymer-based coatings have been extensively reported as reservoirs in medical devices for delivery of therapeutic agents, but control over drug release remains a challenge. Here, a simple but uncommon assembly strategy for EPD polymer coatings was proposed to improve drug release without introducing any additives except the EPD matrix polymer precursor. The added value of the proposed strategy was demonstrated by developing a novel EPD silk fibroin (SF) coating assembled from pre-assembled SF nanospheres for an application model, that is, preventing infections around percutaneous orthopedic implants via local delivery of antibiotics. The EPD mechanism of this nanosphere coating involved oxidation of water near the substrate to neutralize SF nanospheres, resulting in irreversible deposition. The deposition process and mass could be easily controlled using the applied EPD parameters. In comparison with the EPD SF coating assembled in a conventional way (directly obtained from SF molecule solutions), this novel coating had a similar adhesion strength but exhibited a more hydrophobic nanotopography to induce better fibroblastic response. Moreover, the use of nanospheres as building blocks enabled 1.38 and 21 times enhancement on the antibiotic release amount and time (of 95% maximum dug release), respectively, while retaining drug effectiveness and showing undetectable cytotoxicity. This unexpected release kinetics was found attributable to the electrostatic and hydrophobic interactions between the drug and nanospheres and a negligible initial dissolution effect on the nanosphere coating. These results illustrate the promising potential of the pre-assembled strategy on EPD polymer coatings for superior control over drug delivery.

Introduction

Polymer-based coatings are widely used on the surface of different medical devices, such as drug-eluting stents,[1,2] orthopedic (or dental or otorhinolaryngologic) fixators, and implants,[3,4] to act as reservoirs for sustained release of therapeutic and signaling agents. Compared with other commonly used coating methods, such as plasma spraying, one of the most common methods in the industry today, electrophoretic deposition (EPD) is a solution-based and binder-free technique, which is very suitable to construct homogeneous polymer coatings with high purity and fine-tunable thickness on complex geometries, porous structures, and nonline-of-sight surfaces that medical devices usually possess.[5] EPD has been extensively reported to prepare a wide range of (sensitive) polymer coatings for biomedical applications[6] in a mild aqueous environment at room (or low) temperature, with simple equipment requirements,[7] short processing time,[8] high preparation efficiency,[9] as well as upscaling potential for commercial production.[10]

In the past, the conventional way to functionalize EPD coatings with therapeutic drugs is to mix the drug into matrix polymer precursor solution before deposition of the mixture,[11] but control over release kinetics is often not achieved because of the poor affinity and interaction between the drug and coating matrix.[5] Many recent studies aimed at introducing a variety of other material-based drug delivery vehicles into the coating matrix polymer[9,12−14] (e.g., carbon nanotubes or gelatin nanoparticles in a chitosan matrix coating) or using a cross-linking agent such as genipin[15] between the drug and matrix to improve control over the drug release of EPD coatings.

Although introducing additional chemicals to currently existent material systems is frequently applied in research laboratories to modify material functions, it is not a preferred solution from a point of view of clinical application and commercial production that the introduced chemicals might often increase the complexity and cost of manufacture and raise the uncertainty of biosafety.[16] For EPD coatings, it should be noted that the matrix polymer precursors themselves can be assembled into a variety of nanostructures to regulate drug release.[17] Therefore, we hypothesized that the drug release capability of the EPD polymer coating can be enhanced via assembling structures from pre-assembled nanoarchitectures, without introducing any new chemicals except the polymer precursor of the coating matrix itself.

To demonstrate the added value of this assembly strategy on EPD polymer coatings, a novel silk fibroin (SF) EPD coating assembled from pre-assembled SF nanospheres, instead of being assembled from SF molecules used in the conventional EPD assembly strategy, is developed to improve the delivery drug in an application model, that is, preventing infections around percutaneous orthopedic implants. Although antibiotic-loaded EPD coatings have been applied on various orthopedic implants to prevent infections,[11,12,18−20] the controlled and sustained release of the antibiotics and their long-term effects remain to be a challenge. Compared to other widely used EPD polymer precursors, such as chitosan, SF has recently attracted a lot of research attention to be used for EPD coatings[5,21−23] for biomolecule drug delivery because of its stabilization effect on sensitive biological compounds (e.g., antibiotics),[24] excellent biocompatibility,[25] and hypoallergenicity.[26] Moreover, SF molecules themselves can be easily assembled into SF nanospheres,[27] attractive vehicles for sustained delivery of manifold drugs from antibiotics and anticancer drugs to growth factors.[28]

In this study, SF nanospheres were first pre-assembled by precipitation reaction. To reveal the assembly mechanisms of nanosphere coating, we investigated the ζ-potential and particle size of SF nanospheres as a function of pH, and the coating thickness as a function of EPD processing parameters. In comparison with the conventional EPD SF coatings directly assembled from SF molecules (SFM coating), this nanosphere coating (SFN coating) was characterized in terms of conformation changes, topography, wettability, degradation, adhesion strength, and fibroblastic response. Then, drug release profiles of SFN and SFM coatings were compared, and the underlying mechanisms of superior control over drug release of SFN coating were investigated. These results not only make great progress and improvement for the EPD SF coatings simply by assembling the coatings from pre-assembled nanoarchitectures but also indicate the potential of the pre-assembly strategy on EPD polymer coating field as a simple, additive-free, cost-effective approach to achieve superior control over drug release.

Materials and Methods

Preparation of SF Solution

SF solution was prepared as previously described.[29] Briefly, Bombyx mori silk cocoons provided by Prof. Aichun Zhao (State Key Laboratory of Silkworm Genome Biology, Southwest University, Chongqing, P. R. China) were first degummed in boiled 0.02 M Na2CO3 solution for 30 min and then rinsed with Milli-Q water. After drying, the extracted silk was dissolved in 9.3 M LiBr at 60 °C for 4 h and then dialyzed with Milli-Q water using the dialysis membrane (MW = 3500). Insoluble residues were removed by centrifugation at 5000 rpm for 1 h at 4 °C. Finally, the SF concentration in aqueous solution was adjusted to 8 wt %.

Preparation of SF Nanosphere Suspension

SF nanospheres were synthesized via a precipitation reaction, by which SF aqueous solution (5 wt %) was dropwise added into five times-volume acetone, as previously reported.[30] Then, SF nanospheres were washed with Milli-Q water and centrifuged at 5000 rpm for 1 h for three times to remove acetone. SF nanosphere dispersions were obtained following sonication of nanospheres in Milli-Q water for 5 min at 100% amplitude, and pulse rate of 1 s on, 1 s off, using a sonicator (Emerson Industrial Automation, Branson Europe, Dietzenbach, Germany). The morphology of nanospheres was measured by scanning electron microscopy (SEM) (Sigma-300, Zeiss, Germany).

Characterization of SF Nanosphere Suspension

The ζ-potential and particle size of SF nanospheres were measured using dynamic light scattering (DLS, Zetasizer, Nano-S, Malvern Instruments Ltd., U.K.) by dispersion of nanospheres in HEPES buffer (5 mM) with adjusted pH (using HCl or NaOH). Each value represents an average of three measurements.

Preparation of Substrates

Commercially pure titanium disks (grade 2, Baoji Titanium Industry, China) were utilized as a deposition substrate. To obtain a smooth surface, the Ti plates were manually ground with Grit 600 and Grit 2500 grinding paper (Struers, the Netherlands).[31] Then, they were ultrasonically (VWR, the Netherlands) cleaned in acetone, ethanol, and water for 5 min of each treatment, and finally subjected to argon plasma glow discharge (Radio frequency glow discharge machine, Harrick Scientific Corp., U.S.A.) for 5 min. Rough surfaces were prepared by sandblasting with 0.25–0.50 mm grit. Subsequently, these discs were acid-etched with HCl/H2SO4 for 30 min at 60 °C[32] and then also ultrasonically cleaned with acetone, ethanol, water, and glow discharge. Smooth 316L stainless steel and CoCrMo (Tiger International BioMetals Co., Ltd., China) with a similar treatment were also used as substrates to demonstrate the feasibility of constructing SFN coatings on different typical medical metallic surfaces.

EPD of SFN Coating

Pretreated metal substrates were used as a working electrode, and a parallel pure titanium disk of the same size and shape was used as the counter electrode. For each experiment, fresh 10 mL of SFN dispersion was used. The distance between the positive and negative electrodes was 10 mm. To investigate the EPD parametric control over SFN coating deposition, the EPD process was first carried out by connecting both electrodes to a direct current power supply (model 6614C, Agilent Technologies) with different concentrations from 0.5 to 1.75 wt % at a constant electric field of 5 V/cm for 2 min. In another set of experiments, the same time of 2 min and 1 wt % SFN suspension were applied for different electric fields (3–8 V/cm). Finally, different deposition time (1–10 min) was applied with 1 wt % SFN suspension at a constant electric field of 5 V/cm. During deposition, the SFN suspensions were stirred using a magnetic stirrer (50 rpm). For following material characterization and biological assessment SFN coating, SFN coating was deposited at 5 V/cm for 2 min from 1.0 wt % SF nanosphere suspension. After the deposition, the Ti disks were carefully withdrawn from the solution, rinsed with Milli-Q water three times, and slowly air-dried in a box to prevent coating cracks. Finally, the samples were cross-linked by water vapor annealing in a vacuum desiccator overnight at room temperature.

EPD of SFM Coating

For comparison, SFM coatings were directly assembled form SF molecule solution as the conventional EPD method. Similarly, a fresh 10 mL SF molecule solution was used. The distance between the positive and negative electrodes was 10 mm. EPD was carried out by connecting both electrodes to a direct current power supply with a concentration of 1 wt % at a constant electric field of 5 V/cm for 2 min. After the deposition, the disks were carefully withdrawn from the solution, rinsed with Milli-Q water for three times, and slowly dried in a box. Finally, they were cross-linked by water vapor annealing in a vacuum desiccator overnight at room temperature.

Material Characterization of Coatings

The topography of the coating surfaces was examined by SEM (Sigma-300, Zeiss, Germany). Fourier-transform infrared spectroscopy (FTIR) analysis of the coatings was performed by attenuated total reflectance infrared spectroscopy (UATR two, PerkinElmer, the Netherlands). The roughness and thickness of the coatings were measured using the Profilometer (ProScan, U.S.A.) (n = 3) as the method previous reported.[33] The wettability of different SF surfaces was determined by detecting the static water contact angles of the surfaces with an Optical Tensiometer (Theta Lite, Biolin Scientific, Sweden) (n = 3). For the degradation test, coatings were immersed in 1.5 mL of PBS at 37 °C for 14 days, and PBS solutions were replaced freshly every 24 h (n = 3). At specified time points, the samples were rinsed via agitation in Milli-Q water for 5 times (each for 2 min) and dehydrated in an oven at 60 °C overnight. Immediately following the removal from the oven, the samples were weighed.

The adhesion strength of the coatings was tested using a standard lap shear tensile test (Supporting Information, Figure S1a).[31] Each test specimen was an assembly of a coated substrate and a matching substrate sample with the same dimension bonded together using a thin layer of epoxy adhesive (Loctite 415, USA). The samples were held by grips of a Universal Testing Machine (MTS Systems, 858 mini bionix II, U.S.A.). The pull test was run at a constant cross-head displacement of 0.50 mm/min until failure (n = 3). The results were included only when the sample failed at the coating/substrate interface, as shown by the exposure of the metallic substrate (n = 3, Supporting Information, Figure S1b).

Biological Assessment of Coatings

NIH-3T3 (ATCC, USA), a mouse fibroblastic cell line, was cultured in Dulbecco’s modified essential medium (Gibco, Invitrogen Corp., Paisley, Scotland) supplemented with 10% calf serum (Gibco, Invitrogen Corp., Paisley, Scotland) at 37 °C in a humidified 5% CO2 atmosphere. Coated disks were sterilized by UV light, placed in 24-well tissue culture plates, immersed in cell culture media for 4 h, and then seeded at a density of 5000 cells·cm–2, while cell culture coverslips (CS, Thermo Fisher, U.S.A.) were used as a positive control.

Cells cultured on the SFM and SFN coating surfaces after 24 h were washed with PBS and then fixed with 2 wt % glutaraldehyde in 0.1 M sodium-cacodylate solution for 20 min. Subsequently, cells were dehydrated in a graded series of ethanol (70, 80, 90, 96, and 100% ethanol for 5 min each), followed by observing the morphology of the cells by SEM. For immunofluorescent studies, cells were fixed using 4% paraformaldehyde for 10 min, followed by PBS washing for three times, and then permeated with 0.1% Triton X-100 for 10 min. Subsequently, the cells were blocked with 1% BSA for 30 min and incubated with the anti-vinculin antibody (1:200, ab129002, Abcam, U.S.A.) for 1 h. The Alexa Fluor 647-conjugated goat anti-rabbit secondary antibody (1:500, ab150083, Abcam, U.S.A.) was incubated for 1 h. TRITC-phalloidin (1:2000, P1951, Sigma, U.S.A.) was applied to label F-actin, and DAPI (1:2500, D9542, Sigma) was used to mark cell nuclei. Images of stained samples were captured by fluorescent microscopy (Axio Imager Microscope Z1, Zeiss, Germany).

All the images were analyzed by Image J (NIH, USA). To measure focal adhesions (FAs) area per cell, the grayscale vinculin image was thresholded to produce a black and white image from which the pixels representing FAs were counted and summed, following a step-by-step quantitative FA analysis protocol as previously reported.[34] To measure the cell area and perimeter, the cellular morphology were determined using the thresholding method from F-actin fluorescent images, and then, they were used to calculate the cell area and cell perimeter. The cell shape index (CSI) was calculated using the formula as previously reported[35]where a line and a circle have CSI values of 0 and 1, respectively.

The RNA was extracted using an RNeasy Mini Kit (Qiagen, U.S.A.), according to the manual instruction and reversed to cDNA using TaqMan Reverse Transcription kit (Bio-Rad, U.S.A.). Subsequently, cDNA was added with a Fast SYBR Green Master Mix Kit and complemented by the PRISM 7500 sequence amplification system (Applied Biosystems, USA). Vinculin was tested with GAPDH as the housekeeping gene. The primer sets of genes are listed in Table S1 (Supporting Information). The mRNA levels of target genes were normalized by the level of GAPDH mRNA and calculated via the 2–ΔΔ method.

The total cell number was quantified at specific time intervals (1, 3, and 7 days) using a cell counting kit (CCK-8; Dojindo, Japan). The result of CCK-8 was tested spectrophotometrically (Bio-Tek FL600 microplate fluorescence reader, Biotek, U.S.A.) according to the manufacturer instructions. Three samples per group were tested.

After 24 h of incubation, lactate dehydrogenase (LDH) Assay kit (Thermo Fisher Scientific, U.S.A.) was used for determining cytocompatibility of drug-loaded coatings by measuring the LDH activity released from damaged cells. Cell culture coverslips without and with 5% DMSO were used as positive and negative controls, respectively, while wells added with 2% (v/v) Triton-X100 and only culture medium were used as the high and low controls, respectively. The results from the LDH assay were tested spectrophotometrically according to the manufacturer instructions. Three samples per group were tested. The cytocompatibility was calculated using the formula as previously reported[12]

Detection on Vancomycin Concentration

Concentration of vancomycin in this study was detected by HPLC using a Hitachi HPLC machine, which consisted of a Hitachi L-2130 pump, a Hitachi L-2400 UV detector, a Hitachi L-2200 auto sampler, and a LiChrospher RP-18 end-capped HPLC column (125 mm × 4 mm, particle size 5 μm). The mobile phase consisted of 50 mM ammonia phosphate buffer (pH 3, adjusted with H3PO4); acetonitrile (90/10 v/v) was used for the detection of vancomycin. The supernatant (30 μm) containing vancomycin was injected with a flow rate of 1 mL/min. Then, the concentration of vancomycin was quantified at 196 nm.

Optimization of the Vancomycin Loading Amount

Various weight ratios of vancomycin and SF nanospheres were used for determining the optimum loading capacity for SF nanospheres. First, different concentrations (0.8, 1.6, 2.4, 3.2, and 4 mg/mL) of 5 mL vancomycin solutions were added into 5 mL of SF nanosphere suspensions (20 mg/mL) to reach different weight ratios (0.04, 0.08, 0.12, 0.16, and 0.20) under stirring for 24 h. Then, vancomycin-loaded SF nanospheres were collected by washing with Milli-Q water and centrifugation at 15 000 rpm for 5 min three times to remove any unbound drug. The amount of the unbound drug in the washed supernatant was measured by HPLC. The loading capacity and encapsulation efficiency of vancomycin in SF nanospheres were measured using the following equations

EPD of Vancomycin-Loaded Coatings

After optimization, a weight ratio (drug/protein) of 0.12 was chosen to load drug for the following study. For vancomycin-loaded SFN coating (SFNV coating), 5 mL of vancomycin solution (2.4 mg/mL) was added into 5 mL of SF nanosphere suspension (20 mg/mL) under stirring for 24 h. After the washing and centrifugation step to remove any unbound drug, SFNV was resuspended by 10 mL Milli-Q water. For vancomycin-loaded SFM coating (SFMV coating), 5 mL of vancomycin solution (2.4 mg/mL) was added into 5 mL of SF molecule solution (20 mg/mL) under stirring for 24 h. The SFNV and SFMV coatings were then prepared under the same EPD conditions (electric field of 5 V/cm for 2 min at a total protein concentration of 1 wt %) as the SFN and SFM coatings used above.

Drug Release from Coatings

To investigate the release profiles of SFNV and SFMV coatings, the coated disks were placed into glass bottles (n = 3), followed by addition of 1 mL of PBS. To investigate the influence of ionic strength and detergent concentration on the short-term release of drug from SFNV coatings, SFNV coatings were used for an additional release study in media with different (i) ionic strengths (10, 100, and 1000 mM NaCl solutions in water) and (ii) detergent concentrations (0.01, 0.1, and 1 v/v % Tween 20). The SFNV coatings were placed into glass bottles (n = 3), followed by addition of 1 mL of the abovementioned media. All the glass bottles were then placed on a shaking plate with a shaking rate of 90 rpm at 37 °C to perform the release study. At the specific time points, 0.9 mL of the supernatant was collected and refreshed with the same amount of media. The collected supernatant was stored at 4 °C until the concentrations of vancomycin were determined using HPLC.

Antibacterial Efficacy of the Released Drug from Coatings

To investigate the antibacterial efficacy of the released drug from SFNV and SFMV coatings at different time points, both coated disks were placed into glass bottles (n = 3), followed by addition of 1 mL of PBS. All the glass bottles were then placed on a shaking plate with a shaking rate of 90 rpm at 37 °C. At 24 h before the specific time points, all the PBS was refreshed. At the specific time points, 0.5 mL of the supernatant was collected and refreshed with the same amount of PBS.

The MICs of each collected supernatant was evaluated using the Staphylococcus aureus ATCC25923 strain. In brief, the overnight culture of S. aureus was diluted to cfu/mL in two-times concentrated brain heart infusion broth. The SFNV/SFNM supernatants were twofold serial-diluted with PBS. The maximum dilution was 640-fold. S. aureus dilution was mixed with the undiluted or twofold serial-diluted SFNV/SFNM supernatants, PBS and vancomycin solutions at the ratio of 1:1, and pipetted into a 96-well plate (0.2 mL/well). The PBS solution was used as a negative control, and the vancomycin solutions at the final concentrations of 0.53, 1.06, and 3.12 μg/mL were used as positive controls. The 96-well plate containing all mixtures was placed in a SpectraMax i3 microplate reader (Molecular Devices, San Jose, CA, USA), and the optical density (OD) value of each well was recorded at the wavelength of 600 nm at 37 °C after 10 h. In the negative control (PBS) group, the OD value of S. aureus culture reached maximum after 10 h. The highest fold of dilution where there was no growth or the growth was below 50% of those in the negative control group was registered as the dilution times at the break point of the specific SFNV/SFNM supernatant.

Statistical Analysis

One-way ANOVA was used to determine statistical significance followed by post hoc analysis using the Tukey test. All statistical analyses were performed with GraphPad Prism and Origin software.

Results and Discussion

Electrodeposited Assembly Mechanism of SFN Coating

SF molecules were first pre-assembled into SF nanospheres, via precipitation reaction from acetone into which an SF molecule aqueous solution was added dropwise as previously reported.[30,36] The spherical morphology of the SF nanospheres was confirmed using SEM (Figure c). Then, nanospheres were dispersed in water. The ζ-potential and size of nanospheres were examined as a function of pH via DLS (Figure d). The pH value of SF nanosphere suspension (1 wt %) was 7.5, and the average diameter of nanospheres was around 110 nm, which was in agreement with the size observed by SEM and reported previously using this method.[36] The ζ-potential showed a reversed sigmoidal behavior where the SF nanospheres attained a negative charge of about −30 mV at pH values higher than 5, while the nanospheres had a positive charge of ∼18 mV at pH values below 3. The inflection point was at ∼pH 4, which is in line with the pI of SF molecules (∼4.2).[37] The inflection point indicates that SF nanosphere dispersions will be stabilized at pH > 5 by repulsive interactions between negatively charged SF nanospheres and destabilized at pH < 4. This phenomenon also explains our particle size data, which showed a steep increase at pH < 4. The digital photographs indicated stable dispersion for SF nanospheres at pH > 5, whereas the reduction of pH yields in phase separation by precipitation (Figure e).

Figure 1

EPD assembly mechanism of SFN coating. (a) Pre-assembly vs (b) conventional assembly of SF EPD coatings. (c) Scanning electron micrographs of SFNs. (d) ζ-Potential and particle size of SFNs as a function of pH. (e) Digital photographs of SFNs showing the stability of SFNs aqueous solution as a function of pH. Error bars represent one standard deviation.

Similar to conventional SFM coatings deposited directly from SF molecules solution (Figure a), assembling coatings from SF nanosphere suspension was also driven by the applications of a positive potential (Figure b). Water was oxidized around the surface of the anode, causing a reduction in pH in the vicinity of the metal surface (eq ).

The reduction of pH resulted in protonation of the SF nanosphere surfaces and diminished the repulsive interactions between the nanospheres, thus driving irreversible aggregation and deposition of the nanospheres at the metal surface when the adjacent pH value arrived at around 4 under appropriate EPD parameters.

EPD Parametric Influence on SFN Coating Thickness

Then, we investigated whether the deposition parameters of SF nanospheres also followed the Hamaker equation in a similar manner to EPD of polymer molecules, which predicts a linear increase in deposited mass with increased suspension concentration, electric field, and deposition time.[6]

Figure a revealed the effect of the suspension concentration upon applying a constant electric field of 5 V/cm for 2 min. At concentrations below 0.5 wt %, coatings were not deposited, which may have resulted from the fact that the low concentration of nanospheres prevents aggregation.[31] At concentrations higher than 1.0 wt %, coating thickness increased linearly with increasing concentration. However, suspension stability decreased with increasing concentration, which complicated dispersion of the suspension.[38] Therefore, a moderate SF nanosphere concentration of 1.0 wt % was selected for further studies.

Figure 2

EPD parametric influence on SFN coating thickness. Thickness of the SFN coating as a function of EPD processing parameters, including (a) suspension concentration, (b) electric field, and (c) deposition time. Error bars represent one standard deviation.

From Figure b, it can be concluded that the deposition rate increased with an increasing electric field. At electric fields below 3 V/cm, coatings could not form, which may contribute to the fact that the pH near the metal substrates was not low enough to cause aggregation and deposition (see eq ).[6] However, the SFN coatings became less homogeneous and showed poor attachment to the titanium surface with increasing electric field, which may be caused by the very acute generation of oxygen bubbles at a relatively high electric field (see eq ).[32] When the electric field was higher than 8 V/cm, no deposition was observed anymore. Therefore, a moderate electric field of 5 V/cm was selected in further studies to achieve homogeneous coatings.

Figure c reveals that the thickness of the SF nanosphere layer increases linearly when deposition time was increased from 1 to 6 min. This indicates that within 6 min, (i) the EPD process is Faradaic and (ii) the coating thickness can be linearly controlled by deposition time.[31] However, after 6 min, the process showed self-limitation, as no increase in coating thickness occurred with further prolongation of the coating time. Oxidation of water takes place at the metal/electrolyte interface, whereas deposition occurs at the SF nanosphere layer/electrolyte interface. Consequently, this self-limiting phenomenon coating thickness after 6 min may result from the fact that the distance between these two interfaces increases over time.[18]

Material Characterization of SFN Coating

Figure a shows a typical morphology of an SFN and the conventional SFM coating obtained at comparable conditions. The SFN coating was uniform and revealed a nanostructure resembling the shape of the initial SF nanospheres. The tilted view of the coating interior showed that both SFN and SFM coatings were assembled densely and homogeneously without visible pores.

Figure 3

Material characterization SFN coating. (a) Scanning electron micrographs of coatings. (b) FTIR absorbance spectra of the amide I region (between 1695 and 1595 cm–1) showing the conformational changes during the preparation process of coatings. WA: after water annealing. AD: after air drying. E-gel: electrogel. sol: solution. sus: suspension. (c) Surface roughness of coatings. (d) Surface wettability determined by water contact angle measurements and representative images of water droplets. (e) Remaining mass of coatings immersed in PBS after 1, 3, 7, and 14 days. (f) Adhesion strength of coatings measured by lap shear tensile testing. Error bars represent one standard deviation (*p < 0.05).

FTIR results helped us understand the conformational change of SF molecules during coating assembly. For conventional SFM coating, SF molecules first deposit and formed a gel at positive electrodes (Figure a) because of the local conformational changes from a random coil to helical state (Figure b).[39] Then, this metastable form of SF was then converted to the more stable crystalline b-sheet structures (Figure b) during the air-dry process which transforms the silk gel to a coating (Figure a). At last, water annealing was applied to further induce β-sheet formation inside coating, which ensures that coating was both chemically stable and water-insoluble.[5]

For the SFN coatings, SF nanospheres were first assembled (Figure b), which showed a high amount of β-sheet formation (Figure b). This was attributed to exposure of SF molecules to acetone during the pre-assembly process, which dehydrated the SF and facilitated closer chain packing of the hydrophobic Gly-X repeats.[30] In contrast to the conventional SFM coatings, the conformation of SF was stable and almost unchanged (Figure b) immediately after EPD and subsequent drying and annealing of SFN coatings (Figure b). This indicates that the annealing step can be omitted for SFN coatings.

The water contact angle results showed that SFN coatings were less hydrophilic than conventional SFM coatings (Figure d), which may be caused by the rougher topography (Figure c) and the increased hydrophobic β-sheets in SFN coatings.[32] In PBS, both coatings showed very limited degradation (Figures e and S2, Supporting Information), which was consistent with a previous study.[40] However, there was a noticeable weight loss in SFM coating during the first 24 h, which is due to the fact that soluble peptides of the SF leached out into PBS.[40,41] In contrast, the SFN coating showed no initial weight loss, indicating that the washing step during nanosphere preparation did already remove the soluble components.[36]

The adhesion strength between the coatings and the metallic substrates was investigated in a lap shear tensile test (Figures f and S1, Supporting Information). It could be seen that SFN and SFM coatings exhibited an adhesion strength of 6.66 ± 1.10 and 8.16 ± 1.31 MPa, respectively. However, the change of nanostructures of the SF EPD coatings had no significant influence on adhesion strengths. The strength values observed in this study were similar to or higher than the adhesion strengths of polymer EPD coatings that were reported before being measured by lap shear tensile test (in the range from 1.5 to 8 MPa).[19,42,43] Moreover, the coating constructed on the transcutaneous part of implants does not need to resist high mechanical force such as the ones during press or screw-fit placement.[12] Consequently, we assume that the adhesion strength of coatings was sufficient for our application in this study.

In addition, we anticipate that this novel nanospheres EPD coating might be used on various types of metallic medical implants, considering its feasibility of preparation on different metallic surfaces (Figure S3, Supporting Information).

Biological Assessment of SFN Coating

Fibroblasts were cultured to clarify the cellular responses to these coatings. The commercial cell culture coverslips were used as the positive control. From fluorescent staining of F-actin and vinculin (Figure a), we observed more abundant cytoskeleton organization and FA formation on SFN coating than on SFM coating, which was consistent with more filopodia formation on SFN coating than on SFM coating showed by scanning electron micrographs (Figure a). The morphology of the cells on SFN coating tended to be more elongated (Figure c). The cell area (Figure d), the FA area per cell (Figure b), proliferation activity (Figure e), and vinculin gene expression (Figure f) on SFM coating were smaller than those on the positive control, while these values on SFN coating exhibited no significant difference compared with these values on the positive control. This finding was supported by the upregulated gene expression of RhoA (Figure g) in the cell on SFN surfaces compared with SFM. Previous studies demonstrated that via linking F-actin to the exposed cryptic binding sites of unfolded vinculin at FAs, vinculin triggers a series of phosphorylation events to activate the mechanoresponsive signaling transforming protein, RhoA, which engages in the controls of cytoskeleton dynamics and cell polarity.[44,45]

Figure 4

Cellular response to SFN coating. (a) Cell spreading after 24 h shown as immunofluorescent images of vinculin (purple), F-actin (white), and nucleus (blue), corresponding heatmap, and scanning electron micrographs. Quantitative analysis of (b) FA area per cell, (c) CSI, and (d) cell area. (e) Cell proliferation measured by CCK-8. Relative mRNA expression level of (f) vinculin and (g) RhoA. For each box plot, the box boundaries represent the 25–75% quartiles, and the whiskers represent the minimum and maximum value. Error bars represent one standard deviation (*p < 0.05).

Previous studies also observed that adhesion, spreading, and proliferation of fibroblasts were less on flat silk films than on the flat tissue culture plate.[46] However, this cellular response could be improved by creating surface roughness by incorporating (diameter of 100 nm)[47] or establishing porosity (diameter of 80 nm),[48] whose feature sizes are comparable to those of our SFN coating. The other surface parameter which might influence cell behavior is hydrophobicity. The water contact angle of SFN coating increased from 60 to 76° compared to SFM coatings. However, it has been reported that the fibroblastic cell behavior on surfaces with the water contact angles within this range showed no significant difference, and these moderate hydrophilic material surfaces have the best fibroblastic attachment and spreading compared to more hydrophilic or hydrophobic surfaces.[49] In summary, the SFN coating enhanced initial fibroblastic responses compared to the conventional SFM coating, which might be helpful for early transcutaneous wound healing and could favor preventing infections.[5,50]

Drug Loading onto SFN Coating

As a following step, vancomycin was chosen as a model drug to test our hypothesis that SFN coating provides more control over drug release kinetics than conventional SFM coating. Vancomycin was selected because this antibiotic is one of the most widely used antibiotics in the orthopedic surgery to prevent implant-associated infections.[11] The conventional approach of loading drugs onto the EPD coatings involves blending the drugs with the EPD polymer precursor solution followed by deposition of the mixed solution.[6] To mimic this conventional drug-loading strategy, vancomycin was first mixed with the SF solution in our study (Scheme a). Consequently, the positively charged drug and the negatively charged SF molecules formed polyelectrolyte complexes in the solution.[5]

Scheme 1

Schematic Illustrating Mechanisms of Drug Loading and Release from (a) Pre-Assembled SFNV Coatings vs (b) Conventional SFMV Coatings

For our new EPD coating strategy, we first loaded vancomycin onto the SF nanospheres using a diffusional postloading method (Scheme b). Our data showed that the drug encapsulation efficiency of SF nanospheres decreased with increasing vancomycin content, while the drug loading capacity of nanospheres first increased and then slightly decreased, showing an inflection point at 12% w/w (vancomycin/SFNs) (Figure a). Hence, we selected 12% w/w as the fixed amount of drug loading into the nanospheres, where a maximum loading capacity was reached at 8.3% with the encapsulation efficiency of 69.3%. When vancomycin was loaded at this ratio, the particle size of the nanospheres (Figure b) and the pH of the suspension remained almost unchanged (∼7), whereas the ζ-potential of the nanospheres decreased from −30 to −20 mV (Figure c). This decrease of ζ-potential was attributed to partial compensation of the negative charge of SF by the positively charged vancomycin molecules, indicating that electrostatic interactions might be formed between the anionic groups of SF and cationic groups of vancomycin.[30]

Figure 5

Drug loading onto SFN coating. (a) Encapsulation efficiency and loading content of SF nanospheres as a function of the weight ratio. (b) Particle size and (c) ζ-potential and of SF nanospheres and drug-loaded SF nanospheres. (d) Scanning electron micrographs of SFNV and SFMV coatings. (e) FTIR spectra showing pure vancomycin powder, SFM, SFV, SFN, and SFNV coatings. Error bars represent one standard deviation (*p < 0.05).

Next, we prepared coatings from the vancomycin-loaded SF nanospheres (SFNV coatings) or the vancomycin-loaded SF molecule solution (SFMV coatings), as shown in Scheme . For comparison, we used the same vancomycin/protein ratio (12% w/w) and comparable EPD parameters for both coatings. Both coatings showed no significant difference in thickness (Figure S4). Scanning electron micrographs demonstrated that both coatings were deposited feasibly (Figure d). The FTIR results (Figure e) showed that SFM and SFMV coatings both revealed identical absorbance bands of amide I at 1624 cm–1, amide II at 1513 cm–1, and amide III at 1232 cm–128. Similarly, these three identical characteristic bands were observed at 1619, 1512, and 1228 cm–1 in both SFN and SFNV spectra, respectively. In addition, pure vancomycin, SFMV, and SFNV coatings exhibited an additional identical absorbance band at 1124 cm–1, indicating that vancomycin was successfully loaded into both SFNV and SFMV coatings. Furthermore, no distinct new bands or peak shifts were observed, suggesting that no extra chemical reactions or formation of covalent bonds occurred between SF and vancomycin during EPD.[5] Moreover, the amide I spectra of SFNV (SFMV) coatings (Figure S5, Supporting Information) show similar conformational changes during the preparation process as observed for SFN (SFM) coatings (Figure b), indicating that SFNV (SFMV) coating was assembled by the similar principles as SFN (SFM) coatings as discussed above (Scheme and Figure a,b).

Drug Release from SFN Coating

Subsequently, the release of vancomycin from SFNV and SFMV coatings in PBS was monitored by HPLC. The maximum release amount of vancomycin was 38% higher (p < 0.05) from SFNV coating than from SFMV coating (Figure a). In addition, burst release of 95% vancomycin from SFMV coatings was observed within 1 day (Figure b). In contrast, the sustained release of vancomycin was observed on SFNV coating, and 95% maximum release amount of vancomycin was released for 21 days (Figure b). Moreover, the physiological environment surrounding implants sometimes can experience a decrease in pH because of the infection, and previous studies have demonstrated that this pH decrease can further extend the release time of vancomycin from SF nanospheres.[30] These results confirmed that assembly of EPD coatings from nanospheres not only prolonged drug delivery kinetics but also enhanced the drug release amount considerably.

Figure 6

Drug Release from SFN coating. (a) Vancomycin release profiles shown as the cumulative release amount, and the dashed line indicating the maximum release amount. (b) Vancomycin release profiles shown as a cumulative release percentage and the dashed line indicating 95% maximum release amount with arrows indicating when it arrives. (c) MIC tests showing the antibacterial bioactivity of SFNV and SFMV coatings at different time points. (d) Cytocompatibility of SFNV and SFMV coatings. Vancomycin release kinetics from SFN coatings in media of different (e) ionic strength and (f) detergent concentrations. Error bars represent one standard deviation (*p < 0.05).

Furthermore, to examine the effectiveness of released vancomycin of both EPD coatings, S. aureus, one of the most common pathogenic bacteria in infections associated with surgical implants,[20] was used for our antibacterial tests. The minimum inhibition concentration (MIC) test confirmed the effective drug bioactivity of released vancomycin from SFNV coatings against S. aureus for at least 21 days. In contrast, the released drug from SFMV coatings only evoked an antibacterial effect during the first 3 days (Figure c). In addition, both of the drug-loaded coatings were not cytotoxic to fibroblasts (Figure d).

The first-order, Higuchi, and Korsmeyer–Peppas models were used fitted to the release data of SFNV and SFMV coatings (Figure S6). The correlation coefficients were 0.98–0.99, 0.91–0.97, and 0.95–0.99 for the first-order, Higuchi, and Korsmeyer–Peppas models, respectively. The correlation coefficients and visual inspection of the plots showed that the drug release kinetics from SFNV and SFMV coatings can fit with all three models, but the kinetics were more close to the first-order model release than the others. Furthermore, the values of n (0.29–0.33) of both SF coatings in Korsmeyer–Peppas models were smaller than the critical value of 0.45. This suggests that vancomycin in both coatings was released through a Fickian diffusion mechanism.[51]

Finally, we investigated the underlying mechanism by which SFNV coating facilitated sustained delivery of vancomycin over prolonged time periods. SF macromolecules consist of both positively (R and K) and negatively charged (D and E), hydrophobic (A, G, L, V, W, C, I, M, F, P), and hydrophilic (N, Q, S, T, Y, R, D, E, H, K) amino acids.[52] This composition of SF enables establishment of electrostatic and hydrophobic interactions between positively charged vancomycin molecules and SF carriers. The electrostatic interactions between vancomycin and SF nanospheres (Scheme a) were confirmed by the fact that vancomycin release was enhanced with an increase in the ionic strength of the release medium (Figure e). The electric double layer of the nanospheres was compressed with increasing ionic strength.[53,54] Consequently, positively charged sodium ions entered the Stern layer of nanospheres and competed with vancomycin molecules to form stable complexes,[54] which resulted in an accelerated release of vancomycin.

During the pre-assembly process of SF nanospheres, acetone dehydrates SF and leads to the inward folding of the hydrophobic part of SF molecules.[36] To confirm the formation of hydrophobic interactions between the hydrophobic group of vancomycin and SF nanospheres, detergent Tween 20 was added into the release media because amphiphilic Tween 20 can disrupt such hydrophobic interactions.[54] The moderate enhancement of vancomycin release with increasing Tween 20 concentration (Figure f) was a clear indication that hydrophobic interactions were formed as well between vancomycin and SF nanospheres (Scheme a).

In addition to the electrostatic and hydrophobic interactions, less initial dissolution (degradation) of the nanosphere coatings was observed within the first 24 h as discussed above (Figure e). This may also play an essential role in minimizing the burst release of vancomycin from SFN compared to conventional SFM coatings (Scheme b).

Although we have only focused on SF nanoparticles here, other polymer precursors used in EPD such as chitosan and alginate, have also been intensely investigated to assemble various types of nanovehicles for control release of drugs for biomedical applications.[55−57] We envision that this developed strategy might be applied to those materials to tailor drug release from EPD coatings to meet application-specific needs with using no (or as less as possible) additives. Moreover, this novel EPD strategy might also provide a possibility to tune drug release rate within a wide range by assembling the coatings from different ratios of molecules and pre-assembled nanostructures to meet various clinical applications in future.

Conclusions

We put forward a simple, green, and economic pre-assembly strategy to improve the drug release of polymer EPD coating without introducing any additives except the polymer precursor itself. This feasibility of this concept was demonstrated by developing a novel SF EPD coating assembled from SF nanospheres to improve the delivery drug in an application model, that is, preventing infections around percutaneous orthopedic implants via local delivery of antibiotics. The proposed EPD mechanism of SFN coating involved oxidation of water near the substrate to neutralize SF nanospheres resulting in irreversible deposition. The deposition process and mass could be easily controlled using the applied EPD parameters. Compared to conventional SFM coating, this nanostructured SFN coating had a more rough, more hydrophobic, and more stable surface with better cell response and similar adhesion strength. Most importantly, the use of nanospheres as building blocks enhanced 1.38 times on the maximum drug release amount and prolonged 21 times on drug release time (95% maximum release) while retaining drug effectiveness without detectable cytotoxicity. This superior release form SFN coatings resulted from the electrostatic and hydrophobic interactions between the drug and nanospheres, and less initial dissolution effect on nanosphere coating. These results illustrate the potential of the pre-assembly strategy on EPD polymer coatings used for drug-delivery applications.