Bioactive Hydrogel Coatings of Complex Substrates using Diffusion-Mediated Redox Initiation

Graphical abstract

Iron diffusion-mediated hydrogel crosslinking is a facile method to generate conformable hydrogel coatings with tunable thickness, multilayer structures, and bioactivity.

1. Introduction

Beyond mechanical failure, a lack of appropriate biological interactions including poor tissue integration, thrombosis, or infection accounts for a large fraction of medical device failures. One of the challenges in medical device design is identifying a biomaterial that has both the requisite bulk properties, such as durability, and the biofunctionality to promote desired cellular interactions. For example, small diameter vascular grafts must have requisite biomechanical properties while also maintaining thromboresistance and promoting endothelialization to ensure long-term patency.^1–3 In contrast to a single material that addresses these disparate design goals, surface modification strategies can be used to introduce biofunctionality with minimal effect on bulk properties. This approach allows for decoupling of the design criteria and optimization of each independently. Common surface modification methods include physical techniques such as etching,⁴ grit-blasting,⁵ and machining⁶ as well as chemical techniques such as vapor/plasma deposition,^{7, 8} atomic layer deposition,⁹ and electrochemical deposition.¹⁰ However, these techniques focus on surface roughness or chemistry and thus fail to modulate substrate stiffness, which can strongly influence cell behavior.^11–13

Due to their promising physiological similarity to soft tissues, hydrogels have been developed for cardiovascular^14–18 and musculoskeletal^{19, 20} tissue systems, among others.^{21, 22} Applied as coatings, hydrogels can modify the surface chemistry and stiffness of load-bearing biomaterials without affecting bulk properties. For example, a hydrogel coating of a cardiovascular device can be used to both mask an underlying thrombogenic substrate to prevent clotting and modify the bioactivity and substrate modulus to promote endothelialization.^{1, 17, 23} In the design of a hydrogel coating technique that is adaptable to a range of geometries, it is essential that the coating is both uniform and conformable to the substrate of interest. Surface grafting of hydrophilic polymers to a substrate requires chemical activation of the surface and generates thin coatings in the range of a few nanometers.^{24, 25} These thin coatings do not substantially change the mechanics of the surface, can display incomplete surface coverage, and are prone to damage.^26–28 To generate thicker hydrogel coatings (> 50 μm) that can be used to modulate both chemistry and substrate modulus, liquid precursor solutions are typically applied to the surface using molding or dip-coating and then photocured.^{1, 29, 30} Although this approach provides greater versatility in the hydrogel coating and does not require chemical modification of the construct, it is constrained to the geometries of the mold. As such, fine control of coating thickness and the ability to conform to complex geometries is limited. An adaptable coating methodology that could impart biofunctionality to diverse device geometries without requiring chemical modifications of the construct would have broad application in medical device design.

In 2009, Johnson et al. reported a method to form bioactive hydrogel coatings on 3D hydrogel substrates utilizing the Fenton reaction between hydrogen peroxide and ferrous iron (Fe²⁺).^31–33 In this technique, sucrose is swollen into the hydrogel scaffold and then the construct is placed in a solution of glucose oxidase and Fe²⁺. As the sucrose diffuses out of the construct, it reacts with glucose oxidase to generate hydrogen peroxide, which subsequently reacts with Fe²⁺ to yield peroxide radicals that can be used to initiate radical crosslinking around the substrate. Hume et al. and others reported this method was successful in modifying the surface of their hydrogels, demonstrating the ability to fabricate conformable bioactive hydrogel coatings.^{34, 35} However, this method is not broadly applicable to non-hydrogel systems. Ma et al. developed a redox-initiated crosslinking strategy based on the reaction between Fe²⁺ and S₂O₈²⁻ to form hydrogel coatings using a similar diffusion-based approach.³⁶ Materials were mixed with ferrous iron prior to 3D printing. As the Fe²⁺ leached from these 3D printed constructs, it reacted with the S₂O₈²⁻ and initiated hydrogel formation at the solid–liquid interface with tunable growth kinetics. This method expanded the application of diffusion-based, redox-initiated hydrogel coatings to a much broader set of materials and also enabled the generation of hollow hydrogel objects with complex structures.³⁷ However, the method relies on incorporation of ferrous iron into the material prior to fabrication with potential effect on bulk properties of the substrate as well as additional processing and manufacturing burdens.

We hypothesized that controlled adsorption and desorption of Fe²⁺ from a substrate could be used to generate conformable and diffusion-controlled hydrogel coatings in a similar manner but without requiring material compounding or changing the primary manufacturing method. The first step in our process is to adsorb iron (II) gluconate (IG) as a source of Fe²⁺ onto the substrate through a combination of molecular level adsorption and scaling of salt layers.³⁸ As shown in Fig. 1A, IG-coated meshes are then immersed into a solution of ammonium persulfate (APS) and a macromer solution of polyethylene glycol diacrylate (PEGDA). As the redox reagents diffuse from the surface, persulfate anions are spontaneously reduced by ferrous iron to generate sulfate anions and sulfate radicals (Fig. 1B). Sulfate radicals then initiate the vinyl groups of PEGDA, beginning free radical crosslinking and hydrogel growth from the surface. The diffusion of the IG from the surface provides controlled growth kinetics with chain propagatation leading to inreased hydrogel coating thickness. To terminate the reaction, the substrate is simply removed from the macromer solution and rinsed to remove unreacted macromer. The mild post-fabrication process of adsorbing the reducing agent to the substrate can be applied to a broad range of materials and complex geometries without affecting the bulk properties or manufacturing process. The subsequent coating step enables rapid generation of hydrogel coatings with growth kinetics readily controlled by composition and time. In addition to conformable hydrogel coatings, this method is also amenable to the incorporation of proteins at the construct surfaces to introduce biofunctionality to direct cell behavior.

Fig. 1.

(A) Diffusion-mediated crosslinking, where Fe²⁺ diffuses away from the substrate into S₂O₈²⁻ to generate radicals and crosslink PEGDA hydrogels; (B) reaction between iron gluconate and the persulfate anion to generate sulfate radicals that initiate poly(ethylene glycol) diacrylate (PEGDA) end-groups; (C) hydrogel coatings immediately after crosslinking, after swelling for 24 hours, and after drying and re-swelling.

In this work, we first established the efficacy of our redox diffusion-mediated crosslinking technique to form hydrogel coatings on electrospun meshes. The process was repeated with multiple hydrogel compositions to demonstrate the versatility of the approach. We then explored how the methodology could be used to apply hydrogel coatings to other 3D substrates. Finally, we demonstrated that the process could be used to generate multilayered hydrogel coatings and bioactive coatings that support cell adhesion via protein incorporation. Collectively, this work illustrates a new and versatile method to apply bioactive hydrogel coatings on complex substrates with utility in a broad range of biomedical applications.

2. Experimental

2.1. Materials

Reagents were purchased from Sigma-Aldrich and used without further purification unless otherwise noted.

2.2. Fabrication of Electrospun Meshes

Bionate® Segmented Polyurethane (DSM Biomedical Inc., Berkeley, CA, USA) was dissolved at 25 wt% in dimethylacetamide to generate electrospinning solutions. To fabricate electrospun polyurethane meshes, polymer solutions were pumped at a rate of 0.5 mL h⁻¹ from a 20-gauge blunted syringe needle onto a rotating mandrel positioned 50 cm from the needle tip. The mandrel was charged at −5 kV and the needle tip at +15 kV using a high voltage source. Once the desired thickness was achieved (~0.15 mm), the mandrel was removed, and the mesh was annealed at 70°C overnight. Fiber diameter was characterized with scanning electron microscopy (SEM), Fig. S1.

2.3. Synthesis of PEGDA

Poly(ethylene glycol) diacrylate (PEGDA) was synthesized as described by Browning et al. with minor modifications.¹ Triethylamine (2 mole equivalents) was added dropwise to a 0.1 mmol solution of PEG (3.4 kDa or 6 kDa; 1 mole equivalents) in anhydrous dichloromethane under nitrogen. Acryloyl chloride (4 mole equivalents) was then added dropwise. The reaction was stirred for 24 h after addition. The reaction was then washed with potassium bicarbonate (8 mole equivalents) and dried with anhydrous sodium sulfate. PEGDA was then precipitated in cold diethyl ether, filtered, and dried at room temperature overnight followed by vacuum drying. The degree of acrylation of the product was determined using proton nuclear magnetic resonance (¹H NMR) spectroscopy. Spectra were recorded on a Varian MR-400 400 MHz spectrometer and analyzed using a TMS/solvent signal as an internal reference. Polymers with percentage conversions of hydroxyl to acrylate end groups over 85% were used in this work. ¹H-NMR (CDCl₃): 3.6 ppm (m, –OCH₂CH₂–), 4.3 ppm (t, –CH₂OCO–), 6.1 ppm (dd, –CH=CH₂), 5.8 and 6.4 ppm (dd, –CH=CH₂).

2.4. Iron Gluconate Adsorption and Desorption

Electrospun polyurethane meshes were cut to 10 × 5 mm, weighed (W_m) and processed with a graded ethanol/water soak (70%, 50%, 30%, and 0%; 30 min each) prior to use to ensure wetting and penetration of aqueous solutions. Mesh specimens were placed in 3D printed clamps to prevent wrinkling and then soaked in a solution of 1, 3, or 5 wt% IG in deionized water for 15 min (n = 18). Following IG adsorption, meshes were immersed in methanol for 1 s and then dried under compressed air for 1 min to facilitate uniform drying of the highly porous, fibrous meshes. For desorption studies, coated meshes were immersed in deionized water for 2 min in 10 s intervals in a 96 well plate. IG release was monitored via absorbance at 320 nm using a plate reader (BioTek Synergy 2).³⁹ The amount of IG released from meshes was calculated using a calibration curve of IG in DI water (Fig. S2). Adsorption and desorption of 5 wt% IG was characterized utilizing the same process as above on 3D printed constructs of polylactic acid (PLA), polyethylene terephthalate glycol (PETG), and Onyx (carbon fiber-nylon) excluding the methanol immersion that was not needed to achieve uniform solution draining from the nonporous substrates.

2.5. Composite Fabrication and Characterization

To form hydrogel coatings, 5 wt% IG-coated meshes held in 3D printed clamps were immersed in aqueous solutions of PEGDA (3.4 kDa or 6 kDa, 10 wt% or 20 wt%), and ammonium persulfate (APS, 0.137 wt%, a 1:2 molar ratio to cumulative IG released after 2 min) for 10, 30, 60, or 120 s in a 96 well plate. After fabrication, composites were immediately immersed in water for 5 min to remove unreacted reagents, wicked dry, and dried in a vacuum desiccator for 24 h.

The equilibrium swelling ratio and gel fraction of hydrogel composites were determined for each composition (n = 6). First, the dry mass of the mesh (W_m) and the dry mass (W_i) of the composites were measured after vacuum drying. Specimens were then swollen in deionized water that was changed 4x over 72 h to reach equilibrium swelling and the swollen mass of the composites was measured (W_s). Equilibrium swelling ratio (Q) was calculated as Q = (W_s-W_m)/(W_d-W_m). Specimens were dried under vacuum overnight and weighed to determine the re-dried mass following extraction (W_rd) to characterize gel fraction and leachable content of the hydrogel coatings. Composite gel fraction (GF) was calculated as:

$$GF=\frac{{(W}_{rd}-W_m)}{(W_d-W_m)}\times 100$$

The weight of the mesh was removed from calculations of both swelling ratio and gel fraction in order to investigate solely the hydrogel layer.

Hydrogel coating thickness was measured for each composition (n = 12). After 72 hours of swelling and weighing, the edges of composites were removed with a razor blade to eliminate edge effects. Cross-sections of the composites (n = 4 per sample) were imaged with a stereoscope microscope and thickness was analyzed with ImageJ software (n = 12 measurements per sample).⁴⁰

2.6. Multilayered Hydrogel Coatings and Characterization

To fabricate multilayer composites, immediately after crosslinking an initial hydrogel layer (PEGDA 3.4 kDa 10%, 10 s coat), composites were immersed in a second solution containing PEGDA 3.4 kDa (10 wt%), APS (0.137 wt%) with either polymer coated silver nanoparticles (1.1 wt%, 15 nm, SkySpring Nanomaterials, Inc., Houston, TX) for 25 s or fluorescein isothiocyanate-dextran (FITC-dextran, 0.05 wt%, 2,000 kDa, 27 nm) for 16 s prior to quenching in water. Composites with nanoparticle-containing layers were sectioned and imaged with a stereoscope. Composites with FITC-dextran were sectioned and imaged with a Leica tcs sp8 Laser Scanning Confocal Microscope (excitation = 488 nm; emission = 510–530 nm). Polyurethane meshes displayed autofluorescence at both green and red wavelengths. To generate composite images, fluorescence of meshes was collected (excitation = 514 nm; emission = 550–600 nm).

2.7. Bioactive Hydrogel Fabrication and HUVEC Culture

Bioactive hydrogels were fabricated by covalently incorporating functionalized gelatin into redox coatings. Gelatin was functionalized with acrylate-PEG3400-NHS linkers (JenKem) according to established procedures at a ratio of linker to lysine residues of 1:10.^{17, 41, 42} Two types of samples were prepared on 6 × 12 mm electrospun meshes with protein either incorporated into the bulk solution or added to the surface as a final layer (n = 12 per condition). For bulk composites, 5 wt% IG-coated meshes were immersed in a solution of PEGDA (3.4 kDa, 20 wt%), APS (0.137 wt%), and functionalized gelatin (6 mg/mL) for 30 s. Surface protein composites were prepared by first crosslinking a hydrogel layer on 5 wt% IG-coated meshes with PEGDA (3.4 kDa, 20 wt%), APS (0.137 wt%), then immediately moving composites to a small volume of functionalized gelatin solution (6 mg/mL) with APS (0.137 wt%) for 30 s. All samples were immersed in sterile 10 mM phosphate buffered saline (PBS) supplemented with 1% penicillin/streptomycin after crosslinking. Solutions were exchanged 3x overnight, and samples were exposed to UV sterilization for 30 min during this time. Circular specimens (diameter = 6 mm) were punched from the samples and placed in a 96 well plate. Human umbilical vein endothelial cells (HUVECs) were expanded in HUVEC cell media (Lonza), harvested for use between passage P4-P6, and seeded at 5,000 cells per well in 96 well plates. Cell adhesion was allowed to proceed for 3 h and then specimens were washed 1x with PBS. HUVECs were fixed with 3.7% glutaraldehyde, permeabalized with 0.1% Triton-X, and stained for imaging with rhodamine phalloidin (actin) and SYBR green (DNA). Imaging was conducted with a fluorescence microscope (Nikon Eclipse TE2000-S), and cell adhesion was characterized using ImageJ software (n = 12).⁴⁰

2.8. Statistical Analysis

The data for all measurements are displayed as mean ± standard deviation. An analysis of variation (ANOVA) comparison utilizing Tukey’s multiple comparison test was used to analyze the significance of data among multiple compositions. Linearity of trends established by significantly non-zero slopes was determined utilizing linear regression analysis in GraphPad. Outliers were removed using a ROUT analysis (Q = 0.1%). All tests were carried out at a 95% confidence interval (p < 0.05).

3. Results and Discussion

Developing a method for conformable and bioactive hydrogel coatings would have broad impact in medical device design by providing a means to decouple bulk mechanical durability from surface properties that guide cell interactions. In this study, a technique for applying a conformable bioactive PEGDA hydrogel coatings on complex substrates is described. We hypothesized that controlled desorption of a reducing agent from the surface of a substrate could be used to generate hydrogel coatings with growth rate readily controlled with time and concentration. To this end, we characterized fundamental crosslinking features including desorption rates, hydrogel coating thickness, and compositional control of this new conformable hydrogel coating method.

3.1. Controlled Desorption of Iron Gluconate from Scaffolds

First, the desorption rate of IG from electrospun polyurethane meshes was characterized. Meshes were coated with varying concentrations of IG by changing the concentration of the IG soak solution (1, 3, or 5 wt%) (Fig. 2). As expected, the amount of IG that desorbs from the surface increased over time, indicating that the desorption rate can be used to control hydrogel growth kinetics. In addition, the IG concentration desorbing from the surface was also controlled by the initial concentrations of IG on the mesh substrates. After two minutes, the average cumulative release of IG from the meshes was 0.10 ± 0.03 mg, 0.25 ± 0.07 mg, and 0.47 ± 0.10 mg for 1, 3, and 5 wt% coated meshes, respectively. Differences in mass release of IG between 1 and 3 wt% coated meshes were significant at all time points (p < 0.025), and differences between 3 and 5 wt% were signficant at 30 s and later time points (p < 0.005, Fig. S3). As Fe²⁺ is prone to oxidation over time, we utilized the Ferrozine assay which specifically quantifies Fe²⁺ to validate our results (Fig. S4).^43–45 We expect that the desorption variance is due primarily to small variations in size and thickness of the mesh samples. For more precise control, substrates machined to high specifity and robotic control of submersion and drying could be employed. These results demonstrate that the adsorption and desorption of IG from electrospun mesh substrates is controllable by time and initial soak concentration.

Fig. 2.

Iron gluconate desorption from Bionate® electrospun meshes at an initial adsorption concentration of 1, 3, or 5 wt%. All data represents average ± standard deviation of n = 18.

We initially selected electrospun meshes for these studies because these materials are useful in a range of tissue engineering applications due to their low cost, versatility, and structural resemblance to features of the extracellular matrix.⁴⁶ Further, they have a wide range of applications such as vascular grafts, heart valves, and gastrointestinal patches.^{1, 23, 47, 48} However, medical devices are fabricated from a wide range of polymer materials as well as metals and ceramics.^{49, 50} To demonstrate that consistent adsorption and desorption of IG is translatable to other substrates, we repeated these experiments on 3D printed scaffolds made of poly(lactic acid)(PLA), polyethylene terephthalate glycol (PETG), and Onyx® filaments (Fig. S5). PLA, PETG, and Onyx substrates showed consistent desorption of IG over time with significant differences in IG mass desorbed between 10 and 120 s (p < 0.005). In comparison to electrospun mesh substrates, all 3D printed substrates had a higher release of IG at 10 and 20 s (p < 0.001) and a lower overall cumulative release after 120 s (p < 0.001). Differences between 3D printed substrates were not significant. The differences between mesh and 3D printed substrates were attributed to geometric and surface area effects, as variations in polymer chemistry between the 3D printed substrates did not have a significant effect. These results demonstrate the importance of considering substrate geometry and surface area for crosslinking conditions. We expect that if adsorption and desorption of hydrophilic materials were characterized, material chemistry would have a measurable effect. Overall, adsorption of a reducing agent is a simple and straightforward method for physically controlling the concentration of the redox initiating species at the device surface without requiring material compounding or surface functionalization of the substrate. This is a key advancement over previous studies that utilized diffusion-controlled redox initiation of hydrogel coatings.^{36, 51}

3.2. Preparation and Characterization of Tunable Redox Hydrogel Coatings

After achieving consistant adsorption and desorption of IG from our substrates, we proceeded to fabricate hydrogel coatings and characterize the growth kinetics as a function of the IG desorption and diffusion (Fig. 3). Two molecular weights (PEGDA 3.4 and 6 kDa) were explored, as were two concentrations of PEGDA 3.4 kDa (10 and 20 wt%). Electrospun meshes were coated in 5 wt% IG and immersed in a solution of polymer and APS set at a 1:2 molar ratio to the amount of IG released after 120 s (2.98 × 10⁻³ M IG, 5.96 × 10⁻³ M APS) for 10, 30, 60, or 120 s. Hydrogel coatings formed evenly on both sides of electrospun meshes for all compositions and did not delaminate or show large dimensional changes after reaching equilibrium swelling or after drying and re-swelling (Fig. 3A). The thinnest hydrogel coatings fabricated under these conditions were 80 ± 20 μm for PEGDA 3.4 kDa 10wt% compositions at 10 s immersion times, and the thickest composition was 440 ± 56 μm for PEGDA 6 kDa at 120 s (Fig. 3B). Within the same composition (e.g., PEGDA 3.4 kDa 10 wt%), differences in thickness across time were significant across all time points (PEGDA 3.4 kDa 10 wt%: p < 0.0001; PEGDA 3.4 kDa 20 wt%: p < 0.005; PEGDA 6 kDa 10 wt%: p < 0.05, Fig. 3A–C). These results indicate that hydrogel thickness increased with time in a near-linear process regardless of composition (Fig. S6). Thus, adsorption of the IG reducing agent to the substrate is a successful strategy for controlling hydrogel coating growth kinetics.

Fig. 3.

The effect of immersion time, PEGDA molecular weight, and polymer concentration on hydrogel coatings. (A) Hydrogel coating thickness increase between 30 and 120 s as visualized with bright field microscopy (PEGDA 3.4 kDa 10 wt%, scale bar represents 500 μm). (B) Comparison of PEGDA 3.4 kDa and 6 kDa hydrogel thickness as a function of immersion time with (10 wt%, n = 12). (C) Comparison of 10 and 20 wt% polymer concentration on hydrogel thickness as a function of immersion time (PEGDA 3.4 kDa, n = 12). (D) Hydrogel mass swelling ratios (W_s/W_d) for PEGDA 3.4 kDa 20 wt% across various immersion times (n = 18). (E) Hydrogel swelling ratios averaged across immersion times for PEGDA 6 kDa 10 wt%, PEGDA 3.4 kDa 10 wt%, and PEGDA 3.4 kDa 20 wt% (n = 72). All data represents average ± standard deviation. The * represents significant differences (** corresponds to p < 0.005 and *** corresponds to p < 0.0001) in ANOVA with Tukey’s multiple comparison test.

Each of the compositions tested displayed diffusion-based growth kinetics; however, there were some noted differences between the hydrogel compositions. A small but significant increase in hydrogel thickness was observed for the PEGDA 6 kDa as compared to the lower molecular weight PEGDA 3.4 kDa, with significant differences at all time points (p < 0.05, n = 12) and more pronounced differences at later time points, Fig. 3B (Fig. S7). It was hypothesized that this increase in thickness may be due to increased swelling of the PEGDA 6 kDa hydrogel coating due to a lower crosslink density. The equilibrium swelling ratio of PEGDA 6 kDa was 15.7 ± 1.5 as compared to 12.9 ± 2.5 for PEGDA 3.4 kDa (Fig. 3D, p < 0.0001, n = 24). It is well established that swelling ratio increases with PEGDA molecular weight due to an increase in the average distance between crosslinks.^{52, 53} In contrast, there was only a small increase in thickness with increasing polymer concentration (10 wt% to 20 wt%, 3.4 kDa PEGDA); however, these differences were largely insignificant, Fig. 3C (Fig. S7). Although there was no substantial change in thickness between these compositions, there was a marked decrease in swelling ratio from 12.9 ± 2.5 for the 10 wt% PEGDA to 8.9 ± 1.1 for the 20 wt% PEGDA (p < 0.0001, n = 24). This decrease in swelling ratio with increased polymer concentration is well established in photoinitiated systems with PEGDA and attributed to reduced intramolecular crosslinking.^{52, 54} This indicates that swelling ratio is not the sole predictor of differences in coating thickness between hydrogel compositions. Current studies are probing the effects of compositional variables and redox concentrations on the resulting network formation and growth kinetics.

After establishing temporal control of hydrogel coating thickness, we investigated the effect of immersion time and compositional variables on hydrogel coating network properties using gel fraction and equilibrium swelling ratio. Gel fractions were between 57–86% among all compositions and immersion times (Table S1). Trends across batched immersion times indicate that gel fractions are the highest for PEGDA 3.4 kDa 10 wt% at 77 ± 6% (n = 24, Fig. S8). Swelling ratios were similar across immersion times with no apparent trends in differences across time for all compositions (n = 6, Fig. 3D and Table S1). Swelling ratios are often used as a measure of gel crosslink density, and as such, the maintenance of similar swelling ratios at different thicknesses supports that the gel formation process is relatively unchanged over time. These findings highlight that this hydrogel coating technique can be used to tune hydrogel thickness while keeping the gel properties relatively constant. More rigorous characterization of network properties including mechanical testing with nano- or micro-indentation methodologies and end group conversion is needed to confirm these findings.

The desired thickness and stiffness of hydrogel coatings is specific to the application based on desired cellular interactions; thus, independent control over these factors is highly desirable for translation to multiple applications.¹² For photoinitiated hydrogel coatings, thickness is controlled by the mold used and quite limited in available thicknesses and geometries. In this technique, small adjustments to the immersion time provided fine control over hydrogel thickness without affecting gel fraction or equilibrium swelling. The conditions selected here resulted in hydrogel thicknesses from 80 ± 20 μm to 440 ± 56 μm. It is expected that increased thickness could be achieved by increasing time. Similarly, thicknesses less than 100 microns may be achieved by first reducing the adsorbed IG concentration and then iterating with shorter immersion times. In 2018, Ma et al. developed a hydrogel coating technique based on the leaching of Fe²⁺ out of 3D printed substrates into a solution of APS and monomer (acrylamide, N-isopropylacrylamide, and 2-hydroxyethyl methacrylate).³⁶ In their work, hydrogel coatings of thicknesses ranging from 20 μm to 1.4 mm were demonstrated. They utilized the pH dependence of the reduction of sulfate anions to tune hydrogel thickness with pH. Additionally, Ma and coworkers investigated the effect of the IG wt% on hydrogel thickness and found that lower concentrations lead to thinner coatings. Based on their work and the similar mechanism of the work presented here, we believe the crosslinking platform described here could be additionally tuned with pH and IG concentration.

3.3. Coating Conformability

In order to demonstrate the versatility of this diffusion-mediated redox hydrogel coating method, we fabricated hydrogel coatings on geometries and scales relevant to various medical devices. Specifically, this investigation focused on coating small-caliber, electrospun vascular grafts (4 mm wide, 1 cm long) and 3D printed stents (2 cm wide, 4 cm long, Durable Resin®). Both materials were coated with IG as described previously with the flat mesh substrates (5 wt% IG, PEGDA 3.4 kDa 10 wt%) with the exception of the immersion chamber. Hydrogel coatings successfully formed on the surfaces of both the tube and stent (Fig. 4). The ability to tune hydrogel thickness based on time enabled thin coatings on the tubular materials without occlusion. The hydrogel coating on the 3D printed stent formed only at the point of immersion, marked by the black dotted line in Fig. 4B. Hydrogel thickness on 3D printed objects is expected to be time dependent similar to the electrospun meshes given that IG release from a range of 3D printed materials demonstrated similar release profiles. For both tube and stent, hydrogel coatings were formed evenly without bumps or cohesive failure after drying and swelling. The cohesive properties of the hydrogel coating are based on interdigitation with and encirclement of the coated substrate. Nanoporous meshes provide numerous sites for interdigitation (Fig. 4A) and coatings of 3D printed materials are maintained via encirclement (Fig. 4B). In order to coat flat, non-porous substrates, pre-treatment of the substrate to generate acrylate-reactive sites would likely be needed to enable coating with this methodology. These results demonstrate that IG desorption can be used to generate diffusion-controlled hydrogel coatings on multiple material chemistries and geometries.

Fig. 4.

Hydrogel coatings on various construct geometries. (A) Tubular electrospun mesh. (B) 3D printed stent. Dotted line indicates immersed region. Scale bar = 2 mm.

Medical devices range widely in size from less than a millimeter to a few centimeters. Small and large-scale coatings can be fabricated with photo-initiation strategies but are limited by the necessity of correspondingly small or large-scale transparent molds. Our lab has previously developed hydrogel coated electrospun tubes for application as small diameter vascular grafts for coronary artery disease.¹ We utilized a molding and UV crosslinking strategy to form hydrogel coatings and found that it was difficult to consistently form thin coatings that evenly covered the entirety of the tubes. This redox-based crosslinking technique enables even, thin coatings previously difficult to access, allowing for the fabrication of small diameter vascular grafts that do not risk exposure of the underlying substrate. Many other medical devices like heart valves and stents also have complex shapes that would be difficult to coat with hydrogels using traditional molding-based techniques. The principles of diffusion-controlled redox initiation to control the hydrogel coating growth kinetics from the device surface opens the door for a wide-range of applications in device design.

3.4. Hydrogel Multilayers

We next hypothesized that this process could be adapted to fabricate multilayer hydrogel coatings. When fabricating single-layered hydrogel coatings, radicals are quenched post-crosslinking by soaking composites in water. We hypothesized that multilayer structures could be formed by subsequent immersion in a second polymer solution with APS. This is enabled by continued diffusion of IG retained on the mesh and in the initial hydrogel layer that initiates crosslinking of a second hydrogel layer. As proof of concept, PEGDA 3.4 kDa 10 wt% coatings were immersed in a second solution of PEGDA 3.4 kDa 10 wt% containing either FITC-Dextran 20 kDa or silver nanoparticles for visualization (Fig. 5A). The first layer crosslinked as normal and multiple layers clearly formed as visualized with bright field and fluorescence microscopy (Fig. 5B). The formation of multiple hydrogel layers indicates that IG can diffuse through the first crosslinked network in sufficient amounts for crosslinking to occur. The second hydrogel layer demonstrated different growth times in hydrogel thickness, but the thickness of the second layer can be tuned similarly with time as the first layer. Future studies will investigate the growth kinetics of these subsequent layers in more detail. The clear line between layers visible by the encapsulation of FITC-dextran in Fig. 5B indicates that there is distinct layer separation. The lack of fluorescence in the first layer indicates that there was limited diffusion of the second hydrogel solution into the first hydrogel layer.

Fig. 5.

Fabrication of multilayered hydrogel coatings. (A) Hydrogel multilayers form by transferring hydrogel-coated substrates directly into a second macromer and initiator solution, foregoing a washing step to remove radicals. (B) Multilayered hydrogel coating visualized by entrapping silver nanoparticles or (C) FITC-dextran 2,000 kDa. Scale bar = 100 μm.

Hydrogel coatings can tune both surface chemistry and surface stiffness of underlying substrates to mimic complex cellular microenvironments. The ability to form multilayered hydrogel substrates lends more complexity to substrate modulation, enabling spatial regulation of cellular interactions.³⁴ Hydrogel multilayers have been utilized in drug delivery as controlled release applications^55–58 as well as directly with cells in applications such as pancreatic islet encapsulation.⁵⁹ The ability to generate multilayers with a conformable crosslinking technique enables a wide range of hydrogel functionalities and increases the potential impact of this approach.

3.5. Bioactive Hydrogel Coatings

In order to demonstrate that diffusion-mediated redox initiation could be utilized to fabricate hydrogel coatings for biomedical use, an initial cytocompatibility assay of potential leachable components was performed. We found that neither the uncoated nor coated material had any negative effect on HUVEC viability via an indirect assessment of extractables cytotoxicity (Fig. S9). We then investigated the incorporation of bioactive proteins to establish the utility of this method to generate bioactive coatings and confirmed bioactivity with cell attachment studies. Initially, acrylate-functionalized gelatin was incorporated into the polymer (PEGDA 3.4 kDa, 20 wt%) and APS solution prior to diffusion-mediated crosslinking to form “bulk” bioactive coatings (Fig. 6A). HUVEC adhesion was characterized over four hours on bulk specimens in comparison to hydrogel coatings without incorporated protein (“blank”). Cell adhesion was observed at a density of 83 ± 30 cells per mm² of culture area for bulk coatings (Fig. 6B), significantly higher than cell adhesion on blank coatings at 16 ± 15 cells per mm² (n = 12, p < 0.0001). Although this finding indicates that this approach is amenable to conferring bioactivity to the hydrogel coating, adding protein into the bulk crosslinking solution requires a large amount of protein—in our studies, almost two milligrams per sample. For expensive designer proteins, this poses a significant production cost. Further, only proteins incorporated close to the surface of the hydrogel will be able to interact with cells, meaning most of the protein within the bulk of the hydrogel is unavailable for the desired cell interactions and unnecessary (Fig. 6A). In order to improve the efficacy of the proteins used, functionalized proteins were next incorporated at the surface of composites instead of in the bulk (Fig. 6C) by using the multilayer crosslinking method to react proteins specifically at the surface of hydrogel coatings. This method limits the total amount of protein needed by a factor of five. HUVEC adhesion on surface-functionalized bioactive coatings was 80 ± 20 cells per mm² (Fig. 6D), significantly different from blank specimens (n = 12, p < 0.0001), indicating bioactivity can successfully be incorporated in this manner. There were no significant differences in cell attachment between bulk and surface crosslinking methods despite less overall protein used. Bioactive proteins can therefore be incorporated into diffusion-mediated redox hydrogel coatings either in bulk solution or at the surface with similar results.

Fig. 6.

Cell adhesion with bulk vs. surface protein incorporation and representative HUVEC adhesion images (green = SYBR green, red = rhodamine phalloidin, scale bar = 250 μm). (A) Bulk incorporation and (B) HUVEC adhesion. (C) Surface incorporation and (D) HUVEC adhesion.

4. Conclusion

Hydrogel coatings offer unique opportunities to introduce biofunctionality to medical devices with minimal effect on bulk properties; however, it remains challenging to uniformly apply hydrogel coatings without substantially changing the material compounding or manufacturing process. The studies presented here demonstrate a new redox-based crosslinking methodology that enables the formation of conformable and bioactive hydrogel coatings with tunable thickness and chemistry. Adsorption of the redox reagent IG was successfully applied to multiple surfaces and geometries without changing the underlying material chemistry or fabrication process. This is a key advance over previous diffusion-based redox crosslinking systems that relied on material compounding or were restricted to hydrogel-based systems. Hydrogel coating thickness was then readily controlled by immersion time with desorption and diffusion of the reducing agent initiating hydrogel crosslinking from the surface. We also demonstrated that this method could be extended to generate hydrogel multilayer coatings and incorporation of bioactivity at the surface of hydrogel coatings. Overall, this work provides a versatile method for assembling bioactive coatings with a simple post-fabrication process that is amenable to diverse geometric substrates and chemistries.