Mechanochemistry Activated Covalent Conjugation Reactions in Soft Hydrogels Induced by Interfacial Failure

Abstract

This work reports the development of a mechanochemistry activated covalent conjugation (MACC) reaction that shows areas of interfacial failure in soft hydrogels. Hydrogels are prone to delamination from rigid substrates due to the competition between swelling and adhesion, which can lead to bonding failure in a mechanism similar to crack progapation in harder materials. In this work, reductive amination was shown to occur when a ketone-bearing fluorescein derivative was bonded to an amine-functionalized hydrogel, as both of these moieties were found to be necessary for covalent conjugation into the gel network. For thin, circular polyacrylamide hydrogels, wrinkle patterns and regions of subsequent delamination at the edge of the gel were found to be selectively tagged by the dye. This reaction was then used to explore the effect of gel properties on patterns of interfacial failure. As crosslinker loading increased, the propagation of the delamination front and the area fraction of delamination were both found to increase, as shown by fluorescence images of gels. Increasing the thickness of the gel increased the fraction of delaminated area but did not change its propagation towards the center of the gel. This MACC reaction shows how mechanochemical reactions can be used for fluorescence tagging without incorporating mechanophores into the polymer gel matrix.

Graphical Abstract

Introduction

Hydrogels are polymer networks capable of absorbing water to many times their original size, and they have been utilized for numerous applications, including imparting biocompatibility to hard materials such as implants and forming scaffolds for synthetic tissues.¹ However, exposing a constrained hydrogel to water, for example during a post-synthesis wash, leads to residual stress buildup in the network to accommodate the influx of water molecules.² In response to the stress, the hydrogel will often adopt unstable surface modes such as wrinkling, creasing, and buckling, which in turn can result in delamination from the substrate, and vice versa.^3–7 For example, Vaia and co-workers demonstrated one such phenomenon in PDMS thin film organogels in which selective swelling with an appropriate solvent resulted in out-of-plane buckling and subsequent interfacial failure.⁸ In another study, Takahashi and coworkers demonstrated photolithographic fabrication of polyacrylamide hydrogels with variable aspect ratio to direct delamination buckle patterns.⁹

According to thin film fracture mechanics theory,¹⁰ delamination of a thin film from a substrate can be treated as a interfacial failure problem,¹¹ and thus chemical labeling might be achieved through directed mechanochemical reactions. Typically, mechanochemistry in materials has focused on application of force to tough elastomers such as crosslinked PMMA;^12,13 however, the mechanochemistry of soft materials has garnered interest more recently.^14–21 Notably, Creton and coworkers²¹ demonstrated precise quantification of crack tip stresses in multi-network ethyl acrylate systems via the incorporation of mechanophores. More recent work has shown visualization of crack tip stresses in tough, double network hydrogels via mechanochemical radical induced polymerization.²² In terms of developing new mechanochemistries to label sites of material stress, Moore et al.¹⁸ and Sijbesma et al.¹⁹ have reported sensitive mechanophores in polymer networks that produce a significant colorimetric readout as a chemical response to influx of solvent. Mechanochemical production of chemical species due to delamination and contact mechanics has reported by Grybzowski et al.^23–27 They have shown, through several reports the generation of mechanoactive sites, via contact separation, and application of pressure to drive redox reactions such as bleaching of neutral dyes,²⁴ nanoparticle deposition,²⁵ and conversion of polyaniline to its conductive form.²⁷

In this work, we report a new strategy for labeling areas of interfacial failure using mechanochemistry activated covalent conjugation (MACC). The idea of MACC was inspired by affinity-mediated conjugation reactions in which a freely diffusible agent binds transiently to its target but this is affixed in place by addition of a reagent or light.^28–31 The use of MACC, rather than embedded mechanophore, would free the gel from any effects on its structure caused by the incorporation of the mechanophore and could be used on soft materials that have no inherent mechanochemical label, such as native biological tissue. While our previous work pointed to oxidation as a useful reaction following homolytic scission in water,^15,16 other work has also examined the production of reducing species such as H∙ via mechanochemical activation.^24,32 Reductive amination of imines is a well-known synthetic route to produce stable secondary or tertiary amines from carbonyls and amines.^33,34 While traditional method uses hydride equivalents like sodium borohydride or sodium cyanoborohydride, more recent work has utilized single electron transfer reactions to reduce an imine by one electron, followed by hydrogen atom transfer.^35–37 We thus hypothesized that a hydrogen radical could add to the imine, followed by hydrogen atom transfer from water or another organic source (Figure 1).³⁵ Interestingly, these reduction reactions have generally been reported from ultrasound-driven sonochemical processes rather than direct mechanical actuation of the polymer material. Thus, it may be that cavitation produces more favorable reducing conditions than homolytic scission of bonds in the material.

Figure 1.

(a) Schematic of delaminating gel releasing hydrogen and hydroxyl radicals. (b) Schematic of fluorescein pyruvamide (FPA) binding to the amine-bearing gel network. Hydrogen radical generated from the gel delamination reduces the imine, followed by hydrogen atom transfer from solvent to produce a stable amine. (c) Schematic of procedure utilized for these experiments. A blank polyacrylamide gel is swollen in the presence of FPA, then washed to reveal the locations of its attachment. (d) Fluorescence micrographs gels with (top) and without (middle) 3-aminopropylmethacrylamide; bottom: (+) 3AP gel that was peeled from the surface before delamination wrinkles could form.

Results and Discussion

First, a dye-labeled MACC reagent, fluorescein pyruvamide (FPA), was synthesized by EDC/NHS coupling of 4-aminofluorecein and pyruvic acid (Figure S1). Because this fluorescein derivative is always active, rather than a caged fluorophore, its conjugation to the gel structure could be used to validate the MACC reaction. To synthesize hydrogels, acrylamide was mixed with varying ratios of poly(ethylene glycol 400) crosslinker (PEGDA), as well as 3-aminopropyl methacrylamide (3AP). 3AP was chosen as a simple polymerizable amine, as well as a model for pendant lysines found in proteins. The mixture was initiated with APS/TEMED and known volumes were cast between hydrophobically-modified glass slides and circular coverslips (D = 25 mm). After 2 h polymerization at room temperature, the coverslips, with adhered gels, were peeled off gently and swollen in well plates containing 3 mL FPA solution for 16 h, during which time gels formed wrinkles on the coverslip due to the combination of osmotic stress and the constraints of the coverslip (Figure S2). After swelling, gels were peeled off the coverslip and washed in water for 1 h initially followed by PBS for 48 h (Figure 1c).

Each swollen gel was placed in a petri dish containing 4 v/v% ethylene glycol in water for refractive index matching and imaged using a gel imaging system (Versadoc MP 4000). With FPA, characteristic fluorescence patterns formed around the edges of the gel that corresponded to the delaminated buckles observed in bright field imaging. First, two gels were prepared with identical formulations except for the incorporation of 3AP. The addition of 3AP into the gel network was found to be necessary for permanent retention of FPA (Figure 1d). This result shows that mechanochemical conjugation is actually specific to amines rather than a nonspecific reaction to the polymer network (e.g. by radical abstraction) or noncovalent adsorption. Any changes in adhesion caused by the addition of positively charged 3AP were accounted for in a second experiment, in which the 3AP-containing gel was peeled from the substrate prior to swelling. While delamination patterns were neutralized by peeling the gel off the glass substrate, a background fluorescence was still observed in the gels (Figure 1d), again showing that 3AP provided mechanochemical reactivity with the gel framework. We also confirmed that 3AP could undergo reductive amination with pyruvic acid by ¹H NMR spectroscopy (Figure S3), and removal of oxygen by using a degassed swelling solution increased the overall fluorescence in the gel (Figure S4). Finally, in the presence of a competing primary amine in solution such as glycine (10 mM), fluorescent conjugation reduced significantly (Figure S5).

To test the mechanochemical specificity of FPA reacting with the gel, various fluorescein derivatives were tested as analogous dye systems during swelling. As shown in Figure S6, neither carboxyfluorescein nor fluorescein disodium salt showed any appreciable conjugation to the gel network, as they were completely removed after washing in PBS; 4’-aminofluorecein, the synthetic precursor for FPA, performed similarly. Conversely, dyes capable of nonspecific conjugation, such as FITC, resulted in a nonspecific conjugation throughout the gel even after washing (Figure S6). Thus, it was confirmed that the combination of FPA and amines were necessary for obtaining mechanochemical conjugation. In addition, the lack of conjugation specificity for FITC indicates that the edge detail of FPA is not due to differential swelling of the bound and delaminated parts of the gel. Further, gels conjugated with FPA were incubated in 10 mM HCl (Figure S7) in order to rule out possible reversibility of the conjugation (as one would expect in the case of imine formation). It was observed that while incubation led to near complete fluorescence quenching, the gel regained the original strong fluoresecent signal upon reimmersion in PBS.

Having established the mechanochemical reactivity of FPA towards amines, the next step was to determine which mechanical actions triggered the reaction. Thin hydrogel films weakly constrained to rigid glass substrates experience significant residual stresses when swollen due to the development of biaxial compressive forces,⁷ which can promote interfacial failure at the edges that propagates inwards with incremental loading. To determine which mechanical process – delamination, stress buildup, or both -- was responsible for mechanochemical conjugation, gels were prepared sandwiched either as before with unfunctionalized glass, or with a glass surface that had been functionalized with 3-trimethoxysilylpropyl methacrylate to promote covalent attachment to the surface.⁵ Upon imaging, the gels from the unfunctionalized glass showed the characteristic edge pattern. However, the conjugated gels showed diffuse fluorescence, even though significant² bilateral stress is expected due to the covalent anchoring (Figure 2a). Thus, delamination was primarily responsible for driving mechanochemical conjugation rather than stress buildup. To confirm this result, another experiment was run in which half of each gel was manually detached and the detached side was marked with a notch (Figure 2b). As expected, the tethered half of the hydrogel still experienced constraint-induced compressive forces leading to delamination at the edge. In contrast, the detached half the coverslip was relieved of directional compressive forces, leading to diffuse fluorescence with no distinct pattern. Interestingly, in both cases the mechanochemical labeling of the gel appeared different than the photographs, indicating that sites of interfacial failure cannot necessarily be identified by simply looking at the gel.

Figure 2.

Optical photographs (left) and fluorescence micrographs (right) of gels subjected to altered stress development. (a) Hydrogels that were attached to the surface by covalent acrylate groups developed wrinkles in bright field but did not delaminate. The fluorescence image shows the gel after removal from the glass substrate by razor blade. (b) A hydrogel that was half-detached from the surface showed stress development on the edges and diffuse fluorescence on the detached half.

This ability to use MACC reactions to permanently mark regions of gel delamination allowed us to study how hydrogel debonding from glass develops as a function of hydrogel properties. As gel stiffness increases, the delamination front will move farther from the edge and toward the center of the gel. Lake and Thomas predicted the threshold fracture energy for crack propagation in a crosslinked polymer network scales with the inverse square root of the elastic modulus of the network.¹⁰ This theory can further be extended to the molecular configuration of hydrogels where the moduli of hydrogels inversely scale with the number of monomer units between crosslinks. Thus, interfacial separation of hydrogels from the glass coverslips could be controlled by changing the crosslinking density of the network.¹¹ To test this theory, a series of hydrogels were prepared with a range of PEGDA crosslinks from 0.01 to 0.7 mol% on unfunctionalized glass coverslips. Each set of gels were swollen as before and stained with FPA. Each set showed fluorescent patterns at the edge that changed as a function of crosslinker loading (Figures S8, S9). The images were analyzed (Figures S10, S11) by comparing both the labeled distance from the edge and the labeled fraction in the gel. First, as crosslinker loading increased, the fraction of delaminated gel increased monotonically. Second, for each gel there was a critical radial distance at which the fluorescence became indistinguishable from the rest of the gel, which marked the radius of the gel still adhered to the glass substrate. As crosslinking density increased from 0.01 to 0.7, the radius of adhesion decreased significantly, demonstrating a high extent of delamination. At the same time, the total delaminated area increased, as measured by analysis of the bright fraction of the gel. These measurements were performed for gels with 20 wt% and 40 wt% loading in their initial formulations. In general, the 40 wt% gels had a higher area of delamination than the 20 wt% gels and had smaller radii of adhesion (Figure 3). However, while the dependence of delamination area on crosslinker fraction was similar for each gel, the radius of adhesion did not change significantly with crosslinker fraction for the 20 wt% gels. We attribute this observation to the low moduli (Figure S12) and higher solvent fraction of the 20 wt% gels, which would allow the gel to dissipate energy more effectively without delamination.

Figure 3.

Image analyses of 20 wt% (a-c) and 40 wt% gels (d-f). For each set, (a, d): representative fluorescence micrographs of gels at indicated crosslinker loadings. (b, e): relative radius of adhesion vs. crosslinker loading. (c, f): area fraction corresponding to delaminated gel. For (b, c, e, f): Lines are linear fits of the presented data. Data were obtained in at least triplicate, and the error bars are one standard deviation.

The thickness of the hydrogel also contributes significantly to gel delamination behavior. Generally, the thickness of the film affects the length scale of instability, thereby imparting a unique wavelength of periodicity to the unstable feature. For example, Hayward et al. demonstrated that the size of surface creases formed in high swelling constrained thin film hydrogels scales with the thickness of the film.⁵ We explored this theory in delamination behavior by fabricating hydrogels of varying thicknesses and subjecting them to osmotic swelling. As hydrogel thickness was varied from thin (0.08 mm) to thick (0.5 mm), the length of each instability on the edge of the gel (“arclength”) increased, leading to a larger lateral fluorescent labeling area (Figure 4a). Thin hydrogels showed several periodic delamination sites, leading to a much more frequent occurrence of labelled sites. Conversely, increasing gel thickness produced a larger wavelength of delamination buckles, resulting in a wider labelling sites. Quantitatively, the area of delamination increased with increasing thickness, while the radius of delamination remained fairly uniform (Figures 4b and 4c). Thus, we could conclude that the general trend in surface instabilities, as reported in several bilayer systems, could be extended to delamination buckles at the edge in soft hydrogels.^5,6,38–40

Figure 4.

Delamination behavior with varying thicknesses. (a) Fluorescence images showing gels of thickness (h) 0.08, 0.256 and 0.52 mm, measured prior to swelling. (b) Plot of radius of adhesion vs. gel thickness. (c) Plot of delamination area fraction vs. thickness. Error bars represent one standard deviation.

Conclusion

This work represents, to our knowledge, the first example of a mechanochemical conjugation reaction caused by film interfacial failure. However, mechanochemical reactions should be expected at sites of film delamination. In a thin film-on-substrate system, delamination occurs when an applied external load leads to interfacial failure, which can be treated similarly crack propagation in homogeneous materials. Mechanochemistry at sites of crack blunting was shown by Creton and workers in multiple network hydrogels,^10,17 where the additional networks within the gel were required to obtain sufficient stress buildup in the gel to allow imaging. Here, the MACC reaction is sufficient to directly label the sites of delamination in the gel without additional network formation, most likely because the glass substrate could allow the gel to build up sufficient tension prior to debonding. Imaging of the gel post-conjugation also provides a measure of how delamination occurs in a gel film, which matches the modulus and crosslinking density of the hydrogel film. Finally, we show that differences in gel thickness correlate with the total area of delamination but not necessarily the delamination distance from the edge of the gel.^2,4,5,39

Materials and Methods

General Materials.

Acrylamide (AAm), N-hydroxysuccinamide, 4’Aminofluorescein isomer I, 5(6)-Carboxyfluorescein, Fluorescein disodium salt, FITC, 3-(trimethoxysilyl) propyl methacrylate and Ethylene glycol were purchased from Acros Organics. Pyruvic acid and 3-aminopropyl methacrylamide hydrochloride (3AP) were purchased from Sigma-Aldrich. Tetramethylethylenediamine (TEMED) was purchased from ThermoScientific. Ammonium persulfate (APS) was purchased from Aqua Solutions. Polyethylene(glycol) diacrylate (PEGDA; Mw = 400) was purchased from PolySciences Inc. 1-Ethyl-3-(3-dimethylaminopropyl)-carbodiimide (EDC) was purchased from Advanced Chemtec. Hydrochloric acid (37.5%) and glycine were purchased from Fisher Scientific. Glass microscope slides (3” × 1” × 1.2 mm) were purchased from Fisherbrand. Ethanol (EtOH) was purchased from Decon Laboratories Inc. Water (dd-H2O) was deionized using Milli-Q Advantage A-10 water purification system (Millipore, U.S.A.). When required, degassing was performed by sparging with a continuous argon stream for 15 min for small volumes and 2 h for large volumes. Coverglass slips (Thickness #1, Diameter 25mm) were purchased from Electron Microscopy Sciences. Rain-X hydrophobic spray manufactured by ITW (Global Bands, TX) was purchased from Home Depot.

Synthesis of FPA.

67 μL (0.001 mol, 1 eq) pyruvic acid and 50 mL DMF were combined and stirred in a reaction flask for 5 min. Next, 0.9585 g (0.005 mol, 5 eq) EDC was added to the mixture and stirred for 15 min to ensure proper dissolution. 1.15 g (0.01 mol, 10 eq) NHS was added and the reaction mixture was purged with argon and stirred for 30 min in the dark at RT. 347.3 mg (0.001 mol, 1 eq) 4-aminofluorescein was then added to the mixture, which was stirred in the dark at RT overnight. The solvent was then evaporated in vacuo and fresh dd-H₂O was added to the crude slurry. The precipitate was collected and washed thrice with water using centrifugation at 7200 rcf for 15 min per wash. The product was then lyophilized over 48 h to retrieve a brownish-red powder and used without further purification.

Acrylate Functionalization of Coverslips.

25 mm coverslips were subjected to UV/Ozone for 1 h (PSD Series Digital UV Ozone Cleaner, Novascan). Treated coverslips were then placed in a polypropylene cover-slip holder and placed in a beaker. Next, 90 mL 95% ethanol was added to a separate beaker, into which 5–6 drops of glacial acetic acid were added to achieve a pH of 4.5–5. The mixture was stirred for 5 min, after which 0.5 mL 3-(trimethoxysilyl)propylmethacrylate was added and the mixture stirred another 5 min. The solution was then transferred to the first beaker containing the coverslips. The coverslips were incubated for 1 h and then removed and rinsed with 70% ethanol, dd-H₂O, and isopropanol sequentially. Finally, the coverslips were dried under a gentle air stream and placed in a vaccum dessicator for 10 min to remove trace solvent prior to usage.

Gel Synthesis.

Circular gels (d=25mm) were prepared using the sandwich method as reported in our previous work.¹⁵ Aqueous stock solutions of 3AP (5 mg/ml) and TEMED (0.1 M) were prepared and used for each experiment. A 10 mg/ml stock of APS was prepared fresh before each synthesis to avoid decomposition. In a typical procedure, glass slides were coated with Rain-X by spraying it onto the slide and wiping off the excess solution (x2). Next, 400 and 800 mg AAm (for 20 wt% and 40 wt% gels, respectively) were dissolved in MQ-H₂O in a 2 mL centrifugation tube. To this, PEGDA (loading adjusted as per experiment) was added along with 0.1% or 0.05% (by mol, for 20% and 40% AAm respectively) 3AP and vortexed for 30 s. Finally, APS and TEMED stock solutions were added in the mol ratio 1000:1:0.5 (Acrylamide:APS:TEMED) and vortexed for 30 s. 162 μL of the gel solution was pipetted onto the glass slide and a coverslip (as-is or acrylate-functionalized) was placed over it using tweezers. The gelation was carried out for 2 h in the dark at RT.

After gelation, the slide (not coverslip) was removed, and the gel with coverslip still attached was placed in a well of a 6-well plate. 2.7 mL 50 mM phosphate buffer (pH=7.4) was pipetted into the wells first, followed by 300 uL 10 mM FPA. For gels shown in Figure S3, FPA was replaced with the appropriate fluorescent dye. Swelling was carried out overnight (~16 hours). Gels attached to the coverslips were then removed from the wells, peeled from the coverslip and placed in 3 L distilled water with constant stirring for 1 h with one water change at the half hour. Gels were then placed in freshly prepared 3 L PBS for 48 h with constant stirring.

Measurement of Gel Shear Modulus.

Steady shear measurements were were carried out with an MCR 301 rheometer (Anton Paar) using a parallel (top) plate of 25 mm diameter and a sample gap of approximately 400 μm. The bottom plate of the rheometer containing a Peltier element was used to maintain a temperature of 25 °C for all the tests. Mechanical spectra were recorded with a constant shear rate of 0.1 s⁻¹. The reported moduli values were calculated from the slope of the linear elastic regime in the small strain limit (~2%) of the stress-strain graph. Sample preparation for mechanical characterization involved trimming the swollen samples to a dimension of 25 mm diameter, using a metallic punch cutter (Neiko hollow hole punch set). Prevention of slippage was ensured by applying a normal force of >10N. Dehydration of samples was avoided by maintaining a pool of dd-H₂O around the sample throughout the duration of these tests.

Gel Imaging.

After washing and swelling steps described above, gels were removed from the wash solutions and placed in 4 v/v% ethylene glycol/water solutions in 15 mL petri dishes to remove scattering effects. The petri dishes were then placed in a gel scanner (Versadoc MP 4000 Molecular Imaging System, Bio-Rad;Hercules CA) and images were captured at 0.8s exposure using an Alexfluor-488 filter.

Image Analysis Software.

All images in the text were analyzed using Fiji (NIH, v.1.52p) for Windows-64bit. In order to carry out the radius of adhesion analysis, an additional plugin, Radial Profile Angle, was downloaded from https://imagej.nih.gov/ij/plugins/. Prior to analysis, the image scale was set by calibrating pixel dimension to the actual gel dimension.

Radius of Adhesion Analysis.

A typical radius of adhesion analysis is as follows. TIF images of gels (exposure = 0.8 s) were loaded into Fiji. Each gel in an image was selected as a separate region of interest (ROI) upon which a radial profile angle plugin was used to map intensity over 180^o (Figure S10). The plugin measures the integrated intensity values at preset steps starting from the assigned center to the edge (circumference) of the hydrogel. Next, the radii at each step were normalized to the maximum radius of the gel. Each curve was fit to a simple cubic equation (Figure S10b and S10c). Finally, the radius of adhesion was assigned to the local minima of the intensity v/s normalized radius curve, thereby provivding an estimate of relative extent of delamination (or remaining adhesion). The local minima were calculated for individual gels in each formulation and mean radii were used in the final data set.

Analysis of Delamination Area Fraction.

Individual areas of each gel in an image were selected as separate ROI’s and measured. Next, the images were converted to a RGB image in order to enable manual color thresholding. The threshold values were manually set such that only delamination induced fluorescent regions were selected (Figure S11). Once selected, the rest of the gel was cleared, and area of selection was measured relative to total gel area. Finally, the process was repeated for the remaining formultations at the same thresholded pixel value.