Microgel-Assisted Delivery of Adenosine to Accelerate Fracture Healing

Abstract

Extracellular adenosine plays a key role in promoting bone tissue formation. Local delivery of adenosine could be an effective therapeutic strategy to harness the beneficial effect of extracellular adenosine on bone tissue formation following injury. Herein, we describe the development of an injectable in situ curing scaffold containing microgel-based adenosine delivery units. The two-component scaffold includes adenosine-loaded microgels and functionalized hyaluronic acid (HA) molecules. The microgels were generated upon copolymerization of 3-acrylamidophenylboronic acid (3-APBA)- and 2-aminoethylmethacrylamide (2-AEMA)-conjugated HA (HA-AEMA) in an emulsion suspension. The PBA functional groups were used to load the adenosine molecules. Mixing of the microgels with the HA polymers containing clickable groups, dibenzocyclooctyne (DBCO) and azide (HA-DBCO and HA-Azide), resulted in a 3D scaffold embedded with adenosine delivery units. Application of the in situ curing scaffolds containing adenosine-loaded microgels following tibial fracture injury showed improved bone tissue healing in a mouse model as demonstrated by the reduced callus size, higher bone volume, and increased tissue mineral density compared to those treated with the scaffold without adenosine. Overall, our results suggest that local delivery of adenosine could potentially be an effective strategy to promote bone tissue repair.

Graphical Abstract

1. INTRODUCTION

Bone fractures are prevalent injuries with more than 8 million occurring each year in the United States alone.¹ Although bone tissue possesses an intrinsic regenerative potential, approximately 5–10% of all fractures exhibit impaired healing.^2,3 The incidence and burden of both traumatic bone fractures and fractures with compromised healing are predicted to increase with the aging population. The growing clinical need demands the development of effective therapeutic approaches to promote bone tissue regeneration and healing. Although autografts remain the gold standard for treating bone defects, delivery of growth factors (e.g., BMP, PTH) and small molecules (e.g., agonists of wnt/β-catenin or the prostaglandin E2 EP4 agonist, activators of the BMP/SMAD or PKA signaling) and RNAi therapeutics have been extensively studied to promote bone healing.^4–11 Such approaches that directly activate repair mechanisms have the potential to create translatable regenerative therapies. Toward this, small molecules are increasingly being used to advance regenerative medicine strategies. The application of small molecules could either be used to support cell transplantation therapies (by inducing cell-specific differentiation) or augment endogenous cell-mediated repair. Therapeutic interventions that activate the innate repair mechanism of the native tissue and promote healing are of particular interest.

We and others have demonstrated the key role played by extracellular adenosine and adenosine receptor signaling—A2A and A2B receptors— in bone tissue regeneration and homeostasis^11–17 Adenosine is a naturally occurring nucleo-side with multiple functions relevant to bone tissue healing. Given the key role played by adenosine signaling in bone tissue regeneration and osteogenic differentiation,^16,18–21 we have recently leveraged a biomaterial-based approach to sequester endogenous adenosine at the fracture site following injury.²² Localization of adenosine following injury not only promoted osteogenesis but also promoted early vascularization. Although leveraging endogenous adenosine is an attractive strategy, achieving an adequate amount of endogenous adenosine is challenging. Furthermore, diseases like osteoporosis are associated with diminished levels of endogenous adenosine.¹¹ Hence, strategies that can deliver exogenous adenosine to promote fracture healing are needed. Biomaterial-assisted local delivery of adenosine could mitigate challenges associated with systemic administration of adenosine including its short half-life in circulation. Herein, we have developed an in situ-forming scaffold containing microgels loaded with adenosine to facilitate adenosine delivery at the fracture site. In addition to optimizing microgel loading, we have also used the scaffold system to examine the effect of local delivery of adenosine on fracture healing using a tibial fracture model.

2. MATERIALS AND METHODS

2.1. Materials.

Hyaluronic acid (HA, molecular weight 40 kDa) was purchased from Lifecore, USA. 2-Aminoethylmethacrylamide (2-AEMA) was purchased from Polysciences, USA. N-Hydroxysuccinimide (NHS), sodium hydroxide (NaOH), adenosine, and mineral oil were obtained from MilliporeSigma, USA. 1-Ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC.HCl) and 3-acrylamido phenylboronic acid (3-APBA) were obtained from TCI Chemicals, USA. Dibenzocyclooctyne (DBCO)-PEG4-amine was purchased from Click Chemistry Tools, USA. 11-Azido-3,6,9-trioxaundecan-1-amine was purchased from Lumiprobe, USA. Dialysis bags (Molecular weight cutoff, MWCO 3.5 kDa) were obtained from Spectrum, USA. The ABIL EM90 surfactant was obtained from Universal Preserv-A Chem INC, Germany. Hexane, acetone, ethanol, and dimethyl sulfoxide (DMSO) were purchased from MilliporeSigma, USA; the solvents were of ACS or spectroscopic grade and used as received. UV–visible spectra were recorded via a Genesys 10S UV–Vis spectrometer. A Thermo Electron Nicolet 8700 FTIR spectrometer was used to record the Fourier-transform infrared (FTIR) spectra. A 500 MHz Agilent/Varian VNMRS spectrometer was used to record the NMR spectra.

2.2. Synthesis of 2-Aminoethylmethacrylamide (2-AEMA)-Conjugated Hyaluronic Acid (HA-AEMA).

HA-AEMA was prepared by conjugating 2-AEMA to HA via an amide coupling reaction between the carboxylic acid groups of HA and the primary amine groups of 2-AEMA. In brief, HA was dissolved in 2-(N-morpholino)ethanesulfonic acid buffer (pH 5.5) at a concentration of 10 mg/mL. EDC.HCl (1 equiv with respect to the carboxylic acid groups of HA) and NHS (1 equiv with respect to the carboxylic acid groups of HA) were added to the HA solution at 15 min intervals. 2-AEMA, (1 equiv with respect to the carboxylic acid groups of HA) was then added to the reaction mixture. After 24 h of reaction, the resulting product was purified by dialysis and lyophilized. Successful conjugation of 2-AEMA to HA was verified by FTIR and proton nuclear magnetic resonance (¹HNMR) spectroscopy.

2.3. Synthesis and Characterization of Microgels.

Microgels were generated by inverse emulsion suspension polymerization.¹⁴ To synthesize PBA-functionalized microgels containing adenosine, 3-APBA was dissolved in water at pH ~10, which was adjusted using NaOH. To this solution, an equimolar amount of adenosine was added and vortexed extensively to obtain a clear solution of the 3-APBA-adenosine complex. Next, HA-AEMA and lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) were added to the 3-APBA-adenosine complex to obtain a solution of 0.5 M 3-APBA, 0.5 M adenosine, 5 wt % HA-AEMA, and 0.25 wt % LAP, respectively. The mixture was then transferred into a continuous organic phase of mineral oil containing the 10 wt % ABIL EM-90 surfactant, at the 1:10 (v/v) ratio, under a constant stirring of 50–100 rpm. The suspension was subjected to UV irradiation at ~365 nm for ~10 min to intra-cross-link the microdroplets and obtain stable microgels. The microgels were subsequently washed with a series of hexane (3 × 25 mL), isopropanol (2 × 25 mL), and deionized water (2 × 25 mL) and freeze-dried. These microgels pre-loaded with adenosine are termed HA-PA1. Another group of microgels was used to assess the post-loading of adenosine, which involved formation of microgels using HA-AEMA and 3-APBA followed by incubation in adenosine solution. To examine post-loading of adenosine, PBA-functionalized microgels were synthesized by dissolving 3-APBA in water at an alkaline pH (pH ~10). HA-AEMA and LAP were separately dissolved in water. 3-APBA and LAP solutions were then added to the HA-AEMA solution to obtain a mixture of 0.5 M 3-APBA, 5 wt % HA-AEMA, and 0.25 wt % LAP, respectively. The mixture was then transferred into a continuous organic phase of mineral oil supplemented with the 10 wt % ABIL EM-90 surfactant, at a 1:10 (v/v) ratio under constant stirring (50–100 rpm), and cross-linked by UV irradiation as described earlier. The microgels were washed with a series of hexane (3 × 25 mL), isopropanol (2 × 25 mL), and deionized water (2 × 25 mL) and freeze-dried. Adenosine loading into the microgels was then performed by suspending the microgels in 6 mg/mL adenosine solution in phosphate buffered saline (PBS) (pH ~7.4) and incubating overnight at room temperature. After the incubation, the adenosine-loaded microgels were washed with water (2 × 15 mL) and freeze-dried. These microgels are termed HA-PA2. Microgels without 3-APBA and adenosine were prepared similarly and termed HA-PA0. The microgels were characterized via FTIR spectroscopy for detection of the surface functional group (hydroxyl, C=O, C–H, C=C from HA, PBA, and/or adenosine), UV/visible spectroscopy (UV/vis) for quantification of PBA and adenosine contents, and optical microscopy for particle size characterization.

2.4. Determination of the Adenosine Content in the Microgels.

The adenosine content within the microgels (HA-PA1 and HA-PA2) was measured by using UV/vis spectroscopy. The microgels were incubated in acetate buffer (0.1 M, pH 3.5) for 12 h to extract the adenosine, and the adenosine content in the acetate buffer was then measured at 260 nm against the adenosine standards. Briefly, the microgels were suspended in acetate buffer at 5–10 mg/mL and incubated at 37 °C under constant shaking for about 12 h. After incubation, the microgels were centrifuged down at 5000 rpm for 5 min. The supernatant was collected, and the adenosine content in the buffer solution was measured via a UV/vis spectrophotometer at 260 nm. A standard calibration curve of absorbance versus concentration of adenosine (3.9–125 μg/mL) was generated at 260 nm.

2.5. Synthesis of Clickable HA Polymers (HA-Azide and HA-DBCO).

11-Azido-3,6,9-trioxaundecan-1-amine and DBCO-PEG4-amine were independently conjugated to HA to synthesize HA-Azide and HA-DBCO, respectively. Briefly, HA was dissolved in water at 10 mg/mL. To this, excess EDC.HCl and NHS (1.2 equiv each with respect to the carboxylic acid group of HA) were added at an interval of 15 min to activate the carboxylic acid groups of HA. To synthesize HA-Azide, 11-azido-3,6,9-trioxaundecan-1-amine, 0.5 equiv with respect to the carboxylic acid group of HA, was added to the reaction mixture. The reaction was continued for 24 h, and the resulting product was dialyzed against water and freeze-dried. To synthesize HA-DBCO, DBCO-PEG4-amine (0.5 equiv with respect to the carboxylic acid group of HA) dissolved in DMSO was added to the activated HA solution. The reaction was continued for 48 h, and the resulting product was dialyzed against water and freeze-dried. Successful conjugation of 11-azido-3,6,9-trioxaundecan-1-amine and DBCO-PEG4-amine was verified by FTIR and ¹HNMR spectroscopy.

2.6. Development of Microgel-Embedded Scaffolds.

Scaffolds containing microgels were prepared by mixing the microgels with clickable HA polymers (HA-Azide and HA-DBCO). HA-Azide and HA-DBCO polymers were individually dissolved in water at 40 mg/mL. Freeze-dried microgels were then suspended in the HA-Azide polymer solution, and to this, HA-DBCO solution was added and vortexed to obtain the 3D scaffolds. Scaffolds with two different weight ratios were prepared using the microgels (microgel:HA-Azide:HA-DBCO = 5.0:1.0:1.0 or 7.5:1.0:1.0) and termed sHA-PA0, sHA-PA1, and sHA-PA2 for HA-PA0, HA-PA1, and HA-PA2 microgels, respectively. The gelation time was determined via the vial inversion method, that is, the time at which the polymer–microgel mixture stopped flowing freely. An AR G2 rheometer (TA instruments) with a standard steel parallel-plate geometry of 8 mm diameter was used for the rheological characterization of all the samples at 25 °C. The test methods employed were oscillatory strain sweep, time sweep, and frequency sweep. First, a strain sweep (0.1–10% strain) at 1 Rad/s was used on the microgel (HA-PA1, 5 wt %) or hydrogel samples (HA-PA1 microgel with cross-linkable HA polymers, HA-Azide, and HA-DBCO) to determine the elastic modulus, G′, and the loss modulus, G″. The linear viscoelastic region (LVR) was determined by regions at which no significant changes of G′ were observed in the strain sweep measurements. A time sweep was then performed at 0.5% strain in the LVR at 1 Rad/s to monitor the in situ gelation behavior of the microgel or the hydrogel samples. The test, which was terminated after 2500 s, enabled monitoring of the evolution of the elastic modulus, G′, and the loss modulus, G″, with time. The microgel or hydrogel samples were also subjected to a frequency sweep in the LVR at 0.5% strain to study the viscoelastic performance over a wide range of frequencies (0.1–10 Rad/s). The G′ and G″ values following strain, time, or frequency sweep measurements were obtained directly from the software (Rheology Advantage) in the rheometer. All samples were analyzed in triplicate. For minimally invasive application, the mixture of the microgels and the clickable HA polymers was placed into a syringe following mixing and injected before gelation.

2.7. Release Kinetics of Adenosine from the Scaffolds.

Microgel-based scaffolds (sHA-PA1 and sHA-PA2) were prepared by using adenosine-loaded HA-PA1 and HA-PA2 microgels. Briefly, the microgels and the clickable polymers (HA-Azide and HA-DBCO) were mixed (at the 5.0:1.0:1.0 ratio) in phosphate buffer or alpha-minimum essential medium (α-MEM) containing 10% fetal bovine serum (FBS). The scaffolds were prepared in a cylindrical polypropylene mold by taking 100 μL of the microgel–polymer mixture. The scaffolds were then transferred into 1.7 mL Eppendorf tubes and incubated with either 1 mL of phosphate buffer or 1 mL of α-MEM containing 10% FBS. At predetermined time intervals, 200 μL of the medium (phosphate buffer or MEM medium) was removed from the tube and supplemented with 200 μL of fresh corresponding medium. Adenosine release from the scaffolds was determined by quantifying the adenosine content in the buffer or medium through UV/Vis absorption spectroscopy as a function of time. A standard calibration curve of absorbance (at ~260 nm) versus adenosine concentration (3.9–125 μg/mL) was used to calculate the adenosine content in the released medium.

2.8. Tibial Fracture and Implantation of the Scaffolds.

All animal studies were performed with the approval of the Institutional Animal Care and Use Committee (IACUC) at Duke University and in accordance with the guidelines of the National Institutes of Health (NIH). The tibial fractures were performed as described earlier.²² In brief, C57BL/6J mice (3 month old female procured from the Jackson Laboratory) were sedated with 2% isoflurane and administered with buprenorphine (1 mg/kg, sustained release, ZooPharm). Each mouse was placed in a supine position with the right tibia disinfected. After removal of the skin proximal to the right knee, a 0.7 mm pin was placed from the tibial plateau through the medullary cavity to stabilize the tibia, and a cut was made at the tibial midshaft. The mixture of the cross-linkable polymers (HA-Azide and HA-DBCO) containing ~2 mg of the HA-PA0 or HA-PA1 microgel was injected into the fracture site and allowed to cure. Two drops of bupivacaine (0.5%, Hospira) were applied at the surgical site followed by the wound closure. The mice were sacrificed at 21 days post-fracture, and the right tibiae were disarticulated and trimmed to remove the soft tissue.

2.9. Microcomputed Tomography (μCT).

Fractured tibiae were fixed in 4% paraformaldehyde and scanned using a microcomputed tomography (μCT) scanner (vivaCT 80, Scanco Medical) at 55 keV and 10.4 μm voxel size.²³ The scanned images were reconstructed using μCT Evaluation Program V6.6 (Scanco Medical) with a threshold of 513 and further processed by using μCT Ray V4.0 (Scanco Medical). Fracture healing was determined based on the percentage of bone volume over total volume (BV/TV), bone mineral density (BMD), and tissue mineral density (TMD) in the tibial region within 1 mm proximal and 1 mm distal from the fracture site.

2.10. Histological Staining.

Following μCT imaging, the samples were decalcified in 14% ethylenediaminetetraacetic acid (pH 8.0), embedded in paraffin blocks, and cut into 5 μm-thick sections by using a microtome for histological analyses. The sections were deparaffinized and rehydrated in a series of CitriSolv (Decon Labs, catalog number 1601), a reverse ethanol gradient, and deionized water. To visualize the callus remodeling, tissue sections were stained with 1% Safranin-O (Sigma-Aldrich, catalog number S8884) for 1 h and counter-stained with 0.02% Fast Green (Sigma-Aldrich, catalog number F7258) for 1 min. To assess the osteoclastic activity, the sections were incubated in 50 mM tartaric acid (Sigma-Aldrich, catalog number 228729) solution together with 0.5 mg/mL naphthol AS-MX phosphate (Sigma-Aldrich, catalog number N5000) and 1.1 mg/mL fast red TR (Sigma-Aldrich, catalog number F6760) for 1 h at 37 °C. The sections were then dehydrated in an ethanol gradient, mounted with Cytoseal (Thermo Scientific, catalog number 23–244256), and imaged using a Keyence (BZ-X710) microscope.

2.11. Statistical Analysis.

All data were evaluated for the statistical significance using either two-tailed Student’s t-test or oneway analysis of variance (ANOVA) with post hoc Tukey’s test in GraphPad Prism 8.1.1. A P-value less than 0.05 was considered statistically significant. All the results were presented as mean with standard deviations (n ≥ 3). All experiments were independently reproduced.

3. RESULTS

3.1. Synthesis and Characterization of Modified HA Polymers.

Methacrylated HA (HA-AEMA) was synthesized by reacting 2-AEMA with the HA molecules via an amide coupling reaction between the carboxylic acid group of HA and the amine group of 2-AEMA (Scheme S1). The conjugation of 2-AEMA to HA was confirmed by FTIR spectroscopy, which showed the presence of a peak at 1643 cm^–1 corresponding to C=C bond absorption of 2-AEMA and C=O and N–H stretching frequencies of amide groups at 1675 and 1570 cm^–1,respectively (Figure S1). The degree of conjugation was determined using ¹HNMR spectroscopy. In ¹HNMR spectra, the peaks corresponding to the native HA, such as −NHCOCH₃ protons at 1.9 ppm, C2–C6 protons of the HA disaccharide unit at 3.1–4.0 ppm, and C1 protons at 4.3–4.5 ppm, were compared with the newly appeared peaks at 5.8–6.2 ppm corresponding to the vinyl protons of methacrylamide groups. The degree of substitution, calculated by taking the ratio of the integrations of the vinyl protons of 2-AEMA to the methyl protons of the −NHCHCH₃ group of HA, was found to be 15 ± 2% (Figure S2). Separately, HA molecules were also modified with clickable functional groups (azide and DBCO) to obtain HA-Azide and HA-DBCO and were prepared by conjugating 11-azido-3,6,9-trioxaundecan-1-amine and DBCO-PEG4-amine to HA via amide coupling reactions (Scheme S2). The detailed characterization of the polymers (HA-Azide and HA-DBCO) is provided in the Supporting Information (Figures S3–S5). The conjugations of 11-azido-3,6,9-trioxaundecan-1-amine and DBCO-PEG4-amine to HA were confirmed by FTIR spectroscopy. The spectra showed the presence of peaks at 2097 cm^–1 corresponding to N≡N bond absorption from the azide group of HA-Azide and 2158 cm^–1 corresponding to C≡C bond absorption from the DBCO group of HA-DBCO (Figure S3). The ¹HNMR spectrum of HA-Azide showed the appearance of peaks at 2.89–2.92 ppm and 3.10–3.29 ppm corresponding to the −O(CH₂CH₂)N₃ and −O(CH₂CH₂− protons, respectively, of the 11-azido-3,6,9-trioxaundecan-1-amine moiety in addition to the peaks at 4.46–4.56 ppm corresponding to C1 protons of the sugar ring of HA (Figure S4). The ¹HNMR spectrum of HA-DBCO showed the presence of characteristic peaks at 7.37–7.52 ppm corresponding to the aromatic protons of the DBCO group in addition to the peaks at 4.46–4.55 ppm corresponding to C1 protons of the sugar ring of HA (Figure S5). The degrees of azide and DBCO functionalization were ~35% in HA-Azide and ~33% in HA-DBCO, respectively.

3.2. Synthesis of Boronate Containing Hyaluronic Acid Microgels.

HA microgels containing boronate functional groups were generated via an emulsion suspension copolymerization of HA-AEMA and 3-APBA (Figure 1a).^14,24 Two different types of adenosine loadings, pre-loading and post-loading, were examined to determine the effect of the loading procedure on the adenosine content within the microgels. For the pre-loading, a 3-APBA-ADO conjugate was prepared prior to copolymerization by mixing 3-APBA with adenosine at pH 10, resulting in a stable complex of hydroxyboronate with cis-diol groups of adenosine (Figure 1b).^25,26 Adenosine-loaded microgels, termed HA-PA1, were then generated by copolymerizing the 3-APBA-ADO complex with HA-AEMA in an emulsion suspension (Figure 1c). For the post-loading, HA-AEMA-co-3-APBA microgels (named HA-PA2) were first fabricated, and then, adenosine was loaded by suspending the microgels in adenosine solution in phosphate buffer at pH 7.4 (Figure 1d).

Figure 1.

Microgel synthesis and characterization. (a) Schematic representation of microgel synthesis. Microgels are generated by adding solution of precursors (HA-AEMA and 3-APBA with or without adenosine) into mineral oil upon stirring. The bright-field image shows the size of microdroplets created in the emulsion suspension. Scale bar 200 μm. (b) Reaction scheme showing in situ complexation between 3-APBA and adenosine (ADO) at pH 10 and formation of the 3-APBA-ADO conjugate. (c and d) Schematic representation of HA-PA1 and HA-PA2 microgels and adenosine binding with PBA (highlighted in color). (e) FTIR spectra of the microgels. The data is represented in terms of transmittance (a.u.) vs wavenumber (from 2000 to 1000 cm^–1). (f) Microscopic images of the microgel suspensions. Freezedried microgels were suspended in water, and the images were recorded using optical microscopy. Left panel: HA-PA1 and right panel: HA-PA2. The inset shows the particle size distributions for both HA-PA1 and HA-PA2 microgels. Scale bar 200 μm.

The microgels were characterized via a combination of FTIR, UV/Vis, and optical microscopy. FTIR spectra of the microgels showed peaks at 1611, 1581, 1547, and 1489 cm^–1 which are characteristic C=C stretching frequencies of the benzene ring corresponding to PBA, indicating successful incorporation of PBA moieties (Figure 1e). The extent of PBA incorporation into the microgels was determined from the UV/Vis spectra and found to be ~95% (with respect to the amount of 3-APBA used for copolymerization with HA-AEMA). Microscopic images of the microgels showed spherical particles with a diameter ranging between 50 and 250 μm with most of the particles having a diameter of 130 ± 20 μm for both the HA-PA1 and HA-PA2 microgels (Figure 1f).

3.3. Adenosine Loading Efficacy.

The loading efficacy of the pre-loaded and post-loaded microgels (HA-PA1 and HA-PA2) was determined by incubating the microgels in acetate buffer at pH 3.5. The amount of released adenosine in the buffer was then measured by using UV/Vis spectroscopy. A higher amount of adenosine was detected in the pre-loaded (HA-PA1) microgels compared to that in the post-loaded (HA-PA2) microgels. Specifically, ~466 ± 21 μg of adenosine in 1 mg of HA-PA1 and ~276 ± 40 μg of adenosine in 1 mg of the HA-PA2 microgel were found to be present for a similar PBA content (Figure 2a). Adenosine loading efficiency was also studied by varying the PBA content in the microgels using the pre-loading approach. The adenosine content in the microgels (HA-PA1 group) increased from 76 ± 10 to 450 ± 21 μg/mg when the PBA content was increased from 0.125 to 0.5 M (Figure 2b).

Figure 2.

Adenosine loading efficacy into the microgel. (a) Amount of adenosine encapsulated in the pre-loaded (HA-PA1) and post-loaded (HA-PA2) microgels. (b) Amount of encapsulated adenosine in the HA-PA1 microgel at different PBA concentrations.

3.4. In Situ Scaffold Formation with Adenosine Delivery Units.

To assemble the scaffolds containing microgels, the HA-Azide polymer solution containing either HA-PA1 or HA-PA2 microgels was mixed with the HA-DBCO polymer solution, wherein the click reaction between the DBCO and azide groups formed stable triazole linkages (Figure 3).²⁷ The strain-promoted alkyne–azide cycloaddition (SPAAC) reactions allowed simultaneous gelation and entrapment of the microgels within the network. Three different scaffolds— sHA-PA0 containing HA-PA0 microgels with no adenosine, sHA-PA1 containing HA-PA1 microgels, and sHA-PA2 containing HA-PA2 microgels—were generated. The effect of the microgel entrapment on the gelation was also examined by varying the microgel amount while keeping the amounts of HA-DBCO and HA-Azide constant. Between the two different weight ratios tested (microgel: HA-DBCO: HA-Azide = 5.0:1.0:1.0 or 7.5:1.0:1.0), mixtures with a higher microgel to polymer weight ratio (microgel: HA-DBCO: HA-Azide = 7.5:1.0:1.0) required longer gelation time (~360 s) compared to those with a lower microgel to polymer weight ratio (microgel: HA-DBCO: HA-Azide = 5.0:1.0:1.0) (gelation time ~240 s) (Table S1). The mixture with the 5.0:1.0:1.0 ratio of microgel to polymers resulted in robust 3D structures and was used for subsequent experiments including rheological measurements (Figure 3d). The injectability of the polymer–microgel mixture was further confirmed by extrusion via a 19 gauge needle (Figure 3e and Movie S1). Scaffolds with the 5.0:1.0:1.0 ratio were estimated to contain ~321 μg of adenosine per mg of sHA-PA1 and ~179 μg of adenosine per mg of sHA-PA2.

Figure 3.

Synthesis and characterization of microgel-based injectable scaffolds. Structures of the clickable HA polymers: (a) HA-Azide and (b) HA-DBCO. (c) Microgel-based injectable scaffold. Microgels are entrapped in a polymer network formed by azide and DBCO cross-linking. HA-Azide and HA-DBCO polymers are cross-linked in situ via click chemistry forming a triazole ring (shown in the dotted box). (d) Photograph of microgel-based scaffolds under different conditions. (e) Image showing the injectability of the sHA-PA1 scaffold. Microgel-based scaffolds were injected via a syringe before hydrogelation.

The mechanical properties of the scaffolds were investigated using stress-controlled shear oscillatory rheology for the microgel–polymer formulation and corresponding microgel-only system. Oscillatory time sweep measurements showed a sudden increase in moduli of microgel formulation containing clickable polymer molecules suggesting network formation. No such sharp increase in G′ and G″ was observed for the microgel system (Figure 4a). The mechanical properties of the hydrogels were further investigated by oscillatory frequency sweeps on both gel systems. Across frequencies ranging from 1 to 100 Rad/s, the microgel–polymer system maintained higher G′ values compared to the microgel system (Figure 4b). Furthermore, when measured at 1 Rad/s and 0.5% strain, both the storage and loss moduli for the microgel–polymer system were found to be significantly higher than those of the microgel (Figure 4c). Similar to that in the time sweep experiment, both the groups showed solid-like properties (i.e., G′ > G″) in the frequency sweep measurement. The high mechanical properties observed for the microgel–polymer system are due to the azide-DBCO cross-linking leading to a three-dimensional network formation.

Figure 4.

Rheological properties of the microgel and microgel-containing hydrogel. (a) Evolution of the storage modulus (G′) and loss modulus (G″) as a function of time for the microgel–polymer (HA-PA1:HA-Azide:HA-DBCO = 5.0:1.0:1.0) scaffold (red) and the microgel (5 wt %) (blue) at 25 °C. (b) Frequency sweeps for the scaffold and the microgel at 25 °C. (c) Rheological characterizations of the microgel and microgel–polymer systems at 1 Rad/s and 0.5% strain during frequency sweep measurement.

3.5. Adenosine Release from the Scaffolds.

The release profile of adenosine from the scaffolds, sHA-PA1 and sHA-PA2, was examined using two different release media: PBS and α-MEM containing 10% FBS. Following an initial rapid release, a slow release of adenosine was observed until day 10 in both the media for both sHA-PA1 and sHA-PA2 scaffolds (Figure 5a,b). However, a slightly lower percentage of adenosine was released from sHA-PA1 compared to that released from the sHA-PA2 scaffold (Figure 5c,d). At day 14, approximately 48 ± 2% encapsulated adenosine from sHA-PA1 and 57 ± 2% adenosine from sHA-PA2 were found to be released in α-MEM containing FBS (Figure 5c). The release profiles in PBS and α-MEM were similar, albeit with a slightly higher adenosine release in PBS (Figure 5d). The sHA-PA1 scaffold was subsequently used for the in vivo studies.

Figure 5.

Cumulative release kinetics of adenosine from the microgel-based scaffolds. (a and c) α-MEM supplemented with 10% FBS and (b and d) phosphate buffer of pH 7.4.

3.6. Microporous Scaffold Promotes Fracture Healing in Mice.

The effect of local delivery of adenosine to accelerate fracture healing was evaluated using a tibial fracture model in mice.²⁸ Stabilized fractures were induced unilaterally at tibial midshafts and treated with either sHA-PA0 or sHA-PA1 scaffolds containing ~2 mg of HA-PA0 or HA-PA1 microgels. Animals were unconfined during recovery, and the fracture healing at 21 days post-fracture was evaluated using radio-graphic and histomorphometric analyses.

Figure 6a shows 3D reconstructed images and the corresponding radiographs of the injured tibiae obtained through μCT. μCT images at 21 days showed cortical bridging only in groups treated with adenosine (sHA-PA1), suggesting faster healing compared to those treated with the scaffold devoid of adenosine (sHA-PA0). 3D reconstructed images of the fractured callus further showed an obvious size difference between the control (sHA-PA0) and the treated groups (sHA-PA1) (Figure 6b). Analysis of the μCT images further confirmed callus resorption in the adenosine-treated group which displayed significantly smaller callus volume (Figure 6c). Bone analysis results showed a significantly higher BV/TV and TMD in animals that received adenosine delivery (Figure 6d,e). Quantification of BMD from the μCT images also demonstrated higher values in the sHA-PA1-treated mice than those in the group treated with sHA-PA0 (Figure 6f).

Figure 6.

Adenosine-loaded microgel-based scaffolds promote callus remodeling. (a) 3D reconstructions (intact and cut views) and corresponding radiographs of the fractured tibiae treated with various injectables. Tissues were harvested at 21 days following injury (n = 4 mice for each treatment). White boxes enclose the callus regions. Scale bar: 1 mm. (b) Radiographs, corresponding to 3D reconstructions, of the callus regions of fractured tibiae. Tissues were harvested at 21 days following injury. Scale bar: 1 mm. (c–f) Callus size, bone volume ratios (BV/TV) of calluses, TMD, and BMD at 21 days. (g) Representative Safranin-O staining images of fractured tibiae treated with either the HA-PA0 or adenosine-loaded HA-PA1 scaffold at 21 days. Dotted lines indicate the fracture site and subsequent tissue repair. *p < 0.05, **p < 0.01, and ***p < 0.001.

Treatment-dependent changes in fracture callus resorption were also examined through Safranin-O staining. Safranin-O staining of the tissue sections showed smaller cartilage area within the calluses of animals treated with adenosine-loaded scaffolds (i.e., sHA-PA1) compared to those lacking adenosine. Consistent with the callus resorption, the sHA-PA1-treated group had significantly higher cortical bone bridging along the fracture line (demonstrated by the dashed line), while the sHA-PA0 group showed minimal to no bridging between the epiphyses (Figure 6g). These findings suggest that adenosine delivery at the facture site promoted bone regeneration and healing.

4. DISCUSSION

Microgel-based scaffolds are increasingly being used to promote tissue repair and deliver drugs.^29–33 In this study, we developed an in situ curing scaffold containing microgels as adenosine delivery units to treat fracture healing. The microgels were functionalized with a PBA moiety to facilitate loading and release of adenosine via the reversible boronate ester bond formation between the vicinal diol groups of adenosine and the boronate group of PBA. In aqueous solutions, boronic acids exist in equilibrium between the neutral form and hydroxyboronate ion form.²⁶ Depending on the pH of the solution, the equilibrium can be either shifted from a neutral trigonal form at pH 7.0 or less or to a negatively charged tetrahedral boronate form at pH 8.5 or above.²⁶ Although both forms of PBA moieties form boronic esters with cis-diol compounds, negatively charged PBA can form a more stable boronic ester with the cis-diol groups of adenosine.²⁶ We used the hydroxyboronate anion to react with adenosine prior to microgel formation and thereby increased the adenosine loading into the gel particles. The pre-loaded microgels had an approximately 2 times higher amount of adenosine compared to the post-loaded microgels. Furthermore, the ester bond between the neutral boronic acids and the cis-diol groups of adenosine is hydrolytically less stable than the ester bond involving hydroxyboronate anions.^25,26 This could explain the relatively lower percentage of adenosine release from the HA-PA1 scaffold, at a given time, compared to HA-PA2.

The fractured tibia treated with the adenosine-loaded scaffold resulted in better tissue healing with a significant reduction in callus size and bridging of the bone tissues. This is consistent with previous studies that showed that localization of adenosine at the fracture site promotes bone repair possibly due to adenosine-mediated osteogenic differentiation of osteoprogenitor cells or enhanced vascularization.²² Adenosine is also a potent physiological regulator of immune cells wherein extracellular adenosine contributes to suppression and/or resolution of the pro-inflammatory environment.^34–36 In addition to macrophage polarization and neutrophil trafficking, adenosine also exerts immunomodulatory effects on T cells.³⁷ Recent studies have demonstrated the key role played by the inflammatory phase and immune cell dynamics, especially in the early events of fracture healing, in bone regeneration and repair.^38–40 Thus, it is likely that the adenosine-mediated changes to the immunomodulatory environment might also contribute to bone healing.

5. CONCLUSIONS

In summary, an injectable in situ-forming HA-based scaffold containing microgels loaded with adenosine was developed for local delivery of adenosine to the fracture site. PBA was used to encapsulate adenosine into the microgel via the reversible formation of the boronate ester between PBA and adenosine.The in situ curing scaffold containing microgels was generated through spontaneous in situ cross-linking of clickable polymers which enabled entrapment of the microgel particles within the network. The scaffolds loaded with adenosine, when injected at a fracture site, were shown to improve bone healing in the mouse model of tibial fracture. Overall, the injectable system developed herein offers a promising strategy to manage fracture repair via adenosine signaling and provide refillable depots for adenosine. Although the study described here focused on adenosine delivery for fracture healing, other drug molecules with diol functionality, such as anticancer agents (e.g., doxorubicin), nucleic acids, gluconic acid-containing molecules, and antibiotics like anti-PDL1, can be delivered using the microgels functionalized with PBA molecules.^41–44