Strategies to encapsulate cells in cytocompatible three-dimensional hydrogels with tunable mechanical properties and degradability without harmful gelling conditions are highly desired for regenerative medicine applications. Here we reported a method for preparing poly(ethylene glycol)-co-polycarbonate hydrogels through copper-free, strain-promoted azide–alkyne cycloaddition (SPAAC) click chemistry. Hydrogels with varying mechanical properties were formed by “clicking” azido-functionalized poly(ethylene glycol)-co-polycarbonate macromers with dibenzocyclooctyne-functionalized poly(ethylene glycol) under physiological conditions within minutes. Bone marrow stromal cells encapsulated in these gels exhibited higher cellular viability than those encapsulated in photo-cross-linked poly(ethylene glycol) dimethacrylate. The precise control over the macromer compositions, cytocompatible SPAAC cross-linking, and the degradability of the polycarbonate segments make these hydrogels promising candidates for scaffold and stem cell assisted tissue repair and regeneration.
Biocompatible hydrogels play an important role in biomedical research and pharmaceutical applications.1–8 They have been used as protein microchips,9 drug and gene delivery carriers,10 ophthalmic prostheses,11 and scaffolds for encapsulating cells to facilitate either the investigation of cell–extracellular matrix interactions or tissue regenerations.12–21 Naturally occurring biopolymers, such as collagens, fibrin, alginate, agarose, hyaluronan, and chondroitin sulfate, as well as synthetic polymers, such as poly(ethylene glycol) (PEG), poly(vinyl alcohol) (PVA), poly(N-isopropylacrylamine) (PNIPAAM), have been used for regenerative medicine applications.12
It has been recognized that the chemistry, microstructure, and physical properties of hydrogel tissue scaffolds have significant influences on the fate of their resident cells.22–24 From this perspective, synthetic hydrogels present unique advantages over naturally occurring hydrogels due to the broader tunability of the properties of the former.25–27 Challenges still exist, however, for the translation of existing synthetic hydrogels for biomedical uses. For instance, the gelling of most physically cross-linked hydrogels requires substantial changes in environmental conditions (e.g., pH, temperature, ionic strength), which could be detrimental to the in situ encapsulated cells. In addition, the integrity of these physically cross-linked cell–gel constructs could be difficult to maintain in vivo. On the other hand, the cytotoxicity of cross-linking reagents and initiators used for chemically cross-linked hydrogels can negatively impact the viability and long-term fate of the encapsulated cells.28–30 In general, chemical cross-linking conditions and chemically cross-linked networks deemed to be cytocompatible are still limited.31 Among them, PEG–based hydrogels formed by photo-initiated radical polymerization of (meth)acrylated PEG macromers have been utilized most for the encapsulation and support of tissue-specific differentiation of stem cells. Major limitations associated with photo-cross-linked PEG gels include the intrinsic heterogeneities of the network structures due to the uncontrolled radical polymerization process and the varied degrees of cytotoxicity of the aqueous radical initiators utilized (e.g., I-2959 and VA-086).29 Alternative in situ cross-linking strategies involving disulfide-bond formation or Michael addition reactions between thiols and acrylates or vinyl sulfones can eliminate the need for radical initiators, but still suffer from potential interference from the thiol residues widely present within the tissue environment. Thus, a hydrogel system that can be cross-linked under physiological conditions without external perturbations or cross-reactivities with cellular or tissue environments is highly desired. Finally, for tissue regeneration applications, these hydrogels should also ideally possess adequate mechanical properties and exhibit tunable degradation rates that potentially match those of the tissue integrations.
Herein, we report a facile method for preparing PEG macromers flanked with aliphatic azido-functionalized biodegradable polycarbonate blocks (Scheme 1 a), which are subsequently cross-linked with DBCO-terminated PEG macromers (Scheme 1 b) using copper-free SPAAC (Scheme 1 c). The choice of SPAAC click chemistry32, 33 as the in situ cross-linking strategy is inspired by the high fidelity and orthogonality of the reaction as well as its compatibility with physiological conditions. Equally important, compared with the copper-catalyzed [3+2] azide–alkyne cycloaddition (CuAAC) reactions more broadly applied to the functionalization of hydrogels, including PEG,34 PVA,35 and hyaluronan,36 copper-free SPAAC presents significant advantages in terms of short-term cytocompatibility and long-term biocompatibility. Although more biocompatible metal catalysts are being developed for CuAAC,37 their safety for in vivo tissue engineering applications remains unknown. In contrast, SPAAC has already been utilized for live-cell imaging,38 in vivo metabolic labeling in Caenorhabditis elegans,39 zebrafish,40, 41 and mice.42 Thus, it is not surprising that SPAAC has quickly caught the attention of the polymer/hydrogel and tissue-engineering communities.43, 44
Azido-functionalized polycarbonate blocks, which serve as SPAAC cross-linking sites, are grafted to the ends of hydrophilic PEG by organocatalytic ROP of an easy-to-prepare functional cyclic carbonate monomer, AzDXO, recently developed by us.45 ROP of cyclic monomers has been widely used for preparing biodegradable polymers. Recent progress on the development of organocatalysts for ROP46 offers great potential for preparing hydrogels with reduced toxicity compared with traditional metal catalysts.47, 48 The previously demonstrated (co)polymerization versatility of AzDXO by living organocatalytic ROP under mild conditions (e.g., RT) and the facile functionalization of the side chains of the resulting polycarbonate P(AzDXO) obtained by SPAAC45 under physiological conditions have opened the possibilities of adjusting the mechanical, biochemical, and degradation properties of the PEG-co-P(AzDXO) hydrogels for regenerative medicine applications. This study presents the preparation, mechanical property characterization, and cytocompatibility of hydrogels cross-linked by SPAAC.
Results and Discussion
The AzDXO monomer was synthesized in two steps, as previously described.45 Living/controlled ROP of AzDXO (0.1125 M) was initiated by varying amounts of PEG6k, PEG10k, and PEG20k with the organocatalyst DBU in dichloromethane (0.01 M) at RT (Scheme 1 a). The conversion of monomer reached about 90 % in 4 h. Upon termination with benzoic acid, the PEG–P(AzDXO)2m macromer products were purified by repeated precipitation from dichloromethane in ethyl ether with an overall yield of >95 %. As shown in Table 1, the P(AzDXO) block lengths (or degree of polymerization (DP)) in the resulting tri-block macromers PEG–P(AzDXO)2m, as determined by 1H NMR spectroscopy (Figure 1), were close to the theoretical values (2m). GPC results revealed very narrow polydispersities for all PEG–P(AzDXO)2m (PDI: 1.02–1.09) with various PEG and P(AzDXO) block length combinations. The water solubility of macromer PEG–P(AzDXO)2m was dependent on the overall length of the P(AzDXO) blocks (Table 1). When the overall P(AzDXO) block length was shorter than eight repeating units, the resulting PEG–P(AzDXO)2m was soluble in water regardless of the length of the PEG block. In contrast, the triblock macromers became insoluble in water when the DP of AzDXO was above the critical number of eight regardless of the length of the initiating PEG (MW 6000 (6k), 10 000 (10k), or 20 000 (20k)). Similar phenomena were observed in other amphiphilic triblock copolymers initiated by PEG. For example, Hiemstra et al. found that triblock PEG–(PLA)2 (PLA=poly(lactic acid)) copolymers were insoluble in water when the lactic acid repeating units were greater than 22, regardless of the PEG block length investigated.49
Table 1. Characterization of PEG–P(AzDXO)2m macromers.[a]
[a] Naming of the samples reflects the approximate copolymer compositions, including the averaged molecular weight of the initiating PEG and the degree of polymerization of the polycarbonate blocks. [b] Degree of polymerization of AzDXO determined by 1H NMR spectroscopy. [c] Number-averaged molecular weight calculated by 1H NMR spectroscopy. [d] Number-averaged molecular weight per azido group. [e] Number-averaged molecular weight and polydispersity index (PDI) determined by gel permeation chromatography (GPC) by using an evaporative light-scattering (ELS) detector.
Linear or 4-arm DBCO-terminated PEG macromers, PEG–(DBCO)x (x=2 or 4), were synthesized by end-capping PEGs of various molecular weights and architectures (PEG6k, PEG20k or 4-arm-PEG10k) with DBCO acid by esterification. The reactions were carried out in anhydrous dichloromethane with DIPC and DPTS catalysts. The reactant ratio of PEG/DBCO acid/DIPC/DPTS was maintained at 1:1.5:5:0.25/hydroxyl end groups. Complete esterification of the PEG end groups was supported by the disappearance of the characteristic 13C NMR signal for CH2OH at δ=61.4 ppm in the 13C NMR spectra of PEG–(DBCO)x (Figure S1 in the Supporting Information). The catalysts and byproducts were readily removed by washing with water and subsequent precipitation in diethyl ether, or by sequential dialysis precipitation against diethyl ether and water. Overall high yields of >90 % were obtained. As representatively shown in Figure 2, the 1H NMR signal at δ=2.17 ppm, corresponding to the methylene protons of CH2COOH of the DBCO acid, shifted to 2.05 ppm upon esterification with PEG20k and a new signal at δ=4.22 ppm, corresponding to the methylene protons of CH2OCO in the resulting PEG20k–(DBCO)2< appeared. The signal at δ=6.07 ppm in PEG20k–(DBCO)2 and that at δ=6.67 ppm in DBCO acid corresponded to the respective amide protons; the chemical shifts and intensities of which varied significantly with concentration and the water content of the NMR solvent. The 13C (Figure S1 in the Supporting Information) and 1H NMR integrations (Figure 2) supported the successful attachment of DBCO acid to all hydroxyl termini of PEG.
Water-soluble PEG–P(AzDXO)2m and PEG–(DBCO)x were dissolved in cell culture media and, in the presence of cell suspension, readily mixed to form elastic cell–hydrogel constructs under physiological conditions (Scheme 1 d). For this study, six different hydrogels were prepared with two PEG–P(AzDXO)2m macromers and three PEG–(DBCO)x macromers with different structures, compositions, and molecular weights.
To study the SPAAC cross-linking process between the PEG–P(AzDXO)2m macromers and the PEG–(DBCO)x macromers and to determine the shear moduli of the cross-linked gels, time-sweep oscillatory rheology tests were carried out on the six formulations of the respective macromers (Figure 3). Briefly, solutions of PEG–(DBCO)2 or 4-arm-PEG–(DBCO)4 (10 w/v %) were first loaded onto the bottom plate of 20 mm parallel plates equipped with a Peltier heating unit (AR-2000 Rheometer, TA Instruments), before an aqueous solution of PEG–P(AzDXO)2m (10 w/v %) was added and rapidly mixed on the plate with a pipette. The mixtures were equilibrated at 37 °C between the plates for 1 min prior to the test to ensure consistency among various formulations. As shown in Figure 3, both the storage moduli (G′) and loss moduli (G′′) of the mixtures increased with time and the recorded values leveled off after 300 s, suggesting that SPAAC cross-linking was completed within a matter of minutes. The sol–gel transition point, defined as the point where G′ increased to cross with G′′, was not observed when using this testing protocol; this is likely to be as a result of the rapid occurrence of SPAAC cross-linking within the first minute of mixing the respective macromers. Indeed, rigorous quantitative comparisons of the gelling rates among the various fast-gelling formulations would be challenging without significant modification of the mixing and testing protocols. Qualitative observation of the gelling process by tilting the vials upon mixing the respective macromers revealed that the gelling rate followed the following trend: 4-arm-PEG10k–(DBCO)4>PEG6k–(DBCO)2>PEG20k–(DBCO)2 upon mixing with PEG–P(AzDXO)2m. All formulations started to gel in less than 1 min by the vial tilting test (Scheme 1 d), which is consistent with results derived from the time-sweep oscillatory rheology tests (Figure 3). Such a fast gelling rate is beneficial for applications of the macromers as injectable hydrogel formulations to repair tissue defects for which containment of the hydrogel within the local environment is critical.
The equilibrium shear modulus of the hydrogel can be tuned by adjusting the length between the reactive groups (PEG block length) or macromer structures (linear vs. 4-arm). As shown in Figure 3, at the same weight content (10 w/v %), hydrogels cross-linked from PEG–(DBCO)x and PEG6k–P(AzDXO)4 (solid symbols) exhibited higher storage moduli than those cross-linked from PEG–(DBCO)x and PEG20k–P(AzDXO)4 (open symbols), suggesting that the storage modulus inversely correlated with the PEG length between the P(AzDXO) blocks. In all 6 formulations, the hydrogel cross-linked by PEG6k–P(AzDXO)4 and 4-arm-PEG10k–(DBCO)4 exhibited the highest storage modulus throughout the gelling process, with an equilibrium G′ value approaching 6.0 kPa. This is largely owing to the highest chemical cross-linking density accomplished by the four-armed DBCO cross-linker and the PEG–P(AzDXO)2m with the shortest PEG block length. In addition to chemical cross-linking density, the degree of physical entanglement also played an important role in the storage modulus of the gel, especially when both macromers contained sufficiently long PEG blocks. For instance, the gel cross-linked by PEG20k–P(AzDXO)4 and PEG20k–(DBCO)2 exhibited a higher modulus than that cross-linked by PEG20k–P(AzDXO)4 and PEG6k–(DBCO)2. Overall, the equilibrium shear moduli of the six gel systems decrease in the following order: PEG6k–P(AzDXO)4+4-arm-PEG10k–(DBCO)4 (6 KPa)>PEG6k–P(AzDXO)4+PEG6k–(DBCO)2 (3.5 kPa)>PEG6k–P(AzDXO)4+PEG20k–(DBCO)2 (2.7 kPa)>PEG20k–P(AzDXO)4+4-arm-PEG10k–(DBCO)4 (1.3 kPa)>PEG20k–P(AzDXO)4+PEG20k–(DBCO)4 (0.7 kPa)>PEG20k–P(AzDXO)4+PEG6k–(DBCO)2 (0.5 kPa).
The cytocompatibility of PEG–P(AzDXO)2m and PEG–(DBCO)x macromers and the respective click hydrogels were evaluated in vitro. BMSCs cultured in the presence of 10 w/v % of each macromer showed comparable cell viability after 48 h to those cultured without any macromer supplements (Figure S2 in the Supporting Information); this result supports the excellent cytocompatibility of these macromers. Furthermore, we showed that most BMSCs encapsulated by “clicking” the macromer components (Scheme 1 d) remained viable, as supported by the dominant green fluorescent stains for live cells observed after 24 h upon performing live/dead cell staining on the cell–hydrogel constructs (Figure 4 a). No statistically significant difference in the hydrogel storage modulus was detected upon the encapsulation of BMSC in any of the formulations investigated.
An MTT cell viability assay performed on the hydrogel–cell constructs 48 h after cell encapsulation (Figure 4 b) showed that BMSC cells encapsulated in all click hydrogels (106 cells mL−1) exhibited higher viability than those photo-encapsulated in the PEG6k–DMA hydrogel, which was widely used for cell encapsulation in cartilage engineering.50, 51 In our hands, gelation of the 10 w/v % PEG6k–DMA gels in the presence of BMSC (106 cells mL−1) and 0.05 w/v % Irgacure-2959 photoinitiator required irradiation at 365 nm for 10 min. It is known that environmental conditions, such as oxygen level, can lead to variations in the required polymerization time. Consistently more rapid gelation (within 1 min for most formulations) enabled by the SPAAC cross-linking presented herein, coupled with the eliminated need for toxic photoinitiators or UV irradiation, presents a significant advantage.
Cytocompatible PEG-co-polycarbonate hydrogels cross-linked by water-soluble PEG–P(AzDXO)2m macromers with varying PEG block lengths and linear or four-armed DBCO-capped PEG macromers were prepared by copper-free SPAAC. The macromer components were non-cytotoxic and rapid gelation (as fast as <60 s) enabled by copper-free click chemistry allowed the encapsulation of BMSCs with higher cellular viability than the photo-cross-linked PEG6k–DMA gels commonly used for cartilage tissue-engineering applications. The mechanical properties of these gels could be readily tuned by adjusting the macromer structures and the lengths of their constituent polymer blocks. The combination of cytocompatibility and tunable gelling rates and mechanical properties make these “clickable” gels appealing candidates for cell encapsulation strategies and as injectable formulations for minimally invasive tissue repair.
To fully explore the potential of these cytocompatible click gels for regenerative medicine applications, the degradation kinetics of the aliphatic polycarbonate blocks as a function of the gel composition is being investigated. In addition, multipotency of the encapsulated BMSCs is also being analyzed to determine the suitability of these cell–hydrogel constructs for promoting musculoskeletal tissue regeneration.
AzDXO was synthesized as described previously.45 PEG (Mn=6000, 10 000, 20 000 g mol−1, Aldrich) and 4-arm-PEG (Mn=10 000 g mol−1, JenKem Technology) were dried in vacuo in a melted state for 3 h prior to use. DBCO acid was purchased from Click Chemistry Tools (Macon, GA, USA). DBU was purified by distillation with calcium hydride under reduced pressure. Dichloromethane and chloroform were dried by distillation over P2O5 immediately prior to use. All other chemicals were used as received.
Synthesis of PEG–P(AzDXO)2m
PEG–P(AzDXO)2m macromers were prepared by initiating ROP of AzDXO with PEG6k, PEG10k, or PEG20k with catalysis by DBU at RT in CH2Cl2. The AzDXO and DBU concentrations were kept at 0.1125 and 0.01 M, respectively. The amount of PEG was adjusted accordingly to obtain the various copolymer compositions shown in Table 1. In a representative procedure for synthesizing PEG6k–P(AzDXO)4, PEG6k (6.00 g, 1.00 mmol) and AzDXO (0.955 g, 4.50 mmol) were dissolved in CH2Cl2 (38 mL) under an argon atmosphere. A solution of DBU (2 mL; 0.2 M in CH2Cl2) was injected to initiate the polymerization. After 4 h, benzoic acid (0.122 g, 1.0 mmol) was added to neutralize the DBU. The polymer was purified by precipitation in ethyl ether (800 mL). The precipitate was then redissolved in CH2Cl2 (40 mL) and precipitated again in ethyl ether (800 mL), and repeated twice. The macromer was obtained as a white powder and dried under vacuum at RT (6.740 g, 97 %).
Synthesis of PEG–(DBCO)x
The alkyne-containing macromers PEG–(DBCO)x were synthesized by esterification of the hydroxyl ends of the respective linear PEG or 4-arm-PEG with DBCO acid. In a representative synthesis of PEG–(DBCO)x with high Mn, PEG20k (4.0 g, ≈0.2 mmol) was dissolved in chloroform (20 mL) then DBCO acid (234.3 mg, 0.6 mmol), DPTS (29.4 mg, 0.1 mmol), and DIPC (252.4 mg, 2.0 mmol) were added. After 10 h, chloroform (80 mL) was added and the solution was washed with a 0.1 M aqueous solution of NaCl (20 mL) three times. The organic layer was dried with anhydrous sodium sulfate overnight. After filtration, the clear solution was dropped into ethyl ether (900 mL) and the white precipitate was collected by filtration. The solid was redissolved in chloroform (100 mL) and reprecipitated in ethyl ether (900 mL) and repeated twice until no residual catalysts could be detected by 1H NMR spectroscopy. The macromer was obtained as white powder and dried under vacuum at RT (4.05 g, yield=97.4 %).
NMR Spectroscopy and GPC
1H (400 MHz) and 13C NMR (100 MHz) spectra were recorded on a Varian INOVA-400 spectrometer in deuterated chloroform (CDCl3, 99.8 atom % D with 0.03 % v/v TMS). GPC measurements were recorded on a Varian ProStar HPLC system equipped with two 5 mm PLGel MiniMIX-D columns (Polymer Laboratory, Amherst, MA), a UV/Vis detector and a PL-ELS2100 ELS detector (Polymer Laboratory, Amherst, MA). Tetrahydrofuran (THF) was used as an eluent at a flow rate of 0.3 mL min−1 at RT. Mn and PDI were calculated by Cirrus AIA GPC software using narrowly dispersed polystyrenes (ReadyCal kits, PSS Polymer Standards Service, Germany) as calibration standards.
Dynamic rheology tests were performed on an AR-2000 rheometer (TA Instruments) equipped with 20 mm parallel plates and a Peltier heating unit. The gelling process of the various formulations and the evolution of the shear modulus of the hydrogels were studied by oscillatory time-sweep rheology experiments at 37 °C. Aqueous solutions of PEG–P(AzDXO)2 m and PEG–(DBCO)x (10 w/v %) in cell expansion media (α-MEM without ascorbic acid, 20 % FBS) with a 1:1 molar ratio of azide groups to alkyne groups were loaded on the bottom plate sequentially and mixed with a pipette. The mixed solution was kept between the parallel plates for 60 s before the experiment and data collection were initiated to ensure consistency among various formulations. An oscillatory frequency of 1 Hz and a strain of 0.5 % were applied.
PEG–P(AzDXO)2m and PEG–(DBCO)x were dissolved in BMSC expansion media to reach a concentration of 10 w/v %. The solutions were sterile-filtered through a 0.22 μm filter. BMSCs were harvested from the femur and tibia of skeletally mature male rats (Charles River SASCO SD) and enriched by adherent culture as previously described.52 Passage 1 BMSC cells were plated overnight in expansion media, trypsinized, counted, and suspended into the respective macromer solutions (106 cells mL−1). The two BMSC/macromer solutions, PEG–P(AzDXO)2m, and PEG–(DBCO)x, were mixed to a total volume of 50 μL in a 96-well tissue culture plate. Extra expansion media (200 μL) was added to each well after 45 min. As a control hydrogel for BMSC encapsulation, PEG–DMA (Mn ≈6000 g mol−1) was also photo-cross-linked in the presence of BMSC cells. PEG–DMA (10 w/v %) and the photoinitiator Irgacure-2959 (0.05 w/v %) were dissolved in phosphate-buffered saline (PBS; pH 7.4) or BMSC expansion media. The passage 1 BMSC cells were suspended in PEG–DMA/Irgacure-2959 solution (50 μL; 106 cells mL−1) and irradiated with UV light (365 nm) for 10 min. Extra expansion media (200 μL) was added to each well immediately after photo-polymerization. All cell–hydrogel constructs were cultured for 24 and 48 h in humidified incubation (5 % CO2, 37 °C) before being subjected to live/dead cell staining or MTT cell viability assays. A sample size of three was applied to all cell–hydrogel constructs cultured for MTT.
Live and Dead Cell Staining of the Hydrogel–BMSC Constructs
The hydrogel–cell constructs were stained by using a LIVE/DEAD viability/cytotoxicity kit (Molecular Probes), according the vendor’s protocol. Living cells were stained with green fluorescence by intracellular esterase-catalyzed hydrolysis of Calcein AM, and dead cells were stained red by ethidium homodimer-1 after penetrating through the damaged membranes and binding with nucleic acids. The stained hydrogel–cell construct was mounted on a microscope slide and imaged by using a Leica SP laser scanning confocal microscope. Confocal Z-stack images of encapsulated BMSC cells over the depth of 400 μm (20 consecutive 20 μm slices) were overlaid.
MTT Cell Viability Assay
The viability of the BMSCs cultured on tissue-culture plates in the presence of 10 w/v % macromers or those encapsulated in 3D hydrogels were evaluated by using an MTT cell proliferation kit (Roche) after being cultured for 48 h in expansion media (α-MEM without ascorbic acid, 20 % FBS) in 96-well plates. MTT labeling reagent (15 μL) was added to culture media and cell–hydrogel construct (total volume=150 μL) and incubated for 8 h at 37 °C on an orbital shaker. Solubilization solution (150 μL) was then added to each well and incubated at 37 °C on the orbital shaker for 36 h to fully dissolve and release the purple formazan crystals from the 3D hydrogels. The absorbance at 571 nm was read on a MULTISCANFC spectrophotometer (Thermo Scientific). A sample size of three was applied to each construct or culture condition.
This work was supported by the National Institutes of Health (R01AR055615 and R21AR056866).