Incorporation of a silicon-based polymer to PEG-DA templated hydrogel scaffolds for bioactivity and osteoinductivity

Abstract

A scaffold that is inherently bioactive, osteoinductive and osteoconductive may guide mesenchymal stem cells (MSCs) to regenerate bone tissue in the absence of exogenous growth factors. Previously, we established that hydrogel scaffolds formed by crosslinking methacrylated star poly(dimethylsiloxane) (PDMS_star-MA) with poly(ethylene glycol) (PEG-DA) promote bone bonding by induction of hydroxyapatite formation (“bioactive”) and promote MSC lineage progression toward osteoblast-like fate (“osteoinductive”). Herein, we have combined solvent induced phase separation (SIPS) with a fused salt template to create PDMS_star-PEG hydrogel scaffolds with controlled PDMS_star-MA distribution as well as interconnected macropores of a tunable size to allow for subsequent cell seeding and neotissue infiltration (“osteoconductive”). Scaffolds were prepared with PDMS_star-MA of two number average molecular weights (M_ns) (2k and 7k) with varying PDMS_star-MA:PEG-DA ratios and template salt sizes. The distribution of PDMS_star-MA within the hydrogels was examined as well as pore size, percent interconnectivity, dynamic and static moduli, hydration, degradation and in vitro bioactivity (i.e. mineralization when exposed to simulated body fluid, SBF). Finally, cell culture with seeded human bone marrow-derived MSCs (hBMSCs) was used to confirm non-cytotoxicity and characterize osteoinductivity. Tunable, interconnected macropores were achieved by utilization of a fused salt template of a specified salt size during fabrication. Distribution of PDMS_star-MA within the PEG-DA matrix improved for the lower M_n and contributed to differences in specific material properties (e.g. local modulus) and cellular response. However, all templated SIPS PDMS_star-PEG hydrogels were confirmed to be bioactive, non-cytotoxic and displayed PDMS_star-MA dose-dependent osteogenesis.

Graphical Abstract

A tissue engineering scaffold that can inherently guide mesenchymal stem cells (MSCs) to regenerate bone tissue without growth factors would be a more cost-effective and safe strategy for bone repair. Typically, glass/ceramic fillers are utilized to achieve this through their ability to induce hydroxyapatite formation (“bioactive”) and promote MSC differentiation to an osteoblast-like fate (“osteoinductive”). Herein, we have fabricated an interconnected, macroporous PEG-DA hydrogel scaffold that utilizes PDMS_star-MA as a bioactive and osteoinductive scaffold component. We were able to show that these PDMS_star-PEG hydrogels maintain several key material characteristics for bone repair. Further, bioactivity and osteoinductivity were simultaneously achieved in human bone marrow-derived MSC culture, representing a notable achievement for an exclusively material-based strategy.

1. Introduction

Bone tissue regeneration holds potential to be achieved with an instructive scaffold whose chemical and physical properties direct the behavior of associated mesenchymal stem cells (MSCs), even in the absence of exogenous growth factors [1–3]. This “materials-based” approach to bone regeneration would mitigate cost and risks of off-target responses associated with added growth factors [4–6]. Ideally, such a scaffold would be osteoinductive (to stimulate differentiation of cells into the bone-forming lineage), bioactive (to promote bonding with adjacent bone tissue by hydroxyapatite mineralization) and osteoconductive (to permit bone growth on its surface or down into pores) [7–9]. Osteoconductivity may be obtained when scaffolds are produced with optimized pore size (i.e. between 200 and 400 μm) as well as pore interconnectivity [10–12]. Approaches to achieve osteoinductivity and bioactivity have often relied on the incorporation of inorganic and hydrophobic materials, particularly bioactive glasses and ceramics used as fillers [13–16]. However, such composite scaffolds typically suffer from low fracture toughness, limited or low resorption rates and a lack of ideal pore features [7, 17]. Scaffolds that mitigate these issues while achieving these instructive behaviors would represent a significant advancement in bone regeneration.

It has been demonstrated that silicon plays a key role in bone mineralization and gene activation [18, 19], thus prompting the use of certain glasses or ceramics. Instead, we have demonstrated that bioactivity and osteoinductivity could be obtained by introducing silicon-based, inorganic polymer to a conventional organic hydrogel [20, 21]. Specifically, inorganic star poly(dimethylsiloxane) methacrylate (PDMS_star-MA) was introduced to organic poly(ethylene glycol) diacrylate (PEG-DA) to form PDMS_star-PEG hydrogels [20, 22], up to a 30:70 wt/wt % ratio of PDMS_star-MA:PEG-DA. As “biological blank slates”, resisting non-specific protein and cell adhesion, PEG-DA hydrogels have been extensively studied as instructive scaffolds [23, 24]. To control cell adhesion, cell adhesive ligands, such as the peptide RGDS, can be readily introduced [25]. In this way, specific material properties (e.g. stiffness and hydration) may be tuned and readily correlated to associated cellular responses. However, PEG-DA hydrogels lack inherent osteoinductivity and bioactivity. On the other hand, the PDMS_star-PEG hydrogels maintained the biological blank slate nature of PEG-DA hydrogels but also induced osteogenic responses of photoencapsulated bone marrow-derived mesenchymal stem cells (BMSCs) [20–22]. Due to the insolubility of PDMS_star-MA in the aqueous precursor solution, it was heterogeneously distributed as microdroplets throughout the PEG-DA matrix [22]. Thus, to improve PDMS_star-MA distribution, these hydrogels were later prepared via solvent induced phase separation (SIPS) by employing an organic solvent (dichloromethane, DCM) to prepare the precursor solutions [20, 26]. Following sequential UV-cure, removal of the DCM and hydration, the resulting SIPS PDMS_star-PEG hydrogels revealed improved PDMS_star distribution as well as the presence of pores of increased size (up to ~200 μm) with PDMS_star-MA content. However, these pores lacked interconnectivity necessary for cell seeding and eventual neotissue infiltration [20].

Herein, we sought to produce instructive PDMS_star-PEG hydrogels - having controlled porosity and pore interconnectivity as well as enhanced distribution of PDMS_star-MA - by fabrication via SIPS in combination with a fused salt template. Tight control over pore size and maximized pore interconnectivity has been shown to be achieved with fused salt templates [27, 28]. Our recent work has shown that templating is effective to prepare SIPS PEG-DA hydrogels with interconnected macropores of tunable sizes >100 μm [29]; however, these hydrogels remain lacking in osteoinductivity and bioactivity. Thus, we prepared a series of templated SIPS PDMS_star-PEG-DA hydrogels based on different number average molecular weights (M_n) of PDMS_star-MA (2k and 7k). These were incorporated into a DCM-based precursor solution at increasing wt/wt % ratios (0:100, 10:90, 20:80, and 30:70) of PDMS_star-MA to PEG-DA (M_n = 3.4k). The distribution of PDMS_star-MA was evaluated along with the ability to achieve tunable pore size and pore interconnectivity. The effect of templated SIPS hydrogel composition on hydration and modulus was examined, as these material properties are known to influence cellular behavior [30–33]. The bioactivity and degradation rates were also investigated. Scaffolds were further evaluated to confirm non-cytotoxicity and degree of osteogenic capacity after seeding with human bone marrow derived mesenchymal stem cells (hBMSCs).

2. Materials and Methods

2.1. Materials

1-Vinyl-2-pyrrolidinone (NVP), 2,2-dimethyl-2-phenylacetophenone (DMAP), acryloyl chloride, allyl methacrylate, calcium chloride, dibasic potassium phosphate, HC1 (12 M), hexamethyldisilazane (HMDS), magnesium chloride hexahydrate, magnesium sulfate, NaOH, Nile red, NMR grade, deuterated chloroform (CDCl₃), poly(ethylene glycol) 3350 (PEG-3350), potassium carbonate, potassium chloride, silica gel, sodium bicarbonate, sodium chloride (salt), sodium sulfate, triethylamine, trifluoromethanesulfonic acid (triflic acid), tris-hydroxymethyl aminomethane, Triton X-100 (0.1%) and all solvents were obtained from Sigma-Aldrich. HPLC-grade toluene, dichloromethane, and NMR grade CDCl₃ were dried over 4 Å molecular sieves. Salt was sieved (ASTM E-11 Specification, No. 40, 425 μm opening; No. 60 250 μm opening) to obtain the following salt sizes: “small salt” (SS): 181 ± 29 μm, “medium salt” (MS): 268 ± 35 μm and “large salt” (LS): 459 ± 69 μm. Average salt sizes were calculated using ImageJ software, as done previously [29]. Silver nitrate and sodium thiosulfate were obtained from American Mastertech. Fetal bovine serum (16.5%; FBS) was obtained from Atlanta Biologicals. Glutamax, Minimum Essential Medium Alpha (α-MEM), Penicillin-Streptomycin and 1% glutamine (Glutamex) were obtained from Gibco. Phosphate-buffered saline (1X without Ca and Mg; PBS) and Dulbecco’s Modified Eagle Media (DMEM) was obtained from Corning. Octamethylcyclotetrasiloxane (D₄), Pt-divinyltetramethyldisiloxane complex in 2 wt% xylene (Karstedt’s catalyst), and tetrakis(dimethylsiloxy)silane (tetra-SiH) were obtained from Gelest. Activated carbon, buffered formalin, DAPI, dialysis cassettes (3.5k MWCO), anhydrous ethyl ether, and rhodamine phalloidin were obtained from Fisher Scientific. Von Kossa staining kit was obtained from American Mastertech Scientific. Acryloyl-PEG-succinimidyl valerate was obtained from Laysan Bio, Inc. Dulbecco’s phosphate-buffered saline (DPBS) was obtained from Lonza. Peptide RGDS was obtained from Bachem. Cytotoxicity detection kit (LDH) was obtained from Roche, multiplex immunoassay kits were obtained from R&D Systems and the PicoGreen assay kit was obtained from Invitrogen.

2.2. Synthesis

2.2.1. PDMS_star-MA synthesis

PDMS_star-MA (M_n = 2k and 7k g/mol) were prepared as previously reported [20, 22]. First, Silane (SiH)-terminated PDMS_star was prepared by the acid-catalyzed equilibration of D₄ (“2k”: 30.0 g, 101.4 mmol; “7k”: 29.9 g, 101.0 mmol) with tetra-SiH (“2k”: 7.8 g, 23.8 mmol; “7k”: 1.1 g, 3.4 mmol) utilizing triflic acid (60 μl) and subsequent neutralization with HMDS (0.15 g, 0.93 mmol). The resulting PDMS_star-SiH (for “2k” and “7k”) were each precipitated three times in toluene/methanol and dried under reduced pressure, resulting in a colorless liquid. Next, photoreactive groups were added via hydrosilylation to the terminal ends of each arm of PDMS_star-SiH (“2k”: 7.0 g, 2.5 mmol; “7k”: 7.0 g, 0.53 mmol) by Pt-catalyzed [Karstedt’s catalyst (50 μL)] reaction with allyl methacrylate (“2k”: 3.6 mL, 26.8 mmol; “7k”: 0.314 mL, 2.33 mmol) in toluene (30 mL). Both M_n’s of PDMS_star-MA were obtained after a 2 hour activated carbon spin and flash column chromatography. The structures of each M_n of PDMS_star-SiH and PDMS_star-MA were confirmed by ¹H NMR and was in agreement with that previously reported.

2.2.2. PEG-DA synthesis

PEG-DA (M_n = 3.4k g/mol) was prepared as previously reported [22]. PEG-3350 (23.5 g, 7.0 mmol) was dissolved in DCM, then triethylamine (1.95 mL, 14.0 mmol) and subsequently acryloyl chloride (2.27 mL, 28.0 mmol) were added dropwise to the solution and allowed to react overnight. HCl was removed by 2M k₂CO₃ wash in a separatory funnel. The reaction was then dried in the organic phase with MgSO₄ spin and precipitated in anhydrous ethyl ether. After vacuum filtration to obtain the white solid, PEG-DA was vacuum dried to remove residual solvent. The structure was confirmed by ¹H NMR and was in agreement with that previously reported.

2.2.3. ACRL-PEG-RGDS synthesis

ACRL-PEG-RGDS was synthesized as previously reported [21]. The cell adhesion peptide RGDS was reacted with acryloyl-PEG-SVA at a 1:1 molar ratio for 2 hr in 50 mM sodium bicarbonate buffer. ACRL-PEG-RGDS was purified by dialysis, lyophilized, and stored at −20°C under nitrogen (N₂). ACRL-PEG-RGDS was included in all hydrogels utilized for cell culture at 1 μmol/mL of precursor solution.

2.3. NMR

¹H-NMR spectra were obtained on an INOVA 500 500 MHz spectrometer operating in the Fourier transform mode. Solutions of 5% (w/v) CDCl₃ were used to obtain spectra. Residual CHCl₃ served as an internal standard.

2.4. Salt template formation

Salt templates were formed with each of the three average salt sizes: “small salt” (~180 μm), “medium salt” (~270 μm) and “large salt” (~460 μm). Molds for making the salt template and subsequently curing the precursor solution were prepared as follows. One end of a borosilicate glass tube (5/8” OD, 1/2” ID, 1.5” H) was covered with aluminum foil, pushed into a custom Teflon cap (1” OD, 5/8” ID) that contained a small hole (0.05” D) directly through its center and then tightly wrapped with Parafilm to secure the glass tube to the cap (Fig. S1). Salt of the designated size (16 g) was mixed gently with 5 wt% DI water (using a metal spatula) in a 25 mL beaker and 3.5 g of the mixture added to the mold. The salt mixture was then compacted into the mold using a flat-ended glass rod, covered with Parafilm, and centrifuged 5 min at 2000 rpm within a 50 mL centrifuge tube (Eppendorf 5810R centrifuge, A-4-62 rotor). Afterwards, the Parafilm cover was removed and the fused template allowed to air dry at room temperature (RT) for 24 hr within the glass mold.

2.5. Templated hydrogel fabrication

DCM-based precursor solutions were prepared containing 30 wt% total macromer (based on total solution mass) with increasing wt/wt % ratios of PDMS_star-MA to PEG-DA: 0:100 (PEG-DA control), 10:90, 20:80, and 30:70. Photoinitiator solution (30 wt% DMAP in NVP) was added at 100 μL/g of macromer. Solutions were vortexed in a sealed vial for 1 min after addition of each component. The precursor solution (~1.5 mL) was immediately deposited atop the aforementioned dried salt template via syringe. After ~30 s, the aluminum foil at the base of the mold was punctured through the hole in the Teflon cap. Using PVC tubing (3/8” ID, 9/16” OD) held against the top of the glass mold, compressed air was used to gently force the solution throughout the salt template until the solution began to exit through the cap hole. Removal of the air inlet caused the flow of precursor solution to cease. The filled mold was UV-cured for 15 min (12 min standing on Teflon cap and 3 min standing on the glass top) via exposure to a UV-transilluminator (6 mW cm², 365 nm). After air-drying 24 hr to evaporate the solvent, the cylindrical specimen was removed from the mold and sectioned into 1.2 mm thick discs using a vibratome (Leica VT1000S; cutting speed = 1.75 mm/s; frequency = 30 Hz). Only discs from the central portion of the cylinder were used for testing, discarding ~4 mm from each end. Discs were soaked in DI water for 48 hr on a rocker table (100 rpm; water changes 3X daily) to leach the salt template and swell the hydrogels to their final hydrated form (~15 mm diameter, ~1.5 mm thick). Discs were allowed to soak an additional 72 hr prior to testing, for which they were punched to a desired size (8 or 13 mm) from the central region of the disc using a die.

2.6. Sol content

Discs (N = 5; 13 mm diameter) were air dried 30 min. Each disc was placed in an open scintillation vial and dried at RT in a vacuum oven (14.7 psi, 24 hr). Dried discs were weighed (W_d1), placed in a new scintillation vial with 10 mL of DCM, and set on a rocker table (100 rpm) for 48 hr. Removed discs were placed in new scintillation vials, allowed to air dry (30 min) and dried again at RT in the vacuum oven (14.7 psi, 24 h). The final weight was taken (W_d2) and the sol content calculated as [(W_d1 – W_d2)/W_d1] x 100.

2.7. Equilibrium swelling

Discs (N = 5; 13 mm diameter) were each placed in a sealed vial with 15 mL of DI water on a rocker table (100 rpm) for 48 hr at RT. Discs were then removed and weighed (W_s). Discs were dried (14.7 psi, 60 °C, 24 hr) and weighed again (W_d). Equilibrium swelling ratio was calculated as (W_s – W_d)/W_d.

2.8. Distribution of PDMS_star-MA and pore morphology

2.8.1. Confocal laser scanning microscopy (CLSM)

Discs (8 mm diameter) were soaked for 24 hr in 60 mL Nile Red solution, followed by daily changes with PBS for 3 days. Nile red solution contains 75 μL of a solution containing 20 mg Nile Red and 1 mL methanol mixed into 8 mL of DI water and added to 120 mL of PBS, as previously reported [20]. Discs were placed in a coverglass-bottom chamber and imaged on an Olympus FV1000 confocal microscope, equipped with a UPLSAPO 10x/0.4 objective. Excitation and emission were 488 nm and 500-600 nm, respectively. The confocal aperture was set to 1 Airy unit. Z-stacks (80 slices) were acquired with a 4.0 μm step. Confocal zoom and resolution setting resulted in XY pixel size of 2.485 μm. Representative slices of the stacks were exported. The fluorescence images were pseudo-colored green. Bright field imaging was conducted simultaneously and used to measure hydrated pore size. Images were taken of three locations on two different samples for each salt size (20:80 wt/wt% PDMS_star-MA:PEG-DA) with one pore measured per image (N = 6). ImageJ Software was used to calculate pore size, determined by the mean of the diagonal and central cross-section.

2.8.2. Scanning electron microscopy (SEM)

Discs (8 mm diameter) were allowed to air dry (30 min), and dried in a vacuum oven (14.7 psi, 24 hr, RT). Dried discs were subjected to Au-sputter coating (Cressington Sputter Coater 108) and viewed with a field emission scanning electron microscope (FE-SEM; JEOL NeoScope JCM-5000) at an accelerated electron energy of 10 keV. Pore size was measured using ImageJ software, and determined by the mean of the longest and shortest cross-sectional areas of the rectangular pores. Dry pore measurement was conducted by the same method described in section 2.8.1.

2.8.3. Percent interconnectivity

Discs (N = 3; 13 mm diameter) were evaluated to determine interconnectivity with a water wicking procedure adapted from Bencherif et al. [34]. Each disc was soaked for 24 hr in DI water on a rocker table (100 rpm) to ensure no air bubbles existed within pores. Discs were removed and weighed in a glass Petri dish (W_total). A folded Kimwipe was then gently pressed on the disc for 1 min to wick away the interconnected volume, and the disc was weighed again (W_{interconnected}). Percent interconnectivity was calculated as [(W_total – W_{interconnected})/W_total] x 100.

2.9. Mechanical properties

2.9.1. Bulk compressive Young’s modulus and damping – static and dynamic mechanical analysis

Discs (8 mm diameter) were used for the measurement of Young’s modulus (N = 4) as well as dynamic moduli (N = 8) using a dynamic mechanical analyzer (DMA; TA Instruments Q800) with a parallel-plate compression clamp equipped with plates of 40 mm (bottom) and 15 mm (top). Hydrated discs were placed between the plates and several drops of DI water were dropped onto the hydrogel to maintain hydration. Analyses were conducted at RT. For static testing, samples were tested at RT in a controlled strain rate mode (20%/min to 12% strain) with a preload force of 0.1 N. Young’s modulus was obtained from the slope of the linear portion of the stress-strain curve (1-5%). For dynamic testing, samples were tested using a multi-frequency strain mode (1-30 Hz) with an amplitude of 10 μm and preload force of 0.1 N. The obtained values of loss (G”) and storage (G’) were used to calculate tan δ (i.e. damping).

2.9.2. Local Young’s modulus - Atomic force microscopy (AFM)

Discs (N = 2; 8 mm diameter) were used to measure local Young’s modulus using a Dimension ICON AFM (Bruker) with a Nanoscope V controller. A modified silicon nitride probe (NovaScan, USA) having a nominal spring constant of 0.6 N/m with a 5 μm diameter bead attached to the cantilever was used. Standard detector calibration was performed on a clean mica disc to determine deflection sensitivity. Standard thermal tuning calculated a spring constant of ~0.75-0.9 N/m. For each sample, hydrogels were fixed to 15 mm metal AFM specimen discs (Ted Pella, Inc.) with water droplets added to the sample surface to maintain hydration. Indentation was carried out at the pore walls using 1 Hz loading rates. The ramp size was 1 μm and the indentation force was 15 nN. Young’s modulus values were calculated using Nanoscope Analysis software (Bruker) with a Hertzian fit model after baseline correction and boxcar filter. A sample Poisson’s ratio of 0.5 was used. Individual force-displacement curves were taken 6 times at each sampled point of the hydrogel, with 5 total points from 2 hydrogels sampled.

2.10. Accelerated degradation

Discs (N = 3 per time point; 8 mm diameter) underwent accelerated degradation analysis All hydrogel discs were vacuum dried (14.7 psi, 12 hr) and weighed. Each disc was then placed in a 1 dram vial with 1 mL of 0.05M NaOH and stored in a shaking incubator (100 rpm, 37 °C). At 12 hr time points, one disc was removed, dried (14.7 psi, 12 hr), and weighed to determine mass loss percentage. At each time point, solutions for all samples were exchanged with new 0.05M NaOH.

2.11. Bioactivity

2.11.1. Simulated body fluid (SBF) soak

SBF was fabricated as detailed by Kokubo et al. and in previous studies [20, 35]. Individual discs (N = 2; 8 mm diameter) were placed in sealed 50 mL centrifuge tubes containing 40 mL of 1X SBF and incubated in a water bath at 37 °C. Discs were removed after 4 weeks, washed with DI water and vacuum dried (14.7 psi, 24 hr). Dried hydrogels were subjected to SEM imaging (section 2.8.2) and X-ray diffraction spectroscopy (section 2.11.2).

2.11.2. X-ray diffraction spectroscopy (XRD)

Powder X-ray diffraction data was collected on a Bruker D8 diffractometer fitted with a LynxEYE detector. The X-ray source was a 1kW Cu X-ray tube maintained at an operating current of 40 kV and 25 mA (Bragg-Brentano geometry; scan range 3 – 70 °; step size: 0.015 °).

2.12. Cell culture

Bone marrow derived human mesenchymal stem cells (hBMSCs) were obtained from the Texas A&M Institute for Regenerative Medicine. By flow cytometry, hBMSCs tested positive (> 95%) for CD29, CD44, CD49c, CD59, CD73a, CD90, CD105, CD146, CD147, CD166, and HLA-I:ABC and negative (< 5%) for CD3, CD11b, CD14, CD19, CD34, CD36, CD45, CD49b, CD49d, CD79a, CD106, CD184, CD271, HLA-II:DR, and HGFR (c-Met). hBMSCs were expanded in α-MEM supplemented with 16.5% FBS and 1% glutamine. Twenty-four hr prior to seeding, cells at passage 4 were transitioned to DMEM containing 10% FBS and 1% Penicillin/Streptomycin (P/S)).

2.13. Seeding and culture of scaffold constructs

Scaffolds were fabricated with ACRL-PEG-RGDS (section 2.2.3), and cut to 8 mm discs. Scaffolds were seeded with hBMSCs at 5xl0⁶ cells/mL and cultured at 37 °C and 5% CO₂ in DMEM containing 10% FBS and 1% P/S. Culture medium was exchanged every 2 days through the duration of culture. Osteogenic supplements were not utilized so as to assess the intrinsic osteogenic potential of the scaffolds.

2.14. Cytotoxicity

Cytotoxicity assessments were performed after 48 hr culture by measuring the secretion of lactate dehydrogenase (LDH) with a commercial kit. Briefly, cell culture supernatants were collected, reacted with the kit working solution, and absorbance was measured following 25 min of development. Cell morphology and distribution were also evaluated after 48 hr of culture. Scaffolds were washed with DPBS and fixed with 10% buffered formalin. Fixed cells were then washed twice with DPBS and permeabilized with 0.1% Triton X-100. Cell distribution was evaluated after SYBR Green staining while cell morphology was assessed by rhodamine phalloidin and DAPI staining. Confocal images were taken to evaluate both stains using a multiphoton microscope (Zeiss LSM 510Meta).

2.15. Endpoint construct analyses

Samples were collected after 48 hr (cytotoxicity) or 14 days of culture (osteogenic potential) by rinsing them with DPBS and snap freezing with liquid N₂. Scaffolds were stored at −80 °C until analysis. Protein samples were extracted by sample homogenization in lysis buffer and collection of the correspondent supernatant after centrifugation.

2.15.1. Biochemical analyses

hBMSC protein expression was evaluated by means of specific multiplex immunoassay kits following manufacturer’s instructions. Proteins selected for analysis included Collagen 1A1 and osteonectin (secreted protein acidic and rich in cysteine; SPARC) for both 2k and 7k PDMS_star-MA containing scaffolds, and additionally osteopontin (OPN) and bone morphogenic protein-2 (BMP-2) for exclusively 2k PDMS_star-MA containing scaffolds. Samples were properly diluted with the assay buffer and loaded into well plates containing the appropriate bead suspension. Bound analyte was then reacted with detection antibodies, followed by the addition of streptavidin-phycoerythrin. Protein concentrations were determined by the median fluorescence intensity (MFI) using a MAGPIX (Luminex) in comparison to an analyte-specific standard curve. The obtained values were normalized to the DNA content of each construct and assessed via PicoGreen assay.

2.15.2. Histological analyses

Von Kossa staining was used to assess cellular calcium deposition by manufacturer’s protocol. Rehydrated sections were rinsed with distilled water, after which a 5% silver nitrate solution was applied, followed by the exposure of the samples to full-spectrum light for 1 hr. Samples were then sequentially rinsed with distilled water, exposed to 5% sodium thiosulfate for 3 min, and rinsed with distilled water.

2.16. Statistics

Data is reported as the mean ± standard deviation. Data set mean values were compared in GraphPad Prism via ANOVA followed by Tukey’s post hoc test where p-value < 0.05 was considered statistically significant.

3. Results and Discussion

3.1. SIPS/salt fabrication

Templated SIPS PDMS_star-PEG hydrogels were prepared by photocuring a DCM-based precursor solution over a fused salt template followed by sequential evaporation of the DCM solvent, template dissolution and hydration. Series of hydrogels were prepared based on different salt sizes (“small” ~180, “medium” ~270 and “large” ~460 μm) as well as two different M_n’s of PDMS_star-MA (2k and 7k) wherein the wt/wt % ratio of PDMS_star-MA to PEG:DA was systematically increased: 0:100, 10:90, 20:80 and 30:70. In our previous report, templated SIPS PEG-DA scaffolds (i.e. no PDMS_star-MA) were similarly prepared using a combination of SIPS and a fused salt template based on these salt sizes [29]. It was determined that a macromer concentration of 30% (wt/wt) in DCM produced mechanically robust, macroporous hydrogels. This is in contrast to fabrication of conventional hydrogels from aqueous precursor solutions without a salt template, where macromer concentrations of ~5-15% (wt/wt) produce robust hydrogels owing to their low porosity [22, 36]. When we previously prepared non-templated SIPS PDMS_star-PEG hydrogels, the total macromer concentration was only 10% (wt/wt) in DCM [20]. However, in the present study, when the total macromer concentration was increased to 30% (wt/wt), this resulted in extremely hazy precursor solutions, particularly as the concentration and M_n of PDMS_star-MA increased. Moreover, within several minutes, the solution exhibited signs of substantial phase separation, partitioning into a clear (i.e. PDMS_star-MA-rich layer) and cloudy layer (Fig. S2). Thus, it was critical to increase the flow rate of the precursor solution through the fused salt template so that it could be promptly photocured. Previously, the salt template was prepared in a glass vial and the precursor solution distributed through the template by centrifuging for 10 min [29]. Instead, herein, a mold was prepared from a glass tube and a Teflon cap (containing a small hole) secured to the base. After addition of the precursor solution to the top of the salt template, it was quickly pushed through the mold with compressed air and then exposed to UV light. Using this method, templated PDMS_star-PEG hydrogels were prepared with low sol content values (<5 wt%) indicating effective photocuring (Table S1).

3.2. PDMS_star-MA distribution and scaffold morphology

Distribution of PDMS_star-MA within the templated SIPS PDMS_star-PEG hydrogels was examined by CLSM imaging with hydrophobic PDMS_star-MA-rich regions stained by hydrophobic Nile red dye. For the fabrication of non-templated PDMS_star-PEG hydrogels, an aqueous precursor solution produced exclusively discrete PDMS_star-MA-rich microparticles within a PEG-DA matrix [22], whereas SIPS fabrication produced a much more uniform distribution [20]. In general, the templated SIPS PDMS_star-PEG hydrogels exhibited PDMS_star-MA distribution throughout the pore walls (Fig. 1). This was attributed to the improved solubility of PDMS_star-MA in DCM versus in water, even at higher precursor solution concentrations. Since solution phase separation increased with PDMS_star-MA M_n, it was expected that the 2k hydrogel series would exhibit a more uniform distribution of PDMS_star-MA versus the 7k series. Indeed, the 7k series showed the presence of PDMS_star-MA-rich microdroplets that was not observed for the 2k series.

Figure 1.

CLSM images of templated SIPS PDMS_star-PEG hydrogels prepared with increasing amounts of 2k and 7k PDMS_star-MA and a templated PEG control. All hydrogels prepared with medium salt (~270 μm). Green is assigned to PDMS_star-MA-rich regions, stained by hydrophobic dye (Nile red). Photos are adjusted +20% brightness and −20% contrast for clarity. Scale bars = 250 μm.

SEM was used to visually confirm pore interconnectivity of templated SIPS PDMS_star-PEG hydrogels prepared with the different template salt sizes (Fig. 2). While expected differences in pore sizes were apparent depending on template salt size, the measured average pore size was reduced due to specimen shrinkage associated with vacuum drying (Table 1). However, SEM images did reveal pore interconnectivity, a product of the fused salt template. Pore size for hydrated hydrogels was measured using brightfield microscopy (Fig. 2). As anticipated, hydrated pore sizes were swollen beyond the original salt size of the template (Table 1).

Figure 2.

SEM images of templated SIPS PDMS_star-PEG hydrogels fabricated with each salt size [20:80 wt/wt % and 2k PDMS_star-MA] (left and middle column). Representative bright field images were taken during confocal imaging (right column). Scale bars = 200 μm.

Table 1.

Pore size measurements of templated SIPS PDMS_star-PEG hydrogels fabricated with each template salt size [20:80 wt/wt % and 2k PDMS_star-MA] determined by ImageJ software using SEM images (acquired at 100x magnification) for “dry” measurements [i.e. after vacuum drying] and bright field microscopy images for “hydrated’ measurements [i.e. in the hydrated state].

	SMALL SALT	MEDIUM SALT	LARGE SALT
SALT SIZE:	181 ± 29 μm	268 ± 35 μm	459 ± 69 μm
DRY:	124 ± 17 μm	199 ± 13 μm	282 ± 60 μm
HYDRATED:	268 ± 19 μm	367 ± 16 μm	428 ± 60 μm

To quantify interconnectivity of hydrated specimens, a water wicking test was adapted from Bencherif et al. [34]. Templated SIPS PDMS_star-PEG hydrogels (20:80 wt/wt % and 2k PDMS_star-MA) were compared to the analogous non-templated SIPS hydrogels as well as conventionally fabricated hydrogels (i.e. from an aqueous precursor solution) (Fig. 3A) [20, 22]. For the templated hydrogel only, due to its high pore interconnectivity, water was quickly wicked away using a Kimwipe and pore interconnectivity was >50% (Fig. S3). In contrast, conventionally and SIPS fabricated, non-templated PDMS_star-PEG hydrogels demonstrated interconnectivity below 15%.

Figure 3.

Templated SIPS PDMS_star-PEG hydrogels and analogous PEG-DA control material properties: (A) % interconnectivity, (B) equilibrium swelling, (C) damping and (D) local Young’s modulus. All hydrogel scaffolds were fabricated with medium salt (~270 μm). Statistical difference versus the analogous templated SIPS PEG-DA control is represented as * = p <0.05.

3.3. Material characterization

To assess the impact of PDMS_star-MA concentration (0:100, 10:90, 20:80 and 30:70 wt/wt % PDMS_star-MA:PEG-DA) and M_n (2k or 7k), templated SIPS PDMS_star-PEG hydrogels were prepared with the intermediate salt size (medium salt, ~270 μm) and subjected to characterization of key properties – hydration, modulus and degradation rate. Because the hydration (i.e. equilibrium swelling) is known to profoundly impact most hydrogels’ mechanical and degradation properties [37, 38], this was initially measured (Fig. 3B).

All hydrogels exhibited statistically similar swelling ratios (~13) compared to the analogous templated SIPS PEG-DA control. It is hypothesized that the high pore volume and resulting high water content effectively masked any contribution from pore wall hydrophobicity based on PDMS_star-MA content, M_n or its distribution therein. As expected, it was noted that the high pore volume of these templated SIPS hydrogels produced a significantly higher swelling ratio versus the analogous non-templated SIPS PDMS_star-PEG hydrogels (~8) [20].

The bulk compressive modulus of these templated SIPS PDMS_star-PEG hydrogels was also assessed (Fig. S4). All hydrogels displayed similar moduli (~8 kPa) that were not statistically different versus the analogous PEG-DA control. This may be attributed to their similar high bulk hydration (i.e. swelling), resulting from their macroporosity. However, additional mechanical analyses were conducted that were anticipated to reveal differences among hydrogels based on the PDMS_star-MA content and M_n that comprised the pore walls. An indicator of the damping or energy absorption ability of a material, tan δ (based on the ratio of loss [G”] and storage [G’] moduli) was also measured (Fig. 3C). In this case, templated SIPS hydrogels based on 2k PDMS_star-MA produced a reduction in tan δ versus the PEG-DA control whereas those based on 7k PDMS_star-MA did not. The more homogeneous distribution of 2k PDMS_star-MA within the PEG-DA matrix of the pore walls (Fig. 1) is believed to give rise to this difference in damping. Thus, when the PDMS_star-MA is not uniformly distributed (i.e. for 7k), the resulting hydrogels retain more of the native properties of the PEG-DA control. Next, AFM was used to probe the local modulus values for templated SIPS PDMS_star-PEG hydrogels based on 2k PDMS_star-MA. Versus bulk analyses, AFM has been used to measure the local modulus of scaffold of the pore walls at a scale commensurate with cellular interaction [3]. Per Fig. 3D, the uniform dispersion of 2k PDMS_star-MA within the templated SIPS PDMS_star-PEG hydrogels was determined to decrease the Young’s modulus values versus the analogous PEG-DA control. Despite this decrease, however, modulus values (~15 kPa) within the range associated with osteogenesis [31, 39].

The relative differences in degradation rates of templated SIPS PDMS_star-PEG hydrogels versus the analogous PEG-DA control was assessed by measuring mass loss under accelerated (basic) conditions (Fig. S5). Despite the introduction of hydrophobic PDMS_star-MA, no distinguishable differences in mass loss were observed, irrespective of PDMS_star-MA content, at all time points. This is attributed to their similarities in pore size (Fig. 2) and bulk swelling (Fig. 3B) that dominate any contributions from the hydrophobic comacromer. However, based on the known accelerated degradation of porous materials having larger pore sizes [40, 41] and our previous studies with templated SIPS PEG-DA scaffolds of varying pore sizes [29], increasing and decreasing the pore size of these templated hydrogels is expected to increase and decrease degradation rates, respectively.

3.4. Bioactivity

Bioactivity can be examined in vitro by exposing hydrogels to conditions that mimic the implant environment (i.e. ionic concentration, pH and temperature) via soaking in simulated body fluid (SBF, 1X) at 37 °C [35]. A series of templated SIPS hydrogels were prepared with 2k PDMS_star-MA of varying concentrations (0:100, 10:90, 20:80 and 30:70 wt/wt % PDMS_star-MA:PEG-DA) using the intermediate salt size (medium salt, ~270 μm). All PDMS_star-PEG hydrogels exhibited mineralization after 4 weeks of incubation while the analogous PEG-DA control showed none (Fig. 4). XRD confirmed that observed mineral present was hydroxyapatite (HAp) with peaks at 31.7, 45.3, and 56.2 indicating reflections from 112, 222 and 004 crystal planes, respectively [20, 42–44]. Thus, the presence of inorganic PDMS_star-MA introduced bioactivity to the otherwise biologically inert PEG-DA hydrogel.

Figure 4.

Templated SIPS PDMS_star-PEG hydrogel and analogous PEG-DA control following exposure to simulated body fluid (SBF, 1X, 537 °C, 4 weeks). Hydrogels were fabricated with medium salt (~270 μm) and 2k PDMS_star-MA. SEM images (top) displayed HAp growth on all PDMS_star-MA-containing hydrogels. XRD analysis of “20:80 wt/wt %” composition (bottom) confirmed HAp identity by characteristic peaks at 31.7, 45.5 and 56.5. Scale bar = 100 μm.

3.5. Cytotoxicity, cell seeding distribution and cellular morphology

All cell-contacting templated SIPS PDMS_star-PEG scaffolds and the analogous PEG-DA control (prepared with medium salt; ~270 μm) were made with ACRL-PEG-RGDS for controlled cell adhesion and induction. Cytotoxicity was determined by LDH assay after 48 hr culture with hBMSCs (Fig. 5). Normalized to the corresponding templated SIPS PEG-DA control, all PDMS_star-MA-containing hydrogels (2k and 7k) were determined to be non-cytotoxic. To verify consistent cell seeding distribution, SYBR green DNA staining was used to visualize cells. Further staining with DAPI and rhodamine phalloidin indicated that cells were elongated and spread along the pore walls, as expected with the presence of the cell adhesion ligand RGDS (Fig. 5 and S7). Finally, cell morphology and abundance were similar for all PDMS_star-PEG scaffold compositions compared to the PEG-DA control (Fig. 5 and S7). Thus, all compositions were subsequently subjected to 14-day culture for examination of intrinsic osteoinductivity.

Figure 5.

Scaffold non-cytotoxicity after 48 hr culture with hBMSCs, as determined by LDH assay (left), SYBR green stain (middle) and rhodamine phalloidin (red) and DAPI (blue) stacked confocal images (right). Statistically significant difference compared to the control is represented as * = p <0.05. The complete set of images can be found in Fig. S6. All templated SIPS PDMS_star-PEG and analogous PEG-DA control scaffolds were prepared with medium salt (~270 μm). Scale bars = 200 μm.

3.6. Scaffold-induced osteogenesis of hBMSCs

RGD-containing, templated SIPS PDMS_star-PEG scaffolds (prepared with medium salt; ~270 μm) were evaluated for their potential to control hBMSC fate by evaluating the relative presence of ECM molecules relevant to osteogenesis after a 14-day culture. Protein level multiplex analysis was conducted to determine relative levels of collagen 1, the primary collagen found in bone, and osteonectin (secreted protein acidic and rich in cysteine or SPARC), a calcium-binding glycoprotein found in bone (Fig. 6A & B). All templated SIPS PDMS_star-PEG scaffolds displayed evidence of PDMS_star-MA-dose dependent osteogenesis. For those based on 7k PDMS_star-MA, a 30:70 (wt/wt %) was associated with an increase in collagen 1 and SPARC relative to the analogous PEG-DA control. Scaffolds containing 2k PDMS_star-MA also displayed significant increases in collagen I and SPARC but at 20:80 (wt/wt %). However, when 2k PDMS_star-MA content was increased to a 30:70 (wt/wt %), collagen 1 and SPARC production fell below that of the PEG-DA control. Relative DNA levels tested verified that no significant cell death occurred for this particular scaffold (Fig. S7). Thus, we speculate that the more uniform dispersion of 2k PDMS_star-MA at this high concentration (30:70 wt/wt %) may have impacted scaffold surface hydrophobicity, resulting in protein adhesion inhibiting function of RGD receptors.

Figure 6.

Protein-level multiplex immunoassay results for (A) Collagen 1 and (B) SPARC presence after 14-day culture on templated SIPS PDMS_star-PEG and analogous PEG-DA control scaffolds were prepared with medium salt (~270 μm). (C) Von Kossa staining (brown-black color) for mineralization after 14 day culture with hBMSCs on scaffolds prepared with 2k PDMS_star-MA. Due to the intensity of staining, only macro-images could be taken due to over-saturation during imaging. Statistically significant difference compared to the control is represented as * = p <0.05, to the corresponding 10:90 (wt/wt %) as # = p < 0.05, and to the corresponding 20:80 as $ = p < 0.05. Scale bar = 2 mm.

Given the enhanced osteogenesis observed with templated SIPS PDMS_star-PEG scaffolds based on 2k PDMS_star-MA at 20:80 (wt/wt %), further studies were conducted for the 2k scaffold series. Specifically, levels of osteopontin (OPN), a bone ECM protein involved in biomineralization, and bone morphogenic protein 2 (BMP-2), a potent osteogenic growth factor, were analyzed by immunoassay. Although no statistical differences were observed for BMP-2 expression levels, the 20:80 (wt/wt %) scaffold supported a ~1.5-fold increase in OPN levels relative to the PEG-DA control (Fig. S8). However, when 2k PDMS_star-MA levels were further increased to 30:70 (wt/wt %), OPN levels were reduced below that of the 10:90 and 20:80 (wt/wt %) formulations as seen previously with collagen I and SPARC (Fig. 6A&B). Von Kossa staining was used as a visual indicator of matrix mineralization and also revealed considerable enhancement in calcium deposition with increasing 2k PDMS_star-MA content (represented by a brown-black color) (Fig. 6C ). Despite the reductions in collagen 1, SPARC, and BMP-2 noted for the 30:70 (wt/wt %) scaffold, this formulation appeared to support increased mineralization relative to remaining formulations, denoting a possible disconnect between bioactivity and osteoinductivity of the scaffold based on distribution of the PDMS_star-MA.

4. Conclusions

Herein, we describe the preparation of templated SIPS PDMS_star-PEG hydrogels and their utility as instructive scaffolds (absent of exogenous growth factors) for bone regeneration. These hydrogels utilized an inorganic hydrophobic comacromer, PDMS_star-MA, to introduce bioactivity and osteoinductivity. Fabrication with a fused salt template, in conjunction with SIPS employing a DCM precursor solution (followed by drying and hydration), created interconnected macropores of tunable sizes to permit post-fabrication cell seeding and for osteoconductivity. Hydrogels were produced with two M_n’s of PDMS_star-MA (2k and 7k) and varying wt/wt % ratios (0:100, 10:90, 20:80, and 30:70) of PDMS_star-MA to PEG-DA (M_n = 3.4k). CLSM images revealed the improved distribution of the 2k PDMS_star-MA within the PEG matrix, attributed to its superior solubility in the DCM precursor solution. Pore size tunability and interconnectivity for osteoconductivity were confirmed by SEM imaging and a water wicking test. Templated SIPS PDMS_star-PEG hydrogels all had similarly high bulk hydration, resulting in similar bulk Young’s moduli. However, for the 2k-containing scaffolds, a lower modulus was seen at a local level (via AFM probe measurement), but they remained in the range associated with osteogenesis in hydrogels. While absent for the PEG-DA control, bioactivity was confirmed for templated SIPS PDMS_star-PEG hydrogels (2k series) with HAp formation observed (via SEM and XRD) following soaking in SBF. Under accelerated conditions, owing to their similarity in pore size and bulk swelling, these hydrogels exhibited temporal mass loss similar to that of the PEG-DA control. Finally, all RGD-containing templated SIPS PDMS_star-PEG scaffolds were confirmed non-cytotoxic and showed PDMS dose-dependent increases in key indicators of osteogenesis. However, this was dependent on comacromer M_n, wherein the better dispersed comacromer (lower M_n) showed statistical differences at lower concentrations. Von Kossa staining was used to reveal dose-dependent increases in calcium deposition by hBMSCs seeded on 2k-containing scaffolds. Overall, these templated SIPS PDMS_star-PEG hydrogels displayed the potential to serve as instructive scaffolds for bone regeneration without the inclusion of added growth factors. The achievement of the observed bioactivity and osteoinductivity without the inclusion of glass or glass-ceramic fillers but rather a polymeric component (i.e. PDMS_star-MA) is notable.