Gradient scaffold with spatial growth factor profile for osteochondral interface engineering

A gradient template structure was fabricated using thermal sintering and porogen leaching technique developed in our lab [35, 36]. In this method, poly-(85-lactide-co-15-glycolide) (PLGA 85/15, Evonik Industries) microspheres were combined with NaCl porogen to form a 3D, porous structure. To form the microspheres, PLGA 85/15 was dissolved in methylene chloride at a 1:6 ratio and poured into a beaker of 1% polyvinyl alcohol solution spinning at 250 rpm to form microspheres via an oil-in-water emulsion process. The PLGA microspheres were then vacuum filtered to dry, and sieved to obtain spheres in the size range of 355–425 µm. The collected microspheres were then mixed with NaCl (size range 106–212 µm) at different ratios (0%, 5%, 10%, 20%, and 40% NaCl by weight). These different mixtures were sequentially layered upon each other within a metal mold, generating a NaCl porogen gradient through the scaffold. The mold was then thermally sintered at 100 °C for 1 h to form scaffolds of 5 mm diameter and 10 mm length. Once the mold reached room temperature, the structure was collected and leached for porogen removal by soaking samples in deionized water for 2 h [15, 35]. Template structures without porogen and with 30% porogen evenly distributed throughout the structure were also prepared using the same method.

To create an inverse gradient matrix, the available pore space of the developed template scaffold was filled with a hyaluronic acid hydrogel (Glycosil, Advanced Biomatrix). In short, the hydrogel was mixed with polyethylene glycol (PEG) crosslinker (Extralink, Advanced Biomatrix) in a 2:1 ratio, and immediately pipetted onto the more porous end (distal end) of the scaffold, as previously reported [17]. Prior to completely crosslinking, the gel infiltrates downward, filling the scaffold's available pore space. Using this method, an inversely graded matrix is developed that has a distal cartilaginous region and a proximal (less porous) osseous region.

To increase regenerative potential of both cartilage and bone phases, growth factors were incorporated into the inverse gradient matrix. The gel phase was formed with thiol-modified gel and crosslinker to increase the gel phase affinity to bind with external factors. For the template matrix, PLGA polymer was combined with hydroxyapatite (HAp) nanoparticles, a strong promoter of growth factor adsorption [44, 45]. Just as before, PLGA and HAp were dissolved in methylene chloride at a 1:4 ratio to form PLGA:HAp composite microspheres [46]. The composite microspheres were used to form template structures as well as the inverse gradient matrices using the methods described above. The completed PLGA:HAp scaffolds were subjected to x-ray energy dispersive spectroscopy (EDS) analysis to confirm HAp nanoparticle distribution. The scaffolds were sputter coated with gold/palladium for 3 min (E5100, Polaron) and examined using a scanning electron microscope (Nova NanoSEM 450, FEI). EDS was performed using Oxford Aztec Energy Microanalysis System with X-Max Silicon Drift Detector using an accelerating voltage of 15 kV.

Template structures were imaged using cone beam micro-focus x-ray computed tomography (µCT40, Scanco Medical AG, Bassersdorf, Switzerland) to render 3D models for direct quantitation of the porosity gradient [15]. Serial tomographic images were acquired at 45 kV and 177 µA, collecting 2000 projections per rotation at a 300 ms integration time. 3D, 16-bit grayscale images were reconstructed using standard convolution back-projection algorithms with Shepp and Logan filtering, and rendered within a 12.3 mm field of view at a discrete density of 4629 630 voxels mm⁻³ (isometric 6 µm voxels). Segmentation of solid scaffold from open porosity was performed in conjunction with a constrained Gaussian filter to reduce noise, applying a threshold of −100 Hounsfield units (water = 0, air = −1000). Scaffold pore volume was measured directly in a plane transverse to the cylindrical axis at every 6 µm interval, producing a gradient map of pore volume along the length of the scaffold.

To verify infiltration of hyaluronic acid gel into the inverse gradient matrix provided by the PLGA, and to quantify this inverse gradient of the two discrete phases, the same micro-CT imaging technique was used as described above. Since the two constituent phases attenuate x-ray to a low degree and without differentiation, a contrast agent was added to the gel phase in a 1:1 ratio (Histopaque, Sigma-Aldrich) to clearly distinguish the two discrete phases of hyaluronic acid gel and PLGA [47, 48].

To increase growth factor binding, the scaffolds were treated with oxygen plasma. Briefly, PLGA:HAp gradient template structures were placed in the chamber of a CUTE-1MP/R Plasma Processing System (FEMTO Science Inc.) and were individually treated for 5 min at 100 W power, 5 torr processing pressure, and 0.5 torr base pressure.

To characterize the effect of the oxygen plasma treatment, PLGA:HAp composite films were treated with oxygen plasma and then subjected to contact angle measurements. In brief, PLGA and HAp was dissolved in methylene chloride, as described above, and cast into a plastic 100 × 15 mm culture dish covered with Bytec paper (Saint Gobain). The dissolved polymer was poured onto the dish at 4 °C. The composite film was left overnight in 4 °C and further dried in a desiccator. After 24–48 h, 18 mm diameter pieces were bored from the films and were plasma-treated using the same parameters described above. Finally, using a goniometer (Rame Heart, Model 100), the static contact angle measurement was performed for untreated as well as plasma-treated films (n = 5).

To test and model growth factor loading, the PLGA:HAp template scaffold and the crosslinked hyaluronic acid hydrogel were loaded with bovine serum albumin (BSA) fluorescently tagged with either Alexa Fluor 647 or Alexa Fluor 555, respectively (ThermoFisher Scientific). For the template structure, 80 µl of the reconstituted BSA with Alexa Fluor 647 (1 mg ml⁻¹) was added directly onto the scaffold and was allowed to dry overnight. The next day, 20 µl of BSA with Alexa Fluor 555 (1 mg ml⁻¹) was added to 33.3 µl of hyaluronic acid gel prior to addition of PEG crosslinker (2 hyaluronic acid:1 PEG), and the resulting solution was pipetted onto the distal end of the PLGA:HAp template that was previously loaded with BSA with Alexa Fluor 647. Once the gel was crosslinked, the distal and proximal ends of the BSA-loaded matrix were visualized using a Zeiss Confocor 2 Confocal Microscope for model protein distribution.

Inverse gradient matrices with PLGA:HAp composite templates infused with hyaluronic acid were selected for in vitro evaluation. The template scaffold was loaded with BMP-2 by loading 1 ml of BMP-2 solution (2.5 µg ml⁻¹) to the PLGA:HAp template and allowing it to dry and adsorb overnight. The scaffolds were cut into two halves along the transverse plane and labeled as either distal (more porous end of the graded template) or proximal ends (less porous end of the graded template).

For the hydrogel portion of the matrix, hyaluronic acid gel containing 1.25 µg of TGF- β1 was prepared by combining 20 µl of TGF-β1 solution (62.5 µgml⁻¹) with 33.3 µl of hyaluronic acid and 16.7 µl of PEG crosslinker (2:1 ratio, as described in section 2.2). Further, the gel was mixed with 500 000 CD271+ bone marrow-derived human MSCs (hMSCs; passage 5). hMSCs were isolated by concentrating freshly isolated human bone marrow aspirate with proper ethical controls and consent obtained by Lonza (Lonza, Cat #. 1M-125) with the automated Magellan system (Arteriocyte Medical Systems) to obtain concentrated bone marrow aspirate (cBMA) as previously described [49, 50]. Then, MSCs were isolated from the cBMA via magnetic-activated cell sorting for cells expressing the MSC marker, CD271, and cultured until passage 5 [35]. CD271 was used as a marker for MSCs due to its specificity to isolate homogeneous MSC population without going through plastic culture [51, 52]. The flow cytometry cell sorting data and tri-lineage differentiation experiments were conducted as per the established protocols [17]. Afterwards, based on the micro-CT data, the appropriate amount of gel containing hMSCs and TGF-β1 was added to both the distal and proximal portions of the cut scaffold to fill the available pore space (53.7% and 46.3% of the total mixture, respectively). Once crosslinked, distal portions of the scaffolds were cultured in the co-differentiation media, previously established by our group [17]. The proximal portions were treated with the same co-differentiation media with the addition of 100 000 hMSCs in the suspension to mimic the infiltration of bone marrow stromal cells that would occur post-implantation due to its proximity to bone marrow.

After 21 days, both the distal and proximal portions of the scaffolds were assessed for chondrogenesis through glycosaminoglycan (GAG) quantification (n = 3) via dimethyl methylene blue (DMMB) assay [35, 49, 53], and osteogenesis through mineral formation (n = 3) using the alizarin red assay [54]. Briefly, for GAG quantification, samples were removed from culture, digested in a proteinase K solution (1 mg ml⁻¹) for 16 h at 56 °C. Afterwards, 50 µl of the resulting lysate was combined with 200 µl of DMMB solution (16 µg ml⁻¹, pH 3) and absorbance was measured at 520 nm wavelength. For quantifying total calcium deposition, samples were removed from culture, washed with PBS, and fixed with 70% ethanol. The fixed samples were then stained with 40 mM alizarin red solution (pH 4.23) for 10 min. at room temperature. The samples were then washed thoroughly to remove excess dye and the stain was solubilized using a 10% cetylpyridinium chloride solution for 15 min. The solubilized dye was then measured for absorbance at 562 nm wavelength.

To determine whether BMP-2 might deter chondrogenesis or promote osteogenesis in the distal (cartilaginous) region of the matrix, chondrogenic cell pellets were made and treated with increasing concentrations of BMP-2. Briefly, 250 000 hMSCs, passage 5, were placed in microcentrifuge tubes to form a pellet and cultured with one of five media combinations. The control media was chondrogenic media (serum free, high glucose, basal media (Gibco) supplemented with ITS+, 100 μl ml⁻¹ sodium pyruvate, 40 μg ml⁻¹ L-proline, 50 μg ml⁻¹ ascorbate-2-phosphate, 10⁻⁷ M dexamethasone, and 10 ng ml⁻¹ TGF-β1). The remaining four groups were chondrogenic media supplemented with 25, 50, 100, or 250 ng ml⁻¹ of BMP-2.

BMP-2 in the culture media was refreshed once every 3 days when the media was changed. After 21 days of culture, the cell pellets were assessed for both their chondrogenesis via DMMB assay and osteogenesis using alizarin red staining. The scaffolds were also subjected to whole-mount immunofluorescence analysis using primary antibodies (Abcam, Cambridge, MA) to image chondrogenic (Collagen Type II (1:100) and Sox9 (1:50)) as well as osteogenic (Collagen I (1:100) and RUNX2 (1:50)) marker protein expression. Deoxyribonucleic acid (DNA) content was quantified using the PicoGreen double stranded DNA (DNA) assay and used to normalize GAG and calcium quantification.

Statistical analyses were performed using an unpaired t-test for all studies that compared the distal to the proximal end. The analyses to determine the effect of BMP-2 on chondrogenesis used a one-way analysis of variance with Tukey's post-test. Quantitative data were reported as means ± the standard deviation, and a significance level of p < 0.05 was used in all statistical tests performed.

We employed thermal sintering and porogen leaching scaffold fabrication method and varied the amount of porogen along the scaffold length. The resulting scaffolds were imaged using micro-CT to qualitatively and quantitatively characterize the variation in pore volume with respect to the amount of porogen added (figure 1). PLGA template without any added porogen presented a relatively even distribution of pores that fluctuated around 40% pore volume with respect to the longitudinal axis of the scaffold (figure 1(A)). PLGA template with an even amount of porogen (30% by mass) also showed relatively uniform porosity throughout the scaffold, around 45% pore volume (figure 1(B)). However, varying porogen content along the longitudinal axis resulted in a template structure with pore volume that gradually increased from approximately 35% to 65% (figure 1(C)). The gradient nature of the template structure is also clearly visible in the rendered image, where the bottom of the template (without any porogen) is densely packed with PLGA while the top (40% porogen) is relatively open with less PLGA. The increase in pore volume with respect to scaffold height corresponds with increasing amounts of porogen added to the scaffold, which ranged from 0% to 40% by weight.

Zoom In Zoom Out Reset image size

While the PLGA template structure is suitable for bone regeneration, an OC matrix requires bone and cartilage supporting regions in a single matrix system. To achieve this, an inverse gradient matrix was formed by infusing the gradient PLGA template with hyaluronic acid gel. The gel phase served as the secondary phase, which is known to support MSC chondrogenesis in vitro and in vivo [17, 55]. Micro-CT was performed to confirm the creation of an inverse gradient matrix. As seen in figure 2(A), moving from the top to the bottom of the scaffold (0–10 mm), hydrogel content decreased from 50% to 35%. In contrast, the PLGA content increased from 50% to 65%. Representative slices at 0 mm, 4.04 mm, and 8.10 mm clearly allowed visualization of the inverse relationship between the hydrogel and PLGA content along the scaffold length (figure 2(B)). The hydrogel (shown as black) is relatively prevalent in the 0 mm section, while the PLGA (shown as green) is much more abundant in the 8.10 mm section. These results show that the gel phase of the graft (hyaluronic acid) was able to infiltrate the PLGA gradient template to establish an inverse gradient matrix.

Zoom In Zoom Out Reset image size

The presence of HAp in the PLGA:HAp composite microsphere scaffold was analyzed via EDS analysis. As shown in figure 3, composite microspheres showed presence of HAp on the surface of the microspheres, determined by the co-localization of calcium and phosphorus atoms. The HAp nanoparticles formed clusters distributed sparsely along the surface of the PLGA microspheres. The Ca:P ratio was also found to be relatively close to that of HAp (1.57 vs 1.67) which indicates minimal change in chemical composition of HAp, although the results are only semi-quantitative. The PLGA:HAp composites were subjected for plasma treatment, as described in section 2.4. For this purpose, PLGA:HAp films were cast and plasma-treated in order to provide a flat surface for the contact angle measurements (as described in section 2.4). The untreated PLGA films showed a contact angle of 70.16 ± 3.41°, while the treated samples exhibited significantly lower contact angle of 48.72 ± 2.14° (figure 3). This indicates a significant increase in hydrophilicity of the template scaffold through oxygen plasma treatment.

Zoom In Zoom Out Reset image size

Fluorescently labeled model proteins were used to visualize growth factor distribution in the inverse gradient matrix. BSA tagged with Alexa Fluor 647 was used to model osteogenic factor adsorption to the plasma-treated PLGA template structure, and BSA tagged with Alexa Fluor 555 incorporated into the hydrogel was employed to track chondrogenic factor distribution in the gel phase. The template structure and the hydrogel with model proteins were integrated to form the inverse gradient matrix (figure 4). The osteogenic model protein was labeled with red fluorescence while chondrogenic model protein was labeled with green fluorescence. The distal, cartilaginous region of the matrix with an abundance of hyaluronic acid gel showed the presence of both osteogenic and chondrogenic model proteins. The proximal, osseous region of the matrix only showed the presence of osteogenic model protein. Since the amounts of each model protein in PLGA and gel phase is directly proportional to the amounts of each phase present, gradient matrix formation itself can result in spatial distribution of the respective growth factors.

Zoom In Zoom Out Reset image size

To assess growth factor-induced effects of the developed scaffold, inverse gradient matrices loaded with BMP-2 and TGF-β1 were studied in vitro. The BMP-2 loaded matrices were cut in half, loaded with hMSCs and TGF-β1 containing hydrogel, and cultured for 21 days in OC co-differentiation media (figure 5(A)). The proximal half of the cut scaffold was also cultured with cells in suspension to model the infiltration of bone marrow cells post-implantation.

Zoom In Zoom Out Reset image size

After the 21 days of culture, the top (distal; cartilaginous) portion of the graft presented significantly higher GAG content than the bottom (proximal; osseous) portion of the graft (figure 5(B)). The higher GAG expression by cells seeded on the top cartilaginous half of the graft indicated increased chondrogenesis by the seeded hMSCs. However, alizarin red staining and quantification of the cultured scaffolds showed no statistical difference in levels of mineralization for both halves, indicating the same level of osteogenic differentiation in both matrix portions (figure 5(C)). These trends remained after normalization with DNA content (figures 5(D) and (E)).

Although the in vitro characterization demonstrated the ability of the matrix to support chondrogenesis in the distal, cartilaginous region of the matrix, similar amounts of mineralization were seen in both regions of the matrix (figure 5). In addition, the inverse gradient matrix showed localization of TGF-β1 model protein only in the top portion of the matrix, but BMP-2 model protein was prevalent throughout the matrix (figure 4), as it is adsorbed to the template PLGA structure. To determine whether the presence of the BMP-2, an osteogenic factor, caused mineralization in the distal (cartilaginous) region, hMSC cell pellets were cultured in chondrogenic media containing TGF-β1 along with varied amounts of BMP-2 (0, 25, 50, 100, and 250 ng ml⁻¹).

The cultured samples were evaluated for chondrogenesis and osteogenesis via GAG and matrix mineralization quantification, respectively (figure 6). It was found that the addition of BMP-2 in any amount significantly increased the amount of GAG produced by the cells, indicating higher chondrogenic differentiation (figure 6(A)). Increasing the amount of BMP-2 also significantly increased the amount of GAG produced in a dose-dependent manner, up to 250 ng ml⁻¹. Meanwhile, the addition of BMP-2, in any concentration, did not have a significant effect on the amount of mineralization in the cell pellets (figure 6(B)).

Zoom In Zoom Out Reset image size

Immunofluorescence staining was further used to determine the influence of BMP-2 on chondrogenic differentiation. Osteogenic differentiation markers RUNX2 (green) and Collagen type I (Col1; red) were used to visualize osteogenic differentiation (figure 7(A)), and chondrogenic differentiation was visualized using Sox9 (green) and Collagen type II (Col2; red) (figure 7(B)). The cultures without any BMP-2 still showed relatively abundant expression of RUNX2 and Col1, indicating some levels of osteogenic differentiation. The addition of BMP-2 and its effects on RUNX2 expression did not present a distinct trend, but the inclusion of 100 ng ml⁻¹ and 250 ng ml⁻¹ of BMP-2 displayed a dramatic decrease in RUNX2 expression. Meanwhile, the Col1 expression decreased in a dose-dependent manner with lower doses of BMP-2 (25 and 50 ng ml⁻¹), while no expression was observed for higher doses (100 ng ml⁻¹ and 250 ng ml⁻¹) of BMP-2. On the other hand, the cells cultured without BMP-2 did not express much Sox9 or Col2, and were undergoing minimal chondrogenic differentiation. The groups with 50 ng ml⁻¹ and 100 ng ml⁻¹ of BMP-2 also did not show high levels of Sox9 expression. However, the cultures treated with 25 ng ml⁻¹ and 250 ng ml⁻¹ of BMP-2 expressed higher levels of Sox9. As for Col2 expression, all groups treated with BMP-2 expressed higher levels of Col2, with 25 ng ml⁻¹ of BMP-2 group showing the highest level of Col2 expression.

Zoom In Zoom Out Reset image size