Composite Spheroid-Laden Bilayer Hydrogel for Engineering Three-Dimensional Osteochondral Tissue

Materials

Poly(ι-lactic acid) (PLLA; 5.7–8.5 dL/g intrinsic viscosity, Mn 350–500 kDa) was purchased from Samyang (Seoul, Korea). The 1,1,1,3,3,3-Hexafuloro-2-propanol (HFIP) and 4% paraformaldehyde were acquired from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan), and Tris-HCl was acquired from Alfa Aesar (Haverhill, MA). Ethylenediamine (EDA), isopropyl alcohol (IPA), dopamine hydrochloride, fluorescein isothiocyanate (FITC), 2-hydroxy-2-methylpropiophenone, an ApopTag^® Peroxidase In Situ Apoptosis Detection Kit, sodium bicarbonate (NaHCO₃), sodium carbonate (Na₂CO₃), dimethyl sulfoxide (DMSO), Alizarin Red S, Alcian Blue solution, and Trypan Blue solution were purchased from Sigma-Aldrich (St. Louis, MO).

For cell culture and hydrogel, hADSCs, basal medium, fetal bovine serum (FBS), and penicillin/streptomycin (P/S) were purchased from CEFOBIO (Seoul, Korea). Ultra-low cell-adhesive micro-well plates, Labsphero™, were obtained from Labtolab (Daejeon, Korea). Phosphate-buffered saline (PBS), its ion-free version (Dulbecco's phosphate-buffered saline; DPBS), and trypsin/ethylenediaminetetraacetic acid were purchased from Welgene (Gyeongsan, Korea).

GelMA (Gelatin from porcine skin, type A, gel strength: ∼300 bloom, degree of methacrylation: >90%) was acquired from 3D Materials (Busan, Korea), and ultraviolet (UV) spot curing system was from UV SMT (Bucheon, Korea). Growth factors, bone morphogenetic protein-2 (BMP-2), and transforming growth factor beta-1 (TGF-β1), were obtained from PeproTech (Rocky Hill, NJ).

For the in vitro assays, rhodamine/phalloidin, a LIVE/DEAD^® Viability Kit, and CellROX green reagent were purchased from Invitrogen (Carlsbad, CA). Diamidino-2-phenylindole (DAPI) was obtained from Vector Laboratories (Burlingame, CA). Hematoxylin and Eosin (H&E) solution was purchased from BBC Biochemical (Mount Vernon, MA). A Zeba Spin desalting column was purchased from Thermo Scientific (Waltham, MA). Primary antibodies (anti-runt-related transcription factor 2 [RUNX2], osteopontin [OPN], SRY-box transcription factor 9 [SOX9], aggrecan, collagen type 1a [Col1a], and collagen type 2a [Col2a]) and enzyme-linked immunosorbent assay (ELISA) kits (BMP-2, bone sialoprotein [BSP], TGF-β1, and TGF-β3) were purchased from Abcam (Cambridge, United Kingdom).

Secondary anti-mouse/rabbit immunoglobulin G (IgG) biotin-conjugated antibody was obtained from Sigma-Aldrich, and FITC-streptavidin was purchased from eBioscience (San Diego, CA).

Components of real-time quantitative polymerase chain reaction (RT-qPCR; the RNeasy Mini Kit, Maxime RT Premix, and SYBR Premix Ex-Taq) were purchased from Qiagen (Valencia, CA), Intron Biotechnology (Seoul, Korea), and TaKaRa Bio (Otsu, Japan), respectively.

Preparation of composite spheroids

To synthesize the segmented fibers, aligned nanofiber sheets were synthesized by electrospinning (5 mL/h, 23-G needle, 12–14 keV) 10 mL of 3% PLLA solution in HFIP. The collected nanofiber sheets were dried at room temperature for 24 h and cut into four pieces. The pieces were embedded in frozen section compound, frozen at −70°C, and cryosectioned using a Cryostat cryocut microtome (Leica Biosystems GmbH, Wetzlar, Germany). The sectioned fibers, 60–100 μm in length, were selectively collected by sieving, treated with 10% (v/v) EDA in IPA for 1 h, and sequentially washed with IPA, 70% ethanol, and distilled water (DW). For polydopamine (PD) coating, the fibers were immersed in a 2 mg/mL dopamine hydrochloride solution (Tris-HCl buffer, pH 8.5) for 20 min. The PD-coated fibers were sequentially sterilized with 70% ethanol, DW, and UV light before use. Morphology of the fibers was observed by Field emission scanning electron microscopy (FE-SEM; Hitachi S 4800 FE-SEM; Hitachi, Tokyo, Japan).

For sample preparation, the fibers were completely dried at 37°C, attached to a scanning electron microscopy (SEM) holder, and coated with platinum by ion sputter (6 nm/s, 10–15 nm in thickness). Length of each fiber was quantified using ImageJ software (National Institutes of Health, Stapleton, NE).

The hADSCs were cultured in basal medium with 10% FBS and 0.5% P/S under standard culture conditions (5% CO₂ and 37°C). Cells at passage number 4–5 were used for all experiments, and the medium was refreshed every 2 days. We placed 1 × 10⁶ hADSCs and 125 μg of fibers in a LabSphero with 6000 wells (167 cells and 0.0208 μg of fibers per well); 2.5 × 10³ hADSCs and 0.3125 μg of fibers were placed in one well of an ultra-unattached cell plate (SPL Life Science, Pocheon, Korea); and 5 × 10³ hADSCs and 0.625 μg of fibers were placed in the same plate to prepare small, medium, and large spheroids, respectively. The fibers and cells formed a spheroid after 24 h of incubation.

Phase-contrast images of the spheroids were captured using an optical microscope (CKX41; Olympus, Tokyo, Japan), and area of each spheroid was measured from the images. FITC-labeled fibers were prepared by diluting 68 μg/mL of FITC powder in PLLA solution before electrospinning to visualize the incorporation of fibers in spheroids. Small spheroids incorporating the FITC-labeled fibers were cultured for 24 h and fixed in 4% paraformaldehyde for 1 h, washed three times with PBS, and then stained with rhodamine/phalloidin (1:200 in PBS) for 1 h at 37°C.

Each sample was mounted with DAPI mounting medium (nuclei staining) and observed using a confocal laser microscope (TCS SP5, Leica Biosystems GmbH).

Encapsulation of spheroids in hydrogel

GelMA was dissolved in DPBS (7.5% w/v), incubated for 2 h at 37°C, and mixed with 2-hydroxy-2-methylpropiophenone (photo initiator, 0.1% v/v) after filtering (0.2 μm pores). Then, for engineering one hydrogel, 100 μL of hydrogel solution encapsulating 240, 16, and 8 of small, middle, and large spheroids, respectively (total number of cells in a hydrogel is equally 40,000 cells for all groups), were casted in a polytetrafluoroethylene (PTFE) mold (8 mm in diameter, 2 mm in depth, and cylindrical shape). The spheroids were gently agitated in hydrogel solution, and the solution was then immediately irradiated with UV light (350 mW/cm², 10 s, 7 cm of distance from each sample) for gelation. The spheroid-laden hydrogels were cultured for 24 h and then fixed in 4% paraformaldehyde for 20 min, dehydrated by immersion in sequential steps of gradient ethanol (70–100%) and xylene, and embedded in paraffin blocks. The paraffin blocks were horizontally sectioned into 10 μm slices using a microtome (Leica Biosystems GmbH) for H&E staining.

The sectioned specimens were hydrated by following the dehydration steps in reverse (xylene, 100–70% ethanol, and water), stained with Hematoxylin for 2 min and with Eosin for 6 min, dehydrated, and mounted. The H&E-stained images were captured using a phase-contrast microscope (Eclipse Ti2; Nikon, Tokyo, Japan). Vertically sectioned hydrogels were stained with calcein AM for 30 min and observed using a fluorescence microscope (Eclipse Ti2; Nikon) to visualize the distribution of spheroids in hydrogels. The positions of spheroids in each gel were divided from bottom to top (0–20%, 20–40%, 40–60%, 60–80%, and 80–100%), and the number of spheroids in each position was quantified.

The small spheroids were used for further experiments. Spheroid-encapsulating hydrogels with various UV exposure times (10, 30, and 60 s) were prepared following the aforementioned gelation process, and optical images of each hydrogel were captured using a camera. After 1 and 14 days of culture, storage modulus of each hydrogel was measured using a rheometer (HR10; TA Instruments, New Castle, DE) under 0.4 N of axial force and 1% strain, and weight of each sample was measured before and after freeze drying to calculate swelling ratio by the following formula: Swelling ratio (w/w) = (weight of swollen hydrogel − weight of dried hydrogel)/weight of dried hydrogel [(W_s − W_d)/W_d]. Phase-contrast images of each hydrogel with spheroids were captured with the optical microscope, and each sample was subjected to H&E and calcein AM staining as described above. Sprouting length from an encapsulated spheroid on each calcein AM-stained image was measured using ImageJ software. Internal structure of each hydrogel and cells from the encapsulated spheroids were observed by FE-SEM.

Before gene expression analysis, each hydrogel was dissolved in collagenase solution (100 CDU/mL, CDU = collagen digestion units) for 1 h at 37°C. The supernatant was discarded after centrifugation (1200 rpm, 3 min), and messenger RNA (mRNA) was extracted and purified using the RNeasy Mini Kit. Concentration of mRNA was measured using a nano-spectrometer (NanoDrop 2000; Thermo Scientific, Wilmington, DE).

Complementary DNA (cDNA) was synthesized by adding 1 μg of mRNA and diethyl pyrocarbonate (DEPC)-treated water to a cDNA Polymerization Kit (total volume: 20 μL; Maxime RT Premix Kit; Intron Biotechnology). Then, 2 μL of the cDNA solution was mixed with 0.4 μL of SYBR reagents, 0.4 μL of ROX reference dye (50 × ), 6.8 μL of DEPC-treated RNA-free water, and 0.4 μL of target-specific primers for the gene of interest, and the samples were processed for RT-qPCR (StepOnePlus Real-Time PCR System; Applied Biosystems, Foster City, CA), with amplification during 40 cycles of annealing at 95.0°C for 15 s, extension at 60.0°C for 60 s, and a melting curve stage at 60.0°C to 95.0°C in increments of 0.5°C per 5 s.

The primer sequences were as follows: rho-associated protein kinase 1 (ROCK1): forward, 5′-GAA GCT CGA GAG AAG GCT GA-3′, reverse, 5′-TTG TCT GCC TCA AAT GCT TG-3′; mitogen-activated protein kinase 1 (MAPK1): forward, 5′-CCA GAC CAT GAT CAC ACA GG-3′, reverse, 5′-CTG GAA AGA TGG GCC TGT TA-3′; matrix metallopeptidase 2 (MMP2): forward, 5′-ATG ACA GCT GCA CCA CTG AG-3′, reverse, 5′-ATT TGT TGC CCA GGA AAG TG-3′; matrix metallopeptidase 9 (MMP9): forward, 5′-TTG ACA GCG ACA AGA AGT GG-3′, reverse, 5′-GCC ATT CAC GTC GTC CTT AT-3′; and housekeeping gene GAPDH: forward, 5′-GTC AGT GGT GGA CCT GAC CT-3′, reverse, 5′-TGC TGT AGC CAA ATT CGT TG-3′. The relative gene expression from gels treated with UV for 10 and 30 s was normalized to that from the 60 s group. DNA assay was performed for each spheroid-encapsulating hydrogel after 1 or 14 days of culture.

The hydrogels at each time point were lysed in 300 μL of radioimmune precipitation assay (RIPA) lysis buffer (150 mM NaCl, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 150 mM Tris-HCl, pH 7.2), ground using a homogenizer, and reacted with working reagents from a Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen). A spectrophotometer (Varioskan LUX; Thermo Scientific, Waltham, MA) was used to detect fluorescent intensities at an excitation wavelength of 480 nm and emission wavelength of 520 nm. For terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay, the hydrogel samples were embedded in paraffin blocks by following the procedure described above and cut into 10 μm slices. The specimens were hydrated, and DNA fragments were stained using an ApopTag Fluorescein In Situ Apoptosis Detection Kit and a fluorescence microscope. Intracellular oxidative stress level of encapsulated spheroids was analyzed by incubating the hydrogel samples in CellROX staining solution (5 μM in culture medium) for 1 h at 37°C and observing fluorescence with the fluorescence microscope.

TUNEL-positive nuclei and relative CellROX intensity were measured using Photoshop CS6 software (Adobe, San Jose, CA).

Immobilization of growth factors on the fiber

We agitated 1 mg of PD-coated fibers with BMP-2 or TGF-β1 solution (10 μg/mL in Tris-HCl buffer, 4°C, overnight) to synthesize BMP-2- and TGF-β1-immobilized fibers. FE-SEM was used to observe the surface morphology of uncoated fiber, PD-coated fiber, BMP-2-immobilized fiber, and TGF-β1-immobilized fiber, which were named Fiber, PD, BMP-2, and TGF-β1, respectively. To visualize the immobilization of growth factors, BMP-2 or TGF-β1 was dissolved in a reaction buffer (3 μg/mL in 500 mM carbonate, pH 9.5) and reacted with FITC solution (1 mg/mL of FITC powder in DMSO) for 1 h at room temperature to chemically conjugate the BMP-2 and TGF-β1 with the FITC. A Zeba desalt spin column (89889; Thermo Scientific, Rockford, IL) was used to remove the unreacted FITC, and the FITC-labeled growth factors were coated onto PD-coated fibers following the same immobilization procedure. Fluorescent images were then captured using the fluorescence microscope, and phase-contrast images of the same spots were used for background.

The amount of each growth factor in the supernatant after immobilization process was measured using a BMP-2 or TGF-β1 ELISA Kit, and the values were subtracted from the initial amount of growth factor provided to calculate the coating efficiency. High-resolution X-ray photoelectron spectroscopy (Theta Probe Base System; Thermo Scientific, Waltham, MA) was analyzed to characterize surface chemical composition of each fiber.

In vitro osteogenic and chondrogenic differentiation of hADSCs from spheroids within hydrogel

Different groups of hydrogels encapsulating spheroids were prepared using various BMP-2 delivering methods (GFS: without growth factors, GBS-1: 50 ng of BMP-2 diluted in the culturing medium, GBS-2: 50 ng of BMP-2 physically blended in the hydrogel, and GBS-3: 50 ng of BMP-2-immobilized fibers in the spheroids). The hydrogels were cultured for 7, 14, and 21 days and prepared for RT-qPCR analysis following the processes described above. The primer sequences were as follows: RUNX2: forward, 5′-GCA GTT CCC AAG CAT TTC AT-3′, reverse, 5′-CAC TCT GGC TTT GGG AAG AG-3′; Osterix (OSX): forward, 5′-TAA TGG GCT CCT TTC ACC TG-3′, reverse: 5′-CAC TGG GCA GAC AGT CAG AA-3′; Osteocalcin (OCN): forward, 5′-GTG CAG AGT CCA GCA AAG GT-3′, reverse, 5′-TCA GCC AAC TCG TCA CAG TC-3′; OPN: forward, 5′-TGA AAC GAG TCA GCT GGA TG3′, reverse, 5′-TGA AAT TCA TGG CTG TGG AA-3′. Each group was cultured for 14 days, embedded in paraffin, and then sectioned into 10 μm slices following the processes given above.

For immunohistochemistry (IHC) staining, the sectioned hydrogels were hydrated and reacted with anti-RUNX2 and anti-OPN solution (1:100 in blocking buffer, 5% FBS and 0.1% Tween-20 in PBS) at 4°C for 24 h, followed by reactions with anti-mouse IgG biotin-conjugated secondary antibodies (1:100 in PBS) at 37°C for 2 h and FITC-conjugated streptavidin tertiary antibodies (1:100 in PBS) at 37°C for 2 h. The stained samples were then mounted with DAPI mounting medium, and images were obtained using the fluorescent microscope. RUNX2- and OPN-stained DAPI signals were counted to quantify the number of RUNX2- and OPN-positive nuclei. Deposition of mineral calcium was visualized by staining whole hydrogel and then staining histologically sectioned specimen with Alizarin Red S. The samples were fixed in 4% paraformaldehyde for 20 min, hydrated, and immersed in Alizarin Red S working solution (2% in DW, pH 4.2) for 5 min. The images were obtained using the optical microscope.

Deposited calcium ions were extracted by incubating the gels in 300 μL of 0.6 N HCl buffer for 24 h at 37°C, and 5 μL of the supernatant was added to 200 μL of working reagent from the QuantiChrom Calcium Assay Kit (Bioassay Systems, Hayward, CA). Optical densities from each sample were measured using the spectrophotometer at 612 nm. A large spheroid (4 × 10⁴ of hADSCs and 5 μg of BMP-2 immobilized fibers) was placed in a six-well culture plate and cultured with growth medium for 14 days to investigate osteogenic differentiation of cells sprouting from a spheroid containing BMP-2-immobilized fibers. RUNX2 IHC staining was performed by following the previously stated procedures. Images were captured with the fluorescent microscope, and RUNX2-positive cells that spread from a spheroid in designated areas were counted (distance from a spheroid A: 0–1.3 mm, B: 1.3–2.6 mm, and C: 2.6–3.9 mm). Phase-contrast images of the same spots were also captured.

Different groups of hydrogels for encapsulating spheroids were also prepared using various TGF-β1 delivering methods (GFS: without growth factors, GTS-1: 50 ng of TGF-β1 diluted in the culturing medium, GTS-2: 50 ng of TGF-β1 physically blended in the hydrogel, and GTS-3: 50 ng of TGF-β1-immobilized fibers in the spheroids). The hydrogels were cultured for 7, 14, and 21 days and prepared for RT-qPCR analysis following the processes described above. The primer sequences were as follows: SOX9: forward, 5′-GCG GAG GAA GTC GGT GAA GA-3′, reverse, 5′-GTA GAC GGG TTG TTC CCA GT-3′; aggrecan: forward, 5′ -ACA GCT GGG GAC ATT AGT GG-3′, reverse, 5′-GTG GAA TGC AGA GGT GGT TT-3′; chondroadherin (CHAD): forward, 5′-ACC TGG ACC ACA ACA AGG TC-3′, reverse, 5′-TAG CTG GAC AGC TGG TTC CT-3′; and cartilage oligomeric matrix protein (COMP): forward, 5′-AGG ACA ACT GCG TGA CTG TG-3′, reverse, 5′-GTG TCC TTT TGG TCG TCG TT-3′. Each group was cultured for 21 days, embedded in paraffin, and sectioned following the aforementioned processes.

For IHC staining, the sectioned hydrogels were hydrated and reacted with anti-aggrecan and anti-SOX9 solution (1:100 in blocking buffer, 5% FBS, and 0.1% Tween-20 in PBS) at 4°C for 24 h, followed by the IHC reactions described in the previous paragraph. The images were obtained using the fluorescent microscope. Aggrecan- and SOX9-stained DAPI signals were counted to quantify the number of aggrecan- and SOX9-positive nuclei.

The deposition of GAGs was visualized by staining the whole hydrogel and histologically sectioned specimens (paraffin embedding process, 10 μm slices) with Alcian Blue solution. The samples were fixed in 4% paraformaldehyde for 20 min, hydrated, and immersed in Alcian Blue solution (1% in 3% acetic acid, pH 2.5) for 30 min. The images were obtained with the optical microscope. Deposited GAGs were extracted by lysing the gels in 300 μL of RIPA buffer at 4°C for 24 h, and then 20 μL of supernatant was mixed with 200 μL of working reagent (16.0 mg of dimethylmethylene blue [DMMB], 3.0 g of glycine, 1.6 g of NaCl, 95.0 mL of 0.1 M acetic acid in 1 L of DW). Optical densities of each sample were measured using the spectrophotometer at 525 nm.

A large spheroid (4 × 10⁴ hADSCs and 5 μg of TGF-β1-immobilized fibers) was placed in a six-well culture plate and cultured with growth medium for 21 days to investigate chondrogenic differentiation of cells sprouting from a spheroid containing TGF-β1-immobilized fibers. SOX9 IHC staining was performed by following the previously stated procedures. The images were captured using the fluorescent microscope, and SOX9-positive cells spread from a spheroid were counted in designated areas (distance from a spheroid A: 0–1.3 mm, B: 1.3–2.6 mm, and C: 2.6–3.9 mm). Phase-contrast images of the same spots were also captured.

Preparation and characterization of 3D bilayer hydrogel

A PTFE mold for a 3D biphasic hydrogel was newly designed (cylindrical shape, 8 mm in diameter, 4 mm in height: double the height of a monolayer hydrogel). We sequentially put 100 μL of GelMA solution and spheroids into the molds for gelation of the first and second layers, and each layer was UV irradiated for 5, 7, or 10 s. Trypan Blue dye was mixed with hydrogel solution for the first layer to distinguish the layers. Storage moduli of whole biphasic hydrogels and each separate layer were measured using the rheometer.

The biphasic hydrogels were cultured in growth medium for 1 and 14 days, and then they were vertically sectioned for analysis. H&E staining and calcein AM staining of the specimens followed the methods given in previous paragraphs. The interface of bilayer hydrogels was evaluated for uniaxial tensile (pull-off test) and horizontal shear (lap-shear test) strength using a universal testing machine (UNITEST M1; TESTONE Co., Ltd., Sihueng, South Korea) with a 100 N cell. For each test, both sides of the hydrogels were attached to customized acrylic plates (1 × 1 cm and 1 × 4 cm) using 10 μL of commercial cyanoacrylate (Loctite 401; Henkel).

The complexes were fixed into a metal holder within the universal testing machine after 10 min. Tensile and shear stresses were increased at 1 mm/min until the interface of bilayer hydrogel was separated.

In vitro differentiation of hADSCs from spheroids within the bilayer hydrogel

Biphasic hydrogel encapsulating spheroids after treatment with 7 s of UV irradiation for the first and second layers were used for further experiments. Experimental groups were as follows: Biphasic hydrogel composed of GTS-2 and GBS-2 and biphasic hydrogel composed of GTS-3 and GBS-3. The biphasic hydrogels were cultured for 21 days to investigate osteogenic and chondrogenic differentiation of encapsulated cells. Following the Alizarin Red S or Alcian Blue staining procedures described above, whole bilayer hydrogels, and the specimens after paraffin embedding and sectioning of the hydrogels were stained to visualize macroscopic and internal staining images of the bilayer hydrogel. The images were obtained using the digital camera and optical microscope. For RT-qPCR analysis, the biphasic layers were divided into individual layers by sectioning middle area (GBS-2 and GTS-2, and GBS-3 and GTS-3). The osteogenic or chondrogenic gene expression of separated layers from each group was analyzed.

The same primers mentioned in the previous paragraph were used, and the primers for collagens were as follows: Col1a: forward, 5′-ACA GGG CTC TAA TGA TGT TGA-3′ and reverse, 5′-AGG CGT GAT GGC TTA TTT GT-3′; Col2a: forward, 5′-ACA GGG CTC TAA TGA TGT TGA-3′ and reverse, 5′-AGG CGT GAT GGC TTA TTT GT-3′. The sectioned specimens were also used for IHC staining of Col1a and Col2a, and the images were captured using the fluorescent microscope. Col1a- and Col2a-stained DAPI signals were counted to quantify the number of Col1a- and Col2a-positive nuclei. The staining and RT-qPCR followed the same procedures described above. The media culturing each biphasic hydrogel were collected at 1, 4, 7, 10, 14, and 21 days, and the amount of BSP and TGF-β3 in the obtained media was measured using each ELISA Kit.

Statistical analysis

All quantitative data were directly calculated and expressed without preprocessing as the mean ± standard deviation at least three independent experiments. GraphPad Prism 7.0 software (La Jolla, CA) was used to perform one- or two-way analysis of variance followed by Tukey's honestly significant difference test for all investigations, except the experiments comparing two groups, which were analyzed by Student's t-test. Significances are denoted as p < 0.05, p < 0.01, and p < 0.001 (p-values <0.05 were considered significant), and all the sample sizes (n) were calculated from at least triplicate samples (n ≥ 3). All sample sizes and statistical significance are independently denoted in each figure legend.

Preparation and encapsulation of composite spheroids within a GelMA hydrogel

FE-SEM images show the morphology of segmented fibrils (Fig. 2a) of an average length of 57.1 μm (Fig. 2b). All composite spheroids were spherical, irrespective of their size (Fig. 2c, d). The area and diameter of small, middle, and large spheroids were measured and found to be 0.07 ± 0.01, 0.06 ± 0.01, and 0.11 ± 0.04 mm² (Fig. 2d), and 85 ± 5, 272 ± 16, 379 ± 32 μm (data not shown), respectively. A confocal image of the small spheroid demonstrates homogeneous distribution of fibers and hADSCs within the spheroid (Fig. 2e). The spheroids were encapsulated within a GelMA hydrogel under UV exposure (10 s) (Fig. 2f), and H&E staining showed stable localization of spheroid within the hydrogel by forming the number of cell–matrix binding (red arrows) between the hydrogel matrix and encapsulated spheroids (Fig. 2g). Calcein AM staining of vertically sectioned spheroid-laden hydrogels shows homogeneous distribution of the small spheroids in hydrogel, whereas the medium and large spheroids subsided to the bottom of hydrogel during gelation (Fig. 2h). The only small spheroids were evenly positioned throughout hydrogel (Fig. 2i).

OPEN IN VIEWER

FIG. 2.

Preparation and encapsulation of composite spheroids within the GelMA hydrogel. (a) SEM image of a segmented PLLA fibril and (b) the average length of the fibers (n = 25). (c) Phase-contrast images of small, medium, and large spheroids after culturing for 24 h (scale bar = 200 μm), and (d) the average area of each spheroid (n = 6). The statistical analysis was performed by one-way ANOVA. (e) 3D confocal image of a spheroid after culturing for 24 h (blue: DAPI, red: F-actin, green: fibers, and scale bar = 25 μm). (f) Schematic illustration of the hydrogel encapsulating spheroids. (g) H&E-stained image of a cross-sectioned spheroid within the hydrogel (red arrows: binding between the cells and the hydrogel matrix, scale bar = 50 μm). (h) Calcein AM-stained images of vertically sectioned hydrogels encapsulating small, medium, and large spheroids (scale bar = 200 μm), and (i) the distribution of spheroids within the hydrogels (n = 3). p-Values are reported as p < 0.001. 3D, three-dimensional; ANOVA, analysis of variance; DAPI, diamidino-2-phenylindole; H&E, Hematoxylin and Eosin; PLLA, poly(ι-lactic acid); SEM, scanning electron microscopy. Color images are available online.

Macroscopic images of the spheroid-laden hydrogels were shown in Supplementary Figure S1. Thus, the small spheroids (6829 ± 815 μm² of area; Fig. 2d) were used for further investigations.

Characterization of in vitro encapsulated spheroids within the hydrogel depending on UV exposure time

Optical images of each hydrogel show the same cylindrical shape and transparency regardless of UV exposure time (Supplementary Fig. S2a). Storage modulus of the spheroid-laden hydrogels under 10 s of UV exposure was 0.5 ± 0.1 kPa, which was significantly lower than that under 30 and 60 s of UV exposure (1.3 ± 0.1 and 1.5 ± 0.1 kPa, respectively), and no change in modulus was found during 14 days of incubation (Fig. 3a). Swelling ratio of the 10 s UV-exposed hydrogels was greater than that with 30 or 60 s of exposure, and the ratios did not change significantly for 14 days (Fig. 3b). Internal pore structure of the hydrogel with 10 s of UV exposure was larger than with exposure for 30 or 60 s, and cell sprouting from an encapsulated spheroid was observed only in the hydrogel with 10 s of UV exposure (Supplementary Fig. S2b, c). Intensive cell sprouting from the spheroids was found only within the hydrogel with the lowest mechanical strength (10 s of UV exposure) (Fig. 3c). Sprouting cells in the 10 s group filled most of hydrogel area and elongated through hydrogel matrix, which was not found in the other groups (Fig. 3c).

OPEN IN VIEWER

FIG. 3.

In vitro behavior of cells encapsulated within hydrogels depending on the UV exposure time. (a) Storage moduli (n = 6) and (b) swelling ratio (n = 4) of spheroid-encapsulating hydrogels treated with 10, 30, or 60 s of UV exposure after culturing for 1 or 14 days. (c) Phase-contrast images of each hydrogel after 14 days (scale bar = 250 μm), and H&E and calcein AM staining images of the encapsulated spheroids (scale bars = 50 and 100 μm, respectively). (d) Sprouting length (n = 8) and (e) ECM remodeling gene expression (n = 3) from cells in the spheroids encapsulated within each hydrogel. (f) The DNA content from each group at 14 days compared with that at day 1 (n = 4). (g) TUNEL assay and CellROX staining of spheroids encapsulated within each hydrogel (blue: DAPI, green: TUNEL- or CellROX-positive nuclei, and scale bar = 100 μm). (h) TUNEL-positive nuclei per DAPI signal from each group (n = 5). (i) CellROX-positive fluorescent intensities from hydrogels treated with 30 and 60 s of UV exposure relative to that treated with 10 s (n = 10). The statistical analyses were performed by two-way ANOVA for the tests with two or more variables per group and one-way ANOVA for the tests with single variables from each group. p-Values are given as p < 0.05, p < 0.01, and p < 0.001. ECM, extracellular matrix; TUNEL, terminal deoxynucleotidyl transferase dUTP nick end labeling. Color images are available online.

Sprouting lengths within the hydrogels were 256.3 ± 79.3, 63.0 ± 25.3, and 2.4 ± 2.0 μm with 10, 30, and 60 s of UV exposure, respectively (Fig. 3d). RT-qPCR investigation demonstrated that the cells in hydrogel with 10 s of UV exposure showed significantly better ECM remodeling gene expression than those in the other groups (Fig. 3e). For example, expression of ROCK1 and MMP9 from the cells within hydrogel with 10 s of UV exposure was 8.2 ± 1.1 and 13.0 ± 1.7 times greater, respectively, than from those within the 60 s group (Fig. 3e). Furthermore, the cells from spheroids within the lower mechanical-strength hydrogel were more proliferative than those within hydrogels from the other groups (Fig. 3f).

Similarly, the fewer signals from TUNEL-positive nuclei and CellROX were found in hydrogel with 10 s of UV exposure than in the other groups (Fig. 3g). Only 11.2% ± 5.1% of TUNEL-positive nuclei per spheroid and 1.5 times less CellROX-positive fluorescence were detected in the hydrogel with 10 s of UV exposure (Fig. 3h, i). Taken together, the results indicate that the lower mechanical strength in hydrogel enhanced cell sprouting from spheroids, ECM remodeling, and proliferation of hADSCs resistant to apoptosis.

Therefore, the hydrogel with 0.5 kPa of storage modulus was used for further investigations.

Immobilization of growth factors on the fibers

FE-SEM images of the fibers show a roughened surface after PD coating and an even distribution of microscopic granules after BMP-2 or TGF-β1 immobilization (Fig. 4a). The use of FITC-conjugated BMP-2- or TGF-β1-immobilized fibers revealed that the growth factors were homogeneously bound on fiber surfaces, and high-magnification images indicate that the factors were homogeneously distributed on each strand (Fig. 4b). Coating efficiencies of BMP-2 and TGF-β1 on the fibers were 88.5% ± 8.0% and 99.8% ± 0.0%, respectively (Fig. 4c). X-ray photoelectron spectroscopy analysis demonstrated that carbon (C1s, 288 eV) and oxygen (O1s, 533 eV) peaks were common to all fibers, whereas nitrogen (N1s, 399 eV) peak was generated after the PD coating, and sulfur (S2p, 163 eV) peak was specifically found on the BMP-2- or TGF-β1-immobilized fibers (Fig. 4d). High resolution of N1s and S2p peaks for the fibers indicate that the N1s peaks were clearly present in PD, BMP-2, and TGF-β1 groups, and the S2p peaks were found only in BMP-2 and TGF-β1 groups (Fig. 4e).

OPEN IN VIEWER

FIG. 4.

Characterization of the BMP-2- and TGF-β1-immobilized fibers. (a) Surface morphology of the segmented fiber (Fiber), PD-coated fiber (PD), BMP-2-immobilized fiber (BMP-2), and TGF-β1-immobilized fiber (TGF-β1) from FE-SEM (scale bar = 500 nm). (b) Merged images of the fibers immobilized with FITC-labeled growth factors captured by fluorescent and phase-contrast microscopes (scale bar = 50 μm, and scale bar of the high-magnification images [black boxes] = 25 μm). (c) The coating efficiency of BMP-2 and TGF-β1 on the fibers (n = 4). XPS analyses of the fibers: (d) Overall chemical composition, (e) high-resolution analyses for N1s and S2p peaks, and (f) C1s-related spectra from each fiber. BMP-2, bone morphogenetic protein-2; FE-SEM, field emission-scanning electron microscopy; FITC, fluorescein isothiocyanate; PD, polydopamine; TGF-β1, transforming growth factor beta-1; XPS, X-ray photoelectron spectroscopy. Color images are available online.

Similarly, high-resolution carbon spectrum indicates that C-C (285.0 eV), C-O (286.4 eV), and C = O (289.0 eV) were found in all groups, whereas C-N (286.0 eV) was found only after the PD coating, and C-S (284.0 eV) was found only after the immobilization of growth factors (Fig. 4f). The results indicate that growth factors were evenly and stably immobilized on the PD-coated fibers with reasonable efficiency.

In vitro osteogenic and chondrogenic differentiation of hADSCs from spheroids within the hydrogel

A schematic illustration shows different delivery modalities of BMP-2 to the spheroid-laden hydrogels (GFS: without growth factors, GBS-1: 50 ng of BMP-2 diluted in culturing medium, GBS-2: 50 ng of BMP-2 physically blended in hydrogel, and GBS-3: 50 ng of BMP-2-immobilized fibers in spheroids) (Fig. 5a). Expression of early osteogenic markers, RUNX2 and OSX, by the encapsulated hADSCs was significantly enhanced within GBS-2 and GBS-3 groups, compared with GFS and GBS-1 groups, after 7 days of culturing, and only GBS-3 group showed significantly enhanced osteogenic gene expression, compared with the other groups, after 14 and 21 days (Fig. 5b). For example, the cells within GBS-3 hydrogels had 7.4 ± 0.9, 7.8 ± 0.7, 10.0 ± 0.8, and 5.8 ± 1.0 times more RUNX2, OSX, OCN, and OPN expression, respectively, than those within GFS hydrogels after 14 days (Fig. 5b). Similar trends were found in IHC-stained images for quantified ratio of RUNX2- and OPN-positive nuclei per spheroid in GBS-3 compared with the other groups (Fig. 5c, d).

OPEN IN VIEWER

FIG. 5.

In vitro osteogenic differentiation of hADSCs from spheroids within hydrogels. (a) Schematic illustrations of the spheroid-laden hydrogels depending on the delivery modality for the BMP-2. (b) Osteogenic gene expression of cells in GFS, GBS-1, GBS-2, and GBS-3 hydrogels after culturing for 7, 14, and 21 days (n = 4). (c) RUNX2 IHC staining of horizontally cross-sectioned hydrogel groups at 14 days (scale bar = 100 μm) and the RUNX2-positive nuclei from each group (n = 5). (d) OPN IHC staining of horizontally cross-sectioned hydrogel groups at 14 days (scale bar = 100 μm) and the OPN-positive nuclei from each group (n = 5). (e) Alizarin Red S staining of each hydrogel group (scale bar = 4 mm), high-magnification images of the encapsulated spheroids (scale bar = 50 μm), and (f) the amount of deposited calcium from each group (n = 7). (g) Phase-contrast and RUNX2 IHC images of cells sprouting from composite spheroids seeded in a well plate (black dotted circle: area of seeded spheroids, white lines: compartments A, B, and C, and scale bar = 250 μm). (h) RUNX2-positive nuclei of cells sprouting from the spheroid in each compartment (n = 3). The statistical analyses were performed by two-way ANOVA for the tests with two or more variables per group, and one-way ANOVA for the tests with single variables from each group. p-Values are given as p < 0.05, p < 0.01, and p < 0.001. hADSCs, human adipose-derived stem cells; IHC, immunohistochemistry; OPN, osteopontin; RUNX2, runt-related transcription factor 2. Color images are available online.

After 7 days of culture, the IHC staining results demonstrated the significantly greater number of RUNX2-positive and OPN-positive cells within GBS-3 than the other groups and GFS group, respectively (Supplementary Fig. S3a, b).

A greater intensity of reddish color from Alizarin Red S staining was found throughout GBS-3 hydrogel, compared with the other gels, and deposition of calcium granules (dark red) was also found in cross-sectioned GBS-3 hydrogel (Fig. 5e). The greatest amount of mineral calcium (14.3 ± 1.4 μg) was deposited in GBS-3 hydrogel (Fig. 5f). The cells sprouting from a spheroid incorporating BMP-2-immobilized fibers were positively stained for RUNX2, and the ratio of RUNX2-positive nuclei was statistically same in A, B, and C compartments (Fig. 5g, h).

A schematic illustration shows different delivery modalities of TGF-β1 to the spheroid-laden hydrogels (GFS: without growth factors, GTS-1: 50 ng of TGF-β1 diluted in culturing medium, GTS-2: 50 ng of TGF-β1 physically blended in hydrogel, and GTS-3: 50 ng of TGF-β1-immobilized fibers in spheroids) (Fig. 6a). Similar to the results for osteogenic differentiation, analyses for chondrogenesis demonstrated that GTS-3 hydrogel produced greater efficiency of differentiation among cells than the other groups (Fig. 6b–f). In RT-qPCR investigations, GTS-3 hydrogels showed significantly enhanced SOX9 compared with the other groups after 7 days, and expression of all tested genes was significantly higher in the cells within GTS-3 hydrogels than in those within the other hydrogels after 21 days (Fig. 6b). More SOX9 and aggrecan-positive nuclei were detected in the IHC-stained images from GTS-3 hydrogels than in those from the other groups (Fig. 6c, d).

OPEN IN VIEWER

FIG. 6.

In vitro chondrogenic differentiation of hADSCs from spheroids within the hydrogels. (a) Schematic illustrations of the spheroid-laden hydrogels depending on the delivery modality for the TGF-β1. (b) Chondrogenic gene expression from cells in GFS, GTS-1, GTS-2, and GTS-3 hydrogels after culturing for 7, 14, and 21 days (n = 4). (c) SOX9 IHC staining of horizontally cross-sectioned hydrogel groups at 14 days (scale bar = 100 μm) and the SOX9-positive nuclei from each group (n = 5). (d) Aggrecan IHC staining of horizontally cross-sectioned hydrogel groups at 14 days (scale bar = 100 μm) and the aggrecan-positive nuclei from each group (n = 5). (e) Alcian Blue staining of each hydrogel group (scale bar = 4 mm), high-magnification images of the encapsulated spheroids (scale bar = 50 μm), and (f) the amount of deposited GAGs from each group (n = 6). (g) Phase-contrast and SOX9 IHC images of cells sprouting from composite spheroids seeded in a well plate (black dotted circle: area of seeded spheroid, white lines: compartments A, B, and C, and scale bar = 250 μm). (h) SOX9-positive nuclei of cells sprouting from spheroids in each compartment (n = 3). The statistical analyses were performed by two-way ANOVA for the tests with two or more variables per group, and one-way ANOVA for the tests with single variables from each group. p-Values are given as p < 0.05, p < 0.01, and p < 0.001. SOX9, SRY-box transcription factor 9. Color images are available online.

Alcian Blue staining showed a richer blue color for GTS-3 than for the other hydrogels, and cross-sectioned GTS-3 revealed GAG formation around the encapsulated spheroid (bright blue) (Fig. 6e); 334.7 ± 117.6 μg of GAG was measured in GTS-3, compared with 96.0 ± 52.7, 71.8 ± 15.5, and 102.4 ± 20.3 μg in GFS, GTS-1, and GTS-2 groups, respectively (Fig. 6f). IHC staining for hADSCs sprouting from a spheroid incorporating TGF-β1-immobilized fibers showed that the cells were evenly positive for SOX9 (Fig. 6g, h). Collectively, the results suggest that BMP-2 or TGF-β1 delivery within each composite spheroid by using the PD-coated segmented fibrils effectively and specifically directed osteogenic or chondrogenic differentiation, respectively, of hADSCs in hydrogel, and that differentiation was enhanced compared with the control methods, in which growth factors were added to the medium or physically blended into hydrogel.

Preparation and characterization of 3D bilayer hydrogel

A schematic illustration shows sequential procedures used to prepare the bilayer hydrogels (Fig. 7a). Crosslinking of hydrogels under 5 s of UV exposure did not produce completely discrete bilayers, but the other bilayer hydrogels were hierarchically arranged with compatible interface integration (Fig. 7b). Storage moduli of the bilayer hydrogels encapsulating spheroids increased from 0.3 ± 0.0 to 0.7 ± 0.0 kPa with 5 and 10 s of UV exposure on each layer, respectively (Fig. 7c). Storage moduli of top and bottom layers (7 or 10 s of UV exposure) were the same as overall storage modulus for each hydrogel (Supplementary Fig. S4a). H&E-stained images reveal greater cell sprouting from the encapsulated spheroids in each layer with 7 s of UV exposure than with 10 s (Fig. 7d). The spheroids were homogeneously encapsulated within both bilayer hydrogels (Supplementary Fig. S4b), but after 14 days of culture, sprouting of living cells was observed only in 7 s group (Fig. 7e). Collectively, our results suggest that sprouting from spheroids was achieved within the bilayer hydrogel with a modulus of 0.5 kPa.

OPEN IN VIEWER

FIG. 7.

Preparation and characterization of 3D bilayer hydrogels. (a) Schematic illustration of the process for synthesizing a bilayer hydrogel. (b) Optical images from the side and top views of bilayer hydrogels treated with 5, 7, and 10 s of UV exposure per layer (scale bar = 4 mm). (c) Storage modulus of each bilayer hydrogel (n = 6). (d) H&E and (e) Calcein AM staining of vertically cross-sectioned hydrogels treated with 7 and 10 s of UV exposure per layer after culturing for 14 days (dotted line: interfacial border between layers, scale bar = 200 μm). (f) Optical images of the pull-off test for the bilayer hydrogel with 7 s of UV exposure and the tensile stress–strain graph after culturing the hydrogel for 1 or 14 days. (g) Tensile strength from the pull-off test (n = 5). (h) Optical images of the lap-shear test for the bilayer hydrogel, and the shear stress–strain graph after culturing the hydrogel for 1 or 14 days. (i) Shear strength from the lap-shear test (n = 5). The statistical analysis for storage modulus was performed by one-way ANOVA, and the other analyses used Student's t-test. p-Values are given as p < 0.05, p < 0.01, and p < 0.001. Color images are available online.

We then examined interface integration using pull-off and lap-shear tests, which measure resistance to uniaxial and shear stress, respectively (Supplementary Fig. S4c). Intermediate area of the bilayer was separated by uniaxial force in pull-off test for the bilayer hydrogel with 7 s of UV exposure at each layer, and tensile strength increased from 4.8 ± 0.4 (1 day) to 7.0 ± 1.3 MPa after culturing for 14 days (Fig. 7f, g). Similar results were obtained from lap-shear test (Fig. 7h). Calculated shear strength of the bilayer hydrogel increased from 3.9 ± 0.4 (1 day) to 6.3 ± 2.0 MPa (14 days) (Fig. 7i). The results indicate that the bilayer hydrogel was well crosslinked with strong interfacial adhesion during preparation, which was improved during 14 days of in vitro culture, possibly due to intensive cell sprouting (Fig. 7f–i).

In vitro differentiation of hADSCs from spheroids within the bilayer hydrogel

We next confirmed zonal osteogenic or chondrogenic differentiation of hADSCs induced by instructive signals in each layer by culturing entire hydrogel in a general medium. Macroscopic image of bilayer hydrogel after Alizarin Red S staining was strongly positive for GBS-3 layer, and the spheroids within layer showed granulated calcium minerals whereas those within GTS-3 layer were weakly stained (Fig. 8a). In contrast, Alcian Blue staining of showed a greater intensity of blue color for GTS-3 layer, and the spheroids within layer clearly revealed GAGs while those within GBS-3 were barely stained (Fig. 8b).

OPEN IN VIEWER

FIG. 8.

In vitro differentiation of hADSCs from spheroids within the bilayer hydrogel. (a) Alizarin Red S and (b) Alcian Blue staining of bilayer hydrogels composed of GTS-3 and GBS-3 (white dotted line: border between GTS-3 and GBS-3, and scale bar = 1 mm or 300 μm for high-magnified images), and each staining for cross-sectioned specimens of the paraffin embedded bilayer hydrogel (dotted line: border between GTS-3 and GBS-3, and scale bar = 100 μm). (c) Osteogenic and (d) chondrogenic gene expression from cells within the bilayer hydrogel composed of GTS-3 and GBS-3 (n = 3). (e) Col1a IHC staining of vertically sectioned bilayer hydrogels (scale bar = 100 μm) and high-magnification images of the encapsulated spheroids (white boxes; scale bar = 50 μm). (f) The Col1a intensities from each layer (GBS-2, GTS-3, and GBS-3) related to that of GTS-2 (n = 4) and (g) the amount of BSP secreted from the bilayer hydrogels composed of GTS-3 and GBS-3 or GTS-2 and GBS-2 for 21 days (n = 3). (h) Col2a IHC staining of vertically sectioned bilayer hydrogels (scale bar = 100 μm) and high-magnification images of the encapsulated spheroids (white boxes; scale bar = 50 μm). (i) Col2a intensities from each layer (GTS-2, GBS-3, and GTS-3) related to that of GBS-2 (n = 4) and (j) the amount of TGF-β3 secreted from the bilayer hydrogels for 21 days (n = 3). The statistical analyses were performed by two-way ANOVA for qRT-PCR, and one-way ANOVA for the quantification of fluorescent intensities. p-Values are given as p < 0.05, p < 0.01, and p < 0.001. BSP, bone sialoprotein; Col1a, collagen type 1a; Col2a, collagen type 2a; qRT-PCR, quantitative real-time polymerase chain reaction; TGF-β3, transforming growth factor beta-3. Color images are available online.

The spheroids within each GTS-2 and GBS-2 layer similarly demonstrated weak intensity of colors and insignificant color differences from each layer (Supplementary Fig. S5a). The expression of osteogenic markers increased significantly in the cells within GBS-3 layer, whereas expression of chondrogenic markers was significantly upregulated within the GTS-3 layer (Fig. 8c, d). However, the expression of osteogenic and chondrogenic genes by the cells within GTS-2 and GBS-2 layers did not differ significantly, except for OSX and OCN (Supplementary Fig. S5b).

In addition, IHC staining of collagens and an ELISA for the bilayer hydrogel composed of GTS-3 and GBS-3 clearly showed distinct characteristics of each layer (Fig. 8e–j), whereas the characteristics were rarely detected in the bilayer hydrogel composed of GTS-2 and GBS-2 (Supplementary Fig. S5c). Most cells in GBS-3 were positive for Col1a, and high-resolution image clearly shows elongated Col1a, whereas the signals in GTS-3 layer were weak (Fig. 8e). Quantified fluorescent intensity was greatest in the GBS-3 layer (Fig. 8f).

ELISA demonstrated that the amount of BSP secreted from the bilayer hydrogel composed of GTS-3 and GBS-3 for 21 days was 27.5 ± 0.7 ng, which was significantly greater than that secreted from the GTS-2 and GBS-2 bilayer hydrogel (18.0 ± 0.2 ng) (Fig. 8g). Similarly, Col2a signals were found only from cells and spheroids in the GTS-3 layer and were barely detected in the GBS-3 layer (Fig. 8h). The strongest fluorescent intensity was found in the cells in GTS-3 layer (Fig. 8i), and the amount of TGF-β3 secreted from the bilayer hydrogel composed of GTS-3 and GBS-3 for 21 days (40.1 ± 2.1 ng) was greater than that secreted from the GTS-2 and GBS-2 bilayer hydrogel (17.1 ± 1.5 ng) (Fig. 8j).

Taken together, our results suggest that spatially confined delivery of growth factors within the spheroids encapsulated in each layer (GTS-3 and GBS-3) induced zone-dependent differentiation (chondrogenic or osteogenic) of hADSCs without dedifferentiation within bilayer hydrogel structure (Fig. 8). On the other hand, physically blended growth factors in the GTS-2 and GBS-2 bilayer hydrogel showed weak induction of encapsulated cells (Fig. 8 and Supplementary Fig. S5).