Aligned Gelatin Microribbon Scaffolds with Hydroxyapatite Gradient for Engineering the Bone–Tendon Interface

Abstract

Injuries of the bone-to-tendon interface, such as rotator cuff and anterior cruciate ligament tears, are prevalent musculoskeletal injuries, yet effective methods for repair remain elusive. Tissue engineering approaches that use cells and biomaterials offer a promising potential solution for engineering the bone–tendon interface, but previous strategies require seeding multiple cell types and use of multiphasic scaffolds to achieve zonal-specific tissue phenotype. Furthermore, mimicking the aligned tissue morphology present in native bone–tendon interface in three-dimensional (3D) remains challenging. To facilitate clinical translation, engineering bone–tendon interface using a single cell source and one continuous scaffold with alignment cues would be more attractive but has not been achieved before. To address these unmet needs, in this study, we develop an aligned gelatin microribbon (μRB) hydrogel scaffold with hydroxyapatite nanoparticle (HA-np) gradient for guiding zonal-specific differentiation of human mesenchymal stem cell (hMSC) to mimic the bone–tendon interface. We demonstrate that aligned μRBs led to cell alignment in 3D, and HA gradient induced zonal-specific differentiation of mesenchymal stem cells that resemble the transition at the bone–tendon interface. Short chondrogenic priming before exposure to osteogenic factors further enhanced the mimicry of bone–cartilage–tendon transition with significantly improved tensile moduli of the resulting tissues. In summary, aligned gelatin μRBs with HA gradient coupled with optimized soluble factors may offer a promising strategy for engineering bone–tendon interface using a single cell source.

Impact statement

Our 3D macroporous microribbon hydrogel platform with alignment cues zonally integrated with hydroxyapatite nanoparticles enables differentiation across the bone-tendon interface within a continuous scaffold. While most interfacial scaffolds heretofore rely on composites and multilayer approaches, we present a continuous scaffold utilizing a single cell source. The synergy of niche cues with human mesenchymal stem cell (hMSC) culture leads to an over 45-fold enhancement in tensile modulus in culture. We further demonstrate that priming hMSCs towards the chondrogenic lineage can enhance the differential osteogenesis. Relying on a single cell source could enhance zone integration and scaffold integrity, along with practical benefits.

Introduction

Injuries of connective tissues at the interface between soft tissue and bone are prevalent and debilitating.¹ Rotator cuff injury of the shoulder represents one of the most prevalent musculoskeletal injuries, affecting ∼21% of the general population² with an estimated 70–94% failure rate after the repair.³ Tears of the anterior cruciate ligament also have a high reinjury rate. Patients typically experience knee pain 1 year after the surgery.⁴ The bone–tendon interface is characterized by a transition of tissues from tendon, fibrocartilage, cartilage, to bone.

To restore the structures and functions of the bone–tendon interface, it is critical to regenerate the interface with biomimetic zonal organization.⁵ Tissue interfaces serve as transitions between tissues with distinct physical and biological properties, and the interfaces themselves have unique compositions as well.⁶ Gradients minimize stress concentrations and facilitate load transfer.⁵

Tissue engineering approaches that use cells and biomaterials offer a promising potential solution for repair of tissue interfaces. Previous attempts to engineer the bone–tendon interface have relied on use of multiple materials and cell types to mimic the unique properties of each zone. For example, one study used a triphasic scaffold composed of sintered polyglactin knitted mesh sheets seeded with fibroblasts to mimic the tendon, sintered poly(lactic-co-glycolic acid) (PLGA) copolymer microspheres to mimic the fibrocartilage, and sintered PLGA and bioactive glass microspheres seeded with osteoblasts to mimic the bone.⁷ While this approach shows some osteogenesis with expression of an early bone marker, alkaline phosphatase (ALP), it does not mimic cell alignment of tendon tissue and requires the use of multiple cell types.

Another study used a biphasic macroporous scaffold based on silk and fibrin and tested the effect of pore alignment on mesenchymal stem cell (MSC) differentiation toward tendon and cartilage lineages.⁸ However, no bone markers were assessed and cartilage production is low. Another group used methacrylated chondroitin sulfate hydrogel containing magnetic nanoparticles to modulate release of platelet lysate-derived growth factors.⁹ However, this approach used two different cell source and required pre-differentiation of adipose-derived stem cells before encapsulation and also lacks alignment cues.

To better mimic the alignment cues in the tendon zone, one study used electrospun PLGA fibers seeded with fibroblasts,¹⁰ but this approach does not support zonal organization of cell fates and the scaffold lacks macroporosity. To better mimic the transition at the bone–tendon interface, a gradient of calcium phosphate has been coated on electrospun nanofibers.¹¹ However, this approach lacks alignment cues and macroporosity.

To facilitate clinical translation, an ideal scaffold for engineering the bone–tendon interface should be macroporous, contains alignment cues, and supports zone-specific differentiation using one single cell type. Recently, we have reported a technology to fabricate gelatin-based microribbon (μRB) hydrogels with aligned topography and macroporosity that support smooth muscle cell alignment and proliferation in three-dimensional (3D) and guided newly deposited muscle matrix to follow the alignment.¹² The potential of such aligned μRBs for engineering the bone–tendon interface has never been interrogated, yielding a unique opportunity to remove critical bottlenecks in bone–tendon interface tissue engineering.

Unlike other tissues in human body, bone is characterized by high mineral content, with approximately two thirds of bone matrix composed of inorganic minerals.¹³ Hydroxyapatite (HA) is a key component of bone mineral and has been widely used in scaffold design to induce bone differentiation.^14,15 The goal of this study was to develop aligned gelatin μRBs with gradient cues of hydroxyapatite nanoparticles (HA-np) as 3D scaffolds for guiding zonal-specific differentiation of human mesenchymal stem cells (hMSCs) to mimic the bone–tendon interface. By optimizing soluble factors, we further sought to enhance the bone–tendon tissue transition with tissue-mimicking compositions that specifically mimic bone, fibrocartilage, and tendon. Outcomes were characterized using histology, immunostaining, and mechanical testing.

Materials and Methods

Fabrication of aligned gelatin μRB scaffolds

Aligned gelatin-based μRBs were fabricated as previously reported.¹² In brief, type-A gelatin (GelA; Sigma–Aldrich, St. Louis, MO) was dissolved in dimethyl sulfoxide (DMSO; Sigma–Aldrich) at 18% (w/w) concentration, stirred for 3 h at 350 rpm, and then stirred overnight at 125 rpm overnight on hot plate (60°C) to form a viscous GelA in DMSO solution. The GelA solution was transferred to a 30 mL syringe and allowed to age another day and night. Then, GelA solution was wet spun into an ethanol bath (1.8 L) at constant flow rate (3 mL/h) controlled by syringe pump (78–8200; kd Scientific, Holliston, MA) and spun into fibers collected onto a U-shaped collector on a rotating magnet at the bottom of the ethanol bath (750 rpm). The GelA fibers collected precipitated and stayed aligned on the U-shaped collector.

Aligned μRBs were transferred to acetone for 1 h to induce collapse into the μRB shape. To functionalize 15% of primary amine groups of gelatin into photocrosslinkable methacrylate groups, μRBs were reacted overnight with a 0.1 mM solution of methacrylic acid N-hydroxysuccinimide ester (Sigma–Aldrich) in methanol. To retain μRB morphology, methacrylate-coated μRBs were transferred into a 0.1 mM solution of glutaric dialdehyde (Sigma–Aldrich) in methanol for 24 h for internal cross-linking. The unreacted aldehyde in the cross-linked μRBs was quenched using 1 mM solution of L-lysine hydrochloride (Sigma–Aldrich) in phosphate-buffered saline (PBS) at 37°C overnight. Aligned μRBs were then washed three time in ultra-sterile water (Invitrogen, Grand Island, NY) at 37°C for 1 h each before lyophilizing and storing at −20°C.

Cell culture

hMSCs (Lonza) were cultured in growth medium (GM) that composed of Dulbecco's modified Eagle medium (DMEM; Gibco), fetal bovine serum (FBS, 10% v/v; Gibco), penicillin–streptomycin (1% v/v; ThermoFisher Scientific), and recombinant human fibroblast growth factor-basic (10 ng/mL; PeproTech).

Osteogenic medium (OM) was composed of DMEM, FBS (10%), penicillin–streptomycin (1% v/v; ThermoFisher Scientific), dexamethasone (100 nM), ascorbic-2-phosphate (50 μg/mL; Sigma), and beta-glycerol phosphate (10 mM; Sigma). Chondrogenic medium (CM) was composed of high-glucose DMEM (Gibco), 100 nM dexamethasone (Sigma), 50 mg/mL ascorbic-2-phosphate (Sigma), 40 mg/mL proline (Sigma), 5 mg/mL insulin–transferrin–selenium premix (BD Biosciences), 0.001 mg/mL sodium pyruvate, 10 ng/mL recombinant human transforming growth factor beta 3 (PeproTech), 100 units/mL penicillin, and 100 mg/mL streptomycin (Invitrogen).

Cell encapsulation and HA-np incorporation

Freeze-dried μRBs were rehydrated with PBS containing 0.05% lithium phenyl-2,4,6-trimethylbenzoylphosphinate (LAP) photoinitiator at 5% (w/v) of μRB dry weight. Of the PBS with LAP for rehydrating, 70% of the volume contained HA-np of <200 nm (Sigma) at the appropriate concentration (2%, 1%, 0.5%, or 0%). For gradient incorporation of HA-np, this volume was incorporated one section at a time, distributing the HA-np gradually via pipette throughout the entirety of the section and depth of the scaffold, working to maximize the homogeneity of HA-np distribution throughout each individual section. The remaining 30% of the volume of PBS with 0.05% LAP photoinitiator for rehydrating was used to incorporate cells.

Passage 6 hMSCs were encapsulated at 200,000 cells per mg dry weight of the μRB scaffold, pipetting throughout the scaffold and in between μRB fibers to maximize homogeneity of cell distribution before cross-linking. For acellular samples, this 30% of rehydration volume contained PBS with 0.05% LAP photoinitiator alone. Scaffolds were then exposed to ultraviolet light (365 nm, 4 mW/cm², 5 min) for cross-linking and then incubated at 37°C for 30 min to allow for cell attachment before submerging scaffolds in the GM. Scaffolds were maintained on a wire spacer during encapsulation and throughout the culture to maintain fiber alignment.

Cell viability

LIVE/DEAD^® Viability/Cytotoxicity Kit, for mammalian cells (L-3224; ThermoFisher Scientific), was used to evaluate cell viability. Briefly, 2 μM calcein acetoxymethyl ester (for live cells) and 4 μM ethidium homodimer-1 (for dead cells) solutions were prepared. Aligned μRB scaffolds were incubated with the dye solution for 30 min on a rocker at 37°C. Live/dead cells were imaged using fluorescent microscope (Zeiss) and quantified using the intensity of the fluorescent images for n ≥ 10 fields of view.

Histological analyses

To visualize the extracellular matrix (ECM) distribution inside 3D hydrogels and bone formation of hMSCs, immunofluorescence staining was performed. Scaffolds were fixed at the appropriate time point (day 1, 7, 14, or 28) in 4% paraformaldehyde for 2 min at room temperature. For immunostaining, sections were permeabilized with 0.1% Triton X-100/PBS for 30 min and blocked with blocking buffer (3% bovine serum albumin/2% goat serum/PBS) for 30 min at room temperature. Cells were incubated with primary antibodies (rabbit anti-collagen-I, rabbit anti-osteopontin, mouse anti-osteocalcin, rabbit anti-collagen-II, rabbit anti-collagen-X, rabbit anti-scleraxis; all antibodies from Abcam) overnight at 4°C on a shaker.

Cells were washed in PBS three times for 5 min each at room temperature and then incubated with secondary antibodies and dyes (1:300, Alexa 488 goat-anti-mouse [Invitrogen]; 1:300, rhodamine goat-anti-rabbit [Millipore]; 2 μg/mL Hoechst dye [Cell Signaling Technology]) for 1 h at room temperature on a shaker. For cytoskeletal staining, cells were incubated with rhodamine–phalloidin (P1951, 1:100; Sigma) for 1 h at room temperature. Cells were again washed three times with PBS for 5 min each. Sections were imaged on a fluorescent microscope (Zeiss), and whole scaffolds were imaged on a confocal microscope (Leica).

ALP activity was assessed with Fast Blue RR Salt (Sigma). Alizarin red staining was performed to assess mineralization by incubating sections in alizarin red dye for 2 min and washing in water for 1 min and then PBS. Images were colored and overlaid using Fiji software. Images of Alizarin Red S (ARS) and ALP were quantified using color deconvolution in ImageJ, calculating optical density (OD) = log(maximum intensity/mean intensity) of the area of the scaffolds for n ≥ 2 scaffolds and n ≥ 12 fields of view. Fluorescent images were quantified using mean intensity and normalized to nuclear intensity where indicated.

Mechanical Testing

Tensile testing of aligned μRB hydrogels was performed using an Instron 5944 materials testing system (Instron Corporation, Norwood, MA) fitted with a 100 N load cell. At day 28 of culture, aligned μRBs were taken off of the spacer, and then, the ends of the μRB scaffolds were sandwiched between metal grips, using sandpaper to ensure the scaffolds did not slip out of the grips. While under a preload of 5 mN, scaffold cross-sectional area and initial length were recorded with digital calipers and the test machine's position readout, respectively. Scaffolds were pulled at a rate of 1% strain/second to failure. Load and displacement data were recorded at 100 Hz. The tensile modulus was determined from a linear curve fit of the stress versus strain curve from 40% to 60% strain.

Scanning electron microscopy imaging

Variable pressure scanning electron microscopy (VP-SEM, S-3400N; Hitachi) was used to observe HA-np incorporation onto μRB scaffolds. Aligned μRBs were loaded into the VP-SEM chamber where they were gradually cooled from ambient temperature to −20°C, while the chamber pressure was reduced from atmospheric pressure to 50 Pa following a liquid water P/T curve. SEM images were acquired under 15 kV electron beam at ∼7 mm working distance.

Soluble factor optimization

To support maximal mineralization in the “bone” region (containing 2% HA-np) while minimizing mineralization in the “tendon” region (0% HA-np), we performed a soluble factor optimization study by culturing MSC-seeded gelatin μRB scaffolds containing either 2% or 0% HA-np with varying medium compositions and durations. Compositions of OM, CM, and GM are the same as defined earlier. For this experiment, we used cylindrical shaped μRB scaffolds (4.5 mm in diameter, 2 mm in thickness) similar to our previous reports.^16,17

hMSCs were encapsulated a final cell concentration of 10 million cells per mL in μRB scaffolds (7.5% w/v). A total of four media conditions with varying ratio and duration of OM or GM were tested. Alizarin red staining was used to evaluate mineralization at day 28. We found that 7 days in osteogenic media was sufficient to induce strong mineralization, but there was still some undesirable mineralization in the 0% HA-np scaffold. To further minimize the mineralization in the “tendon” portion, we then tested four additional media conditions with varying durations of chondrogenic priming, followed by osteogenic factors and GM, as shown in Figure 5c.

FIG. 5.

Optimizing soluble factor cues to support mineralization with HA-np for bone section, while minimizing undesirable mineralization in gelatin μRB without HA-np for tendon section. Bone formation was assessed using ARS staining at day 28. (a) Schematic of two types of scaffolds used for screening, gelatin μRB containing 2% HA-np or no HA-np. (b) Varying duration and ratio of OM versus GM identified OM for 7 days followed by GM that led to the maximal difference in bone formation in scaffolds containing 2% or 0% HA-np. (c) Quantification of ARS staining at day 28. (d) The effects of chondrogenic priming with varying durations on bone formation by hMSCs in gelatin scaffolds containing 2% or 0% HA-np (scale bar: 200 μm). (e) Quantification of ARS staining at day 21. GM, growth medium; OM, osteogenic medium. Color images are available online.

Statistical analysis

Data are presented as mean ± standard deviation. Unpaired Student's t-test or one-way analysis of variance tests were used for between-group comparisons. p-values <0.05 were considered statistically significant.

Results

Fabrication and characterization of aligned gelatin μRB scaffolds with HA-np gradient cues

To engineer the bone–tendon interface, a scaffold is needed that can provide zonal organization of cell phenotypes, cell alignment and infiltration, and mechanical integrity by seeding a single cell type. Our recently reported gelatin-based μRB hydrogels with aligned topography provide a potential tool for engineering the bone–tendon interface, given their suitability for promoting cell alignment and providing a macroporous 3D scaffold for cell infiltration.¹² Aligned μRB scaffolds were fabricated by wet spinning gelatin into an ethanol bath and collecting the μRBs on a rotating collector (Fig. 1a). Alignment of μRBs was maintained while intercross-linking with methacrylic acid N-hydroxysuccinimide ester, fixing with glutaric dialdehyde, quenching with lysine, freeze drying, and lyophilizing, as previously reported.¹²

FIG. 1.

Fabrication and characterization of aligned gelatin μRB scaffolds with HA-np gradient cues. (a) Schematic of fabrication process of aligned gelatin μRB scaffolds using wet spinning. (b) Gross morphology of aligned gelatin μRBs modified with HA-np (scale bar: 10 mm). (c) Scanning electron microscopy image showed HA-np coating on the surface of μRBs (scale bar: 200 μm). (d) Bright-field images of regions of aligned gelatin μRB scaffold with varying concentrations (2%, 1%, 0.5%, or 0%) of HA-np coating (scale bar: 10 μm). HA-np, hydroxyapatite nanoparticle; μRB, microribbon. Color images are available online.

To promote zonal bone formation, HA-np, an osteoinductive cue, was incorporated to the aligned μRB scaffold in a gradient manner. Two percent HA-np were incorporated on one end of the scaffold (the bone end), and no HA-np were incorporated on the opposite end (the tendon end) (Fig. 1b). HA-np incorporation to the μRB surfaces was verified using SEM imaging (Fig. 1c). To create a gradual transition in cell phenotypes and mechanical properties, HA-nps were incorporated at 2%, 1%, 0.5%, and 0% in four equivalent regions along the aligned μRB scaffold (Fig. 1d). HA-nps were incorporated in the appropriate regions, and the scaffolds were rehydrated before seeding hMSCs throughout the scaffold.

μRB scaffolds with HA-np gradient cues support hMSC viability and alignment

Upon encapsulation, hMSCs attached to the aligned μRB scaffolds with high cell viability in most regions containing up to 1% HA-np (Fig. 2a). Region containing 2% HA-np resulted in some initial cell death, but cells recovered and grew to confluence by day 28 throughout all regions (Fig. 2a, b). Extensive cell spreading along the aligned the μRB scaffolds is evidenced by the confocal imaging of actin filaments at day 28 (Fig. 2c) and the alignment of cells observed in nuclear aspect ratio (Fig. 2d). The aligned μRBs support cell growth and proliferation along each individual μRB, as appreciable through the increase in area of positive staining in the cell viability assay at day 28 compared with the discretized and sparser cells observed at day 1 (Fig. 2a). We hypothesize that this increased cell density and cell–cell contact facilitates new ECM production.

FIG. 2.

Aligned gelatin μRB scaffolds with HA-np gradient support MSC viability and robust alignment in 3D. (a) Cell viability of hMSCs in μRB scaffolds at day 1 and day 28 (green: live, red: dead; scale bar: 200 μm). (b) Quantification of cell viability at day 1 and day 28. (c) Confocal imaging shows extensive MSC spreading that aligns with μRBs after 28 days of culture in osteogenic medium (red: actin, blue: DAPI) (scale bars: 200 μm for 10 × images, 30 μm for 40 × images). (d) Quantification of cell alignment via nuclear aspect ratio showed that all groups support comparable level of MSC alignment. 3D, three-dimensional; DAPI, (4′,6-diamidino-2-phenylindole) blue-fluorescent DNA stain; MSC, mesenchymal stem cell. Color images are available online.

Incorporation of HA-np enhances tensile modulus of MSC-seeded aligned μRB scaffolds

We next assess the effect of HA incorporation on the tensile modulus of resulting bone–tendon tissues produced by the MSCs. MSCs were encapsulated in aligned μRB scaffolds with or without HA-np and cultured in OM for 28 days. Acellular scaffolds containing HA-np were included as controls. Scaffolds were evaluated through imaging and mechanical testing. Tensile testing showed the greatest increase in tensile modulus for the MSC + HA-np group, reaching ∼300 kPa, whereas the MSC without HA-np group only reached ∼100 kPa by day 28 (Fig. 3a).

FIG. 3.

HA-np gradient enhances tensile modulus of MSC-seeded aligned μRB scaffolds after 28 days of culture in vitro. Three groups were tested including acellular aligned μRB + HA-np, hMSC-seeded μRB scaffolds without HA, and hMSC-seeded μRB scaffolds with HA-np gradient. (a) Tensile modulus at day 0 and day 28 (n = 3 per group) (*p < 0.05, ***p < 0.001). (b) Gross morphology of aligned μRB scaffolds at day 28 (scale bars: 5 mm). Acellular group mostly degraded, whereas cellular group containing HA-np showed increased opacity and thickness. hMSC, human mesenchymal stem cell. Color images are available online.

In contrast, acellular HA-np scaffold showed decreased tensile modulus from ∼20 kPa at day 0 to ∼7 kPa at day 28, indicating that the increase in tensile modulus in the cellular/HA group was contributed by the newly deposited tissues by the cells. At the end of 4 weeks in culture, the aligned μRB HA-np scaffolds with hMSCs showed best integrity with great opacity, indicative of new ECM deposition by the cells. In contrast, acellular control was mostly degraded, as appreciable from the macroscopic images (Fig. 3b).

Aligned μRB scaffolds with HA-np gradient induced zonal-specific differentiation of MSCs in OM

To assess the effect of HA-np gradient on zonal-specific differentiation of MSCs, aligned μRB scaffolds with HA-np gradient containing MSCs were cultured in OM for 4 weeks. Staining of ALP and ARS showed an increase in bone formation over time (Fig. 4a). Aligned μRBs with HA-np and hMSCs promoted significant enhancement in ALP compared with acellular scaffolds (Fig. 4a). The staining is stronger on the 2% HA-np end of the scaffold (left) (Fig. 4b). This trend correlates with that of mineralization, which was assessed with ARS staining (Fig. 4a, b). Immunostaining of additional bone markers was also assessed including osteopontin (OP), osteocalcin (OCN), and type I collagen, with positive staining observed throughout the scaffolds (Fig. 4c).

FIG. 4.

Inducing zonal-specific hMSC differentiation to mimic bone–tendon tissue interface using aligned μRB scaffolds in osteogenic medium. (a) Whole scaffold staining of bone markers including ALP (magenta) and mineralization (red) using ARS staining. Left: hMSC-seeded scaffolds at days 1, 7, and 28. Right: acellular scaffold control and cellular control without HA-nps at day 28 (scale bars: 5 mm). (b) Quantification of ALP (left) and ARS (right) staining across conditions using OD. (c) Confocal images of immunostaining of bone markers in different zones of hMSC-seeded μRB scaffolds (2%, 1%, 0.5%, or 0% HA-np): OP (green), OCN (red), and COL-I (magenta); nuclear staining: DAPI (blue) (scale bar: 30 μm). (d) Whole scaffold staining showed a gradient of bone marker expression in response to HA-np concentration; OP (green), OCN (red), DAPI (blue) (scale bars: 5 mm). (e) Quantification of OP, OCN, and COL-I intensity normalized to nucleus staining. (f) Whole scaffold staining for cartilage matrix sGAG using safranin-O at day 28 (scale bar: 2 mm). (g) Quantification of safranin-O staining. (h) Whole scaffold staining for scleraxis, a tendon maker, at day 7 (scale bar: 5 mm). (i) Quantification of scleraxis at day 7 normalized to nucleus staining. ALP, alkaline phosphatase; ARS, Alizarin Red S; COL-I, collagen-I; OCN, osteocalcin; OD, optical density; OP, osteopontin. Color images are available online.

A view of the entire cross section of the whole scaffold showed higher expression of these bone markers in regions containing the higher incorporation of HA-nps (Fig. 4d, e). Staining of safranin-O, a marker for cartilage matrix, showed that our scaffolds support the formation of a cartilaginous region in the middle of the bone–tendon interface (Fig. 4f, g). Immunostaining for type X collagen, a marker for hypertrophic cartilage, was positive as well, with stronger expression in regions containing higher concentration of HA-np (Supplementary Fig. S1a).

To assess the tendon phenotype, we performed immunostaining of scleraxis, an early tendon marker, and found some positive staining at day 7 (Fig. 4h, i). While the aligned μRBs with HA-np thus promoted bone formation of hMSCs, zonal organization, and strong mechanical properties, we wanted to further enhance the zonal organization of bone formation, creating a scaffold that promotes bone formation on the 2% HA-np end of the scaffold alone.

Optimizing soluble cues for maximizing differential bone formation by MSCs in an HA-dependent manner

We hypothesized that by tuning the soluble cues in the culture conditions we could enhance the zonal difference in bone formation—supporting strong bone formation on the 2% HA-np “bone” end of the scaffold, while minimizing bone formation on the 0% HA-np “tendon” end. As the scaffolds had been cultured in complete OM for the full 4 weeks, we next tested if a decreased exposure to osteoinductive soluble factors could enhance this difference between scaffold regions. To screen a variety of culture conditions in a more high-throughput manner, we fabricated small μRB hydrogels with a fraction of the material used in the aligned μRB scaffolds.

We cross-linked μRBs into a hydrogel, following a previously reported protocol developed by our laboratory,¹⁶ and incorporated with hMSCs and 2% or 0% HA-nps to mimic the two ends of the aligned μRB scaffold we are optimizing (Fig. 5a).

We tested the effect of culturing the hMSCs in these hydrogels at the following media conditions: OM for 28 days, OM for 7 days followed by GM for 21 days, 1:1 OM:GM for 7 days followed by GM for 21 days, or 1:9 OM:GM for 7 days followed by GM for 21 days. We found that scaffolds were strongly positive for mineralization stain ARS with and without HA-nps at each of the conditions with at least 1:1 OM:GM for 7 days (Fig. 5b, c). The group with 1:9 OM:GM for 7 days before culture in GM led to much less mineralization in both 2% and 0% HA-np groups. As none of these groups supported the differential mineralization between 2% and 0% HA-np groups desired, we sought to further optimize soluble cues.

Priming cells with chondrogenic factors is a strategy that has been used to modulate stem cell bone formation by inducing the formation of a cartilage template before mineralized bone formation.^18–21 We hypothesize that chondrogenic priming followed by osteogenic factors can reduce undesirable mineralization on the tendon end while still supporting mineralization on the bone end.

To test this hypothesis, we culture MSC-seeded gelatin μRB scaffolds with or without HA-np with varying durations of chondrogenic priming followed by OM culture, then GM for 14 days (Fig. 5d, e). Our results showed that chondrogenic priming significantly reduced undesirable mineralization on scaffolds containing 0% HA-np. Priming in CM for 3–7 days followed by OM culture for up to 14 days was sufficient to maintain robust mineralization by MSCs in scaffolds containing 2% HA-np.

Enhanced zonal organization of hMSCs on aligned μRBs with chondrogenic priming

Using this as a pilot test to optimize media conditions on our aligned μRB scaffold, we chose to move forward with the condition: CM for 3 days followed by OM for 11 days and then GM for the rest of the culture, as it provided the maximum time in OM while still supporting the differential mineralization between conditions (Fig. 6a). Other priming conditions tested could also be suitable, but we chose the one with a shorter CM priming time due to the strong ARS staining on 2% HA-np hydrogels, good morphology, and also the fact that in the aligned μRB scaffold the two sections are co-cultured on a single scaffold and that cell-secreted factors and paracrine signaling between cells on the different regions could alter the cell phenotypes. Given the potency of the HA-np cue and this potential for cell signaling, we opted for the condition with 3 days of CM alone.

FIG. 6.

Chondrogenic priming further enhances zonal-specific differentiation of MSCs in aligned μRB scaffolds that mimic the bone–tendon interface transition. (a) Experimental design: samples were first primed in chondrogenic medium for 3 days, followed by OM for 11 days, and then, GM for additional 14 days, HA-np (2%) were incorporated on half length of the aligned scaffolds, and no HA on the remaining half of the scaffold. (b) MSC-seeded scaffold showed significant higher tensile modulus than acellular control after 28 days of culture (***p < 0.001). (c) Staining of bone markers including ALP (purple) and ARS (red) (scale bars: 0.1 mm). (d) Quantification of ALP and ARS staining. (e–k) Staining of various tissue markers to confirm bone–tendon interface formation. (e) Representative confocal images of bone marker expressions including OP and OCN (OP: green, OCN: red) (scale bars: 50 μm). (f) Whole scaffold staining at day 28: OP (green), OCN (red), COL-I (magenta), nuclei (blue) (scale bar: 1 mm). (g) Quantification of OP, OCN, and COL-I intensity in scaffolds at day 28. (h) Whole scaffold staining and (i) quantification of cartilage matrix by safranin-O staining at day 28 (scale bar: 2 mm). (j) Whole scaffold staining and (k) quantification of tendon maker scleraxis expression at day 7 (scale bar: 5 mm). Color images are available online.

For these reasons, we also decided to incorporate HA-np on only one end of the scaffold, so that half of the scaffold had 2% HA-np and the other half had none, as opposed to seeding HA-np along 75% of the scaffold in a gradient manner (2%, 1%, 0.5%, and 0%) as in the previous experiments (Figs. 3 and 4). The cellular scaffolds maintained integrity throughout the culture. Impressively, the cellular scaffolds showed an over 10-fold higher tensile modulus at day 28 compared with acellular control, reaching ∼450 kPa (Fig. 6b). Such increase is likely due to an improved gradual transition across the length of the scaffold, enabling better distribution of load. Compared with using OM only (Fig. 3a), adding chondrogenic priming further increased the tensile modulus of hMSC-seeded scaffolds with HA-np (Fig. 6b), validating the benefit of optimizing soluble cues.

Furthermore, using the optimized soluble cues removed the undesirable bone formation on the tendon section of the scaffold (Fig. 6c, d), which was an issue using OM alone (Fig. 4). The optimized soluble cues also led to enhanced expression of mature bone markers including OP and OCN on the 2% HA-np section (Fig. 6e–g). Safranin-O staining and collagen X staining showed a cartilage interface in between the bone and tendon section (Fig. 6h, i, Supplementary Fig. S1), which mimics the native bone–tendon interface. Staining of scleraxis, a tendon marker, further confirmed zonal-specific tendon differentiation in scaffold section without HA-np (Fig. 6j, k).

Discussion

In this study, we report an aligned μRB scaffolds with HA gradient for inducing zonal-specific differentiation of MSCs to mimic the bone–tendon interface. We further optimized soluble factor composition and duration to minimize undesirable mineralization in the “tendon” zone while maintaining mineralization in the “bone” zone. In native tendon, highly aligned collagen fibers provide the high tensile strength.²² Previous studies have shown that alignment cues in material substrates led to cell alignment,^10,12,23 which can be used to mimic the cell alignment and phenotype.^24,25 We showed that MSCs exhibit aligned morphology in the μRB scaffolds, whereas HA-np gradient led to zonal-specific differentiation of hMSCs to mimic the bone–cartilage–tendon transition under optimized medium conditions.

The aligned μRBs in our platform not only led to cell alignment (Fig. 2) but also guided aligned organization of the newly deposited ECM by the MSCs, contributing to the great increase in the observed tensile modulus of the resulting tissues. While acellular μRB scaffolds largely degraded after 4 weeks of culture, MSCs containing HA-np μRB scaffolds showed rapid ECM deposition, and resulting tissues demonstrated significant increase in tensile modulus to over 300 kPa (Fig. 3a, b). For the two cellular groups, adding HA-np into gelatin μRB scaffolds significantly enhanced the tensile modulus of the resulting tissues, indicating that HA-np further enhanced the ECM deposition by MSCs in 3D that contributed to the observed higher tensile strength (Fig. 3a, b).

One challenge for the bone–tendon tissue engineering is to induce mineralization in the “bone” zone while minimizing mineralization in the “tendon” zone while using the same medium.

We sought to achieve this by further optimizing the composition and duration of different medium to enhance zonal-specific differentiation of MSCs. We showed that chondrogenic priming for 3 days followed by OM for 11 days was sufficient to support robust mineralization by MSCs in μRB +2%HA-np while minimizing mineralization without HA-np cues in the μRB scaffold (Fig. 5). We also showed that 3 days chondrogenic priming followed by 7 days of osteogenic culture was not sufficient to induce mineralization even with HA-np cues. These results highlight that insoluble scaffold cues and soluble factors can synergize to induce zonal-specific stem cell differentiation to mimic tissue interface.

One limitation of previous strategies on engineering the bone–tendon interface is the need to use multiple cell types and multiphasic scaffolds with sharp transition to mimic zonal-specific tissue phenotype. Our strategy reported here overcomes this limitation by using one stem cell source, and one aligned scaffold with gradient cues to achieve zonal-specific differentiation. Compared with previous strategies, using one single cell source is more attractive for translation with reduced cost and mitigates challenges in obtaining multiple cell sources.

Here, we chose MSCs as they represent an autologous cell source that can be obtained with minimal invasiveness and can differentiate into different desirable tissue lineages found at the bone–tendon interface using soluble factors.²⁶ While the present study used MSCs, other stem cell sources may also be used including adipose-derived stromal cells or pluripotent stem cell-derived progenitors.

In summary, we here report a gelatin μRB scaffold with alignment cues and HA-np gradient that support zonal-specific differentiation of MSCs for bone–tendon interface tissue engineering. Compared with previous strategies, our method is advantageous in that it requires one single cell source and uses one continuous scaffold to mimic tissue transition. Chondrogenic priming followed by OM with optimal duration was found to synergize with the scaffold cues to further enhance the zonal-specific MSCs differentiation.

To accelerate the speed of new tissue deposition that mimics the bone–tendon interface, future studies can also combine the aligned μRB scaffolds with gradient cues with zonal-specific growth factor release. In addition, mechanical stimulation could also be applied during the culture using bioreactors. These strategies could further accelerate the speed of tissue deposition and maturation with faster restoration of the mechanical properties of engineered tissues that match the native tissues. Future studies should also further assess the efficacy for supporting the bone–tendon regeneration using a relevant animal disease model.

Authors' Contributions

Conception of the design of work (A.E.S. and F.Y.), data collection (A.E.S., S.L.J., A.B., and H.S.), data analysis and interpretation (A.E.S., X.T., S.L.J., A.B., H.S., and F.Y.), draft the article (A.E.S. and F.Y.), critical revision of article for final approval (A.E.S. and F.Y.).

Disclosure Statement

No competing financial interests exist.

Funding Information

This study was supported by the National Institutes of Health (NIH) R01DE024772 (F.Y.), the National Science Foundation (NSF) CAREER award CBET-1351289 (F.Y.), California Institute for Regenerative Medicine Tools and Technologies Award RT3-07804 (F.Y.), the Stanford Bio-X Interdisciplinary Initiative Program (F.Y.), the Stanford Coulter Translational Grant, the Stanford SPARK Translational Grant, the Stanford Child Health Research Institute Faculty Scholar Award (F.Y.), the Bio-X fellowship from the Stanford Bio-X program (A.E.S.).

Supplementary Material

Supplementary Figure S1