Integrated Bone Formation Through In Vivo Endochondral Ossification Using Mesenchymal Stem Cells

Autologous bone is still the gold standard for repairing bone defects due to trauma, congenital defects, and surgical resection. However, autogenous bone grafting has significant limitations, including donor pain, risk of infection, and limited bone volume that can be isolated from the patients^1,2,3,4. Numerous biomaterials have been developed as bone substitutes, combining natural or synthetic polymers with mineralized materials such as calcium phosphate or hydroxyapatite^5,6. Bone formation in these engineered materials is usually achieved using the mineralized material as a priming material to allow stem cells to differentiate directly into osteoblasts through the intramembrane ossification (IMO) process⁷. This process lacks the angiogenic step, resulting in insufficient in vivo vascularization of the graft after implantation^8,9,10, and therefore, approaches using such a process may not be optimal for treating large bone defects¹¹.

Strategies applied to recapitulate the endochondral ossification (ECO) process, an innate mechanism in skeletogenesis during development, have been shown to overcome significant problems associated with traditional IMO-based approaches. In ECO, chondrocytes in the cartilage template release vascular endothelial growth factor (VEGF), which promotes vascular infiltration and remodeling of the cartilage template into bone¹². The ECO-mediated approach to osteogenesis via cartilage remodeling and angiogenesis, which is also activated during fracture repair, uses artificially created cartilage tissue derived from MSCs as a priming material. Chondrocytes can tolerate hypoxia in bone defects, induce angiogenesis, and convert a vascular-free cartilage graft into angiogenic tissue. Numerous studies have reported that MSC-based cartilage grafts generate bone in vivo by implementing such an ECO program^{13,14,15,16,17,18,19,20,21}.

An essential requirement for the clinical application of this ECO-mediated approach is how to prepare the desired amount of cartilage graft in a clinical setting. Preparing clinical cartilage of a size that fits the actual bone defect is not practical. Therefore, graft cartilage must form bone integrally when multiple fragments are implanted²². Hydrogels may be an attractive tool for scaling up tissue-engineered grafts for endochondral ossification. Many naturally derived hydrogels support MSC cartilage formation in vitro and ECO in vivo^{23,24,25,26,27,28,29,30,31,32}; however, the optimal support material to meet the clinical application requirements has remained undetermined. Hyaluronic acid (HA) is a biodegradable and biocompatible polysaccharide present in the extracellular matrix of cartilage³³. HA interacts with MSCs via surface receptors such as CD44 to support chondrogenic differentiation^{25,26,28,30,31,32,34}. In addition, HA scaffolds promote IMO-mediated osteogenic differentiation of human dental pulp stem cells³⁵, and scaffolds combined with collagen promote ECO-mediated osteogenesis^36,37.

Here, we present a method for preparing HA hydrogels using bone marrow-derived adult human MSCs and their use for hypertrophic chondrogenesis in vitro and subsequent endochondral ossification in vivo³⁸. We compared the characteristics of HA with those of collagen, a material widely applied in bone tissue engineering with MSCs and a useful material for scaling up artificial grafts for endochondral ossification¹⁷. In an immunocompromised mouse model, HA and collagen constructs seeded with human MSCs were evaluated for in vivo ECO potential by subcutaneous implantation. The results show that HA hydrogels are excellent as a scaffold for MSCs to create artificial cartilage grafts that allow bone formation through ECO.

The protocol is divided into two steps. First, constructs of human MSCs seeded on hyaluronan hydrogel are prepared and differentiated into hypertrophic cartilage in vitro. Next, the differentiated constructs are implanted subcutaneously in a nude model to induce endochondral ossification in vivo (Figure 1).

This protocol uses 4-week-old male nude mice. House four mice in a cage under a 12 h light/dark cycle at 22−24 °C and 50%−70% relative humidity. All animal experiments were conducted in accordance with the guidelines approved by the Institutional Animal Care and Use Committee of Tokyo Medical and Dental University (approval ID: A2019-204C, A2020-116A, and A2021-121A).

1. Preparation of buffers and reagents

1. Prepare mesenchymal stem cell growth medium (MSC growth medium) by adding the supplement kit of MSC growth medium to MSC basal media and store it at 4 °C.
2. Prepare Dulbecco's Modified Eagle Medium and Ham's F12 (DMEM/F-12) medium by adding 10% fetal bovine serum (FBS), 50 µg/mL gentamicin, 200 µM L-alanyl-L-glutamine, and 1 ng/mL basic fibroblastic growth factor to the medium and store it at 4 °C.
3. Prepare chondrogenic medium by adding 1% ITS-G supplement, 0.12% bovine serum albumin, 50 µg/mL gentamicin, 0.35 mM L-proline, 100 nM dexamethasone, and 10 ng/mL TGF-β3 to DMEM just before use.
4. Prepare hypertrophic medium by adding 10 mM β-glycerophosphate, 0.12% bovine serum albumin, 10 nM dexamethasone, 200 µM ascorbate-2-phosphate, 50 nM L-thyroxine, 50 µg/mL gentamycin, and 50 pg/mL IL-1β to DMEM just before use.
5. Coat a 24-well culture dish with 200 µL of melted paraffin pre-heated at 60 °C. Allow to stand at room temperature until the paraffin is set.

2. Expansion of human MSCs

NOTE: Prior to starting the experiments, MSCs with a high potential for MSC chondrogenesis should be selected in micro mass culture as described in¹⁷.

1. According to the manufacturer's instructions, culture primary human bone marrow-derived MSCs (passage 2) with MSC growth medium in a 10 cm dish at a density of 5 x 10³ cells/cm² at 37 °C in 5% CO₂ and 95% humidified air incubator.
2. Change the medium every 2-3 days. Once the cells reach 80%-90% confluency, aspirate the medium from the cell culture dish using a vacuum pump, wash with 5 mL of PBS, and then aspirate the solution with a vacuum pump.
3. Add 1 mL of 0.05% trypsin/0.02% EDTA solution to the dish. Incubate the dish for 5-10 min at 37 °C.
4. Add 1 mL of MSC growth medium to stop the enzymatic reaction. Transfer the cell suspension to a 50 mL tube.
5. Combine the cell suspension from other dishes in the 50 mL tube and keep it on ice.
6. Use the cell counter to count the cells by mixing 10 µL of the cell suspension with 10 µL of 0.4% trypan blue stain.
7. Centrifuge the remaining cell suspension at 220 x g for 3 min at room temperature. Remove the supernatant and add cryopreservation solution at a concentration of 1.0 x 10⁶ cells per mL.
8. Dispense 1 mL of the cell suspension into each cryotube and store at -80 °C for 12 h before transferring to liquid nitrogen. The resulting frozen stocks (passage 3) are used for this protocol.
9. For the experiments, seed the stocked MSCs in a 10 cm dish at a density of 5 x 10³ cells/cm². Culture cells until 80%-90% confluent (passage 4) in DMEM/F-12 medium.

3. Preparation of MSC-encapsulated hydrogels

1. Trypsinize the cells and prepare cell suspension as described in steps 2.3 - 2.6.
2. To make 10 constructs (2.5 x 10⁵ cells per construct), aliquot the cell suspension containing 2.5 x 10⁶ cells into a 1.5 mL microtube.
3. Centrifuge the suspension at 220 x g for 3 min, remove the supernatant with a vacuum pump, and keep it on ice.
4. Bring the thiol-modified hyaluronic acid (HA), thiol-reactive crosslinker, polyethylene glycol diacrylate, and degassed water bottles to room temperature.
5. Under sterile conditions, add 1.0 mL of degassed water to the HA-containing bottle (see manufacturer's instructions) using a syringe with a needle.
6. Vortex occasionally warming to 37 °C until the solution becomes clear, then place on ice. It takes less than 30 min for the solids to dissolve completely.
7. Under sterile conditions, add 0.5 mL of degassed water to the crosslinker bottle using a syringe with a needle. Dissolve by inverting several times, then place on ice.
8. Add 120 µL of the dissolved HA solution to the aliquoted cell pellet (2.5 x 10⁶ cells). Resuspend the cell pellet by pipetting back and forth.
9. Add 30 µL of the dissolved crosslinker solution to the tube containing the HA and cells (step 3.8) and combine by tapping the tube, and then spin down briefly. Combine the HA solution and the crosslinker solution in a 4:1 ratio.
10. Drop 15 µL of the combined solution containing MSCs (2.5 x 10⁵ cells) onto the paraffin-coated 24-well plate and allow it to solidify at 37 °C for 30 min (Figure 2). Due to high viscosity, prepare at least 10% extra amount of the cell/hydrogel mixture than needed.
    ​NOTE: The characteristics of the hydrogel have been described previously³⁹.

4. In vitro differentiation conditions

1. Add 0.5 mL of chondrogenic differentiation medium to a construct of MSC-seeded HA hydrogels. Change the medium every 2-3 days.
2. After 3 weeks, switch to hypertrophic differentiation medium (0.5 mL per well) and culture the constructs for an additional 2 weeks.

5. In vivo implantation of MSCs

1. After a total of 5 weeks of in vitro culture, implant the constructs subcutaneously into the back of nude mice as described previously^38,40.
2. Weigh mice and administer a combination anesthetic prepared with 0.75 mg/kg body weight (b.w.) medetomidine, 4.0 mg/kg b.w. midazolam, and 5.0 mg/kg b.w. butorphanol by intraperitoneal injection as described³⁸.
3. Confirm adequate anesthesia by immobility to surgical stimuli and monitor respiratory depth by slow, regular chest movements and proper oxygen supply by the color of the pink mucosa periodically during the operation. Use veterinary ointment on the eyes to prevent dryness during anesthesia.
4. Place the mouse on a sterile surgical drape in a prone position, sterilize the incision site with 50 ppm hypochlorous acid water (pH 5.0; HAW), and place a perforated surgical drape over the mouse.
   ​NOTE: Sterilize all surgical materials in an autoclave, ethylene oxide gas, or HAW. The surgeons should wash their fingers and wear surgical gowns and gloves.
5. Make two 5 mm long skin incisions 2 cm apart in the shoulder and hip areas along the center line of the spine by using scissors.
6. Insert a spatula subcutaneously through the incision to create a subcutaneous pocket.
7. Insert two to three constructs (~15 µL each, step 5.1) into each pocket with forceps. HA constructs implanted in the same pocket are prone to fusion very frequently. To prevent fusion, insert one HA construct in each pocket.
8. Close the incisions with 4-0 braided silk sutures. Inject the mouse with an antagonist to anesthetics prepared with 0.75 mg/kg b.w. atipamezole as described³⁸.
9. While the mice recover from anesthesia, place the mice in sterile recovery cages containing sterile floor bedding without food and water bottles and continuously monitor body temperature, pulse, and respiration.
10. Do not leave the mice unattended until it has regained sufficient consciousness to maintain sternal recumbency. After full recovery, return the animals to the breeding room with the other animals.
11. Monitor the animals every couple of days during the healing period for any possible complications. Euthanize the mice by carbon dioxide inhalation with little suffering at 4- and 8-weeks post-implantation.
12. Incise the skin at the grafted sites and remove the implanted constructs with forceps. Fix the constructs in 4% paraformaldehyde for 16 h and then subject them to analysis.

6. Statistical analysis

1. Report quantitative data as mean ± standard deviation (SD). Perform statistical analysis using commercial software.
2. Use the Student's t-test or one-way ANOVA followed by Tukey's multiple comparisons test to determine significant differences between groups. p<0.05 and p<0.01 indicate significant differences between two groups.

MSC-encapsulated HA hydrogels were cultured in chondrogenic medium supplemented with TGFβ3, an inducer of chondrogenesis⁴¹ (step 4.1). We compared the properties of HA with those of collagen, which has been shown to be effective in creating MSC-based artificial cartilage grafts for endochondral ossification, as described previously³⁸. Undifferentiated MSCs were not included as negative controls in this study because it has been demonstrated that undifferentiated MSCs require mineralized surfaces as a priming substrate to generate bone by osteogenic differentiation (i.e., intramembrane ossification)⁷.

The size of the hyaluronan hydrogels did not change throughout the in vitro culture period, as judged based on diameters measured under an inverted microscope, whereas the collagen hydrogels used for comparison began to shrink soon after the culture started. To reduce the size difference between the HA and collagen constructs after in vitro culture, the number of MSCs and the volume of the gel used to create the HA hydrogels were halved compared to those used for the collagen hydrogels. However, the wet weight of the HA constructs after 3 weeks of cartilage culture was 4x heavier than that of the collagen constructs.

Chondrogenic and hypertrophic differentiation in vitro.
After chondrogenic differentiation (at week 3 of in vitro culture), sulfated glycosaminoglycans (sGAG)-positive (Safranin O/fast green staining; Figure 3A) and type II collagen-positive extracellular matrix was observed in both constructs, indicating that both HA and collagen hydrogels supported chondrogenesis. However, compared to the collagen construct, the distribution of sGAG, type II collagen (Col II), and type X collagen (Col X) were more homogeneous throughout the tissue in the HA constructs. Differences between the two constructs were also apparent in cell morphology: cells with a chondrocyte-like rounded morphology were distributed throughout the tissue in the HA constructs. In the collagen constructs, however, cells in the periphery showed a heterogeneous morphology, ranging from irregular/stretched to rounded morphology. Consistent with this, the average roundness of cells in the HA constructs was higher than in the collagen constructs, both in the periphery and central regions (34% versus. 77.6% (p<0.01) and 78.3% vs. 100% (p<0.05), respectively; Figure 3B). At 5 weeks of culture, calcium deposition was detected at the outer edges in both constructs (alizarin red; Figure 3C). Furthermore, expression of chondrogenic (Sox-9, aggrecan (ACAN), Col II), hypertrophic (Col X, matrix metallopeptidase 13 (MMP-13)), and osteogenic (type I collagen (Col I), bone sialoprotein (BSP), osteocalcin (OCN)) markers¹⁴ was detected by quantitative real-time RT-PCR, confirming the progression of chondrogenic and hypertrophic differentiation during in vitro culture (Figure 4).

Endochondral ossification of subcutaneously implanted constructs in vivo.
Subcutaneous pockets were created on the back of nude mice, and two to three constructs were implanted in each pocket after hypertrophic differentiation (at week 5 of in vitro culture). At 4 or 8 weeks after implantation, all HA constructs were attached to each other in the implanted pockets (Table 1). In the collagen constructs, adhesion was observed in 60% of the pockets, while in the remaining pockets, the grafts were independent of each other. Hematoxylin and eosin (H&E) staining revealed that in both constructs at 8 weeks post-implantation, osteoid tissue with lamellar morphology was formed in the outer regions of the constructs (Figure 5Aa,f). sGAG staining disappeared in the HA constructs, indicating the loss of their cartilage phenotype (Figure 5Ah). In both constructs, a bone marrow component containing hematopoietic cells and adipose tissue was formed between the inner cartilage and the outer osteoid tissues. In all HA constructs, the bone tissues of the fused constructs were connected and surrounded by joint fibrous tissue and bone marrow developed along the space between the two adherent constructs, indicating that multiple HA constructs tend to form integrated bone tissue (Figure 5Ag). The adhered collagen constructs were similarly integrated in two of the three fused cases, but in the third case, the bone tissue of the two constructs was not connected (Table 1 and Figure 5Ab). In both constructs, vessels identified by CD31-positive endothelial cells⁴² were observed in the bone marrow (Figures 5Ad,i), and the inner cartilage regions were surrounded by TRAP-positive multinucleated cells of the osteoclast lineage, indicating that cartilage tissue was subject to remodeling (Figures 5Ae,j).

Mineral was deposited in the outer osteoid region in both HA and collagen constructs (Figure 5B). While the total mineral density of the new bone was similar between HA and collagen constructs, the mineral volume was significantly higher in the HA constructs than that in the collagen constructs, which may reflect the fact that HA hydrogels did not shrink during in vitro culture, resulting in larger constructs than collagen hydrogels (Figure 5C,D).

Figure 1: Experimental design. The first step of this protocol is to encapsulate expanded MSCs in HA hydrogels to promote chondrogenic and hypertrophic differentiation in vitro. Constructs are cultured in chondrogenic differentiation conditions for 3 weeks, followed by an additional 2 weeks in hypertrophic differentiation conditions. The second step of this protocol is to subcutaneously implant the constructs into nude mice at week 5 of in vitro culture to undergo endochondral ossification in vivo for 8 weeks. Please click here to view a larger version of this figure.

Figure 2: Preparation of MSCs-seeded hydrogels. Drop modified HA/crosslinker solution containing MSCs onto a paraffin-coated 24-well plate and allow to solidify at 37 °C for 30 min. Please click here to view a larger version of this figure.

Figure 3: Histological and immunohistochemical analysis of HA and collagen constructs after in vitro culture²². (A) Constructs at 3 weeks (3W) of in vitro culture were stained for sGAG (safranin O/fast green), type II collagen (Col II), and type X collagen (Col X). (B) The average circularity values of cells at 3 weeks of in vitro culture (n = 5). Open and grey bars indicate collagen and HA constructs, respectively. Groups without a common letter are statistically different (a versus b, p<0.01; b versus c, p<0.05). (C) Constructs at 5 weeks (5W) of in vitro culture were stained for calcium (alizarin red). All pictures were captured with the same magnification (Scale bar: 200 µm). A low-magnification overview of the entire tissues is shown in the insets (Scale bar: 1 mm). Error bars represent mean ± standard deviation (SD). The one-way ANOVA followed by Tukey's multiple comparisons test was used to determine significant differences between groups. This figure has been modified from³⁸. Please click here to view a larger version of this figure.

Figure 4: Gene expression analysis of HA and collagen constructs at 3 and 5 weeks of in vitro culture. (A) Chondrogenic and hypertrophic markers. (B) Osteogenic markers. Values are presented as mean ± standard deviation (SD) (n = 3). Undifferentiated MSCs expressed high levels of Col I and MMP-13 and low levels of Sox-9 and Col X; they did not express (#) ACAN, Col II, BSP, and OCN. Error bars represent mean ± standard deviation (SD). This figure has been modified from³⁸. Please click here to view a larger version of this figure.

Figure 5: Immunohistochemical analysis and calcification of HA and collagen constructs post-implantation at 8 weeks. (A) Constructs were stained for Hematoxylin and Eosin (H&E, a, b, f, and g), safranin O/Fast Green (c and h), endothelial cells (Cluster of Differentiation 31, CD31) (d and i), and Tartrate-resistant acid phosphatase (TRAP, e and j). (a and f) Overview of the entire tissues (scale bar = 500 µm). H&E, safranin O, CD31: scale bar = 200 µm; TRAP: scale bar= 50 µm. Arrows indicate CD31-positive endothelial cells. Abbreviations: c = inner cartilage tissue; o = outer osteoid tissue. (B) MicroCT (µCT) imaging of collagen and HA constructs (main and inset image scale bars = 2 mm). (C,D) The total mineral density (C) and volume (D) of HA and collagen constructs (n = 4). ** indicates significant differences; p <0.01. Four constructs were analyzed per group using µCT. Error bars represent mean ± standard deviation (SD). Student's t-test was used to determine significant differences between groups. This figure has been modified from³⁸. Please click here to view a larger version of this figure.

Figure 6: Strategy for facilitating ECOs using HA-based artificial cartilage. Schematic diagram of ossification and bone marrow development without and with fragmentation (µ-pellets) of implanted HA constructs. (A) HA hydrogels have an excellent fusion tendency to form an integrated bone with vascularization and marrow development between fused grafts in vivo. The resulting implanted tissue, however, remained predominantly in a cartilage state even 8 weeks post-implantation. (B) Given the tendency of HA constructs to fuse to form integrated bone, processing HA constructs into µ-pellets may accelerate the remodeling rate of transplanted tissue and promote bone formation. Micronizing the constructs increases surface area, which may promote the resorption of cartilage tissue and the formation of bone tissue and may also promote angiogenesis and bone marrow development because of the increased space for host cell infiltration. Please click here to view a larger version of this figure.

Table 1: Unification rate when multiple constructs were implanted in a single pocket. Two or three constructs were subcutaneously implanted in a single pocket. Constructs were harvested 4- or 8-weeks post-implantation and histologically examined.

Using appropriate scaffold materials that promote the transition from hypertrophic cartilage to bone is a promising approach to scale up MSC-based engineered hypertrophic cartilage grafts and treat bone defects of clinically significant size. Here, we show that HA is an excellent scaffold material to support the differentiation of MSC-based hypertrophic cartilage tissue in vitro and to promote endochondral bone formation in vivo³⁸. Furthermore, in vivo, HA constructs were shown to facilitate the fusion of multiple constructs implanted in the same pocket to form a single integrated bone. Because the interaction of MSCs with scaffold materials significantly affects cartilage and hypertrophic chondrogenesis, selecting suitable scaffold material for ECOs is important for generating clinically effective tissue-engineered cartilage grafts. Cells embedded in HA hydrogels uniformly exhibit a rounded morphology characteristic of chondrocytes from the center to the periphery, indicating that hyaluronan provides an environment conducive to uniform cartilage formation throughout the hydrogel^21,30. Unlike HA, collagen interacts with MSCs via integrin binding sequences (RGDs); however, the persistence of this interaction prevents MSCs from further differentiating in the chondrogenic pathway⁴³, which may account for the non-typical chondrocyte cell morphology that is prominent in the periphery of collagen constructs.

If multiple grafts unite to form a single bone, this ECO-mediated approach can be extended to more clinically relevant bone defects. When multiple MSC-based and scaffold-free hypertrophic cartilage were applied to a single bone defect, it was reported that the cartilage grafts integrated with the bone defect. However, adjacent grafts did not integrate with each other²². The approach presented here may offer one possible solution to overcome this limitation. Compared to collagen constructs, multiple grafts of HA constructs had a greater tendency to fuse to form a single bone. Furthermore, bone marrow was formed between fused HA constructs, which may be related to the fact that HA supports uniform cartilage differentiation throughout the hydrogel because hypertrophic cartilage secretes VEGF and provides an environment for the hematopoietic stem-cell niche in bone marrow development ⁴⁴.

The key to a successful experiment using this protocol is the quality of MSCs. Confirming that the MSCs to be used have sufficiently high chondrogenic differentiation potential in advance is recommended. Second, the size of each construct should not be too large. In our experience, the volume of a single construct is limited to 10-15 µL. Larger constructs may result in insufficient oxygen and nutrient penetration into the constructs, leading to necrosis and poor differentiation in the center of the constructs. Therefore, gel thickness needs to be reduced to improve oxygen and nutrient penetration. In order to reduce gel thickness, it may be helpful to create thin gels on the porous membrane of Transwell inserts^14,45. Alternatively, because HA constructs tend to fuse and form integrated bone, it may be adequate to implant many smaller constructs rather than a single large construct, as discussed below.

Limitations of this study include the need to accelerate the remodeling rate of the implanted tissues to facilitate bone formation.The implanted engineered tissue remained primarily in a hypertrophic calcified cartilage state even 8 weeks post-implantation. In bone formation via the ECO pathway, host-derived cells play a critical role in promoting ECO^16,46. Providing additional channels in hypertrophic cartilage grafts has been reported to promote vascular invasion and graft mineralization in vivo²⁰. Such channels serve as conduits for the infiltration of host cells, including osteoblasts, osteoclasts, and hematopoietic precursors, to form new tissue within the construct. Alternatively, it has been reported that multiple small fibrin-encapsulated pellets (µ-pellets) composed of chondrogenic differentiated MSCs can be implanted to form integrated bone⁴⁷. Processing HA constructs into µ-pellets could increase the surface area surrounded by TRAP-positive osteoclasts, thus promoting the remodeling rate of cartilage tissue and enhancing osteogenesis (Figure 6)^21,48. Furthermore, implantation of µ-pellets would increase the infiltration space of host cells, thereby facilitating angiogenesis and bone marrow development and generating integral bone formation with abundant vascularization. Thus, considering the tendency of HA constructs to fuse and create integrated bone, this µ-constructed bone regeneration strategy may accelerate the ECO process of hypertrophic cartilage grafts using HA hydrogels.

In conclusion, HA hydrogels are effective in scaling up tissue-engineered grafts with fewer cells and gels than collagen hydrogels. In addition, HA hydrogels have excellent fusion sensitivity and produce integrated bone with vascularization and marrow development between fused grafts. Therefore, modification of HA constructs to promote host cell infiltration and graft remodeling may lead to the development of implantable materials that can further promote vascularization and bone marrow development. With further optimization, this approach could be a promising alternative to current bone therapies in maxillofacial and orthopedic surgery.