Engineering Injectable Bone Using Bone Marrow Stromal Cell Aggregates

Abstract

With the increasing popularity of minimally invasive surgery, to develop an injectable bone would be highly preferable for the repair of bone nonunions and defects. However, the use of dissociated cells and exogenous carriers to construct injectable bone faces several drawbacks. To circumvent these limitations, we first harvested a cell sheet from rabbit bone marrow stromal cells using a continuous culture method and a scraping technique. The obtained sheet was then cut into fragments of multicellular aggregates, each of which was composed of a certain number of cells, extracellular matrix, and intercellular connections. The aggregates showed apparent mineralization properties, high alkaline phosphatase activity, increased osteocalcin content, and upregulated bone markers, implying their in vitro osteogenic potential. Then, serum-free medium (the control group), dissociated cell suspension (the cell group), and suspension of multicellular aggregates (the aggregate group) were injected subcutaneously on the back of the nude mice to evaluate ectopic bone formation. The results revealed that the aggregate group showed significantly larger and denser bone at the injection sites than the cell group, whereas bone formation did not occur in the control group. Additionally, when injecting them locally into the mandibular fracture gap of delayed healing in a rabbit model, we observed the most improved bone healing in the aggregate group. More cells survive and retain at the injection sites in the aggregate group than that in the cell group postoperatively. Our study indicates that the multicellular aggregates might be considered a promising strategy to generate injectable bone tissue and improve the efficacy of cell therapy.

Introduction

T he repair of bone nonunions and defects remains a significant clinical problem [1]. Fortunately, recent advances in regenerative medicine have enabled cell-based therapies to be a potential solution for bone regeneration [2,3]. However, an open operation is often required to implant the engineered bone. Compared with the surgical transplantation method, bone repair via injection technique has obvious advantages such as easier administration, lower risk, less scar formation, minor invasion, and higher reproducibility [4]. Previous studies have showed that injection of autologous bone marrow is a simple and safe method to treat nonunion [5]. However, only small amounts of osteoprogenitor cells are present in bone marrow and the concentration of progenitor cells is low, compromising the efficacy of treatment [5]. To get sufficient numbers of cells, desired cells must be expanded in vitro. Upon confluence, they are usually collected from culture dishes with the treatment of proteolytic enzyme, which results in the loss of their extracellular matrix (ECM), cell–matrix interactions, and cell–cell connections. All these components are critical for cell growth, survival, differentiation, and tissue development [6 –8]. In addition, the current cell therapy technique involves local injection of the dissociated cells. Nevertheless, after injection into damaged tissue, a major portion of the dissociated cells may leak out of the transplantation site or be washed out by the vasculature and lymphatic system [9,10]. Moreover, mechanical insult during cell collection and injection may also impede cell viability and retention within the damaged tissue [10]. To improve retention of the dissociated cells and to provide cells with specific substrate for bone deposition, many injectable biomaterials, such as alginate, collagen, agarose, hyaluronate, fibrin, and pluronic, have been used as cell carriers [11 –13]. However, these products are far from ideal because of some drawbacks including potential immunogenicity, elevated inflammatory response, possible toxicity of degradation products, unforeseeable gelation kinetics, fluctuating degradation rate, and uncontrollable cell–biomaterial interactions [12 –14].

Before an ideal carrier was found, a scaffold-free approach may be an attractive alterative to circumvent the aforementioned problems. Recently, cell sheet technique has been proven effective for the engineering of bone tissues [15 –18]. Moreover, the cell sheet was found flexible, contractible, and deformable after detachment, and its thickness could be controlled [17–18]. Thus, we hypothesize that fragments of a thin cell sheet, which are composed of a certain number of cells, their endogenous ECM, and intercellular connections, can be used to engineer injectable bone tissue and consequently improve the efficacy of cell-based therapy. To test the hypothesis, we first harvested a cell sheet from rabbit bone marrow stromal cells (BMSCs) using a continuous culture system and a scraping technique, instead of enzymatic digestion. The obtained construct was then cut into fragments of multicellular aggregates (with appropriate size) and suspended in serum-free medium. Next, the aggregates were injected into the subcutaneous space of a nude mouse model and the delayed union of a rabbit mandibular model, respectively, to investigate the in vivo osteogenic capability. The dissociated cells with the same concentration and the serum-free medium with the same volume were used as control.

Materials and Methods

Experimental animals

Twelve male athymic mice (6 weeks old, weighing 18–23 g) and 9 New Zealand rabbits (6 months old, with an average weight of 2.5 kg) were used in this study. Our experiments were performed following the guidelines of the Chinese government for the care and use of laboratory animals. All protocols were approved by the Animal Welfare Committee of Fourth Military Medical University.

Cell culture and preparation of multicellular aggregates

BMSCs were isolated from the ilium marrow of adult rabbits. Iliac bone grafts were split in half and the marrow was flushed out with low-glucose Dulbecco's modified Eagle's medium (DMEM) (Gibco, Invitrogen Corp.) containing 10% fetal bovine serum (Gibco) and 0.272 g/L l-glutamine (Sigma-Aldrich). Clumps of cells were broken up to become a homogeneous cell suspension by repeated pipetting. The suspension was then plated in 100-mm cell culture dishes and the standard medium was added to fill each dish up to 15 mL. The cells were incubated at 37°C with 5% CO₂ in 100% humidity. After 3 days, the medium and all floating cells were removed, and new medium was added to the remaining adherent cells, which were defined as BMSCs [19]. The medium was changed every 3 days until the cells reached confluence. The cells were then collected by trypsinase treatment, counted with a hemocytometer, and subcultured at a proportion of 1:3.

A cell sheet was prepared by seeding BMSCs of the first passage into 100-mm dishes at a density of 1 × 10⁴ cells/cm² and adding osteogenic medium consisting of high-glucose DMEM plus 10% fetal bovine serum (Gibco), l-ascorbicacid-2-phosphate (50 μg/mL), β-glycerophosphate (10 mM), and dexamethasone (100 nM) (all from Sigma). The cells were continuously cultured for 2 weeks without passage and the osteogenic medium was replaced every 2 days during this period. After 2 weeks' culture, a thin sheet formed, which could be easily detached with a cell scraper. The obtained sheet was cut into small fragments with sizes of up to 1 mm in diameter with a scalpel. The resultant fragments of multicellular aggregates were then suspended in 0.6 mL of serum-free DMEM for use (Fig. 1). The total number of cells in multicellular aggregates from each cell sheet (55 cm²) was measured by DNA analysis using Hoechst 33258 Fluorometry as previously described [18,20]. Briefly, the lifted cell sheet was digested with 2 mL of 10 mM ethylenediaminetetraacetic acid (pH 12.3) and was lysed by a 10-min sonication in an ice bath. Samples (0.5 mL) were then incubated with 1.5 mL of Hoechst 33258 (100 ng/mL; Sigma-Aldrich) working solution for 10 min at room temperature. Fluorescence of the sample mixture was then measured on a DyNAQuant 200 (Hoefer). Based on the linear relationship between fluorescence value and a known number of dissociated cells (counting with a hemocytometer), a conversion factor was gained to calculate the cell number in a cell sheet.

FIG. 1.

Schematic illustrations of the procedures for constructing the BMSC aggregates for injection. BMSC, bone marrow stromal cell.

As control, the dissociated cells of the first passage were cultured in the same osteogenic medium and subcultured with the method described above. After 2 weeks, the third passage of cells were trypsinized, collected, counted, and resuspended with serum-free medium at a concentration equal to that of multicellular aggregates.

Characterization of cell aggregates

For histological, histochemical, and immunohistochemical staining, samples of multicellular construct were fixed in 4% buffered paraformaldehyde, dehydrated in graded alcohols, embedded in paraffin, and sliced into sections with thickness of 5 μm. Several sections were stained with hematoxylin–eosin (H&E) for light microscopy to observe tissue structure. To identify formation of collagen matrix, some sections were stained with Picrosirius Red (0.1% Sirius Red in saturated picric acid) and analyzed using polarization microscopy. Further, we carried out immunohistochemical staining of collagen type I with specific mouse (anti-rabbit) monoclonal antibody (Bioss). In addition, histochemical staining by Alizarin Red was performed to evaluate their osteogenic mineralization.

For scanning electron microscopy, samples were fixed with 2.5% glutaraldehyde for 2 h followed by serial dehydration with ethanol. After this, the samples were air dried, sputtered with gold, and inspected by S-3400N scanning electron microscopy (Hitachi) at 5 kV. To measure the elemental composition of the formed constructs, some dry samples were transferred to propane that was cooled in liquid nitrogen, embedded in Lowicryl K4 in vacuum, and ultraviolet polymerized at −40°C. Elemental analysis was performed using an energy-dispersive X-ray spectrometer (EMAX-2770, Horiba Ltd.). For transmission electron microscopy, some dehydrated samples were embedded in epoxyresin and sliced into ultrathin sections. The sections were then mounted on copper grids and examined under JEM-2000EX (Jeom).

Alkaline phosphatase activity and osteocalcin measurements

Alkaline phosphatase (ALP) activity of multicellular aggregates was measured with an ALP diagnostic kit (Sigma). After the suspension of aggregates (0.2 mL) was homogenized with 1 mL Tris buffer (pH 7.4; Sigma) and sonicated, the cell lysate (0.1 mL) was added to each well of a 24-well plate, and then 0.15 mL of p-nitrophenol phosphate solution (Sigma) and 0.15 mL of ALP buffer solution (Sigma) were added. After incubation at 37°C for 30 min, the mixture was added with 1 mL of 0.5 N NaOH (Gutian Medicine) to stop the reaction and the absorption at 405 nm was measured by spectrophotometry according to the manufacturer's protocol. ALP activity was normalized by cell numbers. ALP activity was expressed as the amount of p-nitrophenol (nM) released by per million cells per minute. As control, ALP activity of equal number of dissociated osteogenic cells (osteoinduction for 2 weeks) was also measured. Osteocalcin (OCN) content of both the multicellular aggregates and the dissociated cells was also measured. Twenty-four hours prior to measurement, the culture medium was replaced with serum-free osteogenic medium. The OCN content in the medium was measured using an ELISA Kit (Biomedical Technologies) following the manufacturer's protocol and then normalized by the total cell number of the sample. Each group for ALP and OCN measurements involved six cell sheets and each measurement was repeated twice to get the average.

Reverse transcription–polymerase chain reaction for osteogenic gene transcription

Total RNA was extracted from the harvested cell sheet and dissociated cells using the Trizol reagent (Invitrogen), and complementary DNA was synthesized using an isolated mRNA kit (Shanghai ShengGong Biological Technology) following the manufacturer's instructions. The mRNA expression of several osteogenic markers was semiquantitated by polymerase chain reaction (PCR) according to the method described previously with minor modifications [21]. The primers were as follows: (1) β-actin, 180 bq, forward 5′-GGACCTGACCGACTACCTCA-3′ and reverse 5′-GGCAGCTCGTAGCTCTTCTC-3′; (2) OCN, 195 bq, forward 5′-CAGGAGGGCAATAAGGTAGT-3′ and reverse 5′-TTTAGGGCAGCACAGGTC-3′; (3) osteopontin, 249 bq; forward 5′-GCTCAGCACCTGAATGTACC-3′ and reverse 5′-CTTCGGCTCGATGGCTAGC-3′; (4) ALP, 358 bq; forward 5′-CTGTCTTGAATGCCTTGGATA-3′; reverse 5′-AGGGCTTCTTGCCTTGACT-3′; (5) collagen type I, 553 bq, forward 5′-TTCTATTGGTCCCGTCGGT-3′ and reverse 5′-GCTGAGTCTCAGGTCGCG-3′ [21]. Multiplications of both the area and the density of bands were analyzed by AlphaView v.1.3.0 software (Alpha Innotech). The value of each marker was standardized against that of β-actin, which served as the internal control.

Athymic mouse model

Ectopic injection

A total of 36 injection sites of 12 male athymic mice were randomly divided into three groups (n = 12 in each group): the control group (serum-free medium); the cell group (dissociated BMSCs); and the aggregate group (multicellular aggregates). Shortly after the mice were anesthetized with 3% isoflurane gas, we injected 0.6 mL of medium or suspension into the subcutaneous sites of the back of mice through an 18-gauge needle. The injection required no skin incisions and no sutures. In the cell and aggregate groups, the final cell number injected per site was ∼9.8 × 10⁶ cells. At 4 and 8 weeks after injection, six mice in each group were sacrificed with carbon dioxide.

Computed tomography scanning

The newly formed tissue in mice was radiologically examined using a computed tomography (CT) scanner (Lightspeed Ultra 16; General Electric) with a slice thickness of 0.6 mm at a voltage of 120 kV and an electric current of 120 mA. The radiological density of the new mass was measured using the method described in our previous study [17]. The mean density of each mass was then calculated. Based on the serial scanned images, 3D isosurface images were generated using 2004 GE Medical System and the volume of the newly formed mass was calculated using the VoXim software (IVS-Solutions AG).

Histological and histomorphometrical examinations

After CT scanning, the newly formed masses were harvested and observed macroscopically. The weight of the constructs was also measured. For histological observation, the constructs were fixed in 4% buffered paraformaldehyde for 24 h and split into two parts. One half of the specimen was decalcified in 5% formic acid for 3 weeks, embedded in paraffin, sectioned to 4 μm thickness, and stained with H&E. The other half was dehydrated in graded alcohols, embedded in polymethylmethacrylate, cut into 10-μm-thick sections using a microtome (Leica), and stained with Goldner's trichrome. Four sections were selected from each specimen for histomorphometrical examination as reported previously [17]. By NIH image analysis, three high-resolution, low-magnification micrographs were randomly selected from each section and analyzed twice by two unbiased examiners who were blinded to the experimental. The cross-sectional area created by mineralized bone (green staining) was measured and expressed as relative percentage of the total area. In addition, blood vessels were identified in H&E-stained tissues as previously described and blood vessels present in the formed tissue were analyzed for their total number [22].

Rabbit model

Orthotopic injection

For treatment of the delayed bone healing, a rabbit mandible model was used as described previously with some modifications [23 –25]. Briefly, after anesthesia, the rabbit was placed in a supine position and the inferior aspect of mandible was shaved and disinfected with 0.5% iodophors. Then mandible was exposed through a submandibular incision, and a custom-made fixation device (Zhongbang Titanium Biomaterials Corporation) was fixed with 10 screws (1.5 × 5 mm). Subsequently, a 2-mm fracture gap was made between the first molars and the mental foramen using a fissured bur without cooling with water and the periosteum was destroyed by cauterization at 3 mm proximally and distally to the gap. Finally, the wound was sutured in two layers. Postoperatively, animals received soft food and intramuscular 100 kU penicillin prophylactically. Four days after the osteotomy, 0.2 mL of serum-free medium containing either 3.3 × 10⁶ dissociated cells (the cell group), aggregates (3.3 × 10⁶ cells, the aggregate group), or no cells (the control group) was injected into the fracture gap of the cell-donor rabbit. Eighteen mandibular osteotomies (n = 6 per group) from nine rabbits were performed. All animals were sacrificed by an intramuscular overdose injection of ketamine and xylazine at 6 weeks after injection.

BrdU was used to label the cells for later identification in the animal study. Upon reaching 60% confluence, cells of the first passage were labeled with 20 μM of 5-bromo-2-deoxyuridine (BrdU; Sigma) for 48 h. In both groups, about 90% of the BMSCs were positively labeled with BrdU after the 2-day cultivation.

Microfocus CT scanning and analysis

After the removal of fixation device, the osteotomy region, including 5 mm of neighboring normal bone, was harvested from each mandible. The specimens, embedded in a holder, underwent morphological and quantitative examination by using a microfocus CT (Micro-CT) (eXplore Locus SP) with a 1-mm-thick aluminum filter at a voltage of 80 kVp and an electric current of 80 mA. The resolution in all three spatial dimensions was 14 μm and the voxel size after reconstruction was 28.87 × 28.87 × 28.87 μm³. Based on the serial scanned images, 3D isosurface images were reconstructed using GE Healthcare's Microview software (version 2.1.2). A threshold used in this study was 0–2400 Housefield units for bone tissue based on the threshold calculations for samples of rabbit mandible. Bone volume fraction (the ratio between the bone volume and the total volume) and bone mineral density (the mass of bone per unit volume) within the region of interest on the fracture gap were calculated, respectively.

Histological and histomorphometrical examinations

After the Micro-CT scan, all 18 samples (n = 6 per group) were immersed in 4% buffered paraformaldehyde for 48 h and decalcified in 5% formic acid for 3 weeks. Samples were then sectioned in the axial plane and stained with H&E. The extent of mineralized bone in each group was evaluated using the histomorphometrical method described above. The cross-sectional areas created by the newly formed bone were measured and the obtained data were expressed as relative percentages of the total cross-sectional area. Additionally, histomorphometrical examination of blood vessel present in the fracture gap was performed.

BrdU immunohistochemical staining was also performed to identify the injected cells that maintained in the injection site. Briefly, the 4-μm sections were deparaffinized, rehydrated, and subjected to immunolabeling with a mouse monoclonal anti-BrdU antibody (Boster). Subsequently, a fluorescein-conjugated goat anti-mouse IgG (KPL) was used as the secondary antibody to identify positively labeled cells following the manufacturer's instructions. The number of BrdU-labeled cells was counted under fluorescence microscopy and the results were expressed as the mean number of labeled cells per unit area (mm²).

Statistical analysis

All measurements were expressed as mean value with its standard deviation (mean ± SD). A paired t-test was used to analyze difference in ALP activity, OCN content, reverse transcription (RT)-PCR, bone weight, volume, radiological density, and histomorphometrical analysis of the ectopic constructs between the cell group and the aggregate group. Data from orthotopic injection in rabbit model were analyzed by one-way analysis of variance of post-hoc Tukey testing (SPSS, version 13.0). Statistically significant values were set at P < 0.05.

Results

Fabrication of cellular aggregates

After seeding at a density of 1 × 10⁴ cells/cm² and culturing in osteogenic medium for 2 weeks, BMSCs formed a thin membrane-like multicellular construct, with an average of 9.8 × 10⁶ cells (55 cm²). The cell sheet could be lifted from the culture dish with a cell scraper. After detachment, the sheet underwent spontaneous contraction, but no cellular morphological changes were observed (Fig. 2a, b). Histological examination showed that the harvested sheet was composed of several layers of cells, which were distributed by ECM (Fig. 2c, d), showing morphological similarities to the native tissues. Red birefringence in the Picrosirius staining under polarization microscopy suggested that abundant collagen matrix had formed and the presence of type I collagen was further confirmed by immunohistochemical staining (Fig. 2e, f). The formation of calcium deposits, an osteoblast-specific event, was revealed by Alizarin Red staining (Fig. 2g).

FIG. 2.

Morphology and histological analysis of the cell sheet. (a) Gross appearance of multicellular construct after physical detachment from a flask. (b) Morphology of the hyperconfluent cells. (c) H&E staining for cell sheet composed of multilayer cells and extracellular matrix. (d) BMSCs within the construct labeled with BrdU. (e) Picrosirius Red staining of the formed construct analyzed with polarization microscopy. (f) Formation of collagen type I (arrow, dark brown area) was confirmed by immunohistochemical staining. (g) Calcium deposits (arrow, dark red area) were revealed by Alizarin Red staining. (h) Morphology of the BMSC aggregates after injection through the needle. (i) Morphology of the dissociated BMSC suspension after injection. Scale bars: (a) 3 cm; (b, c, e–f) 100 μm; (d) 25 μm. H&E, hematoxylin–eosin.

After cutting into small fragments with diameter up to 1 mm, the cellular aggregates could be collected as a concentrated suspension for injection. Compared with the cells collected by trypsinase treatment, these aggregates are easier to handle and harvest by scraping. The cellular aggregates could readily pass through the 18G needle without clogging the channel of the needle during injection. After injection, the aggregates still maintained the concentrated suspension and no changes in cell morphology were observed. The produced aggregates, microparticle-like multicellular constructs, were composed of a certain number of spread cells, whereas the cell suspension consisted of many dispersed cells (Fig. 2h, i).

ALP activity and OCN content

Measurement of ALP activity and OCN content was performed to confirm the maintenance of the osteogenic phenotype of BMSCs in both groups. As shown in Fig. 3a, b, both ALP activity and OCN content of the aggregate group was significantly higher than that in the same number of dissociated cells (P < 0.05).

FIG. 3.

ALP activity, OCN content, and gene expressions of osteogenic marker in different groups (cell: the cell group; aggr: the aggregate group). (a) Lysate showed significantly elevated ALP activity after osteoinduction (n = 6, ^* P < 0.05). (b) OCN content of the aggregate group was significantly higher than that in the same number of dissociated cells (n = 6, *P < 0.05). (c) Expressions of mRNA for ALP, OCN, collagen type I (COL I), and OPN in the two groups using the same osteogenic medium. (d) Expression of mRNA for all four markers was upregulated in the aggregate group than the cell group (n = 6, *P < 0.05). ALP, alkaline phosphatase; OCN, osteocalcin; OPN, osteopontin.

Gene expression of osteogenic markers

As observed in Fig. 3c, although BMSCs from both two groups expressed ALP, OCN, collagen I, and osteopontin after 2-week cultivation in osteogenic medium, the expression levels of mRNA for all four osteogenic markers were significantly higher in the aggregate group compared with those in the cell group (P < 0.05).

In vivo osteogenic capability in the athymic mouse model

Injection procedure and gross view

At both time points after in vivo injection, hard bone-like mass could be clearly visible at the injection site in the aggregate group, whereas only hard tissue with small volume was observed in the cell group (Fig. 4). In contrast, injection of serum-free medium did not result in the formation of any visible tissue. Table 1 showed that the average weight and volume of constructs from the aggregate group was significantly heavier and larger than those from the cell group.

FIG. 4.

CT scan and gross appearance of the construct after subcutaneous implantation. (a–d) Representative CT images of mice showing ectopic mineralization at 4 and 8 weeks after injection of the dissociated cells (arrowhead) and the aggregates (arrow). (a) 3D CT image at 8 weeks. (b) 3D CT image at 8 weeks. (c) Coronal image at 4 weeks. (d) Coronal image at 8 weeks. (e) The prepared suspension of multicellular aggregates for injection. (f) Gross appearance of the representative mice at 8 weeks after injection. (g) Gross appearance of the harvested sample in the cell group at 4 weeks. (h) Sample was harvested from the site in the aggregate group at 4 weeks. (i, j) Gross appearance of the samples harvested at 8 weeks after cell and aggregate injection, respectively. CT, computed tomography; 3D, three dimensional.

Table 1.

Analysis of Weight, Volume, and Radiological Density and Histomorphometrical Analysis of the Ectopic Constructs

	4 weeks after injection		8 weeks after injection
Variable	Cell	Aggregate	Cell	Aggregate
Volume (mm³)	32.8 ± 5.2	198.4 ± 31.6^a	25.8 ± 4.8	176.7 ± 29.0^a
Weight (mg)	21.3 ± 5.1	215.2 ± 42.6^a	27.0 ± 5.2	253.8 ± 46.6^a
Radiological density (Hu)	246.5 ± 20.7	298.0 ± 28.5^a	403.1 ± 48.6	391.9 ± 32.5
Vessel density (vessels/cm²)	7.7 ± 2.6	8.5 ± 3.5	12.2 ± 4.0	18.0 ± 4.1^a
Bone/total area (%)	19.5 ± 4.9	28.3 ± 7.2	46.1 ± 8.7	67.5 ± 11.2%^a

Results are shown as mean ± standard error of mean (n = 6).

P < 0.05 compared with the cell group after the same time points.

CT scan

Both coronal and 3D CT images revealed that no ectopic mineralization was identified in the control group at each time point. In other two groups, ectopic mineralization was radiologically distinguishable in the back of mice at both 4 and 8 weeks after implantation (Fig. 4a–f). The radiological density of the newly formed tissue was measured to express the extent of mineralization. As shown in Table 1, in both groups, the mineralization extent of the formed mass in the aggregate group was low at 4 weeks and significantly increased at 8 weeks. By 4 weeks, the average radiological density for the newly formed mass in aggregate group was significantly higher than that in the cell group (P < 0.05). However, no significant difference was detected between the two groups at 8 weeks. Interestingly, the average radiological density for the 8-week ectopic constructs in both groups was similar to that of the native spine of mice (442 ± 30 HU; P > 0.05).

Histological and histomorphometrical examinations

The gross appearance of specimens revealed that hard bone-like construct had formed in both the cell group and the aggregate group (Fig. 4g–j). In the control group without cells (serum-free medium), mass formation did not occur at both time points. Representative histological image of the samples in two cell-treated groups is presented in Fig. 5. All these implants were positive for Goldner staining, indicating that the newly formed tissue was mineralized (Fig. 5e–h). The quantitative result is presented in Table 1. Histomorphometrical examination of the 4-week samples demonstrated ∼28.3% ± 7.2% and 19.5% ± 4.9% bone in the aggregate group and the cell group, respectively. By 8 weeks after implantation, mineralized bone occupied 67.5% ± 11.2% and 46.1% ± 8.7% of the total cross-sectional area in the aggregate group and the cell group, respectively. The percentage of new bone area in the aggregate group was significantly higher than that in the cell group at 8 weeks (P < 0.05). Similarly, the densities of total blood vessels were higher in the aggregate group at 8 weeks, when compared with those in the cell group, but there was no statistical difference between the two groups at 4 weeks (Table 1).

FIG. 5.

Histological examination of the harvested construct. (a–d) H&E staining of the ectopic bone structure (arrow) in the cell group at 4 and 8 weeks and in the aggregate group at 4 and 8 weeks, respectively. (e–h) The presence of mineralized bone tissue (arrow, in green) was observed in Goldner's trichrome staining in the cell group at 4 and 8 weeks and in the aggregate group at 4 and 8 weeks, respectively. Scale bars: (a–d) 100 μm; (e–h ) 200 μm.

Enhancing bone healing of delayed mandibular union in the rabbit model

Micro-CT scan and analysis

Similar to the gross appearance of the specimen (Fig. 6 a–c), the micro-CT images of the middle section in the transverse plane demonstrated that new bone formation was more extensive in the aggregate group, when compared with that in the control group and the cell group (Fig. 6d–i). Quantitative analysis of the region of interest on the fracture gap showed that the bone volume fraction of the aggregate group was significantly higher than that of the other two groups. Similarly, the aggregate group had higher bone mass density when compared with others (Table 2).

FIG. 6.

The gross and micro-CT result of bone healing in delayed mandibular union. (a–c) The gross appearance of the harvested mandible in different groups. (d–f) The horizontal micro-CT scanning of the fracture in different groups. (g–i) 3D reconstruction image of micro-CT for the fractured bone in different groups. Scale bars: (d–f) 5 mm. Micro-CT, microfocus computed tomography.

Table 2.

Quantitative Analysis of Delayed Mandibular Union in a Rabbit Model

	Control	Cell	Aggregate
Bone volume fraction (%)	11.8 ± 4.4	28.0 ± 6.1^a	51.4 ± 8.9^b
Bone mass density (mg/cm³)	21.3 ± 5.1	55.2 ± 10.6^a	81.0 ± 15.5^b
New bone/total volume (%)	10.8 ± 3.6	29.8 ± 5.7^a	47.2 ± 8.4^b
Vessel density (vessels/cm²)	9.0 ± 3.2	11.8 ± 3.6	16.2 ± 4.5^a
BrdU-positive cells (cells/mm²)	0	54.0 ± 11.0^a	173.0 ± 26.0^b

Results are shown as mean ± standard error of mean (n = 6).

P < 0.05 compared with the control group.

P < 0.05 compared with the cell group.

Histological and histomorphometrical examinations

Six weeks after injection, bony union was observed in the aggregate group, whereas only a partial bone bridge spanned the fracture gap in the cell group. There was no bridging bone formation at the fracture site in the control group (Fig. 7a–c). The extent of new bone formation in each group was evaluated using histomorphometrical analysis. As presented in Table 2, the quantification results demonstrated that the aggregate group had a significantly higher fraction of new bone compared with others. The densities of total blood vessels in fracture gaps in the aggregate group were higher than those in the control group, but not higher than those in the cell group. Immunofluorescence staining revealed the locations of the labeled cells in both cell-treated groups, as opposed to no labeled cells in the control group (Fig. 7d–f). At 6 weeks following transplantation, many BrdU-positive cells were observed at the sites of injection in the aggregate group, whereas only a few positive cells were found in the cell group. The former was significantly higher than the latter (Table 2).

FIG. 7.

Histological examination of the representative slices of bone healing. (a–c) Decalcified sections stained with H&E in different groups (arrows indicate margins of host bone [hb] and newly formed bone [nb]). (d–f) Immunofluorescence image showing more BrdU-positive cells (arrow) in the aggregate group (f) than the cell group (e). No BrdU-positive cells were present in the control group (d). Scale bars: (a–c) 250 μm; (d–f) 100 μm.

Discussion

The present study demonstrates that BMSCs can be used to generate injectable multicellular aggregates. We also present a novel approach to engineer injectable bone without the need for exogenous carriers. This approach exhibits several advantages compared with currently used methods for tissue engineering. First, the expanded cells were harvested together with their endogenous ECM and cell–cell connections. Second, the approach did not use exogenous carriers and therefore minimized the drawbacks associated with them. Third, our engineered bone would be biocompatible and not foreign to the host because it can be entirely derived from autogenous cells [26].

Several methods have been reported to fabricate injectable cell aggregates and to engineer tissue successfully. For example, Wang et al. [27] cultured a layer of confluent cells on a culture dish coated with a thermoresponsive hydrogel, chipped the sheet using a stainless screen, harvested cell aggregates by decreasing the hydrogel temperature, and injected them to treat myocardial infarction. Chung and Park [28] generated injectable adipose tissue by using cellular aggregates prepared from suspended cells and porous microspheres. However, these methods still involved the use of special equipment or exogenous biomaterials. Further, those studies are restricted to soft tissue regeneration. To our knowledge, our study was the first one to detail the use of injectable cell aggregates for bone tissue engineering.

Compared with the dissociated cells that underwent osteogenic culture under the same conditions for the same time period, aggregates exhibited higher level of ALP activity in both quantitative measurements and RT-PCR analysis. This interesting finding may be due to the maintenance of intercellular contacts in aggregates [29]. Likewise, the presence of OCN, collagen type I, and osteopontin was also higher in aggregates than that in the dissociated cells. These observations indicate that the cell aggregates exhibit better potential of osteogenic differentiation than the dissociated cells. The results are consistent with previous reports investigating the effect of ECM on cell differentiation [30,31]. They have demonstrated that the osteogenic differentiation of BMSCs seeded on mineralized ECM is more likely to be maintained even in the absence of osteogenic culture medium [31], suggesting that the mineralized ECM may supply superior osteogenic environment to the stem cells.

Recent studies have provided convincing evidence that nonunion is closely accompanied with lower quantity and quality of stem cells, indicating cell-based therapies are potential strategies for delayed union treatment [32]. In the present study, fractures in the control group did not achieve union at 6 weeks, suggesting a delayed bone healing. In contrast, after injection of the cell aggregates and the dissociated cells into the fracture gap, respectively, bone healing was significantly enhanced, especially in the former. These results indicate that the aggregates would be more efficient in improving bone regeneration than the dissociated cells. There are several possible explanations for this finding. The first stress is that the presence of endogenetic ECM naturally synthesized in vitro. The tissue volume occupied by ECM in cellular aggregates is significantly larger than that in the dissociated cells, with increasing volume of the new tissue. In addition to providing physical supports and volume, ECM has been known to have many biological effects on cells, such as functioning as adhesive substrate, presence of some growth factors and their receptors, storing growth factors, and sensing mechanical signals [6,33]. Further, cell–cell and cell–ECM interactions are also critical in the skeletal development, which is a dynamic process that requires coordinated intercellular activities among osteoblasts, osteocytes, and osteoclasts to maintain skeletal homeostasis [34,35]. Moreover, cell retention and survival at the sites of transplantation is crucial to the efficacy of cell therapy. As observed in the immunofluorescence staining, more labeled cells were retained at the injection sites in the aggregate group than the cell group. This could be due to a larger physical size, the presence of ECM, and the capability to paracrine antiapoptotic factors in cellular aggregates [36]. These factors may contribute to the promotion of cell viability and retention by impeding cell leakage and clearance via the vascular and lymphatic system, decreasing mechanical insult during injection, providing adequate matrix cues, and protecting against apoptotic cell death [10].

Based on the interesting findings in our study, the multicellular aggregates could be potentially applied to many clinical situations, such as delayed bone healing, nonunions, and distraction osteogenesis. Another potential use of our aggregates is as a minimally invasive cell delivery system.

Our study had some limitations and additional studies should be done. Because of obvious differences in anatomy, physiology, and mechanical properties between human and rabbit bone [37], it is difficult to extrapolate the results from studies performed in rabbits to human clinical trials [37,38]. Therefore, further studies in large animal models would be needed before its clinical applications [38]. Another drawback is that the resultant aggregates usually show a heterogeneous size. As uniformity of size is important for obtaining reproducible results and injectable potential, the ideal size of the multicellular aggregates needs further optimization. Theoretically, it should be not only small enough to pass through a needle and get enough nutrient supply in vivo, but also large enough to entrap into the tissue interstices and thus prevent cell loss. Third, because of its soft properties, uncontrollable shape, low mechanical strength, and limited spatial support at the early stage, cell aggregates alone cannot be used to repair large bone defects.

Conclusion

We have described a novel approach to engineer injectable multicellular aggregates that can result in significantly more in vivo bone formation and cell retention when compared with the disassociated cells. Although further investigations in large animal models that would better simulate the clinical practice are needed, we present a promising strategy to strengthen the BMSCs' osteogenic potential and to revolutionize cell therapy in clinical settings.

Footnotes

Acknowledgments

We thank Dr. Yanqing Wang and engineer Jun Wang for their assistance in CT and micro-CT evaluation. We also thank Miss Junjun Kang and Prof. Chunmei Wang for their assistance in microstructural investigation.

Author Disclosure Statement

The authors state that they have no conflicts of interest.

References

De Long

Jr. , Einhorn

, Koval

, McKee

, Smith

, Sanders

, Watson

. 2007. Bone grafts and bone graft substitutes in orthopaedic trauma surgery. A critical analysis. J Bone Joint Surg Am, 89:649–658.

Meijer

, de Bruijn

, Koole

, van Blitterswijk

. 2007. Cell based bone tissue engineering. PLoS Med, 4:e9.

Vacanti

, Bonassar

, Vacanti

, Shufflebarger

. 2001. Replacement of an avulsed phalanx with tissue-engineered bone. N Engl J Med, 344:1511–1514.

Yamada

, Nakamura

, Ito

, Kohgo

, Hibi

, Nagasaka

, Ueda

. 2008. Injectable tissue-engineered bone using autogenous bone marrow-derived stromal cells for maxillary sinus augmentation: clinical application report from a 2–6-year follow-up. Tissue Eng Part A, 14:1699–1707.

Hernigou

, Poignard

, Beaujean

, Rouard

. 2005. Percutaneous autologous bone-marrow grafting for nonunions. Influence of the number and concentration of progenitor cells. J Bone Joint Surg Am, 87:1430–1437.

Rozario

, Desimone

. 2010. The extracellular matrix in development and morphogenesis: a dynamic view. Dev Biol, 341:126–140.

Discher

, Mooney

, Zandstra

. 2009. Growth factors, matrices, and forces combine and control stem cells. Science, 324:1673–1677.

Chen

, Fitzgerald

, Zimmerberg

, Kleinman

, Margolis

. 2007. Cell-cell and cell-extracellular matrix interactions regulate embryonic stem cell differentiation. Stem Cells, 25:553–561.

Teng

, Luo

, Chiu

, Shum-Tim

. 2006. Massive mechanical loss of microspheres with direct intramyocardial injection in the beating heart: implications for cellular cardiomyoplasty. J Thorac Cardiovasc Surg, 132:628–632.

10.

Karoubi

, Ormiston

, Stewart

, Courtman

. 2009. Single-cell hydrogel encapsulation for enhanced survival of human marrow stromal cells. Biomaterials, 30:5445–5455.

11.

, Lou

, Tang

, Ng

, Zhang

, Yu

, Dai

. 2005. Evaluation of different scaffolds for BMP-2 genetic orthopedic tissue engineering. J Biomed Mater Res B Appl Biomater, 75:289–303.

12.

Temenoff

, Mikos

. 2000. Injectable biodegradable materials for orthopedic tissue engineering. Biomaterials, 21:2405–2412.

13.

Griffith

, Naughton

. 2002. Tissue engineering—current challenges and expanding opportunities. Science, 295:1009–1014.

14.

Logeart-Avramoglou

, Anagnostou

, Bizios

, Petite

. 2005. Engineering bone: challenges and obstacles. J Cell Mol Med, 9:72–84.

15.

Zhou

, Chen

, Ho

, Woodruff

, Lim

, Hutmacher

. 2007. Combined marrow stromal cell-sheet techniques and high-strength biodegradable composite scaffolds for engineered functional bone grafts. Biomaterials, 28:814–824.

16.

Chen

, Zhou

, Barnabas

, Woodruff

, Hutmacher

. 2007. Engineering tubular bone constructs. J Biomech, 40,Suppl. 1:S73–S79.

17.

, Ren

, Liu

, Chen

, Zhang

, Xue

, Mao

. 2010. Engineering scaffold-free bone tissue using bone marrow stromal cell sheets. J Orthop Res, 28:697–702.

18.

, Ren

, Chen

, Liu

, Zhang

, Xue

, Mao

. 2010. Reconstruction of rabbit critical-size calvarial defect using BMSC sheet. Ann Plast Surg, 65:259–265.

19.

Prockop

. 1997. Marrow stromal cells as stem cells for non hematopoietic tissues. Science, 276:71–74.

20.

, Leong

, Hutmacher

. 2005. The challenge to measure cell proliferation in two and three dimensions. Tissue Eng, 11:182–191.

21.

Han

, Sun

, Zhang

, Tang

, Dai

. 2008. Ectopic osteogenesis by ex vivo gene therapy using beta tricalcium phosphate as a carrier. Connect Tissue Res, 49:343–350.

22.

Kaigler

, Krebsbach

, West

, Horger

, Huang

, Mooney

. 2005. Endothelial cell modulation of bone marrow stromal cell osteogenic potential. FASEB J, 19:665–667.

23.

Erdogan

, Esen

, Ustün

, Kürkçü

, Akova

, Gönlüsen

, Uysal

, Cevlik

. 2006. Effects of low-intensity pulsed ultrasound on healing of mandibular fractures: an experimental study in rabbits. J Oral Maxillofac Surg, 64:180–188.

24.

Hochuli-Vieira

, Cabrini Gabrielli

, Pereira-Filho

, Gabrielli

, Padilha

. 2005. Rigid internal fixation with titanium versus bioresorbable miniplates in the repair of mandibular fractures in rabbits. Int J Oral Maxillofac Surg, 34:167–173.

25.

Peters

, Toben

, Lienau

, Schell

, Bail

, Matziolis

, Duda

, Kaspar

. 2009. Locally applied osteogenic predifferentiated progenitor cells are more effective than undifferentiated mesenchymal stem cells in the treatment of delayed bone healing. Tissue Eng Part A, 15:2947–2954.

26.

Williams

. 2008. On the mechanisms of biocompatibility. Biomaterials, 29:2941–2953.

27.

Wang

, Chen

, Lin

, Hwang

, Hsieh

, Lai

, Yeh

, Chang

, Sung

. 2008. Direct intramyocardial injection of mesenchymal stem cell sheet aggregates improves cardiac functions after infarction. Cardiovasc Res, 77:515–524.

28.

Chung

, Park

. 2009. Injectable cellular aggregates prepared from biodegradable porous microspheres for adipose tissue engineering. Tissue Eng Part A, 15:1391–1400.

29.

Donahue

, Li

, Zhou

, Yellowley

. 2000. Differentiation of human fetal osteoblastic cells and gap junctional intercellular communication. Am J Physiol Cell Physiol, 278:C315–C322.

30.

Datta

, Holtorf

, Sikavitsas

, Jansen

, Mikos

. 2005. Effect of bone extracellular matrix synthesized in vitro on the osteoblastic differentiation of marrow stromal cells. Biomaterials, 26:971–977.

31.

Thibault

, Scott Baggett

, Mikos

, Kasper

. 2010. Osteogenic differentiation of mesenchymal stem cells on pregenerated extracellular matrix scaffolds in the absence of osteogenic cell culture supplements. Tissue Eng Part A, 16:431–440.

32.

Tseng

, Lee

, Reddi

. 2008. Nonunions and the potential of stem cells in fracture-healing. J Bone Joint Surg Am, 90,Suppl. 1:92–98.

33.

Kleinman

, Philp

, Hoffman

. 2003. Role of the extracellular matrix in morphogenesis. Curr Opin Biotechnol, 14:526–532.

34.

Civitelli

. 2008. Cell-cell communication in the osteoblast/osteocyte lineage. Arch Biochem Biophys, 473:188–192.

35.

Stains

, Civitelli

. 2005. Cell-to-cell interactions in bone. Biochem Biophys Res Commun, 328:721–727.

36.

Miyahara

, Nagaya

, Kataoka

, Yanagawa

, Tanaka

, Hao

, Ishino

, Ishida

, Shimizu

, Kangawa

, Sano

, Okano

, Kitamura

, Mori

. 2006. Monolayered mesenchymal stem cells repair scarred myocardium after myocardial infarction. Nat Med, 12:459–465.

37.

Pearce

, Richards

, Milz

, Schneider

, Pearce

. 2007. Animal models for implant biomaterial research in bone: a review. Eur Cell Mater, 13:1–10.

38.

Reichert

, Saifzadeh

, Wullschleger

, Epari

, Schütz

, Duda

, Schell

, van Griensven

, Redl

, Hutmacher

. 2009. The challenge of establishing preclinical models for segmental bone defect research. Biomaterials, 30:2149–2163.