The efficacy of lapine preconditioned or genetically modified IL4 over-expressing bone marrow-derived mesenchymal stromal cells in corticosteroid-associated osteonecrosis of the femoral head in rabbits

Abstract

Cell-based therapy for augmentation of core decompression (CD) using mesenchymal stromal cells (MSCs) is a promising treatment for early stage osteonecrosis of the femoral head (ONFH). Recently, the therapeutic potential for immunomodulation of osteogenesis using preconditioned (with pro-inflammatory cytokines) MSCs (pMSCs), or by the timely resolution of inflammation using MSCs that over-express anti-inflammatory cytokines has been described. Here, pMSCs exposed to tumor necrosis factor-alpha and lipopolysaccharide for 3 days accelerated osteogenic differentiation in vitro. Furthermore, injection of pMSCs encapsulated with injectable hydrogels into the bone tunnel facilitated angiogenesis and osteogenesis in the femoral head in vivo, using rabbit bone marrow-derived MSCs and a model of corticosteroid-associated ONFH in rabbits. In contrast, in vitro and in vivo studies demonstrated that genetically-modified MSCs that over-express IL4 (IL4-MSCs), established by using a lentiviral vector carrying the rabbit IL4 gene under the cytomegalovirus promoter, accelerated proliferation of MSCs and decreased the percentage of empty lacunae in the femoral head. Therefore, adjunctive cell-based therapy of CD using pMSCs and IL4-MSCs may hold promise to heal osteonecrotic lesions in the early stage ONFH. These interventions must be applied in a temporally sensitive fashion, without interfering with the mandatory acute inflammatory phase of bone healing.

1. Introduction

Osteonecrosis of the femoral head (ONFH) is a debilitating disease that has the potential for collapse of the femoral head and subsequently, progression to osteoarthritis of the hip [1]. In the United States, an estimated 20,000 new patients are diagnosed with ONFH every year [2], and the cumulative number of cases with ONFH is 300,000 to 600,000 [3]. In reports from Asian countries, the estimated number of new patients with ONFH is 2200 annually in Japan [4], and the cumulative number of patients is estimated to be 8.12 million in China [5]. The peak age of patients with ONFH is 40–50 years; males are affected more than females [4-6].

Although the etiology is still not clarified, it is well known that glucocorticoids, alcohol, and local trauma are related to ONFH [1]. The basic mechanism of ONFH involves circulatory interruption and dysfunction of endothelial progenitor cells for angiogenesis.

Core decompression (CD) is a joint preserving surgery that is commonly performed in the early stage (pre-collapse) ONFH [1]. However, one meta-analysis showed that the success rate of CD for ONFH was only 65% [7]. In a recent review, cell-based treatment holds promise for augmenting the therapeutic effect of CD [8]. CD with local injection of bone marrow aspirate concentrate (BMAC) has been shown to biologically augment CD; a recent meta-analysis demonstrated encouraging clinical outcomes for the early pre-collapse stage of ONFH [9]. However, BMAC is not mesenchymal stromal cells (MSCs) only; it mostly contains several types of mononuclear cells including, lymphocytes, macrophages, and other cells [10]. In addition, MSCs are a very small fraction of the cell population in bone marrow (0.001%–0.01%) [11], and BMAC can concentrate MSCs between 1.6 and 5 times at baseline [12,13]. Furthermore, the number of MSCs in BMAC is affected by several factors such as patient’s age [14], harvesting and manipulation methods [15]. These findings indicate that the available number of MSCs in BMAC is limited. Thus, the mechanisms by which BMAC improves the outcome of ONFH have not been fully elucidated. Using endothelial progenitor cells (EPCs) may be one of the potential cell sources for cell therapy for ONFH [16,17]. However, a recent study demonstrated that both EPCs and MSCs facilitated angiogenesis and osteogenesis in a rabbit model of steroid-associated ONFH, but efficacy showed no significant differences between these two cell types [17]. In addition, EPCs cannot directly affect bone regeneration because they cannot differentiate into osteoblasts.

The therapeutic potential of immunomodulation to facilitate bone healing has generated increasing interest [18]. One of the strategies is the preconditioning of MSCs to empower their immunosuppressive properties [18]. Preconditioned MSCs (pMSCs) are cultured first with a pro-inflammatory cytokine(s) or under hypoxic conditions to mimic the inflammatory environment, prior to their use. Previous in vitro studies showed that preconditioning of MSCs by tumor necrosis factor-alpha (TNFα) [19-21], TNFα + lipopolysaccharide (LPS) [22], IL6 [21], or IL17F [21] enhanced osteogenic differentiation. Another issue is the resolution of inflammation using anti-inflammatory cytokines such as IL4, especially in the case of chronic inflammatory bone disease. For example, in vivo studies using a wear particle-induced chronic inflammation model demonstrated that IL4 prevented bone loss and accelerated bone formation by modulating local macrophage polarization to an M2 phenotype [23-25]. In addition, injection of genetically-modified over-expressing IL4 MSCs (IL4-MSCs) into the murine distal femoral bone marrow cavity increased bone mineral density (BMD) of the distal femur [26]. Moreover, implantation of IL4-MSCs enhanced bone formation in a murine long bone critical-size defect model [27]. However, the therapeutic potential of immunomodulation for augmentation of CD in the early stage ONFH has not been studied.

Injection of cell suspension (without a carrier) into a bone tunnel for augmentation of CD suffers from a potential risk of leakage from the bone tunnel after injection. Collagen-based hydrogels (HGs) closely mimic the native tissue microenvironment and have been extensively used as 3D scaffolds for cell encapsulation [28,29]. Crosslinked collagen HG is not injectable while uncrosslinked collagen HG precursor solution is an injectable liquid that may leak from the defect site after injection and dilution with body fluid before gelation [30-32]. Injectable alginate HGs are biocompatible and stable in-situ, but they lack adhesive ligands that are required for cell survival and proliferation within a 3D matrix [33]. Short peptides have been used to impart cell adhesion to alginate HGs [34-36]. However, short peptides lack the specificity and full function of extracellular matrix proteins including collagen [37,38]. In the present study, we used an in-situ stable injectable HG based on both collagen and alginate for delivery of rabbit MSCs to the femoral head following CD.

The aim of the present study is to investigate the efficacy of adjunctive cell-based treatment of CD using pMSCs or IL4-MSCs with injectable HG for the treatment of steroid-associated ONFH in rabbits. We hypothesized that pMSCs would accelerate new bone formation and IL4-MSCs would decrease the number of necrotic cells within the lesion.

2. Materials and methods

2.1. In vitro study

2.1.1. Establishment of IL4-MSC

Rabbit bone marrow-derived MSCs were purchased from Cyagen Biosciences (Santa Clara, CA). MSCs were expanded with MSC growth medium (α-Minimal Essential Medium [α-MEM, Thermo Fisher Scientific, Rockford, IL] supplemented 10% MSC certified fetal bovine serum [FBS; Life Technologies, Pleasanton, CA], 1% antibiotic and antimycotic solution [100 units of penicillin, 100 μg of streptomycin, and 0.25 μg of amphotericin B per milliliter, Life Technologies]) until passage 5. IL4-MSCs were generated by infecting MSCs with the lentiviral vector carrying the rabbit IL4 gene under the cytomegalovirus promoter. In brief, the rabbit-IL4-overexpressing transfer plasmid (pCDH-CMV-rIL4-EF1a-copGFP) was constructed by insertion of rabbit IL4 coded genes into the control transfer plasmid (pCDH-CMV-MCS-EF1a-copGFP) (System Biosciences, Palo Alto, CA). The control lentivirus vectors or rabbit-IL4-overexpressing lentivirus vectors were produced in human embryonic kidney 293 T cells (ATCC, Manassas, VA) by co-transfecting with the control or rabbit-IL4-overexpressing transfer plasmid, packaged plasmid (psPAX2), and enveloped plasmid (pMD2G VSVG) using a calcium phosphate transfection kit (Takara Bio USA, Inc., Mountain View, CA) with 25 mmol/L chloroquine. The virus was mixed in serum-free medium with 6 μg/mL of polybrene (Sigma Aldrich, St. Louis, MO) and was infected into rabbit MSCs with culture for 6 h at the multiplicity of infection (MOI) of 80 [39]. The infected cells were confirmed as GFP positive cells under the fluorescence microscope at 4 days after infection. Both MSCs and IL4-MSCs from passage 8 were used for the in vitro and in vivo studies after testing the IL4 secreting level by ELISA.

2.1.2. Cell proliferation in the MSC, pMSC, IL4-MSC, and IL4-pMSC groups

The experimental outline is shown in Figure S1. MSCs or IL4-MSCs were seeded in T75 flasks and cultured with MSC growth medium for 1 day. For the MSC and IL4-MSC groups, the medium was replaced by fresh MSC growth medium and cells were cultured for 3 days. For the pMSC and IL4-pMSC groups, the cells were cultured with the MSC preconditioning medium (MSC growth medium containing TNFα [20 ng/mL] and LPS [20 μg/mL]) for 3 days. The cells in all four groups were washed with Dulbecco’s phosphate-buffered saline (DPBS) three times, then the cells were trypsinized and seeded onto a 6-well plate (2 × 10⁴ cell/well) with the MSC growth medium. The samples including the supernatants and cells were collected at day 1, 3, and 7. For the samples at day 7, the medium was replaced with fresh medium at day 3. The supernatants were stored at −80 °C and were used for the other experiments.

Cell proliferation was analyzed using the Quant-iT PicoGreen dsDNA Assay Kit (Invitrogen, Carlsbad, CA). After washing three times with DPBS, 2 mL of DNase/RNase free water was added into each well. Cellular DNA was released by three freeze-thaw cycles and stained with PicoGreen according to the manufacturer’s protocol. Fluorescence was measured at 480/520 nm wavelength using a plate reader (SpectraMax M2e Microplate Reader; Molecular Devices, San Jose, CA), and the amount of dsDNA was calculated.

2.1.3. IL4 enzyme-linked immunosorbent assay (ELISA)

The supernatants at day 3 in the IL4-MSC and IL4-pMSC groups were used. The expression level of IL4 was measured using the IL4 ELISA Kit (R&D Systems, Minneapolis, MN). The manufacturer’s protocols were followed, and the optical absorbance at 450 nm was determined with a plate reader. In addition, the IL4 expression level-dsDNA ratio was calculated using the value for IL4 expression divided by the amount of dsDNA.

2.1.4. Osteogenic differentiation

Cells in all four groups were used. After culture with growth medium or preconditioning medium for 3 days, cells were trypsinized and seeded onto the 24-well plates with 4 × 10⁴ cell/well for alkaline phosphatase (ALP) staining and with 2 × 10⁴ cell/well for Alizarin Red staining. Cells were cultured in osteogenic medium (α-MEM supplemented with 10% MSC certified FBS, 1% antibiotic and antimycotic solution, 10 mM β-glycerophosphate [MP Biomedicals], 50 mM l-ascorbic acid [Sigma-Aldrich], 100 mM Vitamin D3 [Sigma-Aldrich], and 100 nM dexamethasone [Sigma-Aldrich]). The medium was replaced with fresh medium twice per week.

ALP staining was performed at 2 weeks. Cells were fixed with 4% paraformaldehyde (PFA) for 30 min, washed three times with DPBS, and stained with ALP substrate solution (1-StepTM NBT/BICP substrate Solution, Thermo Fisher Scientific) at 37 °C overnight. After washing two times with DPBS and drying, photographs of the whole well were taken using a BZ-X 810 digital microscope (Keyence, Osaka, Japan), and the percentage of stained area per well was analyzed using ImageJ 1.53e (NIH).

Alizarin Red staining was performed at 4 weeks. Cells were fixed in 2% PFA for 10 min and were stained with 40 mM pH 4.1–4.3 Alizarin Red S (Sigma-Aldrich) for 15 min. After washing three times with DPBS, the plates were photographed, and the percentage of stained area per well was analyzed using ImageJ [40].

2.2. Fabrication of HGs

2.2.1. HG preparation

Sodium alginate (alginate, 500 GM) was purchased from Pfaltz & Bauer Inc. (Waterbury, CT). Alginate was dissolved in Dulbecco’s Modified Eagle’s Medium (DMEM) with high glucose and without glutamine and calcium (hereafter referred to as calcium free DMEM medium, Thermo Fisher Scientific) at 1% (w/v) concentration, and the solution was sterilized using a 0.22 μm PES filter (Millex, Millipore Sigma). Then, 8 μL of sterile sodium hydroxide (NaOH) (1 N) was added to 1 mL of alginate solution, and the solution was stored at 4 °C. Calcium sulfate (CaSO₄) (Sigma-Aldrich) was sterilized using a 25 kGy dose of electron beam (E-beam) based on ISO 11137–2:2006 standard as described previously [41]. DI water was sterilized using a 0.22 μm PES filter. 100 mg of sterile CaSO₄ was mixed with 1 mL of sterile DI water and the mixture was vortexed for 5 min at room temperature. Calcium free DMEM medium was added to collagen stock solution at 1:1 vol ratio at 4 °C. Collagen type I from rat tail (Corning Inc, NY) was used for preparing the HG. Then, 20 μL of vortexed CaSO₄ suspension was added to 1 mL of collagen solution and pipetted to make a homogeneous collagen precursor solution. Rabbit MSCs were suspended in collagen precursor solution at 0.5, 1, or 5 million cells/mL density. The cell density for in-vivo experiments was 5 million cells/mL. The collagen precursor solution was then added to the alginate solution at 1:1 vol ratio and pipetted to make a uniform cell suspension. The cell-laden HG was stored on ice before use (Fig. 1 and Video 1).

Fig. 1.

The injectable HG.

Video 1.

Download video file ^{(11MB, mp4)}

Supplementary data related to this article can be found online at https://doi.org/10.1016/j.biomaterials.2021.120972

2.2.2. Viability and proliferation of rabbit MSCs in injectable HG

50 μL of the rabbit MSC-laden HG was injected onto wells of a 24-well plate and incubated for 30 min at 37 °C. 1 mL of DMEM medium (Life Technologies) supplemented with 10% MSC certified FBS and 1% antibiotic and antimycotic solution (Life Technologies) was then added to each well and the plate was incubated at 37 °C and 5% CO₂. Cell viability was evaluated using Calcein AM/Ethidium homodimer-1 Live/Dead assay kit (Thermo Fisher Scientific). 1 day after cell encapsulation, gels were stained with Calcein AM (2 μM) and Ethidium homodimer-1(4 μM) to visualize live/dead cells, and imaged using a Zeiss AxioObserver Z1 fluorescent microscope. To quantify the cell viability, the images were divided into smaller squares and the number of live and dead cells were manually counted, as previously described [42]. For cell proliferation measurement, at each time point (1, 7, 14, and 21 days after cell encapsulation), the gel samples were transferred into new wells and incubated in 500 μL of DMEM medium supplemented with 1 mg/mL of collagenase (Thermo Fisher Scientific) for 1 h at 37 °C, to digest the gel matrix. 250 μL of Triton solution (3% in PBS) was then added to each well and the attached cells were scraped using a CytoOne cell scraper (USA Scientific Inc, Ocala, FL). The cell suspension was then sonicated in a bath sonicator to rupture the cell membrane. The lysed cell suspension was centrifuged for 15 min at 2000×g. The amount of double-stranded DNA in the supernatant of cell lysate was measured using Quant-iT PicoGreen DNA assay according to the manufacturer’s instructions [43].

2.3. In vivo study

2.3.1. Animal surgery

All animal experiments were approved and performed following our institution’s Animal Care and Use Committee guidelines. Thirty-six male mature New Zealand rabbits (West Oregon Rabbit Company, OR), 5–6 months of age and weighing from 4.0 to 4.5 kg, received 20 mg/kg of methylprednisolone acetate (MPS: Depo-Medrol®, Pfizer Inc, NY) intramuscularly once, four weeks before surgery. Rabbits were divided into 6 groups (n = 6 in each group): 1) the CD group, 2) the HG group, 3) the MSC group, 4) the pMSC group, 5) the IL4-MSC group, and 6) the IL4-pMSC group. Note, for Groups 3 to 6, all cells were encapsulated with HGs identical to those used in Group 2.

Rabbits were anesthetized with xylazine (4 mg/kg) and ketamine (40 mg/kg), and additional inhalation anesthesia using isoflurane was given. Buprenorphine SR (0.15 mg/kg) and enrofloxacin (25 mg/kg) were injected. In the CD group, a 20 mm skin incision was made over the proximal lateral thigh, and the vastus lateralis muscle was dissected [44]. A small hole was created at the distal end of the third trochanter using a 2-mm-diameter round burr. Under fluoroscopic guidance, a guide-wire was inserted from the hole proximally towards the center of the femoral head, then a 3-mm-diameter bone tunnel was created using a cannulated drill bit to reach 2 mm from the surface of the femoral head (Figure S2); in the HG group, 200 μl of HG was injected into the bone tunnel in the femoral head; in the MSC and IL4-MSC groups, 1 million of MSCs or IL4-MSCs mixed with 200 μl of HG were injected using a 21G needle, respectively; in the pMSC and IL4-pMSC groups, MSCs or IL4-MSCs were cultured in preconditioning media for 3 days prior to injection.

The wound was closed with a non-absorbable suture. Antibiotics were administered for 2 days, postoperatively. No perioperative complications were observed. Rabbits were kept in cages and allowed free activities. Femurs were harvested eight weeks after surgery.

2.3.2. MicroCT analysis

The proximal femurs were scanned using a microCT (SkyScan 1276 micro-CT system, Bruker, Kontich, Belgium) with 20 μm resolution at 2016 × 1344, Al 1 mm, 85 kV, 200 μA, with 2 average frames at every 0.4° angle step. The data were reconstructed using the 1.6 version of NRecon software and analyzed by GEMS MicroView software (eXplore MicroView v.2.5, Analysis Plus, GE Healthcare, Toronto, Canada). For the analysis, BMD (mg/mm³) and bone volume fraction (BVF) inside and outside the CD in the femoral head were evaluated. In brief, an 11-mm diameter X 6-mm length cylindrical region of interest (ROI) was positioned to cover the entire femoral head, and a 3-mm diameter cylindrical ROI was co-centrally positioned inside the CD area at the bottom of the 6-mm thickness of the femoral head (Figure S3): the total volume (TV, mm³), bone volume (BV, mm³), bone mineral content (BMC, mg), BMD and BVF were measured. The area outside the CD was defined as the entire region of the femoral head excluding the area inside the CD, and BMD and BVF were calculated using the data for the entire femoral head minus the data inside the CD. A threshold value of bony tissue was determined by a phantom. After microCT scanning, specimens were stored in a −80 °C freezer for subsequent biomechanical testing.

2.3.3. Biomechanical analysis

For mechanical testing, a Materials Testing System fitted with a 2 KN load cell (5944 Instron Corporation, Norwood, CA) was used. Specimens were defrosted and kept moist with PBS throughout the whole process of testing. The testing was performed at room temperature.

First, indentation testing of the femoral head surface was performed (Figure S4A) [44]. Specimens were cut 60 mm distal to the femoral head and were aligned vertically in an aluminum block and potted with polymethylmethacrylate (PMMA). Furthermore, PMMA was packed under the inferior surface of the femoral neck to prevent bone deflection/fracture during testing. Specimens were secured in a swivel vise with the diaphyseal axis oriented 17° from vertical in the frontal plane, resulting in 17° lateral direction of the load vector on the femoral head. This loading direction approximates the frontal plane angle at peak load during a variety of daily activities in humans [45]. The 1.6 mm-diameter indenter was aligned with the Ligamentum Teres in the sagittal plane and used to indent the peak surface of the femoral head. Following the application of a 1 N compressive preload, specimens were loaded at a displacement rate of 10 mm/min until 0.5 mm displacement or 300 N to avoid a fracture at the femoral neck. Load and displacement data were recorded at 100 Hz, and stiffness was calculated from the linear portion of load vs. displacement curve.

After indentation testing, the femoral head was cut into two 4-mm-thick segments using a 0.5-mm-hand saw with a hand-made saw guide set perpendicularly along with the bone tunnel (Figure S4B). The proximal segment was used for histological analysis, whereas the middle fragment was used for another indentation test. A 2.3-mm-diameter indenter was positioned to indent vertically at the bone tunnel (Figure S4C). A 1 N compressive preload was applied and then the specimens were loaded at 10 mm/min until failure was observed. The loading protocol and measurement of stiffness were the same as the first indentation test.

2.3.4. Histological analysis of the osteonecrotic area

The proximal fragments of the femoral head specimens were used. The segments were fixed in 4% paraformaldehyde (pH 7.4), decalcified in 0.5 M ethylenediaminetetraacetic acid (EDTA) (pH 7.4), and embedded in optimal cutting temperature (OCT) compound to obtain 8-μm-thick longitudinal frozen sections along with the direction of the bone tunnel. After hematoxylin and eosin (H&E) staining, five fields within the subchondral bone area in the femoral head were randomly selected and viewed with 200× magnification using a BZ-×810 digital microscope (Keyence). Necrosis of the lesion was determined histologically when the following two findings were present: 1) empty lacunae and pyknotic nuclei of osteocytes within the subchondral bone, 2) bone marrow cell necrosis, fatty bone marrow without hematopoietic cells, scant bone marrow, or reparative tissue such as accumulating multi-nuclear cells, granulation tissue, fibrosis, or appositional bone formation with osteoblast-like cells around the osteonecrotic lesion within the region of bone marrow [46,47]. In addition, to accurately evaluate the breadth of the osteonecrotic changes, two blinded investigators individually counted the total cell numbers and the number of empty lacunae in the trabecular bone manually. The percentage of empty lacunae was calculated.

2.3.5. Histochemical and immunohistochemical analysis for osteogenesis and angiogenesis

The proximal fragments of the frozen sections were also stained with tartrate-resistant acid phosphatase (TRAP) and CD31 (Platelet Endothelial Cell Adhesion Molecule-One; PECAM-1). For the TRAP staining, the TRAP histochemical staining kit (Sigma Aldrich) was used. For CD31 staining, sections were treated with proteinase K (S3020, Dako Cytomation, CA) and endogenous peroxidase activity blocked with 0.3% hydrogen peroxide treatment. A mouse monoclonal anti-rabbit CD31 antibody (diluted 1:800, NB600-562; Novus Biologicals, Littleton, CO) was used as a primary antibody. Secondary antibody (Histofine Simple Stain MAX-PO (M), 414132 F, Nichirei Bioscience, Tokyo, Japan) and 3,3′-diaminobenzidine (DAB) solution (34,002, Thermo Fisher Scientific) were used to visualize the bound primary antibodies. The sections were then counterstained with hematoxylin.

To evaluate the area outside the CD, five fields within the subchondral bone area were randomly selected and photographed under 200× magnification in both TRAP and CD31 staining. The percentage of TRAP positive staining area per trabecular bone was quantitatively evaluated. The total positive staining area and trabecular bone area were measured using ImageJ [44]. The color threshold of each parameter was determined by two investigators by consensus. The number of CD31 positive stained microvessels was counted by two investigators individually.

2.4. Statistical analysis

Data are reported as mean ± standard deviation. All the in vitro experiments were done in triplicate. A one-way ANOVA with post-hoc Turkey test was used for the in vitro studies, microCT analysis, and mechanical testing. A multivariable mixed-effects model was used for the histological analyses. The internal reliability of the percentage of empty lacunae and the number of CD31 positive stained microvessels between two investigators was assessed by intraclass correlation coefficients (ICC). Statistical significance was set as p < 0.05. The statistical analysis was conducted using Prism 8.4.3 (GraphPad Software) and R version 4.0.3. Power analysis using previous data [48] indicated that a sample size of six per group would provide 80% statistical power to detect significant differences among the groups (α = 0.05, β = 0.20).

3. Results

3.1. Results of in vitro study

3.1.1. Cell proliferation was accelerated in the IL4-MSC and IL4-pMSC groups

The fold change of the amount of dsDNA compared to day 1 was 6.1 ± 0.5 fold at day 3 and 25.1 ± 5.8 fold at day 7 in the MSC group, 3.9 ± 0.2 fold at day 3 and 11.6 ± 0.8 fold at day 7 in the pMSC group, 34.7 ± 3.5 fold at day 3 and 57.2 ± 8.9 fold at day 7 in the IL4-MSC group, and 21.6 ± 1.6 fold at day 3 and 27.6 ± 1.2 fold at day 7 in the IL4-pMSC group, respectively (Fig. 2A). Cell proliferation of the IL4-MSC group was significantly greater than all other groups at days 3 and 7 (p < 0.005, in each). In addition, cell proliferation of the IL4-pMSC group was significantly greater than the MSC and pMSC groups at day 3 (p < 0.005, in each) and was higher than the pMSC group at day 7 (p < 0.05). Moreover, cell proliferation of the MSC group was significantly higher than the pMSC group at day 7 (p < 0.05).

Fig. 2.

Results of in vitro studies (A) Cellular proliferation was measured by PicoGreen dsDNA quantitation assay. Cell proliferation of the IL4-MSC group was significantly higher than all other groups at days 3 and 7 (p < 0.005, in each). In addition, cell proliferation of the IL4-pMSC group was significantly greater than the MSC and pMSC groups at day 3 (p < 0.005, in each) and was greater than the pMSC group at day 7 (p < 0.05). Moreover, cell proliferation of the MSC group was significantly higher than the pMSC group at day 7 (p < 0.05). a: the IL4-MSC group vs the MSC group, p < 0.005, b: the IL4-MSC group vs the pMSC group, p < 0.005, c: the IL4-MSC group vs the IL4-pMSC group, p < 0.005, d: the IL4-pMSC group vs the MSC group, p < 0.005, e: the IL4-pMSC group vs the pMSC group, p < 0.005, f: the MSC group vs the pMSC group, p < 0.05 (B) IL4 expression level at day 3 in the IL4-MSC group was significantly higher than IL4-pMSC group (175.0 ± 22.5 ng/mL vs. 72.4 ± 13.7 ng/mL, p < 0.005). (C) IL4 expression level-dsDNA ratio at day 3 in IL4-MSC group was significantly higher than IL4-pMSC group (3.59 ± 0.48 vs. 0.64 ± 0.53, p < 0.005). (D, E) ALP staining demonstrated that the percentage of stained area in both the MSC and pMSC groups was higher compared to both the IL4-MSC and IL4-pMSC groups (p < 0.01). Similarly, both the MSC and pMSC groups showed higher percentage of Alizarin Red staining area compared to both the IL4-MSC and IL4-pMSC groups (p < 0.01). Furthermore, the pMSC group demonstrated higher percentage of Alizarin Red staining area than the MSC group (p < 0.01).

3.1.2. The IL4 secretion level in the IL4-MSCs group was higher than IL4-pMSCs group

IL4 expression level at day 3 in the IL4-MSC was significantly higher than IL4-pMSC group (175.0 ± 22.5 ng/mL vs. 72.4 ± 13.7 ng/mL, p < 0.005) (Fig. 2B). Moreover, the IL4 expression level-dsDNA ratio at day 3 in IL4-MSC group was significantly higher than IL4-pMSC group (3.59 ± 0.48 vs. 0.64 ± 0.53, p < 0.005) (Fig. 2C).

3.1.3. Osteogenic ability was increased in pMSC but reduced in IL4-MSC and IL4-pMSC groups

ALP staining demonstrated that the percentage of stained area in both the MSC and pMSC groups was higher compared to both the IL4-MSC and IL4-pMSC groups (p < 0.01) (Fig. 2D). Similarly, both the MSC and pMSC groups showed a higher percentage of Alizarin Red staining area compared to both the IL4-MSC and IL4-pMSC groups (p < 0.01) (Fig. 2E). Furthermore, the pMSC group demonstrated a higher percentage of Alizarin Red staining area than the MSC group (p < 0.01).

3.2. HGs characterization

Fig. 3A shows images of live (green) and dead (orange) rabbit MSCs encapsulated in the injectable HG with 0.5, 1, and 5 million cells/mL densities 1 day after encapsulation. The quantified cell viability 1 day after encapsulation is shown in Fig. 3B. The viability of rabbit MSCs encapsulated in injectable HG after 1 day was 93%, 95%, and 92% for 0.5, 1, and 5 million cells/mL cell densities, respectively. The DNA content of rabbit MSC-laden injectable HG (Fig. 3C) with 0.5, 1, and 5 million cells/mL cell densities increased by 2.9, 2.5, and 1.5 folds, respectively, from day 1 to day 21. Although the proliferation rate decreased with increasing cell density, the DNA content of rabbit MSC-laden HG with 5 million cells/mL was significantly higher than that of HG with 1 million cells/mL or 0.5 million cells/mL at each time point. For in-vivo experiments, rabbit MSC-laden injectable HG with 5 million cells/mL density was used.

Fig. 3.

Cell proliferation and viability of MSCs encapsulated in the injectable HG. (A) Live (green) and dead (orange) rabbit MSCs encapsulated in the injectable HG with 0.5 million cells/mL, 1 million cells/mL, and 5 million cells/mL cell densities 1 day after encapsulation. (B) Quantified viability of rabbit MSCs encapsulated in the injectable HG with 0.5, 1, and 5 million cells/mL cell densities 1 day after encapsulation. (C) DNA content of rabbit MSC-laden injectable HG with cell densities of 0.5, 1, and 5 million cells/mL incubated in growth medium over 21 days. Scale bar: 100 μm. Error bars: means ± 1 SD for n = 3.

3.3. Results of in vivo study

3.3.1. BMD and BVF outside the CD in the pMSC and IL4-pMSC groups were increased

Representative images of the femoral head in each group on microCT are shown in Fig. 4A. For the area outside the CD, BMD and BVF in the pMSC group were significantly higher than in the CD and MSC groups (p < 0.05) (Fig. 4B). Furthermore, BMD and BVF in the IL4-pMSC group were significantly higher than the CD group (p < 0.05). Interestingly, for the area inside the CD, BMD and BVF in the IL4-MSC group were significantly lower than the HG group (Fig. 4C).

Fig. 4.

Representative images of MicroCT and data analysi For the area outside the CD, bone marrow density (BMD) and bone volume fraction (BVF) in the pMSC group were significantly higher than in the CD and MSC groups (p < 0.05). Furthermore, BMD and BVF in the IL4-pMSC group were significantly higher than in the CD group (p < 0.05). Interestingly, for the area inside the CD, BMD and BVF in the IL4-MSC group were significantly lower than in the HG group (p < 0.05). *p < 0.05, *p < 0.01, *p < 0.005.

3.3.2. No difference in mechanical strength was observed

The indentation test demonstrated that the stiffness of the femoral head surface showed no significant differences among all groups (Fig. 5A). In addition, the stiffness of the bone tunnel had no significant differences among all groups (Fig. 5B).

Fig. 5.

The results of mechanical testing. (A) The indentation test demonstrated that the stiffness of the femoral head surface showed no significant differences among all groups. (B) In addition, the stiffness of the bone tunnel had no significant differences among all groups.

3.3.3. The percentage of empty lacunae was less in the IL4-MSC group

Empty lacunae and pyknotic osteocytes within the subchondral bone area were seen in all groups (Fig. 6). However, the IL4-MSC group showed that normal osteocytes were widely distributed in the trabecular bone, and empty lacunae were infrequent. In contrast, all other groups demonstrated widespread empty lacunae in the trabecular bone. The ICC for the two investigators counting the percentage of empty lacunae was r = 0.93 (95% confidence interval: 0.91–0.94, p < 0.001). The percentage of empty lacunae in the IL4-MSC group was 33.7% ± 18.8%, which was significantly lower compared to the other groups (CD group: 55.6% ± 13.9%, HG group: 55.6% ± 13.9%, MSC group: 57.8% ± 28.4%, pMSC group: 56.8% ± 19.9%, IL4-pMSC group: 53.5% ± 17.1%, p < 0.05).

Fig. 6.

Representative images of H&E staining of the femoral head and the data analysis Empty lacunae and pyknotic osteocytes were seen within the subchondral bone area in all groups. However, the IL4-MSC group showed that normal osteocytes were widely distributed in the trabecular bone, and empty lacunae were less. In contrast, all the other groups demonstrated that empty lacunae widespread in the trabecular bone. The percentage of empty lacunae was significantly less in the IL4-MSC group (p < 0.05). Scale bar: 100 μm *p < 0.05, **p < 0.01.

3.3.4. pMSC, and IL4-pMSC groups had greater number of CD31 positive stained microvessels

The ICC for the two investigators counting the number of CD31 positive stained microvessels was r = 0.89 (95% confidence interval: 0.87–0.91, p < 0.001). The CD, pMSC, and IL4-pMSC groups had a greater number of CD31 positive stained microvessels than the HG group (p < 0.05) (Fig. 7). The IL4-MSC group had a lower number of CD31 positive stained microvessels than IL4-pMSC group (p < 0.05).

Fig. 7.

Representative images of CD31 staining of the femoral head and the data analysis The CD, pMSC, and IL4-pMSC groups had a greater number of CD31 positive stained microvessels than the HG group (p < 0.05). The IL4-MSC group had a lower number of CD31 positive stained microvessels than the IL4-pMSC group (p < 0.05). Scale bar: 100 μm *p < 0.05, *p < 0.01.

3.3.5. IL4-MSC group had less TRAP positive stained area

The IL4-MSC group had less TRAP positive stained area than the CD, MSC, and pMSC groups (p < 0.05)(Fig. 8).

Fig. 8.

Representative images of TRAP staining of the femoral head and the data analysis. The IL4-MSC group had less TRAP positive stained area than the CD, MSC, and pMSC groups (p < 0.05). Scale bar: 100 μm *p < 0.05, *p < 0.01.

4. Discussion

In the current study, one vial of rabbit bone marrow-derived MSCs was purchased from Cyagen Biosciences. MSCs were expanded, and we used all four types of MSCs (MSCs, pMSCs, IL4-MSCs, and IL4-pMSCs) from passage 8 for all in vitro and in vivo experiments. pMSCs with exposure of TNFα and LPS for 3 days before their use accelerated osteogenic differentiation in vitro. Furthermore, in vivo studies using a model of corticosteroid-associated ONFH in rabbits demonstrated that the injection of pMSCs encapsulated with injectable HGs into the bone tunnel facilitated angiogenesis and increased BMD and BVF in the femoral head. In contrast, in vitro and in vivo studies demonstrated that IL4-MSCs accelerated their proliferation and decreased the percentage of empty lacunae in the femoral head. Therefore, adjunctive cell-based treatment of CD using pMSCs and IL4-MSCs may have utility in healing the osteonecrotic lesion, which may potentially improve the outcome of CD for early stage ONFH. These interventions must be applied in a temporally sensitive fashion without interfering with the mandatory acute inflammatory phase of bone healing.

The viability of rabbit MSCs encapsulated in the injectable HG was in the range or higher than MSC viability in other injectable HGs. For example, the viability of MSCs in an injectable alginate HG modified with RGD peptide was 90% [49]. The viability of placenta-derived MSCs in collagen HGs was over 90%, 24 h after encapsulation [50]. The results of this study showed that the viability of rabbit MSCs in injected collagen/alginate-based HGs was over 92% one day after encapsulation. The results also showed that rabbit MSCs encapsulated in the injectable collagen/alginate-based HG had 1.5–2.9-fold growth in cell number over 21 days, depending on the initial cell density (0.5–5 million cells/mL). The proliferation rate of MSCs in collagen/alginate-based HG in our study was higher than that of MSCs in RGD-peptide modified alginate HG (2–20 million cells/mL initial cell density) with no growth over 14 days [51,52]. We used rabbit MSCs that were transfected with lentiviral vectors. Bakhita et al. showed that rabbit MSCs transfected with a lentiviral vector reduced their proliferation [53]. In another study, human MSCs encapsulated in the same injectable collagen/alginate-based HGs had 5.5–26.8 fold growth in cell number over 21 days [54]. Thus, the relatively slower 1.5–2.9-fold growth in cell number over 21 days is probably the result of the transfection of the cells using lentiviral vectors.

For preconditioning of MSCs, various pro-inflammatory cytokines such as TNFα, IL6, IL17A have been used to facilitate osteogenic differentiation [18]. However, the osteogenic ability of preconditioned MSCs may be influenced not only by pro-inflammatory cytokine exposure, but also by the tissue of origin of the MSCs and the species selected. Preconditioning of MSCs treated by TNFα accelerated the secretion of prostaglandin E2 (PGE2) and vascular endothelial growth factor (VEGF) [55]; these growth factors have been shown to induce osteogenesis and angiogenesis, respectively. Previously, we demonstrated that preconditioning of murine bone marrow-derived MSCs by a combination of TNFα and LPS enhanced osteogenic differentiation by increasing ALP activity and matrix mineralization; however, TNFα alone or a combination of TNFα and IFNγ did not enhance osteogenic differentiation [22]. Thus, based on our previous experiments, for the current study, rabbit bone marrow-derived MSCs were preconditioned by a combination of LPS and TNFα.

In our in vitro studies, pMSCs accelerated osteogenic differentiation; in addition, in vivo studies using a rabbit model of steroid-associated ONFH demonstrated that pMSCs increased BMD and BVF and the number of CD31 positive microvessels. Thus, pMSCs have the potential to facilitate bone healing of ONFH by increasing osteogenesis and the formation of new blood vessels. However, pMSCs did not affect the percentage of empty lacunae in the femoral head. Moreover, pMSCs did not increase the stiffness at the femoral head surface and bone tunnel in mechanical testing. These findings suggest that cell therapy is insufficient to improve the mechanical strength of the femoral head. Thus, the combination of cell-based therapy together with a biomechanically sound scaffold might optimize treatment.

IL4 protein is degraded quickly within a few hours [56]. Spiller et al. [57] fabricated controlled release scaffolds that sustain the release of IL4 for over 6 days, but the release of IL4 protein did not last for 2 weeks. In contrast, our previous study demonstrated that IL4-MSCs survived in the bone marrow cavity for at least 28 days [26]. Furthermore, the production of IL4 was sufficient to be biologically effective for at least 2 weeks. Our previous study using a murine femoral explant organ culture model [26] demonstrated that both murine IL4-MSCs and unmodified (control) MSCs increased the expression levels of IL10, TNFα, and IL1β after LPS exposure but no significant differences were confirmed between these types of MSCs. These findings indicated that IL4-MSCs induce the production of the above downstream mediators.

IL4-MSCs accelerated cellular proliferation in vitro and decreased the percentage of empty lacunae in the femoral head in vivo. Jin et al. [58] reported that IL4 gene polymorphisms were related to steroid-associated ONFH. Previous in vivo studies showed that osteocyte apoptosis occurred via the STAT1/caspase3 signaling pathway in steroid-associated ONFH [59]. The inhibition of STAT1/caspase3 signaling pathway could be accomplished using fludarabine [60,61] or luteolin [62] resulting in a decreased of incidence of ONFH. IFNγ [63] and TNFα [64] activate the STAT1/caspase3 signaling pathway and polarize macrophages to the M1 subtype [65]. Jin et al. [66] demonstrated that curcumin, a naturally derived drug that suppresses inflammation, inhibited inflammatory mediated apoptosis of osteocytes in part through inhibition of M1 polarization through the JAK1/2-STAT1 pathway and prevented the development of ONFH in mice. It is well known that IL4 polarizes macrophages from the M1 to the M2 phenotype [67]. Wu et al. [68] injected IL4 intra-abdominally following MPS in mice and showed that the incidence of osteonecrosis was decreased. They also demonstrated that the administration of IL4 maintained the number of M2 macrophages and reduced the number of M1 macrophages and downregulated the STAT1/caspase3 signaling pathway, resulting in decreased osteocyte apoptosis. In reference to our study, IL4-MSCs may suppress inflammation polarizing macrophage phenotype from M1 to M2, which inhibited the activation of STAT1/caspase3 signaling pathway in osteocytes, inhibiting osteocyte apoptosis. However, IL4-MSCs also decreased osteogenic differentiation in vitro. In addition, the in vivo study confirmed less bone formation in the bone tunnel and less vascularization and osteoclastogenesis in the femoral head with injection of IL4-MSCs. Several reports have shown that IL4 inhibits osteogenic differentiation of MSCs [39,69] and angiogenic differentiation [69]. Importantly, an initial and optimal transient stage of acute inflammation is a crucial event for bone healing to occur [18] and continued cross-talk between MSCs and macrophages is essential for the healing process [70]. In fact, an MSC-macrophage coculture model using MSCs [71] or preosteoblastic MC3T3 cells [72] with M1 macrophages demonstrated increased bone mineralization and ALP activity. Moreover, the addition of IL4 at 72 h to polarize the M1 to the M2 macrophage phenotype further increased mineralization, compared to adding IL4 earlier [72,73]. A murine long bone critical-size defect model demonstrated that the implantation of IL4-MSCs within scaffolds at 3 days after primary surgery in a critical size bone defect enhanced bone formation [27]. These findings indicated that the injection of IL4-MSCs into the bone tunnel at the primary surgery may impair bone healing. Although the rabbits in the IL-MSC group had no post-operative complications, excessive expression of IL4 might possibly induce adverse effects such as the risk of an infection [74,75] and induction of allergenic reactions [76]. We did not observe these adverse effects in our study.

We expected that IL4-pMSCs would have the advantages of both pMSCs and IL4-MSCs. However, IL4-pMSCs showed similar results as pMSCs in vivo. It may be that cellular proliferation and the expression level of IL4 were decreased after preconditioning of the IL4-MSCs.

In this study, we did not track the transplanted cells. We previously performed CD with transplantation of iron-labeled bone marrow cells in patients with ONFH and tracked the cells on MRI; iron-labeled bone marrow cells were detected at one week after surgery, and the signal of iron-labeled cells declined over time but could still be detected at 6 months after surgery [77]. In addition, another study demonstrated that the signal of the luciferase labeled-MSCs injected into femoral bone marrow cavity can be detected more than 4 weeks after their transfection in mice [26]. Thus, these findings indicate that transplanted MSCs survive in the femoral head after CD.

Our studies have several limitations. First, we used a rabbit model of steroid-associated ONFH to simulate a human model of early stage ONFH. This model is the one most commonly used, however, there are no universally accepted animal models for ONFH [44,48,78]. Early stage ONFH is clinically defined as a pre-collapse stage of the femoral head [79,80]. In addition, the corticosteroid-associated ONFH model that we used develops at 2–4 weeks after steroid injection [47,81]. In our study, all femoral heads showed no collapse, and we performed the surgery at 4 weeks after steroid injection. Based on these points, we defined this model as “early-stage” ONFH. Radiological examinations to evaluate ONFH were not performed at 4 weeks after MPS injection, just before surgery. Previous studies have shown that 6 weeks are needed to see the initial onset of changes in ONFH on magnetic resonance imaging, despite the presence of histological changes [82,83]. Second, the time course for our intervention, i.e. one month of steroid injection is short. Future studies might extend the initial time period prior to surgical intervention, and harvest the specimens after longer time periods. Third, the use of pMSCs and IL4-MSCs are experimental interventions that cannot be translated to the clinical setting at this time, as the MSCs are more than “minimally manipulated”. However, as of 2017 update, almost 2600 gene therapy clinical trials have been completed, are ongoing or have been approved worldwide [84]. In addition, for example, a clinical trial using hypoxic preconditioning of autologous MSCs for ischemic stroke has been reported [85]. Clinical translation of our findings would demand further experiments to demonstrate the safety and efficacy of the cells and methodology in vivo, and validate these concepts in a large animal model of ONFH. Finally, the expression profiles and immunomodulation were not evaluated among four types of MSCs. However, our previous in vitro and in vivo studies using mice demonstrated that both pMSCs using TNFα plus LPS and IL4-MSCs polarized macrophages from a pro-inflammatory M1 phenotype to an anti-inflammatory M2 phenotype [22,27,86-88].

In conclusion, pMSCs exposed to TNFα and LPS for 3 days before their use accelerated osteogenic differentiation in vitro. Furthermore, injection of pMSCs encapsulated with injectable HGs into the bone tunnel facilitated angiogenesis and increased BMD and BVF in the femoral head in vivo, using a model of corticosteroid-associated ONFH in rabbits. In contrast, in vitro and in vivo studies demonstrated that IL4-MSCs accelerated their proliferation and decreased the percentage of empty lacunae in the femoral head. Therefore, adjunctive cell-based therapy of CD using pMSCs and IL4-MSCs may hold promise to heal osteonecrotic lesions in the early stage of ONFH. These interventions must be applied in a temporally sensitive fashion without interfering with the mandatory acute inflammatory phase of bone healing.